chemical changes and off-flavor development in lactose ... · chemical changes and off-flavor...

1

Chemical changes and off-flavor

development in lactose-

hydrolyzed UHT milk during

storage

PhD Thesis By

Therese Jansson

August 2014

Aarhus University

Faculty of Science and Technology

Department of Food Science

Kirstinebjergvej 10

5792 Aarslev

Denmark

AARHUS UNIVERSITY

Mejeribrugets forskningsfond

3

Main supervisor

Senior scientist Hanne C. Bertram


Aarhus University

Co-supervisor

Professor Lotte B. Larsen


Aarhus University

Post doc Morten R Clausen


Aarhus University

Opponents

Associate professor Merete Edelenbos (chairman)


Aarhus University

Associate professor Karsten Olsen


University of Copenhagen

Senior researcher Josefina Muños Belloque

Instituto de Investigacion en Ciencias de la Alimentacion

I

Acknowledgments

This PhD thesis is a result of three years of work at the Faculty of Science and Technology, Department of Food Science, in the group of Food, Metabolomics and Sensory (FMS) at Aarhus University. The project is collaboration between Aarhus University and Arla Foods Amba, and was financed by Aarhus University, Arla Foods Amba, Food Future Innovation (FFI) and Mejerbrugets forskningsfond (MFF).

Hanne C. Bertram, my main supervisor, thank you for believing in me, for helping me in my scientific as well as my personal development and for guiding me through the project during these three years. Thank you Hanne, for your acknowledgment and patience, it means a lot to me. To my co-supervisor Lotte B. Larsen, thank you for your understanding, support within the protein area and the good and fruitful discussions we have had. Morten R. Clausen, my second co-supervisor, a BIG thank you, for ALWAYS being there for me in good and bad situations. I am grateful for your endless patience and your willingness to challenge me within the field. Thank you for listening and taking me to a scientifically higher level and finally for not beating me in 5 km .

Nina Eggers, thank you for being you, for being such a good support and technician, for believing in me and for challenging my perspectives to take difficult decisions and thereby grow as a person. I will always be grateful and you will always be my friend.

I am very grateful for the help with analyses of proteins and for the nice discussions from Hanne B. Jensen, Steffen Nyegaard, Hanne S. Møller and Trine K Dalsgaard.

I want to express my gratitude to Arla for financing parts of the project. I want to thank the dairy in Esbjerg and a special thanks to the process technician Mikael Christensen for supplying me with milk samples. I would like to acknowledge Henrik J. Andersen, Anja Sundgren and Colin Ray for all the nice discussions we have had. Special acknowledgment to Colin Ray for your incredible engagement, and for welcoming me in the laboratory in Stockholm. Thank you for introducing me to new methods and for sharing your knowledge.

I am very much grateful to Kristian M. Kristensen at Department of Agroecology- Climate and Water, for helping me with all the statistics in this project. Thank you, Tina L. Magaard, for the help with all the minor but important details in the final version of the thesis. Finally, thank you Mikael Agerlin Petersen at department of Food Science, University of Copenhagen, for the constructive comments on the laboratory work.

II

I also want to thank Vibe Bach, Ulrik K. Sundekilde, Sidsel Jensen, Line H. Mielby and Birgitte Foged, for your great support and engagement within your specific fields, and of course for being fantastic personalities. Thank you Line, for being such a great support in the end.

Thank you Valentin Rauh and Nanna S. Villumsen for the nice ’dairy science’ seminars we have had, for your support, and for the nice discussions about milk and life.

Thao Le, Hanne B. Jensen, Ulrik K. Sundekilde, Morten R. Clausen, Nina Eggers, Hanne C. Bertram and Sidsel Jensen are greatly acknowledged for proof reading this thesis.

Thank you all the colleagues at the Department of Food Science in Aarslev, for the inspiring and positive working environment. I am very glad that I got the opportunity to work together with all of you.

I want to thank all the athletes in OA which I have shared almost all my free time with in Odense. You have become my Odense family and given me the energy to fulfill this work. A special thanks to the runners and the girl team, you all inspire me.

Helle Svensson, thank you for always being there, for all the running together and for believing in me. Thank you, Maiken Sjøholststrand and Cecilie Tang-Brock, for sharing your family with me and Simone Glad for the ice-cream evenings.

Helena Eriksson, you are everything, without you I would not be here today.

Tack till pappa som alltid trott på mig.

Therese Jansson,

Augusti 2014

III

Abstract

The awareness of lactose intolerance has increased tremendously the last years

and a comparable increased demand for lactose-hydrolyzed dairy products has

been observed. Lactose-hydrolyzed milk is produced by reducing the lactose

content by filtration to approximately 60% of original level of lactose, followed

by an enzymatic hydrolysation of the remaining lactose in the milk by the

enzyme preparation, consisting mainly of β-galactosidase. To increase the shelf-

life of the milk, the milk is heat treated, and one heat treatment is ultra-high

temperature (UHT) heating. During heating several reactions in the milk are

initiated, one of them is the Maillard reaction. This reaction has a major effect

on the appearance and quality of the milk and must be avoided in milk to keep

the high quality during storage.

The overall aim of the present study was to elucidate the chemical changes that

occur in lactose-hydrolyzed UHT milk during nine months of storage. These

chemical changes were expected to be the reason for the formation of off-flavor

observed in lactose-hydrolyzed milk after approximately 90 days of storage. The

specific aims of the study were 1) to compare lactose-hydrolyzed and

conventional UHT milk during storage, 2) to identify specific compounds

correlated to the Maillard reaction, 3) to investigate the changes in protein

profile in lactose-hydrolyzed and conventional UHT milk during storage, and

finally 4) to evaluate how the chemical changes affected the sensory profile of the

milk.

Paper 1 describes the method applied to analyze volatile compounds in the milk.

Dynamic headspace sampling in combination with gas chromatography mass

spectrometry (DHS-GC-MS) was used due to its holistic way of analyzing

volatiles. The study showed that DHS-GC-MS was a useful analytical method,

and in total 24 volatile compounds were identified in the milk. The results

revealed a variation between batches of the same milk type, especially in the

level of ketones. Moreover, the ketone concentration was observed to be higher

in the conventional UHT milk compared to the lactose-hydrolyzed UHT milk.

Paper 2 included several methods to investigate the chemical changes in milk

during storage. Among others, DHS-GC-MS, reversed phase high pressure liquid

chromatography mass spectrometry (RP HPLC MS), and a fluorescamine assay

were used to investigate volatile compounds, furosine, and free amino terminals

respectively. The major results were that a lower initial level of 2-methylbutanal,

furosine and the free amino terminals were observed in lactose-hydrolyzed milk

compared to the conventional milk. However, during storage the level of these

IV

compounds increased significantly in the lactose-hydrolyzed milk, which was

due to increased reactivity of the lactose hydrolysation products glucose and

galactose.

Paper 3 elucidates the protein hydrolysation in the milk by HPLC MS and 1H

NMR spectroscopy. Elevated levels of free amino acids were identified in the

lactose-hydrolyzed milk during storage while the levels remained constant in the

conventional milk. Proteolysis were observed in all milk types during storage,

however, the protein hydrolysis was more evident in the lactose-hydrolyzed milk

compared to the conventional milk, especially in β-CN and αs1-CN after 90 days

of storage.

Paper 4 evaluated the flavor, taste and aroma of the milk with sensory

descriptive analysis, which was correlated to the volatile compounds analyzed

with DHS-GC-MS. The stale flavor increased in all milk types and could be

correlated with increased level of ketones and the aldehydes 2-methylbutanal,

octanal and decanal. A higher level of bitter taste was observed in the lactose-

hydrolyzed milk than in the conventional milk, and a correlation with the level of

free amino terminals analyzed with the fluorescamine assay revealed a good link

between bitter taste and the level of free amino terminals in the milk.

In conclusion, this present study showed that lactose-hydrolyzed UHT milk is

more vulnerable towards chemical changes compared to conventional UHT milk.

This is due to increased reactivity of glucose and galactose, and due to the

proteolytic activity present in the enzyme preparation added to the lactose-

hydrolyzed milk. The proteolytic enzymes hydrolyses the proteins to bitter

tasting peptides and to amino acids that are included in the Maillard reaction to

form flavor compounds such as Strecker aldehydes.

V

Resume

Bevistheden om laktose intolerance er steget de sidste par år og en tilsvarende

efterspørgelse af laktose-hydrolyserede produkter er blevet observeret.

Princippet i produktion af laktose-hydrolyseret mælk er, at først reducere

mængden af laktose ved filtrering til omkring 60% af den oprindelige mængde

laktose i mælken, efterfulgt af en enzymatisk hydrolysering af den resterende

laktose med enzymet β-galactosidase. For at øge holdbarheden på mælken,

varmebehandles den, og en af de varmebehandlinger der bruges er ultra-high-

temperature (UHT). Under varmebehandlingen initieres en kaskade af

forskellige reaktioner, en af disse er Maillard reaktionen. Denne reaktion har en

betydlig effekt på mælkens udseende og kvalitet og bør undgås for at beholde

den høje kvaliteten på mælken under opbevaring.

Det overordnede formål med dette projekt var at undersøge de kemiske

forandringer der forekommer i laktose-hydrolyseret UHT mælk under ni

måneders opbevaring. Disse kemiske forandringer forventedes at være årsagen

til dannelsen af off-flavorer opdaget i laktose-hydrolyseret mælk efter omkring

90 dage opbevaring. De specifikke delmål med projektet var 1) at sammenligne

laktose-hydrolyseret og konventionell UHT mælk under opbevaring, 2) at

identifiere specifikke forbindelser der var korrelerede til Maillard reaktionen, 3)

at undersøge ændringerne i protein profilen i laktose-hydrolyseret og almindelig

UHT mælk under opbevaring, og til sidst 4) at evaluere hvordan de kemiske

ændringer påvirkede den sensoriske profil af mælken.

Artikel 1 beskriver den analytiske metode anvendt til at analysere de flygtige

forbindelser i mælken. Dynamisk headspade sampling i kombination med gas

kromatografi mass spektrometri (DHS-GC-MS) anvendtes for dens holistiske

måde at analysere på. DHS-GC-MS viste sig at være en anvendelig analytisk

metode, og der identifieredes i alt 24 flygtige forbindelser i mælken.

Resultaterne viste, at en variation mellem batch af den samme mælketype

forekom, specielt i niveauet af ketoner. Derudover, observeredes en højre

koncentration af ketoner i den almindelige UHT mælk sammenlignet med den

laktose-hydrolyserede mælk.

Artikel 2 inkluderede adskillige metoder til at undersøge de kemiske ændringer i

mælken under opbevaring. Bland andet anvendtes DHS-GC-MS, reversed phase

high pressure liquid chromatography mass spectrometry (RP HPLC MS), og et

fluorescamine assay for at undersøge de flygtige forbindelser, furosine og de frie

amino terminaler. De primære resultater var, at et lavere start niveauet af 2-

methylbutanal, furosine og de frie amino terminaler var observeret i den laktose-

VI

hydrolyserede UHT mælk sammenlignet med den almindelige UHT mælk, dog

øgedes koncentrationen af disse forbindelser signifikant i den laktose-

hydrolyserede mælk under opbevaring, grundet en højere reaktivitet af laktose

hydrolyserings produkterne glukose og galaktose.

Artikel 3 belyste protein hydrolysen i mælken ved at anvende HPLC MS og

kernemagnetisk resonans (NMR) spektroskopi. Øgede mængder af frie amino

syrer var konstateret i den laktose-hydrolyserede mælk under opbevaring mens

mængde forblev konstant i den almindelige UHT mælk. Proteolyse var påvist i

alle mælke typerne under opbevaring, men protein hydrolyseringen var mere

tydelig i den laktose-hydrolyserede mælk i forhold til den almindelige mælk,

specielt i β-CN og αs1-CN efter 90 dages opbevaring.

Artikel 4 evaluerede flavor og smag i mælken ved brug af sensorisk deskriptiv

analyse, som korreleredes med de flygtige forbindelser i mælken, analyseret med

DHS-GC-MS. Den hengemte flavor øgedes i alle mælke typer og korreleredes

med et forøget niveau af ketoner og følgende aldehyder: 2-methylbutanal,

octanal og decanal. Et forhøjet niveau af bitter smag var bemærket i den laktose-

hydrolyserede mælk i forhold til den almindelige mælk, som korreleredes med

niveauet af frie amino terminaler analyseret med fluroscamine assayet. Der blev

fundet en god korrelation mellem bitter smag og niveauet af frie amino

terminaler i mælken.

Konklusionen på dette projekt er, at laktose-hydrolyseret UHT mælk er mere

sårbar i forhold til kemiske forandringer sammenlignet med almindelg UHT

mælk. Dette skyldes en øget reaktivitet af glukose og galaktose og en øget

proteolytisk aktivitet til stede i enzym præparationen som tilføjedes til den

laktose-hydrolyserede mælken. Det proteolytiske enzyme hydrolyserer

proteinerne i mælken til bitre peptider og til amino syrer som kan inkluderes i

Maillard reaktionen og danne flavor forbindelser såsom Strecker aldhyder.

VII

List of publications

Paper 1 Volatile Component Profiles of Conventional and Lactose-

Hydrolyzed UHT Milk – A Dynamic Headspace Gas

Chromatography-Mass Spectrometry Study

Jansson, T., Jensen, S., Eggers, N., Clausen, M. R., Larsen, L. B.,

Ray, C., Sundgren, A., Andersen, H. J., Bertram, H. C.

Dairy Science and Technology (2014) 94:311-325

Paper 2 Lactose-hydrolyzed milk is more prone to chemical

changes during storage than conventional UHT milk

Jansson, T., Clausen, M. R., Sundekilde, U., Eggers, N., Nyegaard,

S., Larsen, L. B., Ray, C., Sundgren, A., Andersen, H. J., Bertram, H.

C.

Journal of Agricultural and Food Chemistry (2014) 62 (31):7886-

7896

Paper 3 Chemical and proteolysis-derived changes during long-

term storage of lactose-hydrolyzed UHT milk

Jansson, T., Jensen, H. B., Sundekilde, U., Clausen, M. R., Eggers,

N., Larsen, L. B., Ray, C., Sundgren, A., Andersen, H. J., Bertram, H.

C.

Journal of Agricultural and Food Chemistry (2014) Submitted

Paper 4 Storage-induced changes in the sensory properties of

conventional and lactose-hydrolyzed milk explained by the

development of volatiles

Jensen, S., Jansson, T., Eggers, N., Clausen, M. R., Jensen, H. B.,

Larsen, L. B., Ray, C., Sundgren, A., Andersen, H. J., Bertram, H. C.

Food Research International (2014) Submitted

Fremtiden er laktosefri

Jansson, T., Clausen, M. R., Larsen, L. B., Bertram, H.

Mælkeritidende (2012) 5:8-9

IX

Abbreviations

AGEs; Advanced glycated end products; α-CN, alpha-casein; β-CN, beta-

casein; κ-CN, kappa-casein; CONVI, Conventional indirect UHT; DHS,

Dynamic headspace sampling; DMDS, Dimethyl disulfide; DMTS, Dimethyl

trisulfide; ESL, Extended shelf life; GC-MS, Gas chromatography-mass

spectrometry; LA, Lobry de Bruyn-van Ekenstein; α‐LA, Alfa-lactalbumin; β-

LG, Beta-lactoglobulin; LHD, Lactose-hydrolyzed direct UHT; LHI, Lactose-

hydrolyzed direct UHT; LPH, Lactase-phlorizin hydrolase; MS, mass

spectrometry; NF, Nanofiltration; NMR, Nuclear magnetic resonance; OH,

hydroxyl; PLS, Partial least square; RP HPLC, Reversed phase high pressure-

liquid chromatography; Rt, Retention time; SDS-page, Sodium dodecyl sulfate

polyacrylamide gel electrophoresis; SIM, selected-ion monitoring; SPME, Solid

phase micro extraction TAG; Triacylglycerol; Tg, Glass transition temperature;

TIC, Total ion current; TSP, Sodium trimethylsilyl-[2,2,3,3-2H4]-1-propionate;

UF, Ultrafiltration; UHT, Ultra high temperature; US, United States; UV,

Ultraviolet

XI

Table of contents

Acknowledgments .................................................................................................... I

Abstract ................................................................................................................. III

Resume .................................................................................................................... V

List of publications .............................................................................................. VII

Abbreviations ........................................................................................................ IX

Introduction ............................................................................................................. 1

Aims and hypotheses .......................................................................................... 3

Outline of the thesis ............................................................................................ 3

Chapter 1. Milk ........................................................................................................ 5

1.1 Carbohydrates ................................................................................................ 5

1.2 Proteins .......................................................................................................... 7

1.3 Lipids ............................................................................................................. 9

Chapter 2. Heat treatment ..................................................................................... 11

2.1 Heat treatment ............................................................................................. 11

2.2 UHT Process ................................................................................................. 11

2.3 Direct and indirect heat treatment of milk .................................................. 13

2.4 The effect of heat treatment on proteins ..................................................... 14

Chapter 3. Lactose-hydrolyzed milk ...................................................................... 15

3.1 Lactose Intolerance ...................................................................................... 15

3.2 Production of lactose-hydrolyzed milk ........................................................ 16

3.3 β-galactosidase ............................................................................................. 17

Chapter 4. Analytical methods .............................................................................. 19

4.1 Dynamic headspace sampling gas chromatography-mass spectrometry ... 19

4.1.1 Analysis of volatile compounds by dynamic headspace sampling gas

chromatography-mass spectrometry ............................................................. 19

4.1.2 Optimization of DHS-GC-MS method .................................................. 21

4.2 1H NMR spectroscopy ................................................................................. 25

4.3 Methods for characterization of proteins ................................................... 25

4.3.1 RP HPLC MS analysis of intact and hydrolyzed proteins .................... 25

4.3.2 RP HPLC MS for analysis of furosine .................................................. 27

XII

4.3.3 Fluorescamine assay for analysis of free amino terminals .................. 27

4.4 Sensory descriptive analysis ....................................................................... 28

Chapter 5. The Maillard reaction ......................................................................... 29

5.1 Maillard reaction theory .............................................................................. 29

5.2. Maillard reactions in conventional and lactose-hydrolyzed UHT milk .... 32

5.2.1 Early stage Maillard reaction................................................................ 32

5.2.2 Advanced stage Maillard reaction ........................................................ 35

5.2.3 Final stage Maillard reaction ............................................................... 37

Chapter 6. Proteolytic activity .............................................................................. 39

6.1 Proteolytic activity ....................................................................................... 39

6.2 Proteolytic activity in conventional, filtered and lactose-hydrolyzed UHT

milk .................................................................................................................... 39

Chapter 7. Lipid oxidation and thermal degradation .......................................... 43

7.1 Introduction to lipid oxidation .................................................................... 43

7.2 Effect of lactose-hydrolysis on lipid oxidation ........................................... 43

Chapter 8. Sensory characteristics ....................................................................... 47

8.2 Sensory and chemical analysis of milk ....................................................... 47

8.2.1 Aroma and flavor of milk...................................................................... 47

8.2.2 Sensory descriptive analysis and chemical analysis of conventional

and lactose-hydrolyzed UHT milk during storage ....................................... 48

Chapter 9. Effect of heat treatment and storage on lactose-hydrolyzed milk ..... 53

Chapter 10. Conclusion and future perspectives ................................................. 59

10.1 Conclusion ................................................................................................. 59

10.2 Future perspective ..................................................................................... 62

11. References ........................................................................................................ 65

Paper 1 ................................................................................................................... 79

Volatile component profiles of conventional and lactose-hydrolyzed UHT

milk- a dynamic headspace gas chromatography-mass spectrometry study .. 79

Paper 2 .................................................................................................................. 95

Storage-induced changes in the sensory characteristics and volatiles of

conventional and lactose-hydrolyzed UHT processed milk ............................. 95

Paper 3 ................................................................................................................. 107

XIII

Chemical and proteolysis-derived changes during long-term storage of

lactose-hydrolyzed UHT milk ......................................................................... 107

Paper 4 ................................................................................................................ 143

Storage-induced changes in the sensory characteristics and volatiles of

conventional and lactose-hydrolyzed UHT processed milk ........................... 143

1

Introduction

Introduction

Milk contains 87% water but is one of the most nutrient-dense foods, containing

the caseins (CN) and whey proteins, lipids, the disaccharide lactose, minerals

and vitamins (Walstra et al. 1999). In addition, milk contains bioactive peptides

(Choi et al. 2012), immunoglobulins and oligosaccharides which make the milk a

good nutrient source for infants as well as for adults.

Approximately 70% of the world population is lactose-intolerant. Lactose-

intolerant people lack the production of lactase by the highly specific epithelial

cells found in the small intestine. The lack of the lactase enzyme leads to an

inability to hydrolyze lactose to glucose and galactose, which in turn is the cause

of unpleasant problems, such as abdominal pain and diarrhea (Heyman 2006).

Lactase persistence is genetically determined and the ability to produce lactase

has developed from positive natural selection in countries where milk became

domesticated over 7000 years ago (Ingram et al. 2009; Sverrisdóttir et al. 2014).

Due to the high prevalence of the world population suffering from lactose-

intolerance, lactose-hydrolyzed milk is now commercially produced, containing

0.01% lactose compared to the original level 4.8%. The production of lactose-

hydrolyzed milk includes filtration that reduces the level of lactose in the milk to

approximately 60% of original level, followed by addition of a β-galactosidase

enzyme (lactase) preparation that hydrolyzes the remaining lactose to glucose

and galactose. The enzyme can originate from bacteria, yeast or fungi and the

most used enzyme is isolated from the yeast Kluyveromyces or the fungi

Aspergillus (Ansari and Satar 2012). The enzymes used for hydrolyzing lactose in

milk are differentiated by their purity, and a highly pure enzyme is expected to

contain only β-galactosidase. Mittal et al. (1991) investigated the importance of

the enzyme preparation purity and observed a higher proteolytic activity in the

less pure preparation (Mittal et al. 1991).

In order to increase the shelf-life of milk, different heat treatments are applied

such as pasteurization (72 °C, 15 sec), extended shelf-life (ESL) (130-145, <1 sec)

and ultra-high-temperature (UHT) (135-150 °C, 2–20 sec.) treatments (Walstra

et al. 1999; Kessler 2002; Rysstad and Kolstad 2006). In UHT milk the bacteria

are killed and most of the indigenous enzymes inactivated, which gives the milk

a long shelf-life at ambient storage temperature. The market for UHT treated

milk is mainly in Europe as well as Asia. Asia does not have a tradition for daily

intake of milk, however, the positive effects associated with milk intake have

increased the market considerably in Asian countries and the demand for milk

products has increased tremendously in the past years (Beghin 2006; Fuller et

2

Introduction

al. 2006). However, most of the population in Asia is lactose-intolerant, which

increases the demand for high-quality lactose-hydrolyzed milk products.

The enzyme preparations used in the lactose-hydrolyzed milk has shown to

contain side-activities that hydrolyze the proteins in the milk during storage

(Mittal et al. 1991; De et al. 2007). The hydrolysis of proteins may have major

effects on texture, taste and flavor due to formation of peptides, which can be

bitter-tasting (McKellar et al. 1984; Harwalkar et al. 1993; Gomez et al. 1997),

and free amino acids that can be generated by the hydrolysis process. The lysine

residues and amino terminals of intact proteins and peptides, as well as free

amino acids, can participate in the Maillard reaction as well as other reactions.

The Maillard reaction is a non-enzymatic, heat-initiated reaction between ε-

amino groups of proteins, peptides or free amino acids and the carbonyl group of

reduced carbohydrates present in the milk. The Maillard reaction has major

effects on the appearance and quality of milk and may decrease the shelf-life of

the milk. Therefore, the Maillard reaction is not desired in milk and should be

avoided if possible. It is well established that the lactose hydrolysates, i.e.

glucose and galactose reacts with proteins to a higher degree than lactose itself

(Kato et al. 1986; Kato et al. 1988; Meltretter et al. 2009; Naranjo et al. 2013),

which consequently increases the prevalence for formation of Maillard reaction

products in lactose-hydrolyzed UHT milk as compared with conventional UHT

milk during storage.

The shelf-life of ambient temperature stored lactose-hydrolyzed UHT milk is

three months, which is remarkably lower than the eight month shelf-life of

conventional UHT milk. The shelf-life is limited, due to an observed formation of

off-flavor, which is not well defined or described, but decreases the consumer

acceptance of the milk. The sensory attributes, and primarily flavor, are crucial

for the consumer acceptance of milk and should not be compromised (Thomas

1981; Oliver 1993; Adhikari et al. 2010). Therefore, it is essential to investigate

this off-flavor formed in lactose-hydrolyzed UHT milk, to increase the

understanding of its origin and how to decrease its occurrence. In the present

study chemical changes in the volatile, protein, metabolic and proteomic profiles

were investigated in lactose-hydrolyzed, filtered and conventional UHT milk to

determine the chemical changes that takes place. In addition, a sensory

descriptive analysis was applied to elucidate how the chemical changes affected

the sensory characteristics of the milk during storage.

Since the beginning of this project in 2011 the production and sale of lactose-

hydrolyzed milk has increased tremendously in Denmark, and the market is

expected to continue to grow nationally and internationally. In Denmark alone,

3

Introduction

the sale of lactose-hydrolyzed milk increased with 500% in 2011 to 2012, and the

sale of lactose-hydrolyzed products continued to increase in 2013 with 113%, and

a 70% increase is expected in 2014 to 2015 (Thalbitzer 2012; Tüchsen 2014).

Aims and hypotheses

The overall aim of this PhD project was to elucidate the chemistry behind the

formation of off-flavor in lactose-hydrolyzed UHT milk and to identify

differences in the volatile, metabolic, proteomic and sensory profile between

lactose-hydrolyzed and conventional UHT milk during storage.

The specific aims of the study were

1. to compare lactose-hydrolyzed and conventional UHT milk

during storage

2. to identify specific compounds linked to the Maillard reaction

crucial for the changes in chemical composition in lactose-

hydrolyzed UHT milk using DHS-GC-MS and RP HPLC MS

3. to investigate the changes in the protein profile in lactose-

hydrolyzed and conventional UHT milk during storage, using

RP-HPLC MS and fluorescamine assay

4. to evaluate how chemical changes in the lactose-hydrolyzed UHT

milk affect the sensory characteristics compared to conventional

UHT milk using sensory descriptive analysis

It is hypothesized that lactose-hydrolyzed UHT milk is more prone to Maillard

reactions compared to conventional milk due to a higher reactivity of glucose

and galactose with proteins. In addition, it is hypothesized that the enzyme β-

galactosidase used in the present study contains side activities from proteolytic

enzymes hydrolyzing the proteins in the lactose-hydrolyzed milk that contribute

to off-flavor formation. Finally, it is hypothesized that a higher level of Maillard

reaction products and formed peptides change the sensory characteristics in

lactose-hydrolyzed milk compared to conventional UHT milk.

Outline of the thesis

The overall purpose of this thesis is to present the results obtained in this PhD

study in relation to the existing research within the field. In chapter 1 a brief

overview of the major milk components (carbohydrates, protein and lipids) are

4

Introduction

discussed, in chapter 2 common heat treatments applied to milk products is

considered and in chapter 3 the occurrence of lactose intolerance and the

properties of the enzyme β-galactosidase is discussed. In chapter 4 the

methods used to analyze the milk in the present study is briefly discussed.

However, the dynamic headspace sampling gas chromatography-mass

spectrometry method was optimized prior to the analyzing of volatile

compounds in the milk and will therefore be thoroughly discussed. The co-

authors on the papers have contributed to the RP-HPLC MS analysis of the

proteins, the HPLC analysis of furosine and the 1H NMR spectroscopy analysis of

the metabolites (paper 2, 3 and 4). In chapter 5 the Maillard reaction and

glycosylation are described and the results obtained from the present study

related to the Maillard reaction product formation are presented (paper 2 and

3). In chapter 6, proteolysis is reviewed and differences in proteolytic activity

between the lactose-hydrolyzed and conventional UHT milk are addressed

(paper 2 and 3). Chapter 7 gives an overview of the lipid oxidation reactions

present in the milk (paper 1, 2 and 3). Chapter 8 presents the relevance of

aroma and flavour changes in milk and the results from the sensory descriptive

analysis of lactose-hydrolyzed and conventional UHT milk (paper 4).

Furthermore, in chapter 9 the effect of processing of lactose-hydrolyzed UHT

milk is discussed in a holistic perspective including the most important results

from paper 1, 2, 3 and 4. Finally, in chapter 10, conclusions and future

perspectives of the results obtained in the PhD study are presented.

An overview of the workflow in this PhD project is presented in Figure 1.

Figure 1. Overview of the milk and analyzing methods included in the present

study.

5

Chapter 1. Milk

Chapter 1. Milk

In this chapter the carbohydrates, proteins and lipids present in the milk will be

reviewed.

1.1 Carbohydrates

The carbohydrate in milk is mainly lactose which is a disaccharide consisting of

the monosaccharides glucose and galactose bound together by a β-1,4-glycosidic

bond. The main purpose of lactose in milk is to supply the infant with energy, in

addition, lactose retains the osmotic pressure between the milk and blood. The

concentration of lactose in bovine milk is approximately 4.8% (Fox 2009), but

varies throughout the lactation with an observed higher content of lactose in the

early stage of lactation (Walstra et al. 1999). Moreover, milk also contain trace

amounts of glucose and galactose (Sundekilde et al. 2013a).

Lactose, glucose and galactose are reducing carbohydrates, and therefore

contain a free carbonyl group such as an aldehyde (Figure 2). The D-form of the

monosaccharides can have three different steric structures, an open-chain form

and two different cyclic forms (pyranose). When the carbohydrate molecule is

dissolved in water the cyclic forms are converted to each other via the open-

chain form, by the reaction between the aldehyde (C1) and the alcohol (C5) in

the molecule. The cyclization leads to an asymmetric centre and therefore the

cyclic form consists of two isomers, the α- and β-anomer. The conformations are

present at different levels depending on the thermodynamically most preferable

and stable form. The most preferable and stable forms for glucose and galactose

are the cyclic chair conformation of the β-anomer, and the least preferable is the

acyclic, open-chain aldehyde, form (Yaylayan et al. 1993; Belitz et al. 2001). The

chair conformation is most preferable due to the equatorial positions of hydroxyl

(OH) and CH2OH groups, which decreases the interaction between the groups

as compared to the axial positions, and thereby increases the stability of the

molecule. In addition, the configuration of carbon three and four in the glucose

and galactose molecule have shown to be important for their reactivity. It has

been revealed that galactose has an equatorial chair configuration at carbon

three and four, while glucose has the less energetically preferable axial chair

configuration (Kato et al. 1986; Kato et al. 1988) (Figure 3). Consequently,

when galactose is condensed with a protein, this galactose-protein product is

more prone to degrade into polymers compared with similar glucose-protein

products. In the lactose molecule, the galactose molecule is attached to the

6

Chapter 1. Milk

glucose at the hydroxyl group at carbon four which limits further degradations

towards brown compounds and polymers.

The rotation between the different conformations occurs at a specific reaction

rate K which is highly temperature and somewhat pH dependent, and the

concentration of the open-chain aldehyde in the solution increases with

increasing temperature and is preferred at alkaline pH (Walstra et al. 1999). The

reactivity of the carbohydrate molecule is highly dependent on the acyclic form,

thus, the higher level of the acyclic form the higher reactivity of the carbohydrate

(Bunn and Higgins 1981; Yaylayan et al. 1993; Brands et al. 2002; Naranjo et al.

2013). The level of acyclic form of the carbohydrates present in milk is as

follows: galactose>glucose>lactose (Hayward 1977). Furthermore, Naranjo et al

(2013) discussed that the reaction rate constant for the condensation of a

specific carbohydrate and a protein were affected by the molecular mobility of

the carbohydrate which was dependent on the glass transition temperature (Tg)

(Naranjo et al. 2013). Thus, the reaction rate of the carbohydrate increases with

increasing difference between the storage temperature and Tg. Consequently,

lactose with higher Tg than glucose and galactose will react with a protein with a

lower reaction rate compared with glucose and galactose at room temperature.

O

CH2OH

OH

OH OHOH

H

OH- O

O-

CH2OH

OH

OH

OH

OH H

O

CH2OH

OH

OH OH

OH

H+ OH

-

glucose glucoseopen chain aldehyde

Increased T

2<pH>7

Figure 2 Mutarotation of glucose.

OH

CH2NH-P

O H H

OH

OHH

O

OH

CH2NH-P

O H H

OOH

H

RO

OH

CH2NH-P

O H H

OHO

HHO

Galactose Glucose Lactose

Figure 3 Configuration of sugar-protein Amadori product. R=Galactose (Kato et

al. 1986; Kato et al. 1988)

7

Chapter 1. Milk

Carbohydrates give a sweet taste to milk, and different carbohydrates can be

perceived differently. At a scale of perceived sweetness where saccharose is the

reference substance (relative sweetness 100), fructose has a high sweetness

(relative sweetness 114), followed by glucose (relative sweetness 69), galactose

(relative sweetness 63), and the disaccharide lactose (relative sweetness 39) is

one of the least sweet carbohydrates (Belitz et al. 2001).

The carbohydrates are included in different reactions during extensive heating,

in acid and alkaline conditions. Glucose, mannose and fructose are in

equilibrium via the 1,2-enediol, and fructose is in equilibrium with 2,3-enediol,

this enolization and the subsequent isomerization is termed as the Lobry de

Bruyn-van Ekenstein rearrangement (LA-transformation). Lactose is also

included in the LA-transformation and is in equilibrium with the 1,2-enediol and

lactulose. The formed enediols may react further via β-elimination to the highly

reactive dicarbonyl compounds.

Carbohydrates are also included in the Maillard reaction, which is discussed in

chapter 5.

1.2 Proteins

The two major protein groups in milk are caseins (CN) and whey proteins, and

milk protein comprises approximately 3.5% (w/w) of bovine milk composition

(Walstra et al. 1999). In total, 80% (w/w) of the proteins are caseins, which are

divided into four main groups; αS1-casein (αS1-CN), β-casein (β-CN), αS2-casein

(αS2-CN) and κ-casein (κ-CN), where αS1-CN and β-CN are the most abundant

(Wang et al. 2009). The remaining 20% of the proteins is whey proteins,

primarily β-lactoglobulin (β-LG) (50%) and α-lactalbumin (α-LA) (20%), but

also immunoglobulins and enzymes (Farrell et al. 2004). CN has a low degree of

secondary and tertiary structure, and most of the casein molecules are found as

aggregates forming micelles, which are kept together by hydrogen bonds,

hydrophobic interactions and electrostatic interactions (Dalgleish and Corredig

2012). β-LG is a globular protein that depending on pH and temperature can be

present as dimer or monomer. The primary structure of β-LG is composed of 162

amino acids comprising five cysteines forming two disulfide bonds and a free

thiol group (Cys121) which are hidden in a hydrophobic cleft of the protein and

thereby not available for interactions with other proteins in the native form

(Kontopidis et al. 2004).

8

Chapter 1. Milk

Correlations between peptides and bitter taste have been observed in UHT milk

(McKellar 1981; McKellar et al. 1984) (paper 4), and the nature of the amino

acids included in the peptides has a major role for the perceived taste. The

protein itself is not bitter, however, especially β- and αS1-CN contain a high level

of hydrophobic amino acids, which, depending on conformation and length of

the polypeptide they are located in, contribute to a bitter taste (Guigoz and

Solms 1976; Lemieux and Simard 1992; Gomez et al. 1997; Kilara and Panyam

2003). Properties correlated to bitterness, in addition to the hydrophobicity, is

aromatic side-chains, the presence of ammonium groups and the configuration

of the α‐carbon atom and the amino acids in the peptide. Among the bitter

amino acids are isoleucine leucine, lysine, methionine, phenylalanine, proline,

tyrosine, tryptophan and valine (Lemieux and Simard 1992). The level of

peptides in milk may increase with severe heat treatment and during long-term

storage (Meltretter et al. 2008), which is due to both denaturation of the milk

proteins as well as heat inactivation of inhibitors of the plasmin system in milk

and thereby promoting hydrolysis of the proteins after heat treatment (Rauh et

al. 2014).

The proteins are, as the carbohydrates, included in the Maillard reaction

(discussed in chapter 5). In addition, the proteins are exposed for proteolysis

which is discussed in chapter 6.

9

Chapter 1. Milk

1.3 Lipids

Lipids present in milk are mainly triglycerides which are composed of a glycerol

core based on three carbons. One fatty acid is esterified to each carbon atom

(Figure 4) which can be either a saturated fatty acids, mainly unbranched and

straight hydrocarbon chains, or unsaturated fatty acids, hydrocarbon chain

containing up to four double bonds. In addition, the fatty acids can be replaced

with β-keto acids (Hawke 1966). The triglycerides are apolar and are mainly

found in the core of fat globules which are surrounded by a membrane

containing polar phospholipids and proteins such as glycoproteins and enzymes

(Walstra et al. 1999). Approximately 69% of the fat in milk is saturated, 27%

monosaturated, and 3.3% polyunsaturated fatty acids. Especially, the

unsaturated fatty acids and β-keto acids are of interest due to their high

reactivity during heating and storage (Hawke 1966). The double bond in the

unsaturated fatty acids is vulnerable towards oxidation, furthermore the

hydrogen in the methylene group which is adjacent to two carbonyl groups in the

β-keto acid is highly reactive (Whitfield and Mottram 1992).

Figure 4. Triglyceride consisting of a glycerol core attached to, from the top,

palmitic acid, oleic acid and linolenic acid.

Lipids can be exposed to autoxidation and thermal degradation, which is

discussed in chapter 7.

10

Chapter 1. Milk

11

Chapter 2. Heat treatment


In this chapter the heat treatment of milk will be discussed including a brief

discussion of the results obtained in the present study in relation to different

heat treatments.

2.1 Heat treatment

Heat is applied to milk to kill bacteria and inactivate enzymes and thereby

increase the shelf-life of the milk. The shelf-life is highly dependent on the

severity of the heat applied to the milk, where the combination of temperature

and time is essential. Two gentle heat treatments include pasteurization and

extended shelf life (ESL) treatment. Pasteurization, 70-75 °C for 15-30 seconds,

does not inactivate all the enzymes and therefore the shelf-life is approximately

one week for this type of milk when stored cold. ESL treatment is a more severe

heat treatment (130-145 °C for less than one second) than pasteurization. The

ESL heated milk has almost the same appearance and taste as the pasteurized

milk, but with a longer shelf-life of up to one month when stored at cold

temperatures. Ultra high temperature (UHT) treatment (135-150 °C for 2-20

seconds) (Kessler 2002) is applied to inactivate a high level of enzymes present

in the milk, and UHT-treated milk can be stored at ambient temperatures for

several months. A disadvantage related to heating the milk at high temperatures

is that the nutritional and sensory quality decreases due to denaturation of

proteins, thermal degradations of lipids and reactions between sugar and

proteins (Maillard reaction). Therefore, the combination of temperature and

time of the heat treatment must be as low and short as possible to keep the high

nutritional value and sensory quality of the milk, but still be able to inactivate

the bacteria and enzymes present in the milk (Datta et al. 2002; Elliott et al.

2003).

2.2 UHT Process

An overview of a direct and indirect UHT process of milk is seen in Figure 5.

12


Figure 5. Overview of direct (left) and indirect (right) UHT treatment

The UHT treatment can be divided into direct and indirect heat treatment. In

direct UHT treatment steam is injected or infused in to the milk, and in the

indirect UHT treatment high-performance heat exchangers (plate or tubular) are

used and the milk is not in contact with the heating media (steam or water). The

tubular heat exchanger is preferred for UHT milk compared to plate heat

exchangers due to the capability of higher pressure and temperatures (Walstra et

al. 1999; Tetra Pak 2003). A slightly higher heating temperature (3-4 °C) is

reached for the direct heat treatment to obtain the same bactericidal effect as the

indirect heat treatment, but the direct treatment is gentler to the milk due to the

shorter heating and cooling profiles compared to the indirect treatment (Figure

6). The shorter cooling times can be obtained due to the usage of a vacuum

chamber, where additional water from the applied steam during heating, is also

vaporized and removed (Datta et al. 2002).

Figure 6 Heating profile of

A) direct and B) indirect

UHT treatment (Tetra Pak

2003).

13


2.3 Direct and indirect heat treatment of milk

Differences in milk composition have been observed between the two types of

UHT milk. Lower levels of lactulose, furosine, β-LG denaturation and ketones

have been observed in direct treated UHT milk compared to indirect UHT milk

(Elliott et al. 2003; Perkins and Elliott 2005). In addition, a flavor less cooked

has been reported in direct UHT milk, compared to the cooked flavor of indirect

UHT milk (Datta et al. 2002). However, even if less chemical and sensory

attribute changes have been observed in direct treated UHT milk the indirect

treatment is preferred due to lower processing costs. In direct UHT treatment

the steam used for heating cannot be reused and the energy applied is lost, while

for the indirect UHT treatment the heating media can be reused, which decrease

both the energy for heating and the consumption of water (Peterson et al. 2007).

Few studies have investigated the effect of different heat treatments on lactose-

hydrolyzed UHT milk (Mendoza et al. 2005; Messia et al. 2007). Mendoza et al.

(2005) investigated the level of furosine, lactulose and the carbohydrates

glucose, galactose, fructose and tagatose as a function of heat treatment and time

in lactose-hydrolyzed UHT milk. The milk was heated at 135, 140 and 150 °C for

up to 40 seconds, and an increased level of furosine, fructose and tagatose was

observed in the lactose-hydrolyzed milk with increasing heating temperature

and time. Furthermore, the level of glucose and galactose decreased with

increased temperature and time. In addition, Messia et al. (2007) investigated

the level of furosine, lactulose and fructose in 14 different commercial lactose-

hydrolyzed milk samples exposed to pasteurization and UHT heating. The level

of furosine, lactulose and fructose was higher in the UHT treated lactose-

hydrolyzed milk compared to the pasteurized lactose-hydrolyzed milk. In

addition, the milk exposed to direct UHT treatment applied after hydrolysis of

lactose resulted in a lower level of furosine and lactulose compared to the milk

exposed to indirect UHT treatment applied before lactose hydrolysis.

In the present study milk exposed to direct and indirect UHT treatment applied

after lactose hydrolysis was investigated (paper 2 and 3). No differences in

volatile compounds, furosine or proteolytic activity were observed between the

two heat treatments. However, a minor effect of heat treatment was presented in

paper 4:table 2 where a significantly higher level of glycoprotein was observed

in the indirect heated lactose-hydrolyzed UHT milk compared with the direct

heated lactose-hydrolyzed UHT milk. From these results it was concluded that

the reactions taking place in the lactose-hydrolyzed milk due to increased

reactivity of glucose and galactose, compared to lactose, and increased

proteolytic activity is more important for the chemical changes in lactose-

14


hydrolyzed UHT milk than the severity of heat treatment. The results

considering the volatile compounds and the furosine formation during storage is

further discussed in chapter 5, and the proteolysis results are presented in

chapter 6.

2.4 The effect of heat treatment on proteins

The effect of heat treatment on milk proteins induces is structural changes

including unfolding, rearrangement of disulfide bonds, denaturation and

aggregations as well as glycation (de Wit 2009). The non-globular structure of

CN makes the protein heat stable and CN is therefore only slightly affected by

heat. β-LG, on the other hand, is highly dependent on the environmental

conditions, and aggregates with different morphology and structure are formed

during heating and pH changes (Kontopidis et al. 2004; Da Silva Pinto et al.

2012). When β-LG is heated the monomer form is favored (Zúñiga et al. 2010),

which unfolds and exposes the free thiol group (121Cys) which subsequently

reacts with other disulfide bonds in the same β-LG, another β-LG molecule, α-

LA, κ-CN, αs2-CN, or MFGM proteins leading to the formation of small dimers

and bigger clusters. During storage the disulfide linked aggregates undergo

further aggregations via non-covalent bonding (hydrophobic interactions),

which is favored when the pH and approximate the isoelectric point.

15

Chapter 3. Lactose-hydrolyzed milk


In this chapter lactose-intolerance, the production of lactose-hydrolyzed milk

and the enzyme β-galactosidase will be discussed.

3.1 Lactose Intolerance

In humans lactose is hydrolyzed by the enzyme lactase-phlorizin hydrolase

(LPH) produced by the epithelial cells present in the brush border of the small

intestine (Mantei et al. 1988; Sahi 1994a). The enzyme is essential for the

digestion of lactose and ensures that the monosaccharides glucose and galactose

can be absorbed in the blood stream. The production of LPH is high in infants,

however, the LPH level decrease with age and in approximately 70% of the world

population the enzyme production is down regulated. The production of LPH is

genetically determined and has developed from a strong positive natural

selection the last 10 000 years, when cattle was first domesticated (Itan et al.

2010). The domestication started in the Middle East and was spread to Europe

and parts of Africa, and there are high correlations between lactase persistence

and a history of dairying (Itan et al. 2010). The production of the LPH enzyme is

determined by a gene called LCT, which is regulated by a promoter gene, MCM6,

where many of the mutations associated with lactase persistence is found

(Tishkoff et al. 2007; Ranciaro et al. 2014). Different theories have evoked and

discussed to describe the continuous production of LPH throughout lifetime.

The high correlation between lactase persistence and geographic regions with a

high prevalence of dairying has mainly been explained by a positive selected

mutation in genes associated with lactase persistence. Another theory suggests

that the calcium and vitamin D present in milk could prevent populations from

diseases such as rickets, and therefore the lactose persistence developed (Flatz

and Rotthauwe 1973; Itan et al. 2010). A recent study, made by Sverrisdóttir et

al. (2014) investigated a specific mutation associated with lactase persistence

(LCT-13,910*T) and concluded that calcium and vitamin D in the milk were not

the reason for the positive natural selection, furthermore the study concluded

that other “evolutionary selective pressures” such as starvation have affected the

high prevalence of lactase persistence in Europe (Sverrisdóttir et al. 2014).

Different terminologies, such as hypolactasia, lactose malabsorption,

maldigestion and lactose intolerance are used in relation to a down-regulated,

and thereby, a low production of LPH in the small intestine. In this thesis,

lactose intolerance is used as a term referring to an insufficient production of the

16


enzyme, which gives rise to problems such as abdominal pain, flatulence and

diarrhea (Sahi 1994b). The abdominal problems are results of lactose not being

hydrolyzed in the small intestine, which mean that lactose continues down to the

large intestine. Bacteria present in the large intestine will hydrolyze the lactose

and ferment the monosaccharides, which lead to production of gas and short

chain fatty acids. Furthermore, water will be drawn into the large intestine from

the blood by osmosis to balance the high concentration of lactose, which may

result in diarrhea.

3.2 Production of lactose-hydrolyzed milk

Lactose-hydrolyzed UHT treated milk can be produced in different ways,

however, the production generally includes one or more filtration steps

(ultrafiltration, nanofiltration, osmosis) and an enzymatic hydrolysation (Jelen

and Tossavainen 2003; Harju et al. 2012). The lactose can be hydrolyzed before

(pre hydrolysis) or after (post hydrolysis) the applied heat treatment, and the

milk can be direct or indirect heat treated. In the present study the milk was

posthydrolyzed, and direct or indirect UHT heat treated. The production of

lactose-hydrolyzed UHT milk is similar to the production of conventional UHT

milk (chapter 2:figure 5), however, two additional steps are included. The

first additional step is ultra-filtration (UF) in order to separate the lactose, water

and minerals (permeate) from the proteins and fat (retentate), the permeate is

subsequently nanofiltrated (NF) to separate the lactose (permeate) from the

water and minerals (retentate). The NF retentate is then added to the UF

retentate again and milk with 40% reduced level of lactose is obtained. A part of

the lactose is removed from the lactose-hydrolyzed milk to obtain the same

sweetness of the milk as perceived in conventional milk. The milk is then heat

treated (UHT), followed by the second additional step including the enzymatic

hydrolysis where the enzyme is added to the milk aseptically to hydrolyze the

remaining lactose.

The hydrolysation of lactose is dependent on production time and capacity,

enzyme dose and temperature during the hydrolysis. In the posthydrolyzed

approach the dosing of enzyme is low since the hydrolysis take place in the

package during transportation to the suppliers. However, the aseptic enzyme is

more expensive than the non-aseptic enzyme used in the prehydrolyzed milk. In

the prehydrolyzed milk the enzyme is added batch wise, and a high enzyme dose

is required to increase hydrolysation rate to avoid a long storage time before the

heat treatment (Harju et al. 2012).

17


3.3 β-galactosidase

The enzyme used to hydrolyze lactose in milk is β-galactosidase (lactase), which

has a high specificity for O-glycosyl bonds in lactose. In addition to the natural

production of LPH in the small intestine of mammals, β-galactosidase is found

in bacteria, yeast and fungi, which are used for production of commercial β-

galactosidase. The different sources of β-galactosidase have different pH optima

and can thereby be used in different dairy products such as fermented products

with low pH and in milk with a neutral pH. The enzymes most applied in the

industry are from Kluyveromyces fragilis and Aspergillus niger since their pH

optimum are pH 6.5-7.0 and pH 2.5-4.0, respectively, which is highly suitable

for dairy products (Mahoney 2003).

The enzymes are often intracellular, and the cells must therefore be disrupted to

obtain the β-galactosidase. When the cell is disrupted other enzymes may also be

included in the β-galactosidase preparation, which for that reason must be

purified. To purify the enzyme preparation for β-galactosidase gel filtration, ion

exchange chromatography and affinity chromatography can be used. However,

the preparation never obtains 100% purity and the enzyme preparation

therefore contains impurities, such as arylsulfatase and proteases, which can

affect the dairy product quality (Mittal et al. 1991; De et al. 2007). Mittal et al.

(1991) investigated different commercial lactase and observed that higher lactase

purity decreased the proteolytic activity in the milk during storage, and that the

price of the enzyme correlated with the purity and thereby the quality of the

milk. However, there exist only a limited number of published studies about the

quality of different β-galactosidase enzymes.

18


19

Chapter 4. Analytical methods


4.1 Dynamic headspace sampling gas chromatography-mass

spectrometry

4.1.1 Analysis of volatile compounds by dynamic headspace sampling

gas chromatography-mass spectrometry

In milk, more than 400 volatile compounds belonging to different chemical

classes have been identified (Badings et al. 1981). Some of the volatile

compounds are present in very low concentrations and need to be separated and

isolated prior to the analysis by gas chromatography-mass spectrometry (GC-

MS) (Señorans et al. 1996). Dynamic headspace sampling (DHS) is a solvent-free

extraction method used in combination with GC-MS. It is desirable that the

composition of the volatiles is not altered by the extraction method. However,

this is difficult to achieve and often the method should be optimized according to

the most important compounds in the milk for the specific study, which is

discussed in chapter 4.1.2 as well as in paper 1. DHS-GC-MS (Figure 7) is a

holistic extraction and analysis method for volatile compounds and includes an

extraction, desorption, separation and an identification step. In the extraction

step, an inert gas is purged through or above the sample to collect and transport

the volatiles to trap the volatiles on an adsorbent trap. Subsequently, the volatile

compounds are released from the trap by thermal desorption and transferred

directly to a GC-column to be separated by their volatility (Werkhoff and

Bretschneider 1987; Johns et al. 2005). When the migrated compounds have

reached the end of the GC-column they enter a detector, and often it is a mass

spectrometry detector that identifies the compounds.

Figure 7. Dynamic headspace sampling coupled to adsorbent traps, auto

sampler, coldtrap, gas chromatograph and mass spectrometer

20


An alternative, solvent-free method to extract volatile compounds is solid phase

micro-extraction (SPME), which is an often used method. Both DHS and

headspace SPME require controlled extraction conditions such as temperature,

sample volume and extraction time. The sample volume and extraction time

required can be considerably lower in headspace SPME than in DHS.

Furthermore, SPME does not require a desorption step as DHS, which makes

SPME a time-effective and inexpensive extraction method. For both extraction

methods, the concentration and number of extracted compounds is dependent

on extraction temperature, and increasing temperature increases the

concentration of the extracted compounds (Contarini and Leardi 1994; Vazquez-

Landaverde et al. 2005), although the volatiles can be extracted at room

temperature. Similar to DHS, where different adsorbent traps materials are

applied to collect different volatiles, SPME also has different fiber materials with

different affinity for volatile compounds.

DHS-GC-MS is an attractive method when the compounds in the sample under

investigation are unknown prior to the analysis, and DHS-GC-MS has been

widely used for analysis of volatiles in a range of different food matrixes

(Contarini et al. 1997; Jensen et al. 2011; Bach et al. 2012; Reboredo-Rodríguez

et al. 2012; Nightingale et al. 2012; Jansson et al. 2014). However, since SPME is

considered an inexpensive and time effective extraction method, the outcome of

SPME and DHS-GC-MS has been compared for various food products (Elmore

et al. 1997; Marsili 1999; Contarini and Povolo 2002; Povolo and Contarini

2003; Kanavouras et al. 2005). Contarini et al. (2002) compared the two

extraction methods for discrimination of pasteurized, UHT and sterilized heated

whole milk. In the purge and trap method room tempered milk was purged for

one hour with nitrogen (Contarini and Povolo 2002). In the SPME method, milk

heated to 45 °C was extracted for 30 minutes during stirring. The conclusion

from the study was that the two extraction methods could discriminate the milk

types by 11 volatiles, although the type and concentrations of the volatiles

differed between the methods. However, a more comparable evaluation of the

extraction methods would have been obtained if the same extraction

temperature had been applied for both methods. Povolo and Contarini (2003)

analyzed butter (45 °C) with DHS and SPME extracting the volatiles from the

butter for 1 hour and 30 minutes, respectively. The number of volatiles extracted

with DHS was two times higher as for SPME, which could be ascribed to the

larger surface area of the adsorbent trap (Tenax TA) compared to the fiber used

in the SPME method. Marsili (1999) investigated specific lipid oxidation

aldehydes in milk with DHS and SPME. The milk was heated to 45 °C followed

by the extraction by DHS or SPME. After the DHS extraction the adsorbent trap

21


was dry-purged to remove water vapor, followed by desorption and analysis. A

lower variation between the replicates and a higher linearity coefficient of the

calibration curves for the aldehydes were observed for the SPME method

compared to the DHS. Thus, a higher precision was observed for the SPME

samples compared to the DHS (Marsili 1999). It could be considered if the dry-

purging of the tubes in the DHS extraction affected the precision of the method.

It may be concluded that DHS and SPME are somewhat comparable in

reproducibility and sensitivity. However, DHS may be the preferred method for

trace analysis due to larger surface area of the adsorbent tube which will extract

a higher amount and a higher number of different volatiles than SPME (Elmore

et al. 1997; Povolo and Contarini 2003; Kanavouras et al. 2005). In addition, the

volatile compounds extracted with DHS and SPME may vary between the two

techniques.

4.1.2 Optimization of DHS-GC-MS method

It is desirable that the DHS-GC-MS method used for milk analysis is optimized

to detect as many volatiles in the milk as possible. During the extraction step,

factors such as temperature, gas flow, time and sampling volume is important

for the amount and number of volatiles collected (Vallejo-cordoba and Nakai

1993; Contarini and Leardi 1994). The adsorbent tube material also has

considerable impact on which volatiles that are collected in the extraction step.

In the desorption step the coldtrap material may be considered as well as the

cryofocusing and desorption temperature which may influence the evaporation

and peak shape of the volatiles. It is desirable to avoid water in both the

extraction and desorption step, consequently the adsorbent tubes may be back-

purged, and a two-stage desorption may be applied. In the separation step the

temperature gradient is often optimized to increase the separation of the

compounds with similar volatility as well as to shorten the total analyzing time.

In the identification step selected ion monitoring (SIM) can be applied to search

for specific volatile compounds. SIM may preferable be used for identification of

volatile compounds present at very low concentrations that are of particular

interest. The principle for SIM is that specific ions are selected for identification

of the specific compounds which increases the specificity for the particular

compound and decrease involvement of unwanted compounds in the peaks,

consequently the identification and quantification become more certain.

22


4.1.2.1 Optimization of extraction step of the volatile compounds in

milk

In the present study an extraction temperature of 20 °C was applied so that the

volatile profile was comparable with the sensory descriptive analysis of the same

milk served at 14 °C. Previous studies applying headspace extraction to analyze

milk volatiles extract at higher temperatures e.g. 45 °C to increase the

concentrations and number of identified volatiles (Señorans et al. 1996; Valero

et al. 2001; Vazquez-Landaverde et al. 2005; Vazquez-Landaverde et al. 2006).

Valero et al (2001) extracted UHT milk at 45 °C and found 41 and 55 volatile

compounds in whole and skim milk, respectively. Contarini et al. (1997)

extracted UHT milk with DHS at room temperature and identified 11 volatile

compounds which is considerably lower than the 28 volatile compounds

identified in the present study (table 1). The extraction time, sample volume

and gas flow were also optimized in order to increase the concentration of the

detected volatiles (paper 1).

Different adsorbent tube materials were investigated and the porous polymer

Tenax TA tubes were found to be the adsorbent material that extracted highest

amount of compounds, with high affinity for most of the volatile compounds in

the milk. In addition, Tenax TA has a low affinity for water (Helmig and Vierling

1995), which may disturb the extraction of the volatile compounds in the milk.

After extraction, the adsorbent traps were back-purged with nitrogen in order to

remove water that potentially had been absorbed in the traps during the

extraction. However, the water did not appear to interact with the volatiles and

the back-purging resulted in a decreased concentration of volatiles and was

therefore not included in the extraction method.

4.1.2.2 Optimization of the desorption step of the volatile

compounds in milk

The cryofocusing was investigated and was changed between -10 °C, 1 °C and 30

°C where a higher temperature theoretically ought to decrease the condensed

water in the coldtrap. However, 1 °C was found to be the optimal coldtrap

temperature.

Desorption of the volatiles was also improved in relation to water. A two-stage

tube desorption was evaluated to evaporate water at 100 °C in a first stage and

subsequently desorb the volatiles in a second stage by increasing the

temperature to 250 °C. However, this was found not to be effective for milk as

23


the volatiles were evaporated together with the water in the first step and

thereby lost for identification.

4.1.2.3 Optimization of the analyzing and identification of the

volatile compounds in milk

The temperature gradient was optimized to separate the volatile compounds,

especially the highly volatile compounds eluting in the first part of the

chromatogram. The analyzing time could also be shortened by increasing the

temperature gradient in the chromatogram where no volatile compounds were

present.

The volatiles were analyzed in total ion chromatogram (TIC) mode, followed by

identification. Subsequently, SIM was successfully used in the present study to

analyze volatile compounds that was considered important for the development

of the Maillard reaction in the milk. In table 1 all the identified compounds are

listed as well as the compounds analyzed with SIM mode. In addition, the

specific ions that were used for quantification of the specific compounds are

presented.

24


Table 1. Overview of the volatile compounds identified with GC-MS, the

retention time (Rt) and the ions used for quantification in SIM mode in the

present study (paper 2).

Rt (min) Volatile compound m/z 4.84 3,3-dimethylhexane 6.18 2-methylfuran 82 6.87 2-butanone 7.05 2-methylbutanal 86 8.22 2-ethylfuran 81 9.20 2-pentanone 10.12 Decane 12.14 Toluene 14.48 Dimethyl disulfide 15.47 2-hexanone 15.50 Hexanal 82 27.37 2-heptanone 27.38 Heptanal 70 30.48 Dodecane 32.69 2-pentylfuran 95 34.84 Styrene 36.75 2-octanone 36.94 Octanal 39.07 6-methyl-5-heptene-2-one 40.12 Dimethyl trisulfide 40.75 2-nonanone 40.95 Nonanal 41.14 Tetradecane 42.91 Furfural 43.42 Decanal 45.05 Benzaldehyde 45.25 2-undecanone 170 48.90 6,10-dimethyl-5-undecadiene-2-one

25


4.2 1H NMR spectroscopy

1H NMR spectroscopy is a nondestructive technique without extensive sample

preparation that is used to analyze the presence and molecular structure of

organic molecules and metabolites, such as carbohydrates, organic acids,

vitamins, amino acids and aromatic compounds present in fluids. 1H NMR

spectroscopy can be applied on complex mixtures, however a disadvantage with 1H NMR spectroscopy is that highly abundant compounds in a sample may

overlap with less abundant compounds with similar chemical shift. This means

that the dynamic range is limited. In milk, removing proteins and fat prior to the

analysis by precipitation, filtration or ultracentrifugation will improve the

resolution and detection of low-abundant metabolites.

1H NMR spectroscopy has been applied in studies investigating metabolites in

milk (Hu et al. 2007; Klein et al. 2010; Sundekilde et al. 2013b). Sundekilde et

al. (2013a) assigned a total of 40 milk metabolites including amino acids,

organic acids and carbohydrates (Sundekilde et al. 2013a). In addition, 1H NMR

spectroscopy has been applied to analyze lactose-hydrolyzed milk (Monakhova

et al. 2012; Jansson et al. 2014). Monakhova et al (2012) investigated the lactose

concentration in lactose-hydrolyzed milk with 1D and 2D J-resolved NMR

spectroscopy, and the technique could detect a lactose level as low as 0.05 g/L

with high accuracy. The level of the acyclic form of reducing carbohydrates is

highly important for the extent of reactions taking place in the milk, especially in

relation to Maillard reactions. By using a combination of [1-13C]-labeled

aldohexoses and 13C NMR spectroscopy the proportion of acyclic and cyclic

tautomers of different aldohexoses has been determined (Zhu et al. 2001). Zhu

et al. (2001) found a higher percentage of acyclic form for galactose than for

glucose, which support the fact that galactose is more reactive compared with

glucose (Naranjo et al. 2013).

4.3 Methods for characterization of proteins

4.3.1 RP HPLC MS analysis of intact and hydrolyzed proteins

HPLC is used to separate and analyze non-volatile compounds including

proteins. The HPLC system consists of a stationary and mobile phase and high

pressure is applied to force the solvent through the system (Harris 2007). In

reversed phase chromatography the solvent is polar and the stationary phase

(the column) nonpolar. Proteins will be retained and eluted from the column

according to their hydrophobicity by using a solvent with increasing organic

26


content. Bovine milk proteins have previously been analyzed using reverse phase

chromatography (Bonfatti et al. 2008; Jensen et al. 2012) where the elution

order of the proteins was found as follows: glycosylated fraction of κ-CN and

glycoproteins (presume to comprise all major milk proteins), followed by

unglycosylated fraction of κ-CN, αS2-CN, αS1-CN, β-CN and finally the major

whey proteins β-LG and α-LA. Ultraviolet (UV) detection is the most commonly

applied method to detect the separated analytes (Harris 2007). In the present

study the eluted proteins were detected with UV absorbance at 214 nm

measuring the sum of amide bonds with signal intensity mainly proportional to

the analyte concentration. The separated and detected proteins may

subsequently be characterized by MS analysis (Steen and Mann 2004; El-Aneed

et al. 2009). The chromatographic profile of the conventional UHT milk is

presented in Figure 8.

Figure 8. Protein profile of conventional UHT milk detected by UV absorbance

at 214 nm during storage (paper 3).

HPLC can be used for analysis of proteins and peptides and to elucidate

proteolysis as well as lactosylation of proteins in milk (Le et al. 2006; Czerwenka

et al. 2006; Wedholm et al. 2008). As an example Wedholm et al (2008)

investigated peptides in milk and related the cleavage sites of the peptides with

the type of indigenous milk enzymes present in the milk.

27


4.3.2 RP HPLC MS for analysis of furosine

RP HPLC MS is a convenient and rapid method that can be applied to analyze

furosine in milk. In the present study furosine was analyzed by RP HPLC MS

with isocratic elution (0.06 M sodium acetate buffer) using a Supelco supercosil

LC-8 column, and detected at 280 nm as described in ISO 18329:2004

(International Organization for Standardization and International Dairy

Federation 2004) and in Paper 2. The advantage with isocratic conditions is

that the furosine peak elutes early and thereby the analysis time is reduced

compared with gradient elution (Serrano et al. 2002). However, it has been

suggested that analyzing furosine with gradient elution conditions decrease the

involvement of other compounds and thus increase the purity of the furosine

peak (Delgado et al. 1992). Prior to the HPLC analysis the proteins are

hydrolyzed with acid, which may affect the concentration of Amadori product

that is converted to furosine considerably. Serrano et al. (2002) investigated the

effect of 6M, 8M and 10M acid concentrations on furosine concentration in milk

powder. The study revealed that the furosine concentration increased with

increasing acid concentration, thus a better relation between the Amadori

product and furosine was obtained. In addition to RP HPLC MS, furosine may be

determined by ion-exchange chromatography, gas chromatography and capillary

electrophoresis (Serrano et al. 2002).

4.3.3 Fluorescamine assay for analysis of free amino terminals

Fluorescamine assay is a rapid method to analyze free amino terminals (primary

amino groups) situated on amino acids, peptides and proteins (Udenfriend et al.

1972). The non-fluorescent fluorescamine reacts rapidly with the primary

amines generating stable fluorophors that can be analyzed by its excitation and

emission wavelengths of 400 nm and 485 nm (Beeby 1980). In the present

study, the free amino terminals revealed information about the proteolytic

activity in the milk (paper 2). However, the free amino terminals can also be

used to determine the level of lactosylation which is indicated by a decrease in

the level of free amino terminals (Dalsgaard et al. 2007).

28


4.4 Sensory descriptive analysis

Sensory descriptive analysis was used to define the sensory characteristics of the

milk in the present study. The results from the analysis is presented and

discussed in paper 4. Descriptive analysis is a sophisticated and useful tool to

describe and differentiate products according to their sensory characteristics. In

short, descriptive analysis can be divided into three steps; discussion, training,

and analysis. In the first step in the present study the panelists were presented

for reference samples representing sensory attributes commonly associated with

storage related off-flavors in milk as well as extreme milk samples from the

sample set (paper 4). The panelists agreed upon a list of sensory attributes that

could be used to differentiate the milk samples. In sensory descriptive analysis it

is important that the attributes used do not correlate, and that they are as

descriptive as possible, to facilitate the evaluation for the panelists (Stone et al.

1974). The attributes are divided into different sensory impressions: appearance,

aroma, flavor, taste and touch. Appearance is evaluated by sight, aroma is

perceived orthonasally, flavor is perceived retronasally (Rozin 1982), taste is

determined by taste buds present on the tongue and touch is perceived using the

somatosensory system (Lawless and Heymann 2010). The attributes used to

characterize the milk in paper 4 is presented in table 2. The second step, was

training of the panelists. Training is conducted in order to improve the panelists

performance in form of their ability to differentiate between samples,

reproducibility and consistency. In this step, the panelists receive feedback on

their performance after each training session. Finally, in the third step, the

actual sensory descriptive analysis was conducted (Lawless and Hildegarde

2010).

Table 2. Overview of attributes used to differentiate between conventional,

filtered and lactose-hydrolyzed UHT milk

Attribute Definition Popcorn aroma Freshly made popcorn Boiled aroma Boiled/heated milk, like rice porridge Stale aroma Staled stored aroma Color saturation Saturation of white color Creamy flavor The flavor of double cream Boiled milk flavor Rice porridge Popcorn flavor Popcorn/burned Caramel flavor Toffee, aromatic sweet Stale flavor Stale/stored flavor, open milk in fridge for days Sweet taste Pure sweet taste, appear at once Bitter taste The basic bitter taste Fatty mouth feel Creamy mouth feel

29

Chapter 5. The Maillard reaction


In this chapter the Maillard reaction will be discussed including the results

obtained in the present project in relation to the Maillard reaction. Dynamic

headspace sampling-gas chromatography-mass spectrometry (DHS-GC-MS) was

used to analyze volatile compounds (chapter 4.1) and reverse-phase high

pressure liquid chromatography mass spectrometry (RP-HPLC MS) was used to

analyze furosine and glycated proteins present in the milk (chapter 4.3). In

addition, proton nuclear magnetic resonance spectroscopy (1H NMR) was used

to analyze carbohydrates, formate and acetate in the milk (chapter 4.2).

5.1 Maillard reaction theory

The Maillard reaction was first observed in 1912 by Louis-Camille Maillard

(Maillard 1916) and the complex reaction has been widely reviewed ever since,

but still its products are not yet fully characterized. In food products the Maillard

reaction is highly relevant due to the formation of compounds contributing to

changes in aroma, flavor, taste and color. In milk these changes are unwanted as

they decrease the acceptance and shelf-life of the milk. In addition, the Maillard

reaction has a pronounced effect on the nutritive value in the food. Increased

lactosylation of proteins decrease the level of available lysine, consequently, a

nutritive loss is observed in heated and stored milk (Burvall et al. 1977).

Furthermore, in humans, advanced glycation end-products (AGEs) formed in

the Maillard reaction cascade have been associated with Alzheimer disease and

diabetes (Smith et al. 1994; Thorpe and Baynes 1996). Hodge (1953) divided the

Maillard reaction into an early, advanced and final stage (Hodge 1953). This

three-stage classification is still accepted and will be used to describe the

Maillard reaction in this chapter.

The early stage Maillard reaction in milk involves a condensation of an α-

hydroxy group situated on a reducing carbohydrate and a free amino group on a

protein or peptide (mainly ε-amino group of lysine). The condensated

compounds are unstable (also called Schiff base) and are rapidly rearranged to

more stable Amadori products (Figure 9). The Amadori products that are

derived from lactose, glucose and galactose are ε-lactulosyllysine, ε-

fructoselysine and ε-tagatoselysine respectively (Van Renterghem and De Block

1996; Evangelisti et al. 1999; Erbersdobler and Faist 2001; Mendoza et al. 2005;

Fox 2009). Furthermore, acid hydrolysis and heating of the Amadori products

result in furosine, which is used as heat-load indicator in milk products.

30


However, due to an incomplete conversion of Amadori products to furosine,

furosine cannot be used as a direct measure of the glycated protein

(Erbersdobler and Somoza 2007; Meltretter et al. 2009).

CH:O(CHOH)nCH2OH

+RNH2

RNH

CHOH

(CHOH)n

CH2OH

NR

CH

(CHOH)nCH2OH

NHR

CH

(CHOH)n

CHCH2OH

O

-H2O +H+

NHR

CH

CHOH

(HCOH)nCH2OH

+

COH(HCOH)n

CH2OH

CH

NHR

-H+

NHR

CH2

C:O

(HCOH)n

CH2OHAldose in aldehyde form Schiff base

N-substituted aldosylamine

Cation of Schiff base2,3- Eneaminol N-substituted

1-amino-1-deoxy-2-ketose, keto form

Amadori rearrangement

Amadori product

COH(HCOH)n

CH2OH

CH

NHR

1,2-Eneaminol

Figure 9. Condensation of aldose with amino group leading to Amadori product

via Schiff base and Amadori rearrangement.

In the advanced stage, the Amadori products are degraded via enolation

(Yaylayan and Huyghues-Despointes 1994; Kokkinidou and Peterson 2014). The

Amadori products are in equilibrium with 2,3-eneaminol and 1,2-eneaminol and

depending on pH, the two eneaminols give rise to different, highly reactive,

dicarbonyl compounds. In milk the pH is approximately 6.6, which favors the

2,3-eneaminol formation and generate 1-deoxyosone by amino elimination. The

1,2-eneaminol pathway, at acidic pH, favors the formation of 3-deoxyosone by

water elimination, as well as other different degradation products, such as

furfural and hydroxymethylfurfural. The deoxyosone formation is followed by

fission degradation. Its products are pyruvaldehyde, methylglyoxal, diacetyl,

galactose and reductones such as formic and acetic acid (Ledl and Schleicher

1990; Weenen 1998; Van Boekel 1998; Brands and van Boekel 2001; Davidek et

al. 2002; Davídek et al. 2006; Smuda and Glomb 2013). Methylglyoxal is a

reactive α-dicarbonyl that has received a lot of attention due to its contribution

to flavor in food and its formation of AGEs. Carbonyls can be formed from

carbohydrate degradation, from the Schiff base in the early stage of the Maillard

reaction and from the deoxyosones (Wang and Ho 2012). In presence of amino

acids, the deoxyosones and other carbonyls react with the amino acids which

undergo a decarboxylating transamination resulting in loss of a carbon atom and

formation of Strecker degradation compounds. During the formation of Strecker

compounds, flavor, carbon dioxide as well as formation of new carbonyls are

released (Hofmann et al. 2000; Hofmann and Schieberle 2000; Belitz et al.

2001) (Figure 10). The Strecker compounds contribute to flavor, but can also

react further, via aldol condensation, to give different dimeric products, which

31


contribute to odor in food. In table 3 an overview of the amino acids contributing

to Strecker degradation compounds is shown.

+H2O

OOH

N

O

R

R

R1

CO2

R1

N R

OH R

HH2O

R OH

R NH

HOH

R1

Aldehydes + -dicarbonylsNH2OH

O

R1

R

R

O

O

Figure 10. Formation of Strecker degradation products.

Table 3. Overview of amino acid involvement in the formation of Strecker

degradation aldehydes and secondary reaction products (Whitfield and Mottram

1992; Pripis-Nicolau et al. 2000; Limacher et al. 2008).

Amino acid Strecker aldehydes Secondary reaction products Alanine Acetaldehyde 2-Methylfuran, 2-pentylfuran Cysteine Acetaldehyde Hydrogen sulfide, methanethiol Glycine Formaldehyde Isoleucine 2-Methylbutanal Leucine 3-Methylbutanal

Methionine Methional Dimethylsulfide, Dimethyltrisulfide

Phenylalanine Benzaldehyde, phenylacetaldehyde

Furan, 2-methylfuran, pyrazine

Serine 2-Hydroxyethanal Threonine Lactaldehyde 2-Methylfuran Tyrosine 2-(p-Hydroxyphenyl)ethanal Valine 2-Methylpropanal

In the final stage, the aldoses are condensated, the aldehydes and amines are

polymerized and heterocyclic nitrogen compounds are formed, resulting in high-

molecular-weight compounds, colored products e.g. melanoidins (Hodge 1953;

Brands et al. 2002).

In addition to the more general Maillard reaction cascade described by Hodge

(1953), Yaylayan 1997 have added reactions to the model. Yaylayan (1997) states

that carbohydrates, proteins and amino acids as well as Amadori products can

generate fragmentation pools that react together or independently, forming

different degradation products, which are further included in reactions to form

compounds giving flavor and color (Yaylayan 1997). However, the carbohydrate

fragmentation is favored by the presence of amino acids.

32


5.2. Maillard reactions in conventional and lactose-hydrolyzed

UHT milk

The present study aims to analyze specific Maillard reaction products that could

be important to off-flavor in lactose-hydrolyzed UHT milk, which will be

discussed in this chapter.

1H NMR spectroscopy was applied to analyze carbohydrates present in the milk

in order to investigate if the Maillard reaction changes the level of carbohydrates

in the milk (paper 3). The anomers of glucose at 5.23 ppm and 4.63 ppm,

galactose at 5.26 ppm, and lactose at 5.23 ppm were quantified. No pronounced

change was observed in the levels of carbohydrates which most probably indicate

that the carbohydrates are not a limiting factor in the reaction. A decrease could

be expected at least in the lactose-hydrolyzed milk due to the severe Maillard

reaction occurring. However, a possible explanation for the observation is that in

order to observe carbohydrate degradation severe heat treatment during a long

time is necessary (Van Boekel 1998). 1H NMR spectroscopy did however reveal a

decrease in the level of glucose and galactose in the lactose-hydrolyzed milk

between storage day 240 and 270 (paper 3). As the same pattern was observed

for the free amino acids present in the lactose-hydrolyzed milk (paper 3:table

2:figure 1), this could indicate initiation of carbohydrates and protein

condensation and polymerization, and formation of melanoidins.

5.2.1 Early stage Maillard reaction

Furosine, which is an indicator for the early stage of the Maillard reaction, was

analyzed by HPLC (paper 2). The level of furosine in milk stored for 60 days

was higher in the conventional than in the lactose-hydrolyzed UHT milk

(Figure 11). This indicates that the Amadori products were formed during the

heat treatment since the carbohydrate concentration at the time point of heat

treatment is higher in the conventional milk than in the lactose-hydrolyzed milk.

This suggests that the level of carbohydrates is important for the Amadori

product formation and that the carbohydrates may have been a limiting factor

for the formation of these in the lactose-hydrolyzed milk. Furosine increased in

both conventional and lactose-hydrolyzed milk during storage. However, the

formation rate of furosine in the lactose-hydrolyzed milk was significantly higher

in the lactose-hydrolyzed than in the conventional milk throughout the storage

time. This result indicates that the type of sugar has major impact on the

formation of Amadori products during storage, and that glucose and galactose

33


condense with proteins to a higher degree than lactose. This finding elaborates

with previous studies on sugar types in relation with Maillard reaction rate (Kato

et al. 1988; Naranjo et al. 2013). The Amadori products formed in the milk is

highly dependent on the condensation between the reducing carbohydrates and

the proteins in the milk, which is affected by the configuration, glass transition

temperature and the reactivity of each carbohydrate as discussed in chapter 1.

Naranjo et al. (2013) heated glucose, galactose and lactose with casein and

stored the mixtures at 37 °C for several hours and analyzed the loss of

carbohydrates continuously during the storage time. The rate constants (mol h-1)

were calculated from the curves obtained for the decrease of each carbohydrate

as percentage of the total carbohydrate concentration in an unheated control

sample. Naranjo et al. (2013) found highest rate constant for galactose, followed

by glucose and finally lactose, which indicated that galactose had the highest

reactivity with casein compared with glucose and lactose. This finding in

combination with the increased Tg for galactose and glucose compared to lactose

may describe the increased level of furosine observed in figure 11. Kato et al.

(1988) compared condensation of ovalbumin with glucose and lactose and found

similar condensation between the two carbohydrates with ovalbumin, although

the amount of free lysine and Schiff bases were lower and the amount of furosine

was higher in the glucose samples compared to the lactose samples. These

results suggested a faster and easier Amadori rearrangement in glucose-

containing samples.

Tossavainen and Kallioinen (2008) analyzed furosine in lactose-hydrolyzed and

conventional UHT milk during a three month storage period at 5, 22, 30 and 45

°C. The level of furosine was highly dependent on the storage time and

temperature, where the level of furosine increased with increasing storage

temperature in the conventional and lactose-hydrolyzed UHT milk. The

hydrolysis of lactose had a significant effect on the level of furosine, and for the

milk stored at 22 °C approximately 270 mg/100 g protein was observed in the

lactose-hydrolyzed UHT milk, compared with approximately 90 mg/100 g

furosine in the conventional UHT milk (Tossavainen and Kallioinen 2008).

Messia et al. (2007) analyzed furosine in conventional and lactose-hydrolyzed

UHT milk during a 4 month storage period at 20 °C. The level of furosine in

indirect heated lactose-hydrolyzed UHT milk reached 562 mg/100 g protein

after 4 months of storage while the furosine content was 310 mg/ 100 g protein

in indirect heated conventional UHT milk. Furthermore, Messia et al. (2007)

investigated how direct UHT treatment affected the level of furosine in the

lactose-hydrolyzed and conventional milk. The level of furosine decreased in

34


both milk types, and after 4 months of storage the furosine level was 480 and

126 mg/100 g protein in the lactose-hydrolyzed and conventional milk,

respectively (Messia et al. 2007).

The results obtained from Tossavainen and Kallioinen (2008) and Messia et al.

(2007) are comparable with the results obtained in this study (paper 2). Shown

in figure 11 the level of furosine in the lactose-hydrolyzed milk after 3-4 months

of storage was 400-500mg/100 g protein and the level of furosine in the

conventional milk remained somewhat stable at 250mg/100 g protein. However,

the furosine levels are higher in the present study compared to Tossavainen and

Kallioinen (2008) and Messia et al. (2007), and no effect of type of heat

treatment could be observed for the lactose-hydrolyzed UHT milk. These

differences could probably be explained by different heat treatment parameters

applied to the milk.

Days of storage

0 50 100 150 200 250 300

mg

furo

sin

e/ 1

00

g p

rote

in

0

100

200

300

400

500

600

700

800

900

Conventional indirect UHTLactose-hydrolyzed direct UHTLactose-hydrolyzed indirect UHT

Figure 11. Formation of furosine in conventional and lactose-hydrolyzed UHT

milk.

Glycated proteins were analyzed with RP HPLC MS in this study, as presented in

paper 3. The existence of glycated products were confirmed in all milk types at

all storage times (14-270 days) as at least one or more of the identified major

identified proteins (κ-CN, αS2-CN, αS1-CN, β-CN and β-LG) had mass additions of

+162 Da for one hexose, i.e. glucose or galactose, or +324 Da for lactose itself, as

well as multiple hexoses (data not shown). By the method applied in the present

study, the glycated proteins had similar elution time in the chromatographic

profile which contributed to difficulties to integrate or quantify a single glycated

protein. To quantify the glycated proteins, further separation of the proteins

would have been necessary.

35


Moreover, a total pool of glycated (as a result of initial Maillard reaction) and

glycosylated (formed during the biosynthesis of the proteins in the lactating

epithelial cells) proteins were observed in the chromatographic profile as the

glycoprotein, obtained using UV absorbance (figure 8). The level of the

glycoprotein fraction significantly increased in all milk types during storage, and

significantly more in the indirect lactose-hydrolyzed UHT milk compared to the

other milk types (paper 3:table 2). The slightly higher level of glycoprotein in

the indirect lactose-hydrolyzed UHT milk indicates that sugar type affect the

glycated proteins in the milk, furthermore, the lower heat load from the direct

UHT treatment seems to affect the level of glycoprotein. The glycoprotein

comprised both the fraction of the glycated milk proteins, as well as the O-

glycosylated κ-CN naturally present in the milk. It could be expected that the

lactose-hydrolyzed milk had a higher level of glycoprotein compared to the

conventional milk due to the higher levels of furosine observed in paper 2.

Carulli et al. (2011) investigated reactivity of lactose, glucose and galactose with

α-lactalbumin (α-LA) and β-lactoglobulin (β-LG) and observed increased

glycation rate for glucose and galactose including an increased level of glycated

α-LA, compared to lactose (Carulli et al. 2011).

5.2.2 Advanced stage Maillard reaction

Furfural was identified with DHS-GC-MS and the content remained unchanged

during storage, consequently it was concluded that furfural was formed during

the heat treatment, and that no formation was taking place during storage

(paper 2:figure 5). Furfural is formed via the 3-deoxyosone pathway by

reduction of water. A small concentration of furfural has been detected in milk

(Berg 1993), and it increased with increased severity of the heat treatment

applied, which supports that heat is the main cause for the formation of furfurals

in milk. Furfural can also be formed from carbohydrate degradation (Ferrer et

al. 2002), however, this did not seem to affect the level of furfural in the present

study, probably due to minimal carbohydrate degradation during storage at

room temperature revealed from the 1H NMR analysis (paper 3:figure 2).

Acetic acid and formic acid were analyzed by 1H NMR spectroscopy. The

concentration of acetic (paper 2:figure 9) and formic (paper 3:table 2) acid

increased during storage for all the milk types and the changes were not

significantly different between the milk types. However, the initial level of formic

acid was considerably higher in conventional milk than in the lactose-hydrolyzed

and filtered milk; thus, a clear effect of the level of carbohydrates present in the

milk during heat treatment was seen, since a lower level of lactose was present in

the lactose-hydrolyzed and filtered milk compared to the conventional milk

36


during heating. In conclusion, acetic and formic acid are formed during heat

treatment and increase during storage. Acetic acid is considered to be the major

degradation product of Amadori products and is a marker for the 1-deoxyosone

pathway mainly via hydrolytic β-dicarbonyl cleavage, and formate is also formed

from dicarbonyls (Hodge 1953; Brands and van Boekel 2001; Davídek et al.

2006; Smuda and Glomb 2013). The same trend in acid formation observed in

all the milk types indicates similar deoxyosone degradation. This is intriguing

because a higher amount of Amadori products in the lactose-hydrolyzed milk, as

was indicated by a higher level of furosine, would imply that the level of

deoxyosones formation is higher, which would lead to a higher level of acid

formation in the lactose-hydrolyzed milk which was not observed. Consequently,

it could be considered whether the acid formation is independent of the

concentration and type of carbohydrates present in the milk during storage.

2-methylbutanal was one of the compounds analyzed by DHS-GC-MS in SIM

mode (paper 2). The concentration of 2-methylbutanal in the conventional and

lactose-hydrolyzed UHT milk during storage is presented in figure 12. Similar

formation pattern was observed for 2-methylbutanal as for furosine (figure 11),

thus, a lower initial 2-methylbutanal concentration was observed in the lactose-

hydrolyzed milk, compared to the conventional milk. The concentration of 2-

methylbutanal increased to a much greater extent in lactose-hydrolyzed milk

compared to the constant level observed in the conventional milk during storage.

2-Methylbutanal is a Strecker aldehyde and is formed from the reaction between

a dicarbonyl, such as 1-deoxyosone, and the amino acid isoleucine (Balagiannis

et al. 2009). The result can be ascribed to higher reactivity of glucose and

galactose compared to lactose (Naranjo et al. 2013). This is considered to be an

important result for understanding the specific chemical reactions (e.g. Maillard

reaction) occurring in lactose-hydrolyzed UHT milk during heating and storage.

37


Days of storage

0 50 100 150 200 250 300

2-m

eth

ylbu

tanal [

µg/K

g]

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0


Figure 12. Concentration of 2-methylbutanal in conventional and lactose-

hydrolyzed UHT milk during storage.

2-methylfuran and 2-ethylfuran were observed to increase during storage for all

the milk types. However, the increase was significantly higher in the

conventional milk indicating that the formation of these furans were favored by

lactose and not glucose or galactose (paper 2:figure 6). 2-Methylfuran and 2-

ethylfuran can be formed from the Strecker degradation products of alanine

(acetaldehyde) and threonine (lactaldehyde) as well as the presence of

phenylalanine and degradation of carbohydrates in the milk (Limacher et al.

2008). In addition, the significant higher formation of furans from lactose, and

not glucose or galactose, is supported by (Owczarek-Fendor et al. 2012). Thus,

the hydrolysation of lactose seems to decrease the furan formation in the

lactose-hydrolyzed milk during storage.

5.2.3 Final stage Maillard reaction

No chemical compounds associated with the final steps in the Maillard reaction

were included in this study. However, the sensory descriptive analysis in paper

4 revealed differences in the attribute color saturation (saturation of white color)

which is shown in figure 13. The conventional UHT milk had a higher degree of

color saturation than the indirect lactose-hydrolyzed UHT milk at 3 months of

storage and a significantly higher level was observed in the fresh and 3 months

stored conventional milk compared with the direct lactose-hydrolyzed UHT milk

at the same storage period. No significant difference in color saturation was

observed between the milk types at 4 months of storage. This result indicates

that the color saturation was only important the first three months of storage.

However, during the DHS-GC-MS analyses clear visual differences between the

conventional, filtered and lactose-hydrolyzed UHT milk were observed. The

38


lactose-hydrolyzed milk could easily be separated from the other milk types by

the color; the lactose-hydrolyzed milk had a more yellow-brownish color in an

early stage of storage, while the conventional and filtered milk kept the white

color throughout the storage period. These visual differences observed indicate

that the color saturation may be important after 4 months of storage. Figure 14

reveal that the color saturation increases throughout the storage time in the

lactose-hydrolyzed milk and an increase is seen in the indirect lactose-

hydrolyzed milk after 3 months of storage, while a constant level is observed in

the conventional UHT milk. This pattern indicate that after 4 months of storage

the color saturation in lactose-hydrolyzed milk could be higher than in the

conventional milk if the increase continue during storage supporting the visual

differences during DHS-GC-MS analysis. An increased color formation in the

lactose-hydrolyzed milk indicates a higher protein polymerization for glucose

and galactose and a more advanced Maillard reaction and formation of high

molecular weight compounds, so called melanoidins, compared to lactose in the

conventional milk (Brands et al. 2002). Thus, it could have been interesting to

correlate the formation of color and Strecker aldehydes, and specifically 2-

methylbutanal. The increased browning in lactose-hydrolyzed milk is supported

by (Hayward 1977; Bunn and Higgins 1981; Kato et al. 1986; Kato et al. 1988;

Oliver et al. 2006), Kato et al (1988) found an increased browning in ovalbumin-

galactose>ovalbumin-glucose>ovalbumin-lactose model system.

Figure 13. Color saturation in conventional indirect heated UHT milk (CONVI),

lactose-hydrolyzed direct heated UHT milk (LHD) and lactose-hydrolyzed

indirect heated UHT milk analyzed with sensory descriptive analysis.

39

Chapter 6. Proteolytic activity


In this chapter the proteolytic activity will be discussed including the results

obtained in the present project in relation to free amino terminals, free amino

acids and proteolysis in the milk. Reverse phase high pressure liquid

chromatography-mass spectrometry (RP HPLC MS) was used to analyze the

protein profile, fluorescamine assay was used to analyze free amino terminals,

and 1H NMR was used to analyze free amino acids in the milk, which was

described in chapter 4.

6.1 Proteolytic activity

Modifications on proteins due to proteolytic activity have evoked a lot of

attention, especially the peptides generated after the hydrolysis contributing to

flavor and texture changes which is desired in cheese production but unwanted

in milk (Nielsen 2002; Rauh et al. 2014). Native proteases such as the serine

proteinase plasmin, the cysteine protease cathepsin D, and proteases produced

by psychrotrophic bacteria are present in milk. Cathepsin D has shown to be

increased in milk with increasing somatic cell count in the milk (Larsen et al.

2004). The psychrotrophs produce extracellular enzymes that are highly heat

stable and may survive severe heat treatment such as UHT, plasmin may also

survive heat treatments such as UHT heating while cathepsin D is heat unstable

(Hurley et al. 2000; Hayes et al. 2001). The specificity of the proteases varies,

but are preferentially caseins (Nielsen 2002; Wedholm et al. 2008). The

peptides formed after hydrolysis may be analyzed and related to the specificities

of the different proteases. One of the aims in the present study was to elucidate

the effect of protein hydrolysis in the lactose-hydrolyzed UHT milk during

storage and to obtain information about which proteins that were most affected

by the proteolysis.

6.2 Proteolytic activity in conventional, filtered and lactose-

hydrolyzed UHT milk

Free amino terminals (N-terminals) were analyzed, and quantified as leucine

equivalents, with a fluorescamine assay, as discussed in paper 2. The level of

free N-terminals increased in lactose-hydrolyzed milk during storage while in

the conventional milk the level of N-terminals remained more or less constant

40


throughout the storage period (paper 2:figure 10). This result indicates an

increased proteolytic activity in lactose-hydrolyzed milk.

The protein profiles revealed that proteolytic activity was observed in all milk

types for all intact proteins identified (β-CN, αs1-CN, αs2-CN, κ-CN and β-LG)

during storage, which was presented in paper 3:table 2. Two-dimensional gel

electrophoresis was also conducted on the milk proteins in the present study,

which confirmed a proteolytic activity as well as glycation of the proteins in the

milk (results not shown). The reduction of total intact CN during storage (sum of

β-CN, αs1-CN, αs2-CN, κ-CN) was significantly higher in lactose-hydrolyzed milk

with a 56-61% reduction compared to a 15% reduction in conventional and

filtered milk. The difference between the lactose-hydrolyzed, filtered and

conventional UHT milk became pronounced after 90 days of storage, where the

reduction of total intact CN was observed in the lactose-hydrolyzed milk. The

proteins that contributed most to the reduction in the lactose-hydrolyzed milk

included β-CN and αs1-CN. Figure 14 presents the reduction in total intact CN,

intact β-CN and αs1-CN during storage.

Free amino acids (alanine, isoleucine, leucine, lysine, methionine,

phenylalanine, tryptophan, tyrosine and valine) were analyzed with 1H NMR

spectroscopy and were discussed in paper 2 and 3. With an exception for

methionine, increased level of all identified free amino acids was observed in the

lactose-hydrolyzed milk, while no increase (except for lysine) was observed in

the conventional milk. These results are in good agreement with the results

obtained by the fluorescamine assay and the protein profiles. Lysine increased in

all milk types (paper 3:table 2), which indicates that the heat treatment

applied to the milk did not inactivate all natural present indigenous enzymes.

The lysine increase is most probably due to plasmin, which has high specificity

for lysine, in addition, the cathepsin D was most probably inactivated during the

heat treatment. Phenylalanine, tryptophan and tyrosine could not be identified

in the conventional milk, which could be ascribed to that the indigenous

enzymes primary produces large and small peptide fragments and not free

amino acids (Sousa et al. 2001). The increase in free amino acids in the lactose-

hydrolyzed milk indicates that the side-activity from the enzyme preparation

applied to this milk is ascribed to unspecific enzymes that cleave one amino acid

at a time e.g. amino peptidases (Gonzales and Robert-Baudouy 1996).

The concomitant higher level of free amino terminals, free amino acids and the

reduction in the level of caseins in lactose-hydrolyzed milk compared to the

conventional and filtered milk strongly indicates that this is due to proteolytic

41


activity from the β-galactosidase enzyme preparation added to the lactose-

hydrolyzed milk. The filtered milk was included in the study to elucidate what

effect filtration had on the milk and milk proteins. However, the comparable

behavior of the filtered and conventional milk revealed that the milk was not

affected by this processing step. This is an interesting and important result that

must be investigated further in relation to the type of responsible enzymes as

well as the type of peptides that are generated in the milk during storage.

Furthermore, it is worth noticing that the level of amino acids increased

throughout the storage time in the lactose-hydrolyzed milk, while the reduction

in total intact casein was first observed after 90 days. This indicates that the

protein hydrolysis must reach a specific level to be detected by the RP-HPLC MS.

In addition, the amino acid increase may reflect the total hydrolysis of all the

proteins present in the milk.

The increased proteolytic activity observed in the lactose-hydrolyzed UHT milk

is supported by (Tossavainen and Kallioinen 2007). Tossavainen and Kallioinen

(2007) analyzed UHT milk stored for 4 months at 22 °C by SDS-page and

observed an increased proteolytic activity in the lactose-hydrolyzed milk during

storage compared to the conventional milk. The most hydrolyzed proteins were

αs1-CN, αs2-CN and β-CN and the conclusion from the study was that the enzyme

applied to the milk to hydrolyze lactose contained proteolytic side activities.

Similar results were obtained in another study on extended shelf-life heated

lactose-hydrolyzed milk which revealed an increased proteolytic activity in the

lactose-hydrolyzed milk compared to the conventional milk after two weeks of

storage (Kallioinen and Tossavainen 2009).

42


Tot

al c

asei

n (%

inta

ct p

rote

in)

0

20

40

60

80

Conventional indirect UHT

Lactose-hydrolyzed direct UHT

Lactose-hydrolyzed indirect UHT

Filtered indirect UHT

Alp

ha-s

1-ca

sein

(%

inta

ct p

rote

in)

0

5

10

15

20

25

30

Days of storage

0 50 100 150 200 250 300

Bet

a-ca

sein

(%

inta

ct p

rote

in)

0

5

10

15

20

25

30

35

Figure 14. Relative quantification of total-CN, αs1-CN and β-CN in conventional,

lactose-hydrolyzed and filtered UHT heated milk during storage at 22 °C.

43

Chapter 7. Lipid oxidation and thermal degradation


In this chapter lipid oxidation will be discussed including the results obtained in

the present project in the concentration of aldehydes, ketones and furans in the

milk. DHS-GC-MS was used to analyze the volatile compounds present in the

milk, as described in chapter 4.1.1.

7.1 Introduction to lipid oxidation

Lipid oxidation is, in addition to the Maillard reaction, one of the most

important sequences of reactions taking place in food products. Oxidation of

lipids, primary the unsaturated fatty acids is catalyzed by heat (Vazquez-

Landaverde et al. 2005; Costa et al. 2011), light (Barriuso et al. 2012), enzymes,

transition metals and by microorganisms (Walstra et al. 1999; Shahidi and

Zhong 2010). The oxidation of the fatty acids generates primary oxidation

products including lipid hydroperoxides (LOOH). These peroxides are unstable

and decompose to radicals which react with new lipid molecules and initiate a

propagation process in the so called autoxidation. This process will proceed until

the radicals potentially react with antioxidants or proteins which may terminate

the process (Frankel 1982; Alaiz et al. 1997; Yilmaz and Toledo 2005; Elias et al.

2008; Shahidi and Zhong 2010). The hydroperoxides can also be decomposed to

secondary oxidation products such as aldehydes and ketones.

Psychrotrophic bacteria are favored by cold storage and may contribute to lipid

degradation by the formation of the heat stable extracellular enzymes such as

lipases and proteases. This could be relevant in the milk prior heat treatment

when the milk is stored cold. The bacteria produce these enzymes in the late log

phase and early stationary phase, thus a rather high concentration of bacteria is

needed to produce sufficient levels of enzymes that can degrade the lipids in

milk. However, for UHT treated milk, during the long-term storage at higher

temperatures the bacterial lipases may contribute to the degradation of lipids

(Czerniewicz et al. 2006; Gargouri et al. 2013).

7.2 Effect of lactose-hydrolysis on lipid oxidation

In paper 1 it was revealed that the ketone concentration varied between the

milk types (conventional, filtered and lactose-hydrolyzed UHT milk) as well as

within the batches of each milk type. In addition, the conventional UHT milk

44


had overall a higher concentration of all ketones identified in the milk during

storage compared to the lactose-hydrolyzed UHT milk which was presented in

paper 2:figure 4. A similar formation rate was observed in all milk types

during storage, although the formation rate of 2-pentanone, 2-heptanone and 2-

nonanone were significantly higher in the lactose-hydrolyzed milk compared to

the conventional milk, which was also the most abundant compounds in all milk

types (paper 2:figure 4). Ketones are formed from oxidation of β-ketoacids

and are highly temperature-dependent. Consequently, similar levels of ketones

could have been expected for the indirect heated conventional and lactose-

hydrolyzed milk, while a lower level of ketones may have been expected in the

direct lactose-hydrolyzed milk compared to the indirect heated milk. However,

no difference was observed between the direct and indirect heated lactose-

hydrolyzed milk, indicating that the different heat-load did not affect the

formation of ketones. A possible explanation for the variation within the batches

could be differences in holding times before the heat treatment, as well as

different raw milk quality which contributes to variations in bacterial growth

and bacterial production of enzymes (Gargouri et al. 2013). These enzymes may

affect the stability of the lipids prior to the heat treatment and contribute to a

higher level of lipid degradation during the heat treatment. The higher level of

ketones in the conventional UHT milk compared to the lactose-hydrolyzed UHT

milk, and the similar ketone levels in the lactose-hydrolyzed milk types indicate

that the hydrolysis of lactose affect the ketone formation more than the heat

load.

No significant changes in the level of lipid oxidation derived aldehydes were

observed in any of the milk types during the 9-months storage period (paper 2

and 3). Similar levels were found for the aldehydes in the lactose-hydrolyzed and

conventional UHT milk during storage, probably due to the same concentration

of fat in all milk types. From these results it were concluded that the lactose

hydrolysis did not affect the lipid oxidation derived aldehydes.

Vazquez-Landaverde et al. (2005) investigated volatile compounds in raw,

pasteurized and UHT heated skim and whole milk by SPME (Vazquez-

Landaverde et al. 2005). The study revealed that the level of ketones increased

with increasing heat load and that 2-heptanone and 2-nonanone were two of the

most abundant ketones in all the milk supporting the results observed in paper

2. In addition, the study showed that the aldehydes (hexanal, furfural, heptanal,

octanal, nonanal and decanal) increased with increasing heat treatment and

increasing lipid content, although to a lower degree compared to the ketones.

Contarini et al. (1997) investigated direct heated UHT milk stored for 3 months

45


at room temperature using dynamic headspace GC-MS and observed an

increased concentration in the identified ketones 2-butanone, 2-pentanone and

2-heptanone during storage. In addition, the aldehydes identified (pentanal,

hexanal and heptanal) in the skim milk increased during storage, while they

decreased in the whole milk (Contarini et al. 1997). Valero et al. (2001) found an

increased level of hexanal in whole milk and a decrease in skim milk, including

high variation of the compounds during the storage time. These contradictory

results indicate that the method applied to analyze aldehydes is important for

the content of aldehydes, that the aldehydes are dependent on storage conditions

as well as concentration of lipids.

In paper 2, 2-pentylfuran was observed to increase significantly in conventional

UHT milk compared to the lactose-hydrolyzed UHT milk during storage. 2-

Pentylfuran is found to be derived from linoleic acid (Frankel 1982; Min and

Boff 2002; Perez Locas and Yaylayan 2004; Becalski and Seaman 2005; Hasnip

et al. 2006; Zhang et al. 2012), and the lower formation rate in lactose-

hydrolyzed milk could indicate that an antioxidant effect from the more

progressed Maillard reaction and from the proteolytic activity forming

antioxidative compounds may decrease the formation of lipid oxidation (Morales

and Jiménez-pérez 2001; Clausen et al. 2009; Gu et al. 2009).

46


47

Chapter 8. Sensory characteristics


In this chapter the sensory characteristics of milk will be discussed including

results obtained in the present project. It will be focused on results presented in

paper 4 which related the chemical compounds, in the form of volatile

compounds, carbohydrates and primary amines with sensory characteristics.

Sensory descriptive analysis was used to describe the sensory characteristics of

milk and other analyzing techniques applied were DHS-GC-MS, 1H NMR

spectroscopy and fluorescamine assay (described in chapter 4).

8.2 Sensory and chemical analysis of milk

8.2.1 Aroma and flavor of milk

Aroma and flavor are both very important sensory attributes of milk products

(Thomas 1981; Giusti et al. 2007; Kokkinidou and Peterson 2014) and if the

aroma and/or flavor are altered due to bacterial, enzymatic or chemical changes

the consumer acceptance will be affected. The aroma and flavor of milk are

determined by the content of aroma active volatiles at different concentrations.

Many compounds are found in concentrations below the threshold values but

still contribute to the distinct impression of milk due to mixture enhancement

(Badings et al. 1981; Carunchia and Drake 2007). In lactose-hydrolyzed milk, it

is a challenge to maintain sensory characteristics identical to conventional milk

(Adhikari et al. 2010), since the change in carbohydrates may alter the perceived

sweetness and also affect other sensory characteristics through mixture

enhancement.

The aroma and flavor of raw milk is refined and balanced, generated from a

complex mixture of volatile compounds present at different concentrations.

Dimethyl sulfide, diacetyl, 2-methyl-1-butanol and different aldehydes are

important for the aroma and flavor of raw milk (Belitz et al. 2001; Al-Attabi

2009). When milk is heat-treated the aroma and flavor is changed markedly.

Heat-treated milk is described to have a cooked, heated, stale and flat flavor

(Badings et al. 1981; Zabbia et al. 2012). The aroma and flavor is mainly

determined by sulfide compounds, ketones, aldehydes and lactones, formed

from thermal degradation of proteins, lipid oxidation and the Maillard reaction.

However, the aroma and flavor depends on the heat load applied as well as level

of dissolved oxygen (Thomas et al. 1975; Contarini et al. 1997; Walstra et al.

1999; Vazquez-Landaverde et al. 2005). The aroma and flavor of pasteurized

48


milk is highly accepted among people in Scandinavia, as well as in the US, while

the flavor of UHT milk is less accepted in Scandinavia and US but still popular in

other European countries (Clare et al. 2005; Kokkinidou and Peterson 2014).

As found in the project, the aroma and flavor of milk constantly change during

storage due to the different chemical reactions taking place in the milk. These

chemical changes were discussed in chapter 4, 5 and 6, and paper 2, 3, and

4, and an overview can be seen in figure 15.

8.2.2 Sensory descriptive analysis and chemical analysis of

conventional and lactose-hydrolyzed UHT milk during storage

In the present study, boiled milk flavor (figure 15) was present in a higher level

in conventional UHT milk compared with the two lactose-hydrolyzed milk types

after 3 months of storage (paper 4:figure 1). In addition, boiled milk aroma

increased from fresh (0 months) to 3 months of storage in conventional and

indirect heated lactose-hydrolyzed UHT milk (figure 15). This indicates that the

different heat processes, direct and indirect UHT, did not affect the boiled milk

flavor but had an effect on the boiled milk aroma. In addition, the hydrolyzation

did have an effect on the boiled flavor intensity. Glycated β-LG has a higher heat

stability compared to unglycated β-LG which might explain the observed effect

in boiled flavor of the lactose hydrolysis in the milk (Liu and Zhong 2013).

Tossavainen (2008) used a trained sensory panel to investigate lactose-

hydrolyzed and conventional UHT milk. The sensory analysis revealed similar

development of cooked (boiled), heated and burned flavor in both milk types

suggesting that cooked flavor is mainly dependent on the protein denaturation

and not the type of carbohydrates in the milk (Tossavainen 2008). This result

was inconsistent with the results in the present study. Cooked flavor is directly

linked to the heat processing as a result of the thermal denaturation of the serum

proteins, which expose thiol and sulfide groups that contribute to the formation

of sulfur containing volatiles (Hutton and Patton 1952). A reduced protein

denaturation and a concomitant reduction in cooked flavor has been

demonstrated in milk subjected to direct UHT processing compared with

indirect UHT processing (Al-Attabi 2009). However, this was not observed for

the lactose-hydrolyzed UHT milk in paper 2, 3 and 4.

DMDS and DMTS were the only sulfides identified in the study (paper 2, 3 and

4). DMDS remained constant and below its threshold (7.6 µg/liter water (Belitz

et al. 2001)) value in all milk types throughout the study (paper 3:table 2).

49


DMTS increased in the lactose-hydrolyzed milk during the first 50 days of

storage followed by a decline, while in the conventional milk DMTS decreased

steadily throughout storage (paper 2:figure 7). After 3 months of storage the

conventional UHT milk had a lower level of DMTS compared to the direct

lactose-hydrolyzed UHT milk (paper 4). The decline in DMTS is probably due

to oxidation, or conversion to other sulfide containing compounds (Rerkrai et al.

1987; Al-Attabi 2009). The sulfides may have contributed to the boiled milk

aroma in the first period of storage.

Rerkrai et al. (1987) studied sensory characteristics of direct heat-treated UHT

milk for 24 weeks of storage and found that the cooked flavor decreased while

the stale flavor increased throughout storage (Rerkrai et al. 1987). In this study

stale flavor was evaluated to be more intense in the 3 months stored direct

lactose-hydrolyzed UHT milk compared to the 3 months stored conventional

UHT milk shown in figure 15 (paper 4). The off-flavor observed in the lactose-

hydrolyzed milk was expected to be related to stale flavor formed in the milk

during storage. Therefore, a correlation between methyl ketones, aldehydes and

stale flavor was investigated by partial least square (PLS) regression presented in

paper 4:figure 2. The expected off-flavor could not be entirely related to the

stale flavor formed during storage as the stale flavor intensity was observed to be

highest in fresh conventional UHT milk compared to all the other milk types at

every analyzing time. The PLS regression revealed that 2-butanone, 2-

pentanone, 2-heptanone, 2-nonanone, heptanal, octanal and nonanal were

found to be most important for the stale flavor (paper 4:figure 2). As

discussed in chapter 7 and paper 2 the ketones increased during storage in all

milk types which was also shown in paper 4. In addition, no pronounced

difference in ketone formation during storage was observed between the

conventional and lactose-hydrolyzed milk (paper 2:figure 4). From this result

it was concluded that the ketones do not have a major role in the formation of

off-flavor in lactose-hydrolyzed UHT milk during storage.

The lipid oxidation derived aldehydes identified and presented in paper 2 and 3

did not increase during storage. However, in the results presented in paper 4

the conventional milk had a higher level of hexanal compared to the lactose-

hydrolyzed milk after three months of storage and octanal and decanal

significantly increased during 4 months of storage in all milk types. Based on

these findings, lipid oxidation derived aldehydes are not expected to contribute

greatly to the off-flavor in lactose-hydrolyzed milk.

50


Furthermore, Valero et al. (2001) studied the sensory properties of conventional

UHT milk and found that a stale flavor appeared after three months of storage.

Increased aldehyde concentration during storage has been observed to correlate

with stale flavor, as well as Strecker degradation products (Jeon et al. 1978; Ledl

and Schleicher 1990; Perkins and Elliott 2005; Zabbia et al. 2012). Vazquez-

Landaverde (2005) compared the concentration of aldehydes in UHT,

pasteurized and raw milk. In general, nonanal and decanal were found to

contribute to raw milk flavor and for UHT flavor, octanal, hexanal, 2-

methylbutanal, 3-methylbutanal and 2-methylpropanal were found to be

important, even though they were not as abundant as the ketones in the UHT

milk.

In paper 2, 2-methylbutanal was identified and quantified in lactose-

hydrolyzed and conventional UHT milk. The threshold value of 2-methyltbuanal

is 4 µg/ liter water (Belitz et al. 2001) and in the lactose-hydrolyzed milk 2-

methylbutanal was quantified to 7-8 µg/kg after 9 months of storage while the

concentration remained constant below 1 µg/kg in the conventional UHT milk

throughout the storage period (paper 2:figure 5). Intriguingly, the

concentration of 2-methylbtutanal in the lactose-hydrolyzed milk reached the

threshold value when the milk was 4.5 months old, which is close to the shelf-life

of this milk type. The high concentration of 2-methylbutanal observed in the

lactose-hydrolyzed milk during storage would definitely contribute to the flavor

in the milk.

Even though no decrease in carbohydrate concentrations were observed from

the 1H NMR spectroscopy as discussed in chapter 5 and in paper 3:table 2

the sweet taste decreased in all milk types during storage (figure 15) (paper 4).

This finding can possibly be described by a reduction of available carbohydrates

due to the Maillard reaction (Chávez-Servín et al. 2006) and possibly by the

masking effect from formation of bitter tasting compounds during storage (Beck

et al. 2014). In paper 4, an increased bitter taste was observed in the lactose-

hydrolyzed milk compared to the conventional milk. In addition, the bitter taste

intensity was correlated with the level of free amino terminals which was higher

in the lactose-hydrolyzed milk compared to the conventional milk (figure 16)

(paper 2 and 4). The increased level of free amino terminals in the lactose-

hydrolyzed milk support the results in paper 3 which revealed a reduction in

total intact CN in the lactose-hydrolyzed milk after 90 days of storage, while the

reduction was less pronounced in the conventional and filtered UHT milk

(figure 14). The reduction in total intact CN was primary due to the proteolysis

of β-CN and αs1-CN. This result is supported by McKellar (1981) who

51


demonstrated a correlation between levels of proteolysis and formation of bitter

flavor in UHT milk (McKellar 1981). The bitter flavor may be due to increased

level of hydrophobic bitter peptides and amino acids due to casein hydrolysis

(Guigoz and Solms 1976; Ney 1979; Harwalkar et al. 1993; Gomez et al. 1997;

Kilara and Panyam 2003).

Figure 15. Intensity of sensory attributes during 4 months of storage in

conventional UHT milk (CONVI), lactose-hydrolyzed direct heated UHT milk

(LHD) and lactose-hydrolyzed indirect heated UHT milk (LHI). Lower case

letters indicate statistical difference between different milk at the same storage

time (p<0.05).

52


Figure 16. Correlation between the level of primary amino groups (X-axis) and

bitter taste intensity of three different samples of milk (direct and indirect

heated lactose-hydrolyzed UHT milk and indirect heated conventional UHT

milk) stored for four months.

53

Chapter 9. Effect of heat treatment and storage on lactose-hydrolyzed milk

Chapter 9. Effect of heat treatment and storage on lactose-

hydrolyzed milk

In this chapter the relation between the different compounds identified in the

present study will be discussed. The Maillard reaction, proteolysis and lipid

oxidation are all parameters that have a great influence on the chemical changes

in the milk. This chapter aims to connect the reactions and the most important

compounds identified into a coherent perspective. In figure 17 a cascade of

reactions mainly based on the Maillard reaction compounds identified in this

study, however, additional side activities such as the carbohydrate degradation,

formation of amino acids by proteolysis, and thermal degradation of lipids are

included in the cascade.

54


Protein + Sugar

Imine

1,2-eneaminol

Amadori compound(Lactulosyllysine/fructosyllysine)

2,3-eneaminol

Furosine Acid hydrolysis

Acid pHFurfural 3-deoxyosone

1-deoxyosone Neutral or alkaline pH

1-deoxy-2,4-hexodiulose(2,4-pentadione)

Isomerization

Acetate

Lipids

LysineIsoleucineMethionineValineAlanine

LactoseGlucoseGalactose

Strecker degradation

2-methylbutanal

KetonesAldehydes

AldehydesKetones

Oxidation Thermal degradation

2-pentylfuranOxidation

Lipid hydroperoxides

1,2-enediol Fructose 2,3-enediol

GlyoxalMethylglyoxalGlucosone3-deoxyglucosone

Alkaline hydrolysis

Strecker aldehydes

2-methylfuran2-ethylfuran

DMDS

DMTS

Proteolytic activity

Free amino terminals

Formate + Galactose

Glucose

Figure 17. Overview of the compounds identified in the conventional, filtered

and lactose-hydrolyzed milk. The compounds were identified with GC-MS (blue

rectangles), NMR (orange rectangles), HPLC (green rectangles) and

fluorescamine assay (pink rectangle). A dashed outline indicates increase of

compounds in lactose-hydrolyzed UHT milk, a dotted outline indicates increase

of compounds in conventional UHT milk, a solid outline indicates no difference

between lactose-hydrolyzed and conventional UHT milk during storage.

55


The Maillard reaction is highly dependent on the concentration of carbohydrates

and proteins available in the milk during heat treatment and storage. As was

observed in the formation of the volatile compounds furosine, 2-methylbutanal

and formate, shown in figure 12, 13 and paper 3:table 2, respectively, the

carbohydrate concentration during heat treatment had a major effect on the

initial level of the formation of the compounds. In addition, during storage the

carbohydrate type affected the formation rate tremendously in furosine and 2-

methylbutanal.

The free amino acid isoleucine formed in the milk due to proteolysis had a major

effect on the formation of 2-methylbutanal and in figure 18 it can be seen that

2-methylbutanal and isoleucine showed similar development pattern. Without

the involvement of amino acids in the Maillard reaction, less advanced Maillard

reaction products would be formed, as demonstrated in figure 18 by the

constant low concentration of 2-methylbutanal in the conventional UHT milk

during storage (paper 2:figure 13). Based on this fact it is proposed that the

proteolysis present in the lactose-hydrolyzed milk gives rise to free amino acids

that can be included in the Maillard reaction cascade, such as Strecker

degradation and give rise to flavor-contributing compounds such as 2-

methylbutanal.

56


2-m

eth

ylb

uta

nal

[µ

g/K

g]

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0


Days of storage

0 50 100 150 200 250 300

Iso

leuc

ine

[m

M]

0.00

0.05

0.10

0.15

0.20

Figure 18. GC-MS analysis of 2-methylbutanal and 1H NMR analysis of

isoleucine in conventional and lactose-hydrolyzed UHT milk during storage.

The importance of amino acids in the Maillard reaction has been highlighted in

earlier studies and also reviewed (Izzo and Ho 1992; van Boekel 2006;

Balagiannis et al. 2009). Balagiannis et al. (2009) investigated the importance of

the level of glucose in the formation of Strecker aldehydes, and observed a clear

correlation between the level of glucose and the formation of Strecker aldehydes.

However, the study also revealed that the levels of the individual amino acids

included in the formation of the specific aldehyde were highly important. The

amino acids should have more focus in relation to the formation of flavor

compounds in milk and especially lactose-hydrolyzed milk. However, Izzo and

Ho (1992) as well as van Boekel (2006) have discussed the importance of the

arrangement of amino acids in peptides and proteins, which also may contribute

to the formation of flavor compounds. Martins and van Boekel (2005)

investigated the importance of the initial concentration of glucose and glycine

for the formation of organic acids, glucosones and melanoidins formed in the

Maillard reaction (Martins and van Boekel 2005). No significant difference was

found for the reaction rate when the initial concentration of the carbohydrates

57


was two times higher than the proteins concentration or when the protein

concentration was two times higher than the carbohydrate concentration. This

indicates that the precursor with lowest concentration limits the formation of the

organic acids, glucosones and melanoidins in the Maillard reaction. As already

discussed in this chapter, a lower initial concentration was observed in many of

the identified compounds in the milk as presented in paper 2 and 3, and the

40% lower level of lactose during heat treatment was important for the protein

and carbohydrate condensation. Considering the lactose-hydrolyzed UHT milk,

and the higher level of Maillard reactions taking place during storage, it is

expected that the fact that the milk is heated with a lower lactose concentration

has a positive influence. This may decrease the reaction rate of the formation of

Maillard reaction compounds in the initial storage period, and should decrease

the formation of products in the advanced and final stage of the Maillard

reaction compared to as if the lactose had been hydrolyzed before the heat

treatment.

In the present study, comparing conventional and lactose-hydrolyzed UHT milk,

lipid oxidations are not the major reaction contributing to off-flavor in lactose-

hydrolyzed milk during storage. However, even though the lipid oxidation does

not directly contribute to flavor in the lactose-hydrolyzed milk the lipid

oxidation products may be included in reactions with carbohydrates, proteins as

well as in the Maillard reaction at different stages (Whitfield and Mottram 1992).

Yaylayan (1997) states that all the precursors in milk such as carbohydrates,

amino acids and lipids are divided into pools that can react and degrade to

reactive compounds individually but also across the pools (Yaylayan 1997). This

contributes to a highly complex cascade of reactions.

Zamora and Hidalgo (2005) reviewed the reaction products formed in the

Maillard reaction and lipid oxidation and discussed how the two reactions

interacted and concluded that the Maillard reaction may both increase lipid

oxidation by phospholipid glycation and reduce the lipid oxidation by formation

of antioxidant compounds, and the lipid oxidation may also increase the

formation of Maillard reaction compounds (Zamora and Hidalgo 2005). The

Maillard reaction can in the advanced and final stage generate compounds, such

as melanoidins, that have an antioxidant activity which thereby can reduce the

lipid oxidation (Alaiz et al. 1999). The lipids are included in the Maillard reaction

by the amino group of the phospholipid and by the aldehydes and ketones

formed from the fatty acid oxidation (Zamora and Hidalgo 2005). In addition,

the unsaturated aldehydes may react with hydrogen sulfide and thereby inhibit

the formation of further reactions to sulfide containing compounds that

58


contribute to cooked flavor of the product. The lipid oxidation may also generate

dicarbonyls such as glyoxal and methylglyoxal, as well as radicals, highly reactive

with proteins and Maillard reaction compounds.

An extensive Maillard reaction may decrease the quality of milk considerably.

However, the prevalence of proteolytic enzymes that degrade proteins into

amino acids, included in the Maillard reaction, and bitter tasting peptides may

be just as important for the milk quality during storage than the Maillard

reaction itself. Hence, the enzymatic hydrolysis of lactose into glucose and

galactose contributes to an increased Maillard reaction due to the increased

reactivity of the monosaccharides. However, side-activities from the enzyme

preparation may contribute more to the increased reactions observed in the

lactose-hydrolyzed milk through the formation of amino acids and bitter

peptides.

59

Chapter 10. Conclusion and future perspectives


10.1 Conclusion

The overall aim of the present PhD study was to elucidate differences between

conventional and lactose-hydrolyzed UHT milk in relation to formation of off-

flavor during a nine-month storage period.

Specific Maillard reaction compounds were identified in the conventional and

lactose-hydrolyzed UHT milk by DHS-GC-MS, NMR and RP HPLC MS. The

study revealed that the Amadori-product indicator, furosine, and the Strecker

degradation compound 2-methylbutanal increased in the lactose-hydrolyzed

UHT milk during storage, while the concentration remained constant in the

conventional UHT milk (paper 2). This finding indicates that the hydrolysis of

lactose to the more reactive carbohydrates glucose and galactose appears to have

a major effect on the formation of Maillard reaction compounds. The study also

revealed that the initial concentration of furosine, formate and specific volatile

compounds e.g. 2-methylbutanal were higher in the conventional UHT milk than

in lactose-hydrolyzed and filtered UHT milk. This likely reflects that the

concentration of lactose affects the degree of formation of these compounds

during the heat treatment.

Chromatographic profiling of the intact proteins in the milk revealed that

hydrolysis was present during storage in both lactose-hydrolyzed UHT and

conventional UHT milk, although more apparent in the lactose-hydrolyzed milk

compared to the conventional milk (paper 3). Filtered UHT milk was also

included in the study to follow the processing of the lactose-hydrolyzed milk, and

a similar protein hydrolysis in the filtered and conventional milk was observed.

Consequently, it is concluded that the β-galactosidase enzyme preparation

contains side-activities that causes protein hydrolysis in the lactose-hydrolyzed

UHT milk.

An interesting and concomitant pattern was observed in the formation of 2-

methylbutanal and the hydrolysis of β- and αs1-CN in the lactose-hydrolyzed

UHT milk. After 90 days of storage the level of 2-methylbutanal was higher in

the lactose-hydrolyzed UHT milk than in the conventional milk, and

concomitantly a pronounced decrease in the level of intact β- and αs1-CN was

initiated. Intriguingly, the shelf-life of lactose-hydrolyzed UHT milk is also

approximately 90 days. Thus, a connection seems to be plausible, which implies

60


that these chemical changes in the lactose-hydrolyzed milk may be crucial for the

shelf-life.

Sensory analysis revealed that stale flavor increased in both lactose-hydrolyzed

UHT and conventional UHT milk during 4 months of storage. GC-MS analysis

revealed that the stale flavor was related to an increased level of ketones and

specific aldehydes: 2-methylbutanal, octanal and decanal (paper 4). The

relation between stale flavor and 2-methylbutanal could indicate that 2-

methylbutanal is a decisive compound for the formation of off-flavor observed in

the lactose-hydrolyzed milk since a higher concentration was observed in this

milk type compared with the conventional milk during storage.

Sensory analysis also revealed a higher bitter taste score in the lactose-

hydrolyzed milk than in the conventional UHT milk (paper 4). Intriguingly, the

bitter taste was correlated with the level of free amino terminals in the milk

analyzed by fluorescamine assay, which was higher in the lactose-hydrolyzed

milk compared to the conventional milk. The correlation between bitter taste

and increased level of free amino terminals is an important result in relation to

understand the consequence of an increased protein hydrolysis in the lactose-

hydrolyzed milk. The bitter taste and the higher level of free amino terminals in

lactose-hydrolyzed milk could most probably be related to the more extensive

hydrolysis of the proteins β- and αs1-CN (paper 3).

When comparing two lactose-hydrolyzed milk types exposed to direct and

indirect UHT treatment, respectively, no differences in the formation of Maillard

reaction compounds or protein hydrolysis were observed (paper 2 and 3).

Consequently, it can be concluded that the type of sugar and protein hydrolysis

most probably affect the chemical reactions taking place in the milk to such an

extent that the effect of the difference in the heat load becomes diminishing.

Batch differences were observed within each milk type, which indicates that it is

difficult to produce commercial milk without variation (paper 1). The variation

was mostly apparent in the concentration of ketones, indicating that the

variation is dependent on the lipids in the milk. Batch variation can probably be

reduced by making the time before heat treatment of the milk as short as

possible to avoid degradation of lipids by naturally present lipases or

psychrotrophic produced enzymes. The variation present between different

batches may not be noticeable or important for the consumers. However, when

analytically investigating the milk the variation had considerable impact on the

results.

61


The Maillard reaction is a highly complex reaction including many different

reaction pathways. This study has contributed to an increased understanding of

how Maillard reaction products are generated in milk during storage. In

addition, the study has revealed how protein hydrolysis affects Maillard reaction

product formation, and how the proteolysis may contribute to bitter taste in the

milk. Further investigations are needed to find the relation between milk

constituents and chemical changes that can be linked to the decreased shelf-life

of lactose-hydrolyzed UHT milk.

62


10.2 Future perspective

For future improvements in the production of long-term stored lactose-

hydrolyzed milk, the raw material should be considered as a key parameter to

obtain a high-quality product and to eliminate variation between batch

productions. It is recommended that the raw milk is of highest quality and that

the transportation to and holding time at the dairy are as short and cold as

possible to avoid unnecessary degradation of proteins and lipids in the milk.

Based on the results of the present project it seems advantageous that the

enzyme preparation applied to the milk to hydrolyze lactose is of highest purity,

to avoid side-activities from other enzymes present in the preparation.

Consequently, investigating the enzyme preparation in relation to purity and to

the side activities that contribute to the degradation of the proteins in the milk

could be highly relevant and beneficial for further advancements in the

development of high-quality lactose-hydrolyzed milk products.

This study revealed that especially the hydrolysis of two caseins was enhanced in

the lactose-hydrolyzed milk, and there was a strong indication that this

hydrolysis could be correlated to the development of an increased bitterness.

Therefore these casein degradation products should be investigated further in

order to find the specific peptides that contribute to bitterness. The

characterization of these peptides and the cleavage sites will also contribute to

the identification of the enzymes responsible for the hydrolysis.

Future research in lactose-hydrolyzed milk could include a more specific

analysis of the phenolic compounds in the milk. It is known that the β-

galactosidase preparation may contain the enzyme arylsulfatase, which

hydrolyze alkyl phenols substituted with a sulfate group that can give off-flavor.

In addition, more Strecker degradation compounds should be analyzed and

identified in order to correlate the free amino acids with the formation of

Maillard reaction compounds. It could also be relevant to correlate the bacteria

and enzyme activity in the milk prior to the heat treatment with the ketone

concentration and fatty acid composition in the milk to obtain an increased

understanding for how the raw milk affects the end product.

Another chemical reaction that may pose effect on the milk quality is glycation of

proteins. The present study revealed a molecular weight shift of +162 Da and

+324 Da for the intact proteins that indicated glycation. However, the result did

not reveal which of the proteins that was glycated or the degree of glycation. The

degree of glycation in the different milk types can be correlated to reactions

taking place in the Maillard reaction during storage, and would therefore be

63


highly relevant to analyze. In addition, it could be interesting to elucidate which

proteins in the milk that was most disposed for glycation and if a relation

between degree of glycation and proteolysis could be observed.

For further research it could be very exciting to investigate if the milk genome

could be used to find a type of milk that was suited for long-shelf life, highly

heat-treated, milk. As milk genomic and metabolomic projects have been used to

find specific milk types with high coagulating proteins for cheese production

(Sundekilde et al. 2011), milk with less level of proteins prone for glycation and

hydrolysis could be found and used specifically for UHT milk.

The β-galactosidase enzyme used to hydrolyze the lactose in the milk may

contribute to a formation of oligosaccharides (Gänzle et al. 2008; Coulier et al.

2009; Gosling et al. 2011; Ruiz-Matute et al. 2012). Oligosaccharides have been

shown to have prebiotic effects (Ben et al. 2008) and consequently, the

oligosaccharides contribute to a better micro-flora in the intestine. It could be

relevant to investigate the positive effect of adding β-galactosidase to milk by

investigating the oligosaccharides formed in more detail by NMR spectroscopy.

It is desirable to decrease the Maillard reaction in milk during storage both from

a product and a nutritive perspective. A previous study has shown that

antioxidants may decrease the formation of Maillard reaction compounds in

conventional UHT milk (Kokkinidou and Peterson 2014). Consequently, it would

be interesting to investigate how different antioxidants added to milk affect the

degree of Maillard reaction in lactose-hydrolyzed UHT milk during storage. In

addition, a sensory descriptive analysis could be applied to the study to obtain

information about the antioxidant effect on the sensory characteristics of the

milk. Overall, there are many unexplored factors and possibilities to optimize

lactose-hydrolyzed milk products and several issues to address in future research

within this field, which can be expected to increase as a result of an increasing

interest in lactose-reduced products from consumers world-wide.

64


65

11. References

11. References

Adhikari K, Dooley LM, Chambers E, Bhumiratana N (2010) Sensory characteristics of commercial lactose-free milks manufactured in the United States. LWT - Food Sci Technol 43:113–118.

Alaiz M, Hidalgo FJ, Zamora R (1999) Effect of pH and temperature on comparative antioxidant activity of nonenzymatically browned proteins produced by reaction with oxidized lipids and carbohydrates. J Agric Food Chem 47:748–752.

Alaiz M, Hidalgo FJ, Zamora R (1997) Comparative Antioxidant Activity of Maillard- and Oxidized Lipid-Damaged Bovine Serum Albumin. J Agric Food Chem 45:3250–3254.

Al-Attabi Z (2009) Dissertation:Volatile sulphur compounds in UHT milk. School of Land, Crop, and Food Sciences, The University of Queensland, St Lucia, Qld, Australia.

Ansari SA, Satar R (2012) Recombinant β-galactosidases – Past, present and future: A mini review. J Mol Catal B Enzym 81:1–6.

Bach V, Kidmose U, Bjørn GK, Edelenbos M (2012) Effects of harvest time and variety on sensory quality and chemcial composition of Jerusalem artichoke (Helianthus tuberosus) tubers. Food Chem 133:82–89.

Badings H, Pol J Van der, Neeter R (1981) Aroma compounds which contribute to the difference in flavour between pasteurized milk and UHT milk. In: Schreier P (ed) Flavour’81. Walter de Gruyter, Berlin, pp 683–692

Balagiannis DP, Parker JK, Pyle DL, et al. (2009) Kinetic modeling of the generation of 2- and 3-methylbutanal in a heated extract of beef liver. J Agric Food Chem 57:9916–22.

Barriuso B, Astiasarán I, Ansorena D (2012) A review of analytical methods measuring lipid oxidation status in foods: a challenging task. Eur Food Res Technol 236:1–15.

Becalski A, Seaman S (2005) Furan precursors in food: A model study and development of a simple headspace method for determination of furan. J AOAC Int 88:102–106.

Beck TK, Jensen S, Bjoern GK, Kidmose U (2014) The masking effect of sucrose on perception of bitter compounds in brassica vegetables. J Sens Stud 29:190–200.

66

11. References

Beeby R (1980) The use of fluorescamine at pH 6.0 to follow the action of chymosin on kappa-casein and to estimate this protein in milk. New Zeal J Dairy Sci Technol 15:94.

Beghin JC (2006) Evolving dairy markets in Asia: Recent findings and implications. Food Policy 31:195–200.

Belitz H-D, Grosch W, Schieberle P (2001) Food Chemistry, 4th ed. Food Chem 73:158–402.

Ben X-M, Li J, Feng Z-T, et al. (2008) Low level of galacto-oligosaccharide in infant formula stimulates growth of intestinal Bifidobacteria and Lactobacilli. World J Gastroenterol 14:6564–8.

Berg H (1993) Reactions of lactose during heat treatment of milk: a quantitative study. 1–176.

Bonfatti V, Grigoletto L, Cecchinato A, et al. (2008) Validation of a new reversed-phase high-performance liquid chromatography method for separation and quantification of bovine milk protein genetic variants. J Chromatogr A 1195:101–6.

Brands CM, van Boekel M a (2001) Reactions of monosaccharides during heating of sugar-casein systems: building of a reaction network model. J Agric Food Chem 49:4667–75.

Brands CMJ, Wedzicha BL, van Boekel M a JS (2002) Quantification of melanoidin concentration in sugar-casein systems. J Agric Food Chem 50:1178–83.

Bunn FH, Higgins PJ (1981) Reaction of monosaccharides with proteins:Possible evalutionary significance. Sci New Ser 213:222–224.

Burvall BYA, Asp N, Dahlqvist A, Oste R (1977) Nutritional value of lactose-hydrolysed milk : protein quality after some industrial processes. J Dairy Res 44:549–553.

Carulli S, Calvano CD, Palmisano F, Pischetsrieder M (2011) MALDI-TOF MS characterization of glycation products of whey proteins in a glucose/galactose model system and lactose-free milk. J Agric Food Chem 59:1793–803.

Carunchia MEW, Drake M (2007) The flavor and flavor stability of skim and whole milk powders. In: Cadwallader KR, Drake M, McGorrin RJ (eds) Flavor dairy Prod. American chemical society, Washington, pp 217–251

Chávez-Servín JL, Romeu-Nadal M, Castellote AI, López-Sabater C (2006) Evolution of free mono and disaccharide content of milk based formula powder during storage. Food Chem 97:103–108.

67

11. References

Choi J, Sabikhi L, Hassan A, Anand S (2012) Bioactive peptides in dairy products. Int J Dairy Technol 65:1–12.

Clare D a, Bang WS, Cartwright G, et al. (2005) Comparison of sensory, microbiological, and biochemical parameters of microwave versus indirect UHT fluid skim milk during storage. J Dairy Sci 88:4172–82.

Clausen MR, Skibsted LH, Stagsted J (2009) Characterization of major radical scavenger species in bovine milk through size exclusion chromatography and functional assays. J Agric Food Chem 57:2912–9.

Contarini G, Leardi R (1994) Use of experimental design to optimize the analysis of volatile compounds by dynamic headspace extraction followed by cold trapping and capillary GC. J high Resolut Chromatogr 17:91–94.

Contarini G, Povolo M (2002) Volatile fraction of milk: comparison between purge and trap and solid phase microextraction techniques. J Agric Food Chem 50:7350–5.

Contarini G, Povolo M, Leardi R, et al. (1997) Influence of Heat Treatment on the Volatile Compounds of Milk. J Agric Food Chem 45:3171–3177.

Costa EN, Lacerda ECQ, Santos SMS, et al. (2011) Action of Successive Heat Treatments in Bovine Milk Fatty Acids. J Braz Chem Soc 22:2115–2120.

Coulier L, Timmermans J, Bas R, et al. (2009) In-depth characterization of prebiotic galacto-oligosaccharides by a combination of analytical techniques. J Agric Food Chem 57:8488–95.

Czerniewicz M, Kruk A, Kieczewska K (2006) Storage stability of raw milk subjected to vibration. Polish J food Nutr Sci 15:65–70.

Czerwenka C, Maier I, Pittner F, Lindner W (2006) Investigation of the lactosylation of whey proteins by liquid chromatography-mass spectrometry. J Agric Food Chem 54:8874–82.

Da Silva Pinto M, Bouhallab S, De Carvalho AF, et al. (2012) Glucose slows down the heat-induced aggregation of β-lactoglobulin at neutral pH. J Agric Food Chem 60:214–9.

Dalgleish DG, Corredig M (2012) The structure of the casein micelle of milk and its changes during processing. Annu Rev Food Sci Technol 3:449–67.

Dalsgaard TK, Nielsen JH, Larsen LB (2007) Proteolysis of milk proteins lactosylated in model systems. Mol Nutr Food Res 51:404–14.

Datta N, Elliott AJ, Perkins ML, Deeth HC (2002) Ultra-high-temperature (UHT) treatment of milk: comparison of direct and indirect modes of heating. Aust J dairy Technol 57:211–227.

68

11. References

Davidek T, Clety N, Aubin S, Blank I (2002) Degradation of the Amadori Compound N-(1-Deoxy-d-fructos-1-yl)glycine in Aqueous Model Systems. J Agric Food Chem 50:5472–5479.

Davídek T, Devaud S, Robert F, Blank I (2006) Sugar fragmentation in the maillard reaction cascade: isotope labeling studies on the formation of acetic acid by a hydrolytic beta-dicarbonyl cleavage mechanism. J Agric Food Chem 54:6667–76.

De MPM, van Dijk AA, Edens L, Dekker PJT (2007) DSM Patent: Enzyme preparations yielding a clean taste. 2007:

De Wit JN (2009) Thermal behaviour of bovine β-lactoglobulin at temperatures up to 150°C. a review. Trends Food Sci Technol 20:27–34.

Delgado T, Corzo N, Santa-María G, et al. (1992) Determination of furosine in milk samples by ion-pair reversed phase liquid chromatography. Chromatographia 33:374–376.

El-Aneed A, Cohen A, Banoub J (2009) Mass Spectrometry, Review of the Basics: Electrospray, MALDI, and Commonly Used Mass Analyzers. Appl Spectrosc Rev 44:210–230.

Elias RJ, Kellerby SS, Decker E a (2008) Antioxidant activity of proteins and peptides. Crit Rev Food Sci Nutr 48:430–41.

Elliott AJ, Dhakal A, Datta N, Deeth HC (2003) Heat-induced changes in UHT milks. Aust J dairy Technol 58:3–10.

Elmore JS, Erbahadir MA, Mottram DS (1997) Comparison of Dynamic Headspace Concentration on Tenax with Solid Phase Microextraction for the Analysis of Aroma Volatiles. J Agric Food Chem 45:2638–2641.

Erbersdobler HF, Faist V (2001) Metabolic transit of Amadori products. Nahrung 45:177–81.

Erbersdobler HF, Somoza V (2007) Forty years of furosine - forty years of using Maillard reaction products as indicators of the nutritional quality of foods. Mol Nutr Food Res 51:423–30.

Evangelisti F, Calcagno C, Nardi S, Zunin P (1999) Deterioration of protein fraction by Maillard reaction in dietetic milks. J Dairy Res 66:237–43.

Farrell HJ, Jimenez-Flores R, Bleck G, et al. (2004) Nomenclature of the proteins of cows’ milk--sixth revision. J Dairy Sci 87:1641–74.

Ferrer E, Alegría A, Farré R, et al. (2002) High performance liquid chromatographic determination of furfural compounds in infant formulas changes furnig heat treatment and storage. J Chromatogr A 947:85–95.

69

11. References

Flatz G, Rotthauwe HW (1973) Lactose nutrition and natural selection. Lancet 302:76–77.

Fox PF (2009) Advanced Dairy Chemistry. 3:1–15. Frankel EN (1982) Volatile lipid oxidation products. Prog Lipid Res 22:1–33.

Fuller F, Huang J, Ma H, Rozelle S (2006) Got milk? The rapid rise of China’s dairy sector and its future prospects. Food Policy 31:201–215.

Gargouri A, Hamed H, ElFeki A (2013) Analysis of raw milk quality at reception and during cold storage: Combined effects of somatic cell count and psychrotrophic bacteria on lipolysis. J Food Sci 78:1405–1411.

Giusti AM, Bignetti E, Cannella C (2007) Exploring New Frontiers in Total Food Quality Definition and Assessment: From Chemical to Neurochemical Properties. Food Bioprocess Technol 1:130–142.

Gomez BMJ, Garde S, Gaya P, et al. (1997) Relationship between level of hydrophobic peptides and bitterness in cheese made from pasteurized and raw milk. J Dairy Res 64:289–297.

Gonzales T, Robert-Baudouy J (1996) Bacterial aminopeptidases: properties and functions. FEMS Microbiol Rev 18:319–44.

Gosling A, Stevens GW, Barber AR, et al. (2011) Effect of the substrate concentration and water activity on the yield and rate of the transfer reaction of β-galactosidase from Bacillus circulans. J Agric Food Chem 59:3366–72.

Gu F, Kim JM, Hayat K, et al. (2009) Characteristics and antioxidant activity of ultrafiltrated Maillard reaction products from a casein–glucose model system. Food Chem 117:48–54.

Guigoz Y, Solms J (1976) Bitter Peptides, Occurrence and Structure. Chem Senses 2:71–84.

Gänzle MG, Haase G, Jelen P (2008) Lactose: Crystallization, hydrolysis and value-added derivatives. Int Dairy J 18:685–694.

Harju M, Kallioinen H, Tossavainen O (2012) Lactose hydrolysis and other conversions in dairy products: Technological aspects. Int Dairy J 22:104–109.

Harris DC (2007) High-performance liquid chromatography. Quant. Chem. Anal., 7th ed. W H Freeman and Company, New York, pp 556–587

Harwalkar VR, Cholette H, McKellar RC, Emmons DB (1993) Relation Between Proteolysis and Astringent Off-Flavor in Milk. J Dairy Sci 76:2521–2527.

70

11. References

Hasnip S, Crews C, Castle L (2006) Some factors affecting the formation of furan in heated foods. Food Addit Contam 23:219–27.

Hawke JC (1966) Reviews of the progress of dairy science. J Dairy Res 33:225–243.

Hayes MG, Hurley MJ, Larsen LB, et al. (2001) Thermal inactivation kinetics of bovine cathepsin D. J Dairy Res 68:267–76.

Hayward DL (1977) A symmetry rule for the cricular dichroism of reducing sugars, and the proportion of carbonyl fomrs in aqueous solution thereof. Carbohydr Res 53:13–20.

Helmig D, Vierling L (1995) Water adsorption capacity of the solid adsorbents Tenax TA, Tenax GR, Carbotrap, Carbotrap C, Carbosieve SIII, and Carboxen 569 and water management techniques for the atmospheric sampling of volatile organic trace gases. Anal Chem 67:4380–4386.

Heyman MB (2006) Lactose intolerance in infants, children, and adolescents. Pediatrics 118:1279–86.

Hodge JE (1953) Chemistry of Browning Reactions in Model Systems. 928–943.

Hofmann T, Münch P, Schieberle P (2000) Quantitative model studies on the formation of aroma-active aldehydes and acids by strecker-type reactions. J Agric Food Chem 48:434–40.

Hofmann T, Schieberle P (2000) Formation of aroma-active strecker-aldehydes by a direct oxidative degradation of Amadori compounds. J Agric Food Chem 48:4301–5.

Hu F, Furikata K, Kato Y, Tanokura M (2007) Nondestructive quantification of organic compounds in whole milk without pretreatment by two-dimensional NMR spectroscopy. J Agric Food Chem 55:4307–4311.

Hurley M., Larsen L., Kelly A., McSweeney PL. (2000) The milk acid proteinase cathepsin D: a review. Int Dairy J 10:673–681.

Hutton JT, Patton S (1952) The Origin of Sulfhydryl Groups in Milk Proteins and their Contributions to “Cooked” Flavor. J Dairy Sci 35:699–705.

Ingram CJE, Mulcare C a, Itan Y, et al. (2009) Lactose digestion and the evolutionary genetics of lactase persistence. Hum Genet 124:579–91.

International Organization for Standardization, International Dairy Federation (2004) Milk and milk products - Determination of furosine content - Ion-pair reverse-phase high-performance liquid chromatography method. Int. Stand. ISO 18329 :2004

71

11. References

Itan Y, Jones BL, Ingram CJE, et al. (2010) A worldwide correlation of lactase persistence phenotype and genotypes. BMC Evol Biol 10:36.

Izzo H V, Ho C (1992) Peptide-specific Maillard reaction products : a new pathway for flavor chemistry. 3:3–7.

Jansson T, Clausen MR, Sundekilde UK, et al. (2014) Lactose-Hydrolyzed Milk Is More Prone to Chemical Changes during Storage than Conventional Ultra-High-Temperature (UHT) Milk. J Agric Food Chem 62:7886–96.

Jelen P, Tossavainen O (2003) Low lactose free milk and dairy products - prospects, technologies and applications. Aust J dairy Sci Technol 58:161–165.

Jensen HB, Holland JW, Poulsen N A, Larsen LB (2012) Milk protein genetic variants and isoforms identified in bovine milk representing extremes in coagulation properties. J Dairy Sci 95:2891–903.

Jensen S, Oestdal H, Skibsted LH, et al. (2011) Chemical changes in wheat pan bread during storage and how it affects the sensory perception of aroma, flavour, and taste. J Cereal Sci 53:259–268.

Jeon IJ, Thomas EL, Reineccius GA (1978) Production of volatile flavor compounds in ultrahigh-temperature processed milk during aseptic storage. J Agric Food Chem 26:1183–1188.

Johns DO, Dills RL, Morgan MS (2005) Evaluation of dynamic headspace with gas chromatography/mass spectrometry for the determination of 1,1,1-trichloroethane, trichloroethanol, and trichloroacetic acid in biological samples. J Chromatogr B 817:255–261.

Kallioinen H, Tossavainen O (2009) Changes during storage of lactose hydrolysed extended shelf life (ESL) milk. DMZ, Leb und Milchwirtschaft 130:47–50.

Kanavouras A, Kiritsakis A, Hernandez RJ (2005) Comparative study on volatile analysis of extra virgin olive oil by dynamic headspace and solid phase micro-extraction. Food Chem 90:69–79.

Kato Y, Matsuda T, Kato N, et al. (1986) Browning and insolubilization of ovalbumin by the Maillard reaction with some aldohexoses. J Agric Food Chem 34:351–355.

Kato Y, Matsuda T, Kato N, Nakamura R (1988) Browning and protein polymerization induced by amino-carbonyl reaction of ovalbumin with glucose and lactose. J Agric Food Chem 806–809.

Kessler HG (2002) Food and Bio Process Engineering Dairy Technology, 5th ed. 130–215.

72

11. References

Kilara A, Panyam D (2003) Peptides from milk proteins and their properties. Crit Rev Food Sci Nutr 43:607–33.

Klein MS, Almstetter MF, Schlamberger G, et al. (2010) Nuclear magnetic resonance and mass spectrometry-based milk metabolomics in dairy cows during early and late lactation. J Dairy Sci 93:1539–50.

Kokkinidou S, Peterson DG (2014) Control of maillard-type off-flavor development in ultrahigh-temperature-processed bovine milk by phenolic chemistry. J Agric Food Chem 62:8023–33.

Kontopidis G, Holt C, Sawyer L (2004) Invited review: beta-lactoglobulin: binding properties, structure, and function. J Dairy Sci 87:785–96.

Larsen LB, Rasmussen MD, Bjerring M, Nielsen JH (2004) Proteases and protein degradation in milk from cows infected with Streptococcus uberis. Int Dairy J 14:899–907.

Lawless HT, Heymann H (2010) Physiological and physiological foundations of sensory function. Sens. Eval. food Princ. Pract. Springer, pp 19–56

Lawless HT, Hildegarde H (2010) Descriptive analysis. Sens. Eval. food Princ. Pract., 2nd ed. Springer, New York, pp 227–258

Le TX, Datta N, Deeth HC (2006) A sensitive HPLC method for measuring bacterial proteolysis and proteinase activity in UHT milk. Food Res Int 39:823–830.

Ledl F, Schleicher E (1990) New Aspects of the Maillard Reaction in Foods and in the Human Body. Angew Chemie Int 29:565–594.

Lemieux L, Simard RE (1992) Bitter flavour in dairy products . Il . A review of bitter peptides from caseins : their formation , isolation and identification , structure masking and inhibition 72:335-382.

Limacher A, Kerler J, Davidek T, et al. (2008) Formation of furan and methylfuran by maillard-type reactions in model systems and food. J Agric Food Chem 56:3639–47.

Liu G, Zhong Q (2013) Thermal aggregation properties of whey protein glycated with various saccharides. Food Hydrocoll 32:87–96.

Mahoney RR (2003) Beta -Galactosidase. Handb. food Enzymol.

Maillard LC (1916) Synthèse des matières humiques par action des acides aminés sur les sucres réducteurs. Ann di Chim Ser 5:258–317.

Mantei N, Vifla M, Enzler T, et al. (1988) Complete primary structure of human and rabbit lactase-phlorizin hydrolase : implications for biosynthesis,

73

11. References

membrane anchoring and evolution of the enzyme a 4-fold internal homology, suggesting that they evolved. EMBO J 7:2705–2713.

Marsili RT (1999) Comparison of solid-phase microextraction and dynamic headspace methods for the gas chromatographic-mass spectrometric analysis of light-induced lipid oxidation products in milk. J Chromatogr Sci 37:17–23.

Martins SIFS, van Boekel MAJS (2005) Kinetics of the glucose/glycine Maillard reaction pathways: influences of pH and reactant initial concentrations. Food Chem 92:437–448.

McKellar RC (1981) Development of off-flavors in ultra high temperature and pasteurized milk as a function of proteolysis.pdf. J Dairy Sci 64:2138–2145.

McKellar RC, Froehlich DA, Butler G, et al. (1984) The effect of uncooled storage on proteolysis bitterness and apparent viscosity in ultra high temperature milk.pdf. Can Inst Food Sci Technol J 17:14–17.

Meltretter J, Birlouez-Aragon I, Becker C-M, Pischetsrieder M (2009) Assessment of heat treatment of dairy products by MALDI-TOF-MS. Mol Nutr Food Res 53:1487–95.

Meltretter J, Schmidt A, Humeny A, et al. (2008) Analysis of the peptide profile of milk and its changes during thermal treatment and storage. J Agric Food Chem 56:2899–906.

Mendoza MR, Olano A, Villamiel M (2005) Chemical indicators of heat treatment in fortified and special milks. J Agric Food Chem 53:2995–9.

Messia MC, Candigliota T, Marconi E (2007) Assessment of quality and technological characterization of lactose-hydrolyzed milk. Food Chem 104:910–917.

Min DB, Boff JM (2002) Chemistry and Reaction of Singlet Oxygen in Foods. Compr Rev Food Sci Food Saf 1:58–72.

Mittal SB, Newell G, Hourigan JA, Zadow JG (1991) The effect of protesase contamination in lactase on the flavour of lactose-hydrolyzed milks. Aust J dairy Technol 46:46–47.

Monakhova YB, Kuballa T, Leitz J, et al. (2012) NMR spectroscopy as a screening tool to validate nutrition labeling of milk, lactose-free milk, and milk substitutes based on soy and grains. Dairy Sci Technol 92:109–120.

Morales FJ, Jiménez-pérez S (2001) Free radical scavenging capacity of Maillard reaction products as related to colour and fluorescence. Food Chem 72:119–125.

74

11. References

Naranjo GB, Pereyra Gonzales AS, Leiva GE, Malec LS (2013) The kinetics of Maillard reaction in lactose-hydrolysed milk powder and related systems containing carbohydrate mixtures. Food Chem 141:3790–5.

Ney KH (1979) Bitterness of peptides: amino acid composition and chain length. ACS Symp Ser 115:149–173.

Nielsen SS (2002) Plasmin System and Microbial Proteases in Milk: Characteristics, Roles, and Relationship. J Agric Food Chem 50:6628–6634.

Nightingale LM, Cadwallader KR, Engeseth NJ (2012) Changes in dark chocolate volatiles during storage. J Agric Food Chem 60:4500–7.

Oliver CM, Melton LD, Stanley RA (2006) Glycation of caseinate by fructose and fructooligosaccharides during controlled heat treatment in the dry state. J Sci Food Agric 86:722–731.

Oliver RL (1993) Cognitive , of the Attribute Bases Satisfaction Response Cognition in Satisfaction. J Consum Res 20:418–430.

Owczarek-Fendor A, De Meulenaer B, Scholl G, et al. (2012) Furan formation in starch-based model systems containing carbohydrates in combination with proteins, ascorbic acid and lipids. Food Chem 133:816–821.

Perez Locas C, Yaylayan VA (2004) Origin and mechanistic pathways of formation of the parent furan--a food toxicant. J Agric Food Chem 52:6830–6.

Perkins M, Elliott A (2005) Stale flavour volatiles in Australian commercial UHT milk during storage. Aust J dairy Sci Technol 60:231–237.

Peterson DG, Colahan-Sederstrom PM, Potineni R V (2007) Effect of processing technology and phenolic chemistry on ultra-high temperature bovne milk flavor quality. In: Cadwallader KR, Drake M, McGorrin RJ (eds) Flavor dairy Prod. American chemical society, Washington, pp 253–259

Povolo M, Contarini G (2003) Comparison of solid-phase microextraction and purge-and-trap methods for the analysis of the volatile fraction of butter. J Chromatogr A 985:117–125.

Pripis-Nicolau L, de Revel G, Bertrand A, Maujean a (2000) Formation of flavor components by the reaction of amino acid and carbonyl compounds in mild conditions. J Agric Food Chem 48:3761–6.

Ranciaro A, Campbell MC, Hirbo JB, et al. (2014) Genetic Origins of Lactase Persistence and the Spread of Pastoralism in Africa. Am J Hum Genet 1–15.

Rauh VM, Johansen LB, Ipsen R, et al. (2014) Plasmin Activity in UHT Milk: Relationship between Proteolysis, Age Gelation, and Bitterness. J Agric Food Chem 62:6852–60.

75

11. References

Reboredo-Rodríguez P, González-Barreiro C, Cancho-Grande B, Simal-Gándara J (2012) Dynamic headspace / GC-MS to control the aroma fi ngerprint of extra-virgin olive oil from the same and different olive varieties. Food Control 25:684–695.

Rerkrai S, Jeon IJ, Bassette R (1987) Effect of Various Direct Ultra-High Temperature Heat Treatments on Flavor of Commercially Prepared Milks. J Dairy Sci 70:2046–2054.

Rozin P (1982) “Taste-smell confusions” and the duality of the olfactory sense. Percept Psychophys 31:397–401.

Ruiz-Matute AI, Corzo-Martínez M, Montilla A, et al. (2012) Presence of mono-, di- and galactooligosaccharides in commercial lactose-free UHT dairy products. J Food Compos Anal 28:164–169.

Rysstad G, Kolstad J (2006) Extended shelf life milk-advances in technology. Int J Dairy Technol 59:85–96.

Sahi T (1994a) Genetics and epidemiology of adult-type hypolactasia. Scand J Gastroenterol Suppl 202:7–20.

Sahi T (1994b) Hypolactasia and Lactase Persistence. Scand J Gastroenterol 29:1–6.

Señorans FJ, Tabera J, Herraiz M, Reglero G (1996) A Method for the Direct Isolation and Gas Chromatographic Analysis of Milk Flavor Components Using a Programmed Temperature Vaporizer. J Dairy Sci 79:1706–1712.

Serrano MA, Castillo G, Muñoz MM, Hernández A (2002) Influence of Hydrolysis, Purification, and Calibration Method on Furosine Determination Using Liquid Chromatography. J Chromatogr Sci 40:87–91.

Shahidi F, Zhong Y (2010) Lipid oxidation and improving the oxidative stability. Chem Soc Rev 39:4067–79.

Smith MA, Taneda S, Richey PL, et al. (1994) Advanced Maillard reaction end products are associtated with Alzheimer disease pathology. Proc Natl Acad Sci U S A 91:5710–5714.

Smuda M, Glomb M a (2013) Fragmentation pathways during Maillard-induced carbohydrate degradation. J Agric Food Chem 61:10198–208.

Sousa M., Ardö Y, McSweeney PL. (2001) Advances in the study of proteolysis during cheese ripening. Int Dairy J 11:327–345.

Steen H, Mann M (2004) The ABC’s (and XYZ's) of peptide sequencing. Nat Rev Mol Cell Biol 5:699–711.

76

11. References

Stone H, Sidel J, Oliver S, et al. (1974) Sensory evaluation by quantitative descriptive analysis. Food Technol 28:24–34.

Sundekilde UK, Frederiksen PD, Clausen MR, et al. (2011) Relationship between the metabolite profile and technological properties of bovine milk from two dairy breeds elucidated by NMR-based metabolomics. J Agric Food Chem 59:7360–7367.

Sundekilde UK, Larsen LB, Bertram HC (2013a) NMR-Based Milk Metabolomics. Metabolites 3:204–22.

Sundekilde UK, Poulsen NA, Larsen LB, Bertram HC (2013b) Nuclear magnetic resonance metabonomics reveals strong association between milk metabolites and somatic cell count in bovine milk. J Dairy Sci 96:290–9.

Sverrisdóttir OÓ, Timpson A, Toombs J, et al. (2014) Direct Estimates of Natural Selection in Iberia Indicate Calcium Absorption Was Not the Only Driver of Lactase Persistence in Europe. Mol Biol Evol. 31:975-983

Tetra Pak (2003) Dairy processing handbook, 2nd ed. 227–246.

Thalbitzer F (2012) Salg af laktosefri mælk femdoblet på et år. Landbrugsavisen 19.

Thomas E, Burton H, Ford JE, Perkin AG (1975) The effect of oxygen content on flavour and chemical changes during aseptic storage of whole milk after ultra-high-temperature processing. J Dairy Res 42:285–295.

Thomas EL (1981) Trends in Milk Flavors. J Dairy Sci 64:1023–1027.

Thorpe SR, Baynes JW (1996) Role of the Maillard reaction in diabetes mellitus and diseases of aging. Drugs and Aging 9:69–77.

Tishkoff S A, Reed F A, Ranciaro A, et al. (2007) Convergent adaptation of human lactase persistence in Africa and Europe. Nat Genet 39:31–40.

Tossavainen O (2008) Dissertation: Heat-induced changes in lactose hydrolysed milks. Helsinki University of Technology

Tossavainen O, Kallioinen H (2008) Effect of lactose hydrolysis on furosine and available lysine in UHT skim milk. Milchwissenschaft 63:22–26.

Tossavainen O, Kallioinen H (2007) Proteolytic changes in lactose hydrolysed UHT milks during storage. Milchwissenschaft-Milk Sci Int 62:410–415.

Tüchsen H (2014) Arla vil booste salg af laktosefri produkter. Fødevare Watch http://fødevarewatch.dk//Fødevarer/article6694259, 05072014.

77

11. References

Udenfriend S, Stein S, Böhlen P, et al. (1972) Fluorescamine: a reagent for assay of amino acids, peptides, proteins, and primary amines in the picomole range. Science 178:871–2.

Valero E, Villamiel M, Miralles B, et al. (2001) Changes in flavour and volatile components during storage of whole and skimmed UHT milk. Food Chem 72:51–58.

Vallejo-cordoba B, Nakai S (1993) Using a Simultaneous Factor Optimization Approach for the Detection of Volatiles in Milk by Dynamic Headspace Gas Chromatographic Analysis. 2378–2384.

Van Boekel MAJS (2006) Formation of flavour compounds in the Maillard reaction. Biotechnol Adv 24:230–3.

Van Boekel MAJS (1998) Effect of heating on Maillard reactions in milk. Food Chem 62:403–414.

Van Renterghem R, De Block J (1996) Furosine in consumption milk and milk powders. Int Dairy J 6:371–382.

Vazquez-Landaverde PA, Torres JA, Qian MC (2006) Quantification of trace volatile sulfur compounds in milk by solid-phase microextraction and gas chromatography-pulsed flame photometric detection. J Dairy Sci 89:2919–27.

Vazquez-Landaverde PA, Velazquez G, Torres JA, Qian MC (2005) Quantitative determination of thermally derived off-flavor compounds in milk using solid-phase microextraction and gas chromatography. J Dairy Sci 88:3764–72.

Walstra P, Geurts TJ, Noomen A, et al. (1999) Dairy technology : principles of milk properties and processing. Marcel Dekker, Inc.

Wang J, Zhang Q-H, Wang Z-H, Li H-M (2009) Determination of Major Bovine Milk Proteins by Reversed Phase High Performance Liquid Chromatography. Chinese J Anal Chem 37:1667–1670.

Wang Y, Ho C-T (2012) Flavour chemistry of methylglyoxal and glyoxal. Chem Soc Rev 41:4140–9.

Wedholm A, Møller HS, Lindmark-Månsson H, et al. (2008) Identification of peptides in milk as a result of proteolysis at different levels of somatic cell counts using LC MALDI MS/MS detection. J Dairy Res 75:76–83.

Weenen H (1998) Reactive intermediates and carbohydrate fragmentation in Maillard chemistry. Food Chem 62:393–401.

78

11. References

Werkhoff P, Bretschneider W (1987) Dynamic headspace gas chromatography: concentration of volatile components after thermal desorption by intermediate cryofocusing in a cold trap. J Chromatogr 405:87–98.

Whitfield F, Mottram D (1992) Volatiles from interactions of Maillard reactions and lipids. Crit Rev Food Sci Nutr 31:1–58.

Yaylayan V a, Huyghues-Despointes A (1994) Chemistry of Amadori rearrangement products: Analysis, synthesis, kinetics, reactions, and spectroscopic properties. Crit Rev Food Sci Nutr 34:321–69.

Yaylayan VA., Ismail AA., Mandeville S (1993) Quantitative determination of the effect of pH and temperature on the keto form of d-fructose by FT IR spectroscopy. Carbohydr Res 248:355–360.

Yaylayan VA (1997) Classifiaction of the maillard reaction: a conceptual approach. Trends Food Sci Technol 8:13–18.

Yilmaz Y, Toledo R (2005) Antioxidant activity of water-soluble Maillard reaction products. Food Chem 93:273–278.

Zabbia A, Buys EM, De Kock HL (2012) Undesirable sulphur and carbonyl flavor compounds in UHT milk: a review. Crit Rev Food Sci Nutr 52:21–30.

Zamora R, Hidalgo FJ (2005) Coordinate contribution of lipid oxidation and Maillard reaction to the nonenzymatic food browning. Crit Rev Food Sci Nutr 45:49–59.

Zhang Y, Guo S, Liu Z, Chang SKC (2012) Off-flavor related volatiles in soymilk as affected by soybean variety, grinding, and heat-processing methods. J Agric Food Chem 60:7457–62.

Zhu Y, Zajicek J, Serianni AS (2001) Acyclic forms of [1-(13)C]aldohexoses in aqueous solution: quantitation by (13)C NMR and deuterium isotope effects on tautomeric equilibria. J Org Chem 66:6244–51.

Zúñiga RN, Tolkach A, Kulozik U, Aguilera JM (2010) Kinetics of formation and physicochemical characterization of thermally-induced beta-lactoglobulin aggregates. J Food Sci 75:E261–8.

79

Paper 1

Paper 1

Volatile component profiles of conventional and lactose-

hydrolyzed UHT milk- a dynamic headspace gas

chromatography-mass spectrometry study

Jansson, T., Jensen, S., Eggers, N., Clausen, M.R., Larsen, L.B., Colin, R.,

Sundgren, A., Andersen, H.J., Bertram, H.C.

Dairy Science & Technology (2014) 94:311-325

ORIGINAL PAPER

Volatile component profiles of conventionaland lactose-hydrolyzed UHTmilk—a dynamic headspacegas chromatography-mass spectrometry study

Therese Jansson & Sidsel Jensen & Nina Eggers &Morten R. Clausen & Lotte B. Larsen & Colin Ray &

Anja Sundgren & Henrik J. Andersen &

Hanne Christine Bertram

Received: 16 January 2014 /Revised: 11 February 2014 /Accepted: 11 February 2014 /Published online: 14 March 2014# INRA and Springer-Verlag France 2014

Abstract Lactose-hydrolyzed milk gains still increasing market share, and understand-ing the chemical characteristics of lactose-hydrolyzed milk products is important for thedairy industry. The aim of the present study was to identify and compare volatilecompounds of commercial lactose-hydrolyzed and conventional ultra-high temperature(UHT) milk. For this purpose, the volatile compounds of lactose-hydrolyzed (<1%lactose), conventional (100% lactose), and filtered (60% lactose) UHT-treated milkwere extracted using dynamic headspace sampling and analyzed by gaschromatography-mass spectrometry (GC-MS). A total of 24 volatile compounds wereidentified including ketones, aldehydes, and sulfides. Overall, principal componentanalysis (PCA) showed grouping of the different milk types, with loadings indicatinga higher concentration of ketones in conventional versus lactose-hydrolyzed UHT milk,but PCA also indicated a marked batch-to-batch variation. Elucidation of individualvolatile compounds detected also revealed that the content of ketones in general washigher in conventional UHT milk than in lactose-hydrolyzed milk; however, nosignificant differences in the volatile compound profiles could be identified betweenthe various milk types as a result of the batch-to-batch variation. The present study

Dairy Sci. & Technol. (2014) 94:311–325DOI 10.1007/s13594-014-0164-7

T. Jansson : S. Jensen :N. Eggers :M. R. Clausen : H. C. Bertram (*)Department of Food Science, Research Center Aarslev, Aarhus University, Kirstinebjergvej 10,5792 Aarslev, Aarhus, Denmarke-mail: [email protected]

L. B. LarsenDepartment of Food Science, Research Centre Foulum, Aarhus University, Blichers Allé 20,8830 Tjele, Aarhus, Denmark

C. Ray :A. SundgrenArla Foods Strategic Innovation Centre, Lindhagensgatan 126, 10546 Stockholm, Sweden

H. J. AndersenRoerdrumvej 2, 8220 Brabrand, Aarhus, Denmark

highlights a useful analytical method based on dynamic headspace sampling and GC-MS to profile volatiles important for the flavor characteristics of lactose-hydrolyzed andconventional UHT milk. In addition, the present study reveals that a considerablebatch-to-batch variation exists in industrially produced batches of lactose-hydrolyzedUHT milk, which must be considered an important challenge for the dairy industry.

Keywords Lactose-hydrolyzed milk . UHT . Milk batch variations . Dynamicheadspace sampling . GC-MS

AbbreviationsDHS Dynamic headspace samplingDMDS Dimethyl disulfideDMS Dimethyl sulfideDMSO Dimethyl sulfoxideDMTS Dimethyl trisulfideFWER Family-wise error rateGC Gas chromatographIS Internal standardLME Linear mixed effectLOD Limit of detectionLOQ Limit of quantificationMR Maillard reactionMS Mass spectrometrynd Not detectednq Not quantifiedPC Principal componentPCA Principal component analysisREML Restricted maximum likelihoodTIC Total ion currentUHT Ultra-high temperatureUV Unit variance

1 Introduction

More than 70% of the world’s population is unable to digest lactose as a consequenceof a reduced or nonexistent β-galactosidase activity (Messia et al. 2007; Sahi 1994).This is referred to as “lactose malabsorption” or “lactose intolerance” depending on thedegree of reduction in β-galactosidase activity. These conditions give rise to bacterialfermentation of lactose in the colon, resulting in the production of copious amounts ofgas (a mixture of hydrogen, carbon dioxide, and methane) that causes abdominalsymptoms such as bloating, cramps, and nausea (Zhong et al. 2004). During the pastdecade, several lactose-hydrolyzed milk products have been introduced to the com-mercial market to meet the needs of consumers with lactose intolerance.

High-quality lactose-hydrolyzed milk may be produced using filtration to removeapproximately 40% of the lactose. The remaining lactose is enzymatically hydrolyzedby the addition of β-galactosidase, producing the monosaccharides glucose and

312 T. Jansson et al.

galactose. The decomposition of lactose to glucose and galactose in milk may influencethe type and extent of chemical reactions in the milk taking place during processing andstorage. For instance, Kato et al. (1986, 1988) reported that browning reactions andprotein polymerization, both of which are characteristic for the advanced stages of theMaillard reaction (Brands et al. 2002), proceed more extensively in protein-galactoseand protein-glucose models than in an equivalent protein-lactose model during heating(Brands et al. 2000). Consequently, lactose-hydrolyzed milk is expected to be moresusceptible to Maillard reaction and Strecker degradation during storage thannonhydrolyzed ultra-high temperature (UHT) milk (Messia et al. 2007).

Various forms of thermal processing are used to prevent enzymatic and microbialspoilage and thereby extend the shelf life of milk. One commonly used heat treatmentin the dairy industry is UHT processing (135–150 °C, 2–20 s; Kessler 2002). UHT-treated milk has a long shelf life and may be stored at room temperature for up to8 months. However, it is well established that severe heat treatment and storage atambient temperature lead to the development of off-flavors in milk (Jeon et al. 1978;Rerkrai et al. 1987; Valero et al. 2001).

Extensive investigations have been carried out to characterize the chemical changesgiving rise to unwanted off-flavors in heat-treated milk (Rerkrai et al. 1987; Valero et al.2001; Coppa et al. 2011; Vazquez-Landaverde et al. 2005; Toso et al. 2002; Contariniand Povolo 2002; Contarini et al. 1997). The severity of the heat treatment stronglyaffects the resulting volatile profile of milk. In particular, UHT milk is characterized byan increase in the amount of ketones (Vazquez-Landaverde et al. 2005), aldehydes(Vazquez-Landaverde et al. 2005), and sulfides (Vazquez-Landaverde et al. 2006).

In general, the amount of volatile compounds in milk is low and a concentration step isrequired to perform effective analyses. Dynamic headspace sampling (DHS) in combinationwith chromatographic techniques is a useful method for concentrating and analyzingvolatileswith high reproducibility and sensitivity. Optimally, the volatiles should be collectedunder mild conditions to minimize sample manipulation and the potential for artifactformation (Dirinck et al. 1984). DHS in combination with gas chromatography-massspectrometry (GC-MS) has been frequently used for characterizing the volatiles in milk(Valero et al. 2001; Toso et al. 2002; Vallejo-Cordoba et al. 1993) and is also used in thecharacterization ofmany other foodmatrices (Jensen et al. 2011; Bach et al. 2012). However,to our knowledge, no characterization of the volatile profile of heat-treated lactose-hydrolyzed milk obtained using the combination of DHS and GC-MS has been reported.

The aim of the present study was to identify and compare the volatile fractions ofdifferent types of commercially produced enzymatically hydrolyzed UHT milk (directversus indirect) and conventional UHT milk. To ensure a representative pool ofsamples, several batches of milk were produced in a full-scale commercial dairy plant,which made it possible also to elucidate batch variations in industrially produced milks.

2 Materials and methods

2.1 Chemical references

Decane (99%), tetradecane (99%), dodecane (99%), 2-methylbutanal (95%), hexanal(98%), octanal (99%), nonanal (95%), decanal (98%), 2-butanone (99%), 2-pentanone

Profiles of conventional and lactose-hydrolyzed UHT milk 313

(99%), 2-hexanone (99.5%), 2-heptanone (98%), 2-octanone (98%), 2-nonanone(99%), 6-methyl-5-heptene-2-one (99%), 6,10-dimethyl-5,4-undecadien-2-one (97%),dimethyl disulfide (98%), 4-methyl-1-pentanol (97%), toluene (99.5%), styrene (99%),and 2-pentyl furan (97%) were obtained from Sigma-Aldrich Inc. (Steinheim,Germany). Dimethyl sulfoxide (99%) was purchased from Serva electrophoresisGmbH (Heidelberg, Germany).

2.2 Milk samples

Milk was produced at a Danish commercial dairy plant (Arla Foods, Esbjerg,Denmark). A total of four types of milk (CONVI, LHD, LHI, and FILTI) wereproduced (Table 1) and stored at 22 °C until the day of analysis. CONVI, LHD, andLHI were produced in three batches in September and October 2012 and February2013. For technical reasons, FILTI was included to separate the impact of filtration andenzyme treatment on the formation of volatiles. FILTI was only produced in twobatches in September 2012. The milk was UHT-treated using either direct heat treat-ment (LHD) (steam is injected into the milk to heat it to 140–145 °C for a few seconds,followed by flash cooling; Tetra Pak 2003) or indirect heat treatment (CONVI, LHI,FILTI) (a tubular heat exchanger is used to heat the milk to 140–145 °C for a fewseconds, followed by slow cooling (Tetra Pak 2003)) and packaged aseptically. Lactosein the lactose-hydrolyzed and the filtered milk (LHD, LHI, FILTI) was removed usingultrafiltration, to approximately 60% of the original lactose level. The remaining lactosewas enzymatically hydrolyzed (LHD, LHI) by the addition of β-galactosidase (Jelenand Tossavainen 2003). A total of three replicates of each milk type and each batchwere analyzed 17–21 days after production.

2.3 Dynamic headspace sampling

A 100-mL milk sample was placed in a conical flask with a glass cap insert and 200 μLof a 10 ppm 4-methyl-1-pentanol solution was added to the milk as an internal standard(IS). Adsorbent traps (Tenax TA, 200 g Tenax TA, 35/60 mesh; Markes InternationalLimited, LIantrisant, UK) were attached to the flask, which was subsequently placed in

Table 1 Overview of milk types included in the study, heat treatments, and lactose levels. All milks contained1.5% fat (w/w) and 3.5% protein (w/w)

Milk No. ofbatches

Enzymatichydrolysis

Heattreatment

Filtration Level of lactose(% of originallevel)

Amount oflactose(% w/w)

CONVI Conventional UHT 3 − Indirect − 100 5

LHD Lactose-hydrolyzeddirect UHT

3 + Direct + <1 0.01

LHI Lactose-hydrolyzedindirect UHT

3 + Indirect + <1 0.01

FILTI Filtered UHT 2 − Indirect + 60 3


a thermostatically controlled cabinet at 20 °C. The milk was sparged with nitrogen(40 mL.min) for 45 min while being stirred.

2.4 GC-MS

The adsorbent traps were desorbed (250 °C, 15 min) into a cold trap (U-TIIGCP,Markes International Limited, UK) using a thermal desorption system (ultra-UNITY™,Markes International Limited, LIantrisant, UK). Helium was used as carrier gas at aconstant pressure of 6.985×103 Pa. The desorbed volatiles were focused at 1 °C for5 min and then injected onto a gas chromatography column through a transfer line(140 °C) by raising the temperature of the trap to 300 °C under splitless conditions. Thegas chromatograph (Finnigan trace GC ultra, Thermo, Waltham, MA) was equippedwith a CP-Wax 52CB (Varian, Palo Alto, CA, USA) column (50 m×0.25 mm, filmthickness 0.25 μm). The gradient program included an isothermal step at 32 °C for1 min, a ramp to 35 °C at 0.2 °C.min, a ramp to 40 °C at 0.5 °C.min, a ramp to 60 °C at2.5 °C.min, and a ramp to 220 °C at 10 °C.min, followed by a 15-min isothermal step.

The GC was connected to a single quadrupole mass spectrometer (Finnigan Tracedual-stage quadrupole (DSQ), Thermo, Waltham, MA, USA). The MS transfer linetemperature was 250 °C and the ion source temperature was 200 °C. The massspectrometer was scanned over a mass to charge (m/z) range of 45–650 (4.46 scans/s)and the spectra were obtained using a fragmentation voltage of 70 eV.

2.5 Method optimization

To optimize the sampling parameters, the effect of flow, sample volume, and headspacesampling time were initially studied. Final sampling parameter values were set to40 mL.min, 100 mL, and 45 min, respectively. The sampling temperature was fixedat 20 °C. In addition, several trap and cold trap materials were tested, and the injectionmode and GC temperature gradient were adapted to obtain optimal separation of thevolatile compounds. The final trap material used was Tenax TA and the coldtrapmaterial was U-T11GPC. The optimum parameter values were selected on the basisof peak shape, separation, intensity, and sensitivity.

2.6 Identification and quantitative analysis

A MS database (NIST MS search version 2.0, 2011) was used to identify 24 volatilecompounds. In addition, all of the compounds except 3,3-dimethyl hexane, 2-undecanone, and dimethyl trisulfide were verified by comparison of mass spectral dataand retention times with authentic reference compounds.

The identified volatile compounds were quantified using external standard curvesprepared in duplicates. The four external calibrants, which were eluted in differentregions of the chromatogram, were 2-pentanone (eluting at 10.7 min), hexanal(19.2 min), 2-heptanone (31.9 min), and decanal (eluting at 44.5 min). Analytes elutingin the same retention time window were assumed to have comparable responsecoefficient and equal linear range as the calibrant (Jensen et al. 2011). However, theketone responses reminded more of the nearest ketone external calibrant and werequantified (μg.kg milk) according to the appropriate calibrant regression equation


(Table 2). The regression equation, limits of detection (LOD), and limits of quantifi-cation (LOQ) were calculated from the standard error of the y-intercept and the slope ofthe linear regression, for each standard curve (Table 2) (Dolan 2009a, b).

2.7 Multivariate data and statistical analysis

Multivariate data analysis in the form of principal component analysis (PCA) wasapplied including all quantified compounds to investigate grouping of the samplesusing the Simca software (version 13.0, Umetrics AB, Umeå, Sweden). Data werenormalized against the area of the added IS and unit-variance (UV) scaled before PCAanalysis.

To identify significant differences between the milk samples, univariate statisticalanalysis including linear mixed model and Tukey’s honest significant difference (α=0.05) test was performed using the software R (version 2.15.2, R Foundation forStatistical Computing, Vienna, Austria). The variance-covariance was estimated usingrestricted maximum likelihood (REML) to avoid underestimations of random varia-tions in the data. The statistical model (linear mixed effect, LME) included fixed factorsfor the four milk types (CONVI, LHD, LHI, FILTI) and the 11 batches and took intoaccount the correlation between the three replicates from the same batch. The variancewithin milk types was estimated using a linear model and Tukey’s honest significantdifference (α=0.05), and Bonferroni’s method was used to maintain the family-wiseerror rate (FWER). Compounds below the LOD and LOQ were not included in thestatistical analysis.

3 Results

GC-MS analyses were conducted on the 11 different industrially produced milk batchesincluded in the study. Representative total ion current (TIC) chromatograms obtainedfrom conventional, lactose-hydrolyzed direct and indirect treated UHT milk, andfiltered UHT milk are shown in Fig. 1. A total of 24 volatile compounds includingaldehydes, ketones, hydrocarbons, and sulfides were identified and quantified in themilk samples (Table 3). A comparison of the chromatograms indicates that the con-ventional UHT milk contains a greater amount of the ketones 2-butanone (7.61 min), 2-pentanone (10.6 min), 2-heptanone (31.91 min), 2-nonanone (42.04 min), and 2-

Table 2 Regression equations, limit of detection (LOD), limit of quantification (LOQ), and coefficient ofdetermination (R2) for calibration curves in filtered UHT milk containing 1.5% fat (w/w)

Compound Calibrationconcentrations(μg.kg)

Limit of detection(LOD)(μg.kg)

Limit of quantification(LOQ)(μg.kg)

Regressionequation

R2

2-Pentanone 4.0–79.4 3.8 11.6 y=0.198x 0.997

Hexanal 4.0–31.8 2.0 6.2 y=0.2295x+0.1999 0.994

2-Heptanone 6.4–64.3 9.4 28.4 y=0.2762x 0.985

Decanal 16.3–244.1 13.9 42.1 y=0.0108x+0.0711 0.994


undecanone (46.2 min) compared to lactose-hydrolyzed milks (Fig. 1a–e). In addition,a batch-to-batch variation was observed for the ketones when two batches of directtreated lactose-hydrolyzed (Fig. 1b, c) and indirect treated UHT milk (Fig. 1d, e) werecompared.

PCAwas carried out on the 24 volatile compounds identified, and Fig. 2 shows thePCA score scatter plot (a) and the corresponding loading plot (b) of the volatilecompounds identified in the milk types. In the score plot, principal components (PC)1 and 2 describe 38% and 23.6% of the variation, respectively. The score plot alsoindicates a separation of the lactose-hydrolyzed samples along PC1 and filteredsamples along PC2. However, the separation of the lactose-hydrolyzed milk away fromthe lactose-containing milk is batch dependent, and one batch each of LHD and LHIwas more similar to the CONVI milk. Our data therefore exhibit a significant batchvariation which tends to be more severe for lactose-hydrolyzed milk than for conven-tional milk. The loading scatter plot (Fig. 2b) reveals that the separation of the milktypes along PC1 may be ascribed to differences in the concentration of ketones and thealdehydes octanal, nonanal, and decanal, while PC2 reveals differences in the concen-tration of the sulfides, dimethyl disulfide (DMDS), dimethyl trisulfide (DMTS), anddimethyl sulfoxide (DMSO).

It was observed that CONVI tended to contain a greater amount of ketones thanLHD or LHI milks, which in contrast were characterized by a higher content of thealdehydes nonanal and decanal, if the two outlier milk batches LHD (batch 1) and LHI(batch 3) were excluded. In general, no clear difference was seen between direct and

Fig. 1 TIC-GC chromatograms of a conventional indirect treated UHT milk; b lactose-hydrolyzed directtreated UHT milk; c lactose-hydrolyzed direct treated UHT milk; d lactose-hydrolyzed indirect treated UHTmilk; e lactose-hydrolyzed indirect treated UHT milk; f filtered indirect treated UHT milk


Table3

Average

concentrations

[μg.kg]of

volatilecompounds

inconventio

nal,lactose-hydrolyzed,and

filteredUHTmilk

Rt[m

in]

Volatile

compound

Conventionalindirect

UHT(CONVI)

←Batch

no.→

Lactose-hydrolyzeddirect

UHT(LHD)

←Batch

no.→

Lactose-hydrolyzedindirect

UHT(LHI)

←Batch

no.→

FilteredUHT

(FILTI)

←Batch

no.→

12

31

23

12

31

2

Alkanes

5.06

3,3-Dim

ethylhexane

15.59a

8.86

#20.89b

19.01a

14.38b

21.47c

38.52a

30.16b

23.58c

31.70

2.51*

12.23

Decane

0.51*

0.55*

3.50*

1.27*

1.21*

0.72*

0.96*

2.19*

1.65*

0.27*

0.77*

34.50

Dodecane

0.01*

nd0.88*

<0.01*

0.12*

0.71*

0.52*

0.73*

1.60*

0.98*

2.45*

42.43

Tetradecane

ndnd

ndnd

ndnd

ndnd

nd10.51#

42.73

Totalcontentof

alkanes

16.11

9.41

25.27

20.29

15.71

22.90

40.00

33.08

26.83

43.46

48.46

Aldehydes

8.85

2-Methylbutanal

0.06*

0.27*

0.48*

0.05*

ndnd

ndnd

0.18*

ndnd

19.23

Hexanal

0.02*

nd2.59

#nd

ndnd

ndnd

1.67*

0.45*

1.13*

39.01

Octanal

ndnd

ndnd

ndnd

ndnd

nd0.97*

4.03*

42.17

Nonanal

1.04*

1.35*

nd0.73*

5.71*

nd5.29*

9.83*

nd13.50#

6.49*

44.49

Decanal

ndnd

ndnd

4.20*

nd0.4*

3.30*

ndnd

3.75*

Totalcontentof

aldehydes

1.12

1.62

3.07

0.78

9.91

nd5.69

13.13

1.85

14.92

15.40

Ketones

7.61

2-Butanone

91.43a

73.90b

105.10

c49.13

7.06

#9.62

#6.35

#6.20

#99.11

57.60a

50.50b

10.73

2-Pentanone

21.67a

34.51b

24.25c

15.47

5.36

#5.68

#4.61

#4.58

#20.82

17.79a

17.39a

18.95

2-Hexanone

<0.01*

0.13*

0.73*

ndnd

ndnd

nd0.25*

nd0.01*

31.91

2-Heptanone

56.03a

93.98b

76.09c

49.20a

15.60#

23.47b

14.73#

14.05#

90.64

45.96a

45.30a

38.93

2-Octanone

0.17*

0.38*

0.47*

0.23*

0.04*

0.01*

0.04*

0.05*

0.19*

0.26*

0.39*

40.65

6-Methyl-5-hepten-2-one

0.34*

ndnd

nd0.33*

nd0.26*

0.44*

nd0.23*

0.74*


Table3

(contin

ued)

Rt[m

in]

Volatile

compound

Conventionalindirect

UHT(CONVI)

←Batch

no.→

Lactose-hydrolyzeddirect

UHT(LHD)

←Batch

no.→

Lactose-hydrolyzedindirect

UHT(LHI)

←Batch

no.→

FilteredUHT

(FILTI)

←Batch

no.→

12

31

23

12

31

2

42.04

2-Nonanone

10.68a

17.12b

11.68c

9.55

2.86*

3.57*

3.02*

2.93*

14.35

11.09a

11.46a

46.19

2-Undecanone

0.92*

1.94*

1.10*

0.90*

0.11#

0.09*

0.19*

0.22*

1.33*

1.16*

1.32*

49.71

6,10-D

imethyl-5,9-

undecadien-2-one

(E)

0.09*

ndnd

ndnd

ndnd

nd0.10*

ndnd

Totalcontentof

ketones

181.34

221.96

219.42

124.48

31.36

42.44

29.2

28.47

226.79

134.09

127.11

Sulfides

17.82

Dim

ethyldisulfide

8.70

a5.48

#13.41b

4.92

#6.37

a7.99

b6.72

4.83

#5.70

#3.94

#4.11

#

41.67

Dim

ethyltrisulfide

73.06a

51.74b

136.74

c24.91#

73.17a

77.76a

56.42a

54.91a

33.70#

28.94#

32.17#

46.07

Dim

ethylsulfoxide

7.24*

20.44#

13.99#

5.30*

ndnd

ndnd

1.35*

12.22*

41.63#

Totalcontentof

sulfides

89.00

77.66

161.14

35.13

79.54

85.75

63.14

59.74

40.75

45.10

77.91

Others

14.78

Toluene

4.91

#4.71

#8.14

12.44

1.91*

1.74*

2.44

#2.55

#5.92

#12.30a

12.53a

36.33

2-Pentylfuran

0.32*

0.59*

0.79*

0.44*

0.14*

<0.01*

0.14*

0.21*

0.17*

0.53*

0.93*

37.34

Styrene

1.69*

1.25*

1.80*

1.40*

0.48*

0.07*

0.63*

0.59*

0.46*

0.99*

0.43*

Num

bers1,

2,and3representthebatchnumberforeach

milk

type

(three

replicates

foreach

batch).S

uperscript

letters(a–c)indicatesignificance

(P<0.05)with

inthemilk

types

ndnotdetected,a

steriskconcentration<LOD,n

umbersign

concentration<LOQ


indirect lactose-hydrolyzed UHT (milk types LHD and LHI). The score plot in Fig. 2apartly places lactose-reduced, filtered UHT milk (FILTI) between the CONVI milk typeand the LHD and LHImilk types, whichmay be ascribed to differences in ketones and thealdehydes nonanal and decanal (Fig. 2b). In contrast, FILTI is separated from the othermilk types due to increased concentrations of tetradecane, DMSO, and 2-pentyl furan.

The volatile compounds were quantified based on the four calibration curves(Table 2). The concentrations of the 24 volatile compounds in the milk types aresummarized in Table 3. However, some of the compounds were below the LOQ, andtherefore, the given concentrations are only tentatively quantified. The volatile com-pounds varied widely in concentration within batches of similar milk types, which wasvisually seen in the PCA score scatter plot in Fig. 2a. Generally, ketones represented thechemical class with the highest number of detected compounds. In total, nine ketoneswere identified, and among these, 2-butanone, 2-heptanone, and 2-nonanone werepresent in the highest concentrations. These three methyl ketones were also thecompounds that differed most in concentration between CONVI, LHD, and LHIaccording to the TIC in Fig. 1. In total, five aldehydes were identified, and of these,nonanal was the compound found in the highest concentration in all of the milk typesanalyzed. Of the three sulfides identified (DMDS, DMTS, DMSO), DMTS was found

Fig. 2 a PCA score plot with principal components 1 and 2, describing 38 % and 23.6 % of the total variationin the volatile compound profiles, respectively. Circle A: Conventional indirect UHT milk, square B: lactose-hydrolyzed direct UHT milk, triangle C: lactose-hydrolyzed indirect UHT milk, inverted triangle D: Filteredindirect UHT milk, numbers 1–3 indicate batch number. b Corresponding loading plot. R2=0.739 and Q2=0.492, with three components


to be present in the highest concentration in all milk types. Overall, no significantdifferences could be identified between the milk types (CONVI, LHD, LHI, or FILTI)for any of the detected volatile compounds. However, significant differences withinmilk types were observed due to high batch-to-batch variation in the industriallyproduced samples (Table 3, Fig. 2a).

4 Discussion

Reports on the volatile profile of lactose-hydrolyzed milk are relatively sparse (Jelen andTossavainen 2003; Kallioinen and Tossavainen 2008;Messia et al. 2007; Tossavainen andKallioinen 2007). DHS and thermal desorption combined with GC-MS is a useful andefficient method for collecting and analyzing volatiles (Naudé et al. 2009). In the presentstudy, this method combinedwithmultivariate data analyses was applied to obtain volatileprofiles of lactose-hydrolyzed, filtered, and conventional UHT-treated milk. To ourknowledge, no elucidation of the volatile profile of heat-treated lactose-hydrolyzed milkobtained using the combination of DHS and GC-MS has previously been reported.

A total of 24 volatile compounds were identified and quantified (Table 3). Thenumber of ketones and aldehydes identified is in line with the results obtained byVazquez-Landaverde et al. (2005), who identified eight ketones and nine aldehydeswhen investigating volatile compounds in UHT milk using headspace solid-phasemicroextraction GC. The 24 quantified volatile compounds (Table 3) ranged in con-centration from <0.01 to 105.10 μg.kg milk. The concentrations of the high-concentration compounds in the present study were somewhat higher, and the concen-trations of the low-concentration compounds were lower than previously reported(Valero et al. 2001; Vazquez-Landaverde et al. 2005; Toso et al. 2002; Contariniet al. 1997; Vazquez-Landaverde et al. 2006; Al-Attabi 2009). This divergence inresults is most probably due to differences in analytical and quantification proceduresbut also in the processing of the milk (e.g., heat treatment and packaging).

It is known that aldehydes and ketones contribute to the “stale” note found in UHTmilk (Rerkrai et al. 1987; Vazquez-Landaverde et al. 2005) and this becomes moreevident when the “cooked” flavor begins to decline (Thomas et al. 1975). PCAindicated that the secondary lipid peroxidation products octanal and nonanal wereimportant for the differentiation of LHI and FILTI milks from the other milks(Fig. 2). However, the levels of octanal and nonanal were below LOD (Table 3), andfurther studies are therefore needed to substantiate this finding. The presence of thealdehydes is an indicator of lipid oxidation in the milk, and Vazquez-Landaverde et al.(2005) reported that although the aldehydes were less affected by heat treatment incomparison with ketones, an increase in the total concentration of aldehydes increasedwith the severity of heat treatment.

Ketones were the most abundant volatile compounds in all of the milk types inaccordance with previous findings in conventional UHT milk (Vazquez-Landaverdeet al. 2005; Contarini and Povolo 2002; Contarini et al. 1997). In general, the presentstudy indicates that conventional UHT milk contained higher concentrations of ketonesthan lactose-hydrolyzed or filtered milk (Fig. 2a, Table 3).

The methyl ketones 2-butanone, 2-heptanone, and 2-nonanone were among theketones present in the highest concentrations in both conventional and lactose-


hydrolyzed milk. This is consistent with the results of Vazquez-Landaverde et al.(2005), who concluded that 2-heptanone and 2-nonanone had a major impact on theflavor of heat-treated milk. In heat-treated milk, ketones are mainly products of theheat-initiated decarboxylation of β-oxidized saturated fatty acids or decarboxylation ofβ-ketoacids (Vazquez-Landaverde et al. 2005; Belitz et al. 2004). In general, ketoneformation requires a low activation energy (Schwartz et al. 1966), and therefore, theproduction of these compounds progresses throughout the storage period followinginitiation of heat-induced oxidation (Kochhar 1996). As a result of a marked batch-to-batch variation in the milks, no significant differences in the content of ketones betweenthe different milk types were evident. Intriguingly, 2-methylbutanal, a characteristicMaillard reaction product, was detected in all three batches of conventional UHT milkbut only sporadically seen in the lactose-hydrolyzed milk. 2-Methylbutanal is aStrecker degradation product of isoleucine and is an important indicator for thisreaction having taken place in severely heat-treated milk (Belitz et al. 2004).Adamiec et al. (2001) suggested that Strecker degradation of amino acidsincludes radical reactions, and more radicals may have been produced in theconventional UHT milk than in the lactose-hydrolyzed UHT milk based on thehigher concentration of 2-methylbutanal in the former. The fact that Maillardreaction products formed in the final stage are well-known antioxidants (Yilmazand Toledo 2005) and that the Maillard reaction proceeds more extensively inglucose- and galactose-protein model systems (Kato et al. 1986, 1988) mayexplain a more limited heat-activated lipid oxidation (lower ketone concentra-tion). Furthermore, the presence of antioxidants may have affected the forma-tion of 2-methylbutanal in lactose-hydrolyzed UHT milk, considering that thiscompound may be formed through a radical process as outlined above.

Three sulfide compounds were identified in the present study. Sampling and analysisof sulfides is difficult because of their high volatility and reactivity, which also explainsthe large variation in the concentration of these compounds (Table 3). The concentra-tion of sulfides increases with the severity of heat treatment, primarily due to denatur-ation of β-lactoglobulin (Hutton and Patton 1952), which exposes the sulfur-containingamino acids cysteine and methionine to oxidation or Strecker degradation (Hutton andPatton 1952; Datta et al. 2002). DMDS and DMTS are products of Strecker degrada-tion of methionine, and DMSO is formed when the Strecker degradation productdimethyl sulfide (DMS) is oxidized (Al-Attabi et al. 2008). The presence of sulfidesin milk is an indication of Maillard reaction, and the compounds are characteristic forthe flavor formation of UHT milk (Al-Attabi 2009). Surprisingly, DMS, which previ-ously was found to be the most abundant sulfide in milk (Vazquez-Landaverde et al.2006), was not identified in the present study, probably due to the high volatility of thecompound, the mild collection conditions, and inadequate sensitivity of the method forsulfide compounds.

Four alkanes were identified in the milk (Table 3), and higher concentrations of3,3-dimethylhexane and tetradecane were observed in LHI and FILTI comparedwith UHT and LHD. The origin of alkanes in milk is far from well-understood(Coppa et al. 2011); however, the compounds may be formed from fatty acidoxidation and thermal degradation of amino acids (Lien and Nawar 1973; Lien andNawar 1974; Macku and Takayuki 1991; Whitfield 1992). There are no obviousexplanations why a difference in alkane concentration was observed between the


different milk types in the present study. However, these compounds were consid-ered to have a minor impact on the volatile profile of the different milk types.

Previous studies have reported that direct UHT treatment represents a less severeprocess than indirect UHT treatment (Badings et al. 1981) due to the shorter heatingand cooling times. Perkins et al. (2005) found that the amount of ketones wasapproximately two times higher in 1-week-old indirect UHT-treated milk than in directUHT-treated milk. Badings et al. (1981) found that the compounds primarily respon-sible for the differences between the two heat-treatment procedures were not only 2-heptanone and 2-nonanone, but also DMDS, 2-hexanone, 2-octanone, and 2-undecanone. In the present study, there were no clear differences in the volatilecompound profiles between the two lactose-hydrolyzed milk types LHD and LHI,suggesting that the direct and indirect heat treatment do not have the same effect onlactose-hydrolyzed UHT milk as previously observed for conventional UHT milk.However, the volatile profile of the milk types can be expected to develop differentlyduring storage (Perkins et al. 2005), an aspect which was not investigated in the presentstudy.

Lactose-hydrolyzed milk is an alternative to conventional milk, and therefore, thedairy industry strives to produce lactose-hydrolyzed milk with characteristics resem-bling conventional milk. However, the present study reveals that considerable productdifferences between milk batches exist, which may impair the consumers’ recognitionof the product. Lactose-hydrolyzed milk produced through partial physical removal oflactose and subsequent enzymatic hydrolysis of the remaining lactose is a rather newproduct, and a full understanding of its quality characteristics is important to ensureconsumer acceptance. Consequently, the grouping of conventional, filtered, andlactose-hydrolyzed UHT milk in Fig. 2 is highly relevant in relation to optimizingthe flavor profile of lactose-hydrolyzed milk.

Moreover, significant batch variations were present in all types of milk which isdeemed to be a challenge in the dairy industry. Consumers are relatively sensitive tovariations in flavor (Thomas 1981). The batch variation in two out of six batches of thelactose-hydrolyzed milk was due to differences in the ketone concentrations. This maybe ascribed to a number of factors, including seasonal variations in the rawmilk (Auldistet al. 1998; Ostersen et al. 1997) and differences in processing. The latter seems mostfeasible as industrial-scale thermal processing can subject certain batches of milk to amore severe heat load. This would explain the difference in ketone concentration, as ahigher ketone concentration may result from more severe heat treatment (Vazquez-Landaverde et al. 2005; Contarini and Povolo 2002; Contarini et al. 1997). Similarvariations in ketone concentration between batches and brands have been noted inprevious studies by Perkins et al. (2005) and Vazquez-Landaverde et al. (2005).

5 Conclusion

The present study focused on the analysis of short-term stored lactose-hydrolyzed andconventional UHT milk, being analyzed 2–3 weeks after production at the dairy plant.GC-MS analysis indicated that the content of ketones in general was higher inconventional UHT milk than in lactose-hydrolyzed UHT milk; however, no significantdifferences in the volatile compound profiles could be observed between the milk types


as a result of a batch-to-batch variation. However, it is hypothesized that the volatileprofile of conventional and lactose-hydrolyzed UHT milk will change during storageand that differences in the volatile profiles of the milk will increase during extendedstorage. Further studies are necessary to obtain a comprehensive understanding of howthe volatile profile of lactose-hydrolyzed milk will change during storage, and howthese changes affect the quality of the milk.

Acknowledgments The authors gratefully acknowledge the excellent technical assistance of Birgitte Foged.The present study is part of the Ph.D. work of Therese Jansson and was financially supported by Arla FoodsAmba, Food Future Innovation (FFI), Danish Dairy Research Foundation, and Aarhus University.

References

Adamiec J, Rössner J, Velíšek J, Cejpek K, Šavel J (2001) Minor Strecker degradation products ofphenylalanine and phenylglycine. Eur Food Res Technol 212(2):135–140

Al-Attabi ZHR (2009) Volatile sulphur compounds in UHT milk. The University of Queensland, DissertationAl-Attabi Z, D'Arcy BR, Deeth HC (2008) Volatile sulphur compounds in UHT milk. Crit Rev Food Sci Nutr

49(1):28–47Auldist MJ, Walsh BJ, Thomas NA (1998) Seasonal and lactational influences on bovine milk composition in

New Zealand. J Dairy Res 65(03):401–411Bach V, Kidmose U, Kjeldsen Bjørn G, Edelenbos M (2012) Effects of harvest time and variety on sensory

quality and chemical composition of Jerusalem artichoke (Helianthus tuberous) tubers. Food Chem133(1):82–89

Badings HT, Van der Pol JJG, Neeter R (1981) Aroma compounds which contribute to the difference inflavour between pasteurized milk and UHT milk. Flavour 81:683–692

Belitz HD, Grosch W, Schieberle P (2004) In Food chemistry; 3rd edn. Springer, New YorkBrands CMJ, Alink GM, van Boekel MAJS, Jongen WMF (2000) Mutagenicity of heated sugar-casein

systems: effect of the Maillard reaction. J Agric Food Chem 48(6):2271–2275Brands CM,Wedzicha BL, van Boekel MA (2002) Quantification of melanoidin concentration in sugar-casein

systems. J Agric Food Chem 50(5):1178–1183Contarini G, Povolo M (2002) Volatile fraction of milk: comparison between purge and trap and solid phase

microextraction techniques. J Agric Food Chem 50(25):7350–7355Contarini G, Povolo M, Leardi R, Toppino PM (1997) Influence of heat treatment on the volatile compounds

of milk. J Agric Food Chem 45(8):3171–3177Coppa M, Martin B, Pradel P, Leotta B, Priolo A, Vasta V (2011) Effect of a hay-based diet or different upland

grazing systems on milk volatile compounds. J Agric Food Chem 59(9):4947–4954Datta N, Elliott AJ, Perkins ML, Deeth HC (2002) Ultra-high-temperature (UHT) treatment of milk:

comparison of direct and indirect modes of heating. Aust J Dairy Technol 57(3):11–227Dirinck P, De Pooter H, Willaert G, Schamp N (1984) Application of a dynamic headspace procedure in fruit

flavour analysis. In: Schreier P (ed) Analysis of volatiles. Walter de Gruyter & Co, BerlinDolan JW (2009a) Calibration curves, part 1: to be or not to b? LCGC N Am 27(3):224–230Dolan JW (2009b) Calibration curves, part II: what are the limits? LCGC N Am 27(4):306–312Hutton JT, Patton S (1952) The origin of sulfhydryl groups in milk proteins and their contributions to

“cooked” flavor. J Dairy Sci 35(8):699–705Jelen P, Tossavainen O (2003) Low lactose and lactose-free milk and dairy products—prospects, technologies

and applications. Aust J Dairy Technol 58(2):161–165Jensen S, Oestdal H, Skibsted LH, Larsen E, Thybo AK (2011) Chemical changes in wheat pan bread during

storage and how it affects the sensory perception of aroma, flavour, and taste. J Cereal Sci 53(2):259–268Jeon IJ, Thomas EL, Reineccius GA (1978) Production of volatile flavor compounds in ultrahigh-temperature

processed milk during aseptic storage. J Agric Food Chem 26(5):1183–1188Kallioinen H, Tossavainen O (2008) Changes during storage of lactose hydrolysed extended shelf life (ESL)

milk. Milchwissenschaft 63(4):381–385Kato Y, Matsuda T, Kato N, Watanabe K, Nakamura R (1986) Browning and insolubilization of ovalbumin by

the Maillard reaction with some aldohexoses. J Agric Food Chem 34(2):351–355


Kato Y, Matsuda T, Kato N, Nakamura R (1988) Browning and protein polymerization induced by amino-carbonyl reaction of ovalbumin with glucose and lactose. J Agric Food Chem 36(4):806–809

Kessler HG (2002) Food and bioprocess engineering—dairy technology. Publishing House A, Munich,Germany

Kochhar SP (1996) In: Saxby MJ (ed) Food taints and off-flavours. Springer science and Business mediaDordrecht, Glasgow, UK

Lien YC, Nawar WW (1973) Thermal decomposition of tricaproin. J Am Oil Chem Soc 50:76–78Lien YC, Nawar WW (1974) Thermal interaction of amino acids and triglycerides. Valine and tricaproin. J

Food Sci 39:917–919Macku C, Takayuki S (1991) Headspace volatile compounds formed from heated corn oil and corn oil with

glycine. J Agric Food Chem 39:1265–1269Messia MC, Candigliota T, Marconi E (2007) Assessment of quality and technological characterization of

lactose-hydrolyzed milk. Food Chem 104(3):910–917Naudé Y, Van Aardt M, Rohwer ER (2009) Multi-channel open tubular traps for headspace sampling, gas

chromatographic fraction collection and olfactory assessment of milk volatiles. J Chromatogr A 1216(14):2798–2804

Ostersen S, Foldager J, Hermansen JE (1997) Effects of stage of lactation, milk protein genotype and bodycondition at calving on protein composition and renneting properties of bovine milk. J Dairy Res 64(02):207–219

Perkins ML, Elliott AJ, D’Arcy BR, Deeth HC (2005) Stale flavour volatiles in Australian commercial UHTmilk during storage. Aust J Dairy Technol 60(3):231–237

Rerkrai S, Jeon IJ, Bassette R (1987) Effect of various direct ultra-high temperature heat treatments on flavorof commercially prepared milks. J Dairy Sci 70(10):2046–2054

Sahi T (1994) Genetics and epidemiology of adult-type hypolactasia. Scand J Gastroenterol 29(S202):7–20Schwartz DP, Parks OW, Yoncoskie RA (1966) Quantitative studies on methyl ketone formation in butteroil:

effect of temperature. J Am Oil Chem Soc 43(3):128–129Tetra Pak (2003) In: Teknotext AB (ed) Dairy processing handbook, 2nd edn. Tetra pak processing systems

AB, Lund, SwedenThomas EL (1981) Trends in milk flavors. J Dairy Sci 64(6):1023–1027Thomas EL, Burton H, Ford JE, Perkin AG (1975) The effect of oxygen content on flavour and chemical

changes during aseptic storage of whole milk after ultra-high-temperature processing. J Dairy Res 42(02):285–295

Toso B, Procida G, Stefanon B (2002) Determination of volatile compounds in cows’ milk using headspaceGC-MS. J Dairy Res 69(04):569–577

Tossavainen O, Kallioinen H (2007) Proteolytic changes in lactose hydrolysed UHT milks during storage.Milchwissenschaft 62(4):410–414

Valero E, Villamiel M, Miralles B, Sanz J, Martínez-Castro I (2001) Changes in flavour and volatilecomponents during storage of whole and skimmed UHT milk. Food Chem 72(1):51–58

Vallejo-Cordoba B, Nakai S (1993) Using a simultaneous factor optimization approach for detection ofvolatiles in milk by dynamic headspace gas chromatographic analysis. J Agric Food Chem 41(12):2378–2384

Vazquez-Landaverde PA, Velazquez G, Torres JA, Qian MC (2005) Quantitative determination of thermallyderived off-flavor compounds in milk using solid-phase microextraction and gas chromatography. J DairySci 88(11):3764–3772

Vazquez-Landaverde PA, Torres JA, Qian MC (2006) Quantification of trace volatile sulfur compounds inmilk by solid-phase microextraction and gas chromatography-pulsed flame photometric detection. J DairySci 89(8):2919–2927

Whitfield FB (1992) Volatiles from interactions of Maillard reactions and lipids. Crit Rev Food Sci31(1/2):1–58

Yilmaz Y, Toledo R (2005) Antioxidant activity of water-soluble Maillard reaction products. Food Chem93(2):273–278

Zhong Y, Priebe MG, Vonk RJ, Huang CY, Antoine JM, He T, Harmsen HJ, Welling GW (2004) The role ofcolonic microbiota in lactose intolerance. Dig Dis Sci 49(1):78–83


95

Paper 2

Paper 2

Storage-induced changes in the sensory characteristics and

volatiles of conventional and lactose-hydrolyzed UHT processed

milk

Jensen S., Jansson, T., Eggers, N., Clausen, M.R., Jensen, H.B., Larsen, L.B.,

Colin, R., Sundgren, A., Andersen, H.J., Bertram, H.C.

Submitted to Food Research International

Lactose-Hydrolyzed Milk Is More Prone to Chemical Changes duringStorage than Conventional Ultra-High-Temperature (UHT) MilkTherese Jansson,† Morten R. Clausen,† Ulrik K Sundekilde,† Nina Eggers,† Steffen Nyegaard,‡,

Lotte B. Larsen,‡ Colin Ray,# Anja Sundgren,#, Henrik J. Andersen,⊥ and Hanne C. Bertram*,†

†Department of Food Science, Aarhus University, Kirstinebjergvej 10, DK-5792 Aarslev, Denmark‡Department of Food Science, Aarhus University, Blichers Alle 20, DK-8830 Tjele, Denmark#Arla Foods Strategic Innovation Centre, Lindhagensgatan 126, 10546 Stockholm, Sweden⊥Arla Foods Strategic Innovation Centre, Roerdrumvej 2, DK-8220 Brabrand, Denmark

*S Supporting Information

ABSTRACT: The enzymatic hydrolysis of lactose to glucose and galactose gives rise to reactions that change the chemistry andquality of ambient-stored lactose-hydrolyzed ultra-high-temperature (UHT) milk. The aim of the present study was to investigateand compare chemical changes in lactose-hydrolyzed and conventional UHT milk during a 9 month ambient storage period.Several complementary analyses of volatiles, free amino acids, acetate, furosine, and level of free amino terminals were concluded.The analyses revealed an increased level of free amino acids and an increased formation rate of specific compounds such asfurosine and 2-methylbutanal in lactose-hydrolyzed UHT milk compared to conventional UHT milk during storage. Theseobservations indicate more favorable conditions for Maillard and subsequent reactions in lactose-hydrolyzed milk compared toconventional UHT milk stored at ambient temperature. Furthermore, it is postulated that proteolytic activity from the lactase-enzyme preparation may be responsible for the observed higher levels of free amino acids in lactose-hydrolyzed UHT milk.

KEYWORDS: lactose-hydrolyzed milk, aldehydes, heat-induced changes, Maillard reaction, furosine, long-term storage

INTRODUCTION

Ultra-high-temperature (UHT) treatment and subsequentambient storage of milk have been studied extensively.1−4

The high temperature (130−150 °C, 2−10 s1) destroysbacteria and inactivates most of the enzymes present in milk,and thereby the milk shelf life is extended up to 8 months atambient temperature. However, the severe heat treatment alsocauses protein denaturation and nonenzymatic reactionbrowning (Maillard reactions) and initiates oxidative processesthat deteriorate the quality of UHT products during storage.This deterioration can result in quality defects such as colorchange, flavor change, and gelation that shorten the shelf life ofUHT products.3

As the result of an increased awareness of lactoseintolerance,5 new dairy products without lactose have becomecommercially available in the market. Lactose-hydrolyzed (LH)milk may be produced using ultra- and nanofiltration, whichremoves approximately 40% of the lactose, and enzymatichydrolysis by the addition of lactase. The net result is milk withapproximately 60% of the original carbohydrate content,composed of the reducing sugars glucose (approximately 2.5g/100 mL milk) and galactose (approximately 2.5 g/100 mLmilk) instead of lactose.6 Studies have indicated that lactose-hydrolyzed milk may be more vulnerable to chemical changesduring processing and storage than conventional milk.4,7 Thechanged carbohydrate composition in lactose-hydrolyzed milkwith the formation of glucose and galactose and their higherreactivity with amines (primary protein-bound lysine but alsoother NH− terminals) extend the number of Maillard reactionproducts in lactose-hydrolyzed milk compared to conventional

milk.4,8,9 Additionally, the added lactase enzyme may give riseto unwanted side activities, for example, proteolytic activity.10

In an effort to better understand the reactions that take place inlactose-hydrolyzed milk during storage and how these reactionsaffect the subsequent milk characteristics, the sensory character-istics, proteolytic activity, and role of heat load have all beenstudied in detail. Higher levels of cooked and processed flavor,proteolytic activity, and furosine were observed in lactose-hydrolyzed UHT milk compared with conventional UHTmilk.4,7 The Maillard reactions have been reported to form asmany as 3500 different volatile compounds, includingaldehydes, ketones, and sulfide compounds.11 Some of thesevolatile compounds are formed in lactose-hydrolyzed UHTmilk during heating and may affect the flavor and quality of thecorresponding milk during storage, for example, by shorteningits shelf life compared to the shelf life of conventional UHTmilk. Thus, it is of fundamental importance to investigate thechanges in volatile composition over the entire shelf life of thistype of milk, so that the production of lactose-hydrolyzed milkcan be better understood in relation to shelf life extendinginitiatives. The aim of the present study was to elucidate thedetailed chemical changes in enzymatically lactose-hydrolyzedmilk (both direct and indirect UHT-treated) and conventionalUHT milk during a 9 month storage period. To support thehypothesis that lactose-hydrolyzed UHT milk is more

Received: April 15, 2014Revised: July 6, 2014Accepted: July 14, 2014Published: July 14, 2014

Article

pubs.acs.org/JAFC

© 2014 American Chemical Society 7886 dx.doi.org/10.1021/jf501671z | J. Agric. Food Chem. 2014, 62, 7886−7896

pubs.acs.org/JAFC

susceptible to Maillard induced chemical changes, proteolyticand other non-Maillard chemical changes were observed duringstorage compared to corresponding changes in conventionalUHT milk.

MATERIAL AND METHODSChemical References. Decane (99%), tetradecane (99%),

dodecane (99%), 2-methylbutanal (95%), hexanal (98%), heptanal(95%), octanal (99%), furfural (98%), nonanal (95%), decanal (98%),2-butanone (99%), 2-pentanone (99%), 2-hexanone (99.5%), 2-heptanone (98%), 2-octanone (98%), 2-nonanone (99%), 6-methyl-5-heptene-2-one (99%), 6,10-dimethyl-5,4-undecadien-2-one (97%), 2-undecanone (99%), dimethyl disulfide (98%), dimethyl trisulfide(98%) 4-methyl-1-pentanol (97%), toluene (99.5%), styrene (99%). 2-methylfuran (98%), 2-ethylfuran (99%), 2-pentylfuran (97%),fluorescamine (4-phenylspiro[furan-2(3H),19-phthalan]-3,39-dione)(99%), and D2O containing 0.025% sodium trimethylsilyl-[2,2,3,3-2H4]-1-propionate were obtained from Sigma-Aldrich Inc.(Steinheim, Germany). Trichloroacetic acid, HCl, and water-freeacetone were from Merck (Darmstadt, Germany).Milk Samples. Milk for the study was produced at a commercial

Danish dairy plant (Arla Foods, Esbjerg, Denmark) (Figure 1). Three

types of milk (conventional indirect UHT (CONVI), lactose-hydrolyzed direct UHT (LHD), and lactose-hydrolyzed indirectUHT (LHI)) were produced (Table 1). The milk was UHT-treatedusing either direct heat treatment (LHD) (steam is injected into themilk to heat it followed by flash cooling12) (140−145 °C for 2−3 s) orindirect heat treatment (CONVI, LHI) (a tubular heat exchanger isused to heat the milk, followed by slow cooling12) (140−145 °C for2−3 s) and packaged aseptically in 1 L Tetra Brik cartons. Lactose inthe lactose-hydrolyzed milk (LHD, LHI) was removed (approximately40% of initial lactose content) before heat treatment using ultra- andnanofiltration. The remaining lactose was enzymatically hydrolyzed(LHD, LHI), after heat treatment, by the addition of lactase (MaxilactLAG 5000, DSM, Limburg, The Netherlands).After transportation to the laboratory, the milk samples were stored

in an air-conditioned room at a constant temperature of 22 °C. At days7, 14, 21, 28, 45, 60, 90, 120, 150, 180, 210, 240, and 270 afterproduction, two milk cartons were picked and frozen and stored at−20 °C until analysis. Milk for the NMR spectroscopy analysis andfluorescamine determination was filtered using Amicon Ultra 0.5 mL10 kDa (Millipore, Bedford, MA, USA) spin filters at 10000g for 15min to remove residual lipids and protein prior to freezing. Afterfrozen storage, the milk samples and filtrates were randomized andthawed in 4 °C before final analyses.

Determination of Volatile Compounds by Dynamic Head-space Sampling and GC-MS. The volatile compounds wereextracted using dynamic headspace sampling and analyzed by gaschromatography−mass spectrometry (GC-MS) according to theprocedure described by Jansson et al.13 The volatile compoundswere isolated on adsorbent traps (Tenax TA, 200 g Tenax TA, 35/60mesh (Markes International Limited, LIantrisant, UK)) and desorbed(250 °C, 15 min) into a cold trap (U-TIIGCP, Markes InternationalLimited) using a thermal desorption system (ultra-UNITY, MarkesInternational Limited) and subsequently injected onto a gaschromatograph (Finnigan trace GC ultra, Thermo, Waltham, MA,USA) using a CP-Wax 52CB column (50 m × 0.25 mm, film thickness= 0.25 μm (Varian, Palo Alto, CA, USA)).

To identify the separated volatile compounds, a single-quadrupolemass spectrometer (Finnigan Trace dual-stage quadrupole (DSQ),Thermo) was used. The mass spectrometer scanned over a mass tocharge (m/z) of 45−350 (1.61 scans/s) in TIC mode, and specific m/z values were detected in SIM mode (Table 2). The volatile

compounds were identified using an MS database (NIST MS search2.0, 2011) and verified by comparison of mass spectral data andretention times with those of authentic reference compounds.

Figure 1. Flowchart of production of lactose-hydrolyzed UHT milk(red) and conventional UHT milk (black).

Table 1. Overview of Milk Types Included in the Study, Heat Treatments, and Lactose Levelsa

milkenzymatichydrolysis

heattreatment filtration level of lactose (% of original level) amount of lactose (% w/w)

CONVI conventional UHT − indirect − 100 5LHD lactose-hydrolyzed direct UHT + direct + <1 0.01LHI lactose-hydrolyzed indirect UHT + indirect + <1 0.01

aAll milks contained 1.5% w/w fat and 3.5% w/w protein.

Table 2. Overview of Retention Time (tR), Identified andVerified Volatile Compounds, and m/z Ions Scanned withGC-MS SIM Mode

tR (min) volatile compound m/z

4.84 3,3-dimethylhexane6.18 2-methylfuran 826.87 2-butanone7.05 2-methylbutanal 868.22 2-ethylfuran 819.20 2-pentanone10.12 decane12.14 toluene14.48 dimethyl disulfide15.47 2-hexanone15.50 hexanal 8227.37 2-heptanone27.38 heptanal 7030.48 dodecane32.69 2-pentylfuran 9534.84 styrene36.75 2-octanone36.94 octanal39.07 6-methyl-5-heptene-2-one40.12 dimethyl trisulfide40.75 2-nonanone40.95 nonanal41.14 tetradecane42.91 furfural43.42 decanal45.25 2-undecanone 17048.90 6,10-dimethyl-5-undecadien-2-one

Journal of Agricultural and Food Chemistry Article

dx.doi.org/10.1021/jf501671z | J. Agric. Food Chem. 2014, 62, 7886−78967887

A total of 15 identified volatile compounds were quantified usingexternal standard curves prepared in duplicates in UHT milk. For eachstandard curve, the regression equation, limit of detection (LOD), andlimit of quantification (LOQ) were calculated from the standard errorof the y-intercept and the slope of the linear regression (Table 2).14,15

Determination of Furosine by RP-HPLC. Furosine formationwas determined in milk after acid hydrolysis by RP-HPLC as describedpreviously in ISO 18329:2004 (IDF 193:2004)16 with somemodifications. One milliliter of milk was mixed with 3 mL of 10 MHCl in screw-cap tubes, and nitrogen was bubbled through thesamples for 2 min. The tubes were closed and heated at 110−115 °Cfor 18 h. After heating, the samples were cooled and filtered through afilter paper (Black Ribbon 589/1, Schleicher & Schuell MicroScienceGmbH, Dassel, Germany), and 2 mL of filtrate was filtered through a0.45 μM disposable filter, diluted 5 times in 3 M HCl, and transferredinto HPLC vials. Furosine detection was performed on a Dionex300DX system (Dionex, Germering, Germany) with a DAD detectorand using a Supelco Supercosil LC-8 column (Sigma-Aldrich Inc.)with isocratic elution with 0.06 M sodium acetate buffer at a flow rateof 1 mL/min. The injection volume was 10 μL, and the DADdetection (UV) was at 280 nm.Determination of Free Amino Acids Using 1H NMR Spec-

troscopy. Determination of free amino acids was performed using 1HNMR spectroscopy analysis according to a method adapted fromSundekilde er al.17 The milk filtrates were thawed, and 400 μL was

mixed with 200 μL of D2O containing 0.025% sodium trimethylsilyl-[2,2,3,3-2H4]-1-propionate (TSP) as internal chemical shift reference.A Bruker Avance III 600 spectrometer (Bruker Biospin, Rheinstetten,Germany) equipped with a 5 mm TXI probe was used for the analyses.The sample sequence was randomized prior to acquisition. Standardone-dimensional spectra were acquired using a single 90° pulseexperiment with a relaxation delay of 5 s. Water suppression wasachieved by irradiating the water peak during the relaxation delay, anda total of 64 scans were collected into 32K data points scanning aspectral width of 12.15 ppm. All 1H spectra were initially referenced tothe TSP signal at 0 ppm. Prior to Fourier transformation the data weremultiplied by a 0.3 Hz line-broadening function. The proton NMRspectra were phase and baseline corrected manually using Topspin 3.2(Bruker Biospin). Amino acid concentrations were calculated forvaline, methionine, isoleucine, alanine, and acetate using Chenomxsoftware (Chenomx NMR suite 7.7, Edmonton, Canada).

Determination of Free Amino Terminals by FluorescamineAssay. The level of free amino terminals in the milk was determinedaccording to the method of Dalsgaard et al.18 An amount of 75 μL ofmilk was mixed with 75 μL of 24% trichloroacetic acid (24 g TCA/100mL water) and placed on ice for 30 min to accelerate proteinprecipitation. The precipitated mixture was centrifuged (17049g for 20min). A total of 37.5 μL of supernatant was then added to 1130 μL of0.1 M borate buffer and 375 μL of fluorescamine solution (0.2 mg/mLfluorescamine in water free acetone), and 250 μL was then transferred

Figure 2. Overview of compounds identified in the milk originating from the Maillard reaction, lipid oxidation, and proteolytic activity analyzed withGC-MS (blue rectangles), NMR (orange rectangles), HPLC (green rectangles), and fluorescamine assay (pink rectangle). A dashed outline indicatesincrease of compounds in lactose-hydrolyzed UHT milk, a dotted outline indicates increase of compounds in conventional UHT milk, a solid outlineindicates no difference between lactose-hydrolyzed and conventional UHT milk during storage.



to each well in a microtiter plate. The analysis was carried out, 18 minafter the addition of the fluorescamine, on a Synergy 2 Multi-Modemicroplate reader (Holm & Halby, Brøndby, Denmark) withexcitation at 390 nm and emission at 480 nm. Quantification wasachieved by calculating leucine equivalents using an external leucinestandard curve prepared as follows. A 0.1 M leucine stock solution wasmade (327.5 mg of leucine in 25 mL of 1 mM HCl), and from thestock solution six different concentrations were made (0.00050−0.0030 M leucine) and mixed with 10 mL of 1 mM HCl and analyzedas the milk samples.Statistical Analysis. To identify significant effects of milk type,

storage time, and degree of increase in concentration on the quantifiedvolatile compounds, furosine, and free amino acids, regressions weremade by a linear model (lm) followed by a one-way analysis ofvariance (ANOVA). In addition, a linear mixed model (lme) and ageneralized linear mixed model (lmer), least-squares means(LSMEANS), and Tukey’s honest significant difference (α = 0.05)test was performed using the software R (version 3.0.2, R Foundationfor Statistical Computing, Vienna, Austria).

RESULTSGC-MS, NMR, furosine, and fluorescamine analyses wereconducted on industrially produced conventional UHT milk(100% lactose) and lactose-hydrolyzed UHT milk (<1% oforiginal level of lactose) throughout a storage period of 9months at 22 °C. The lactose-hydrolyzed UHT milk washeated with 60% of the original level lactose, which wassubsequently hydrolyzed to glucose and galactose by theaddition of lactase. The compounds detected in the milk by thedifferent analytical approaches are summarized in Figure 2.Furosine. The development of furosine levels in the milk

after acid hydrolysis was found to be significantly affected bymilk type. A significantly higher initial level of furosine wasobserved in CONVI milk (p < 0.001) (Figure 3) compared to

LHD milk and LHI milk; however, after approximately 60 days,furosine levels started to increase in LHD milk and LHI milk,whereas in CONVI milk furosine levels were constantthroughout the storage period. After 270 days of storage, thefurosine content was about 2.5 times higher in lactose-hydrolyzed milk than in conventional milk.GC-MS. A total of 27 volatile compounds were extracted

from the headspace sampling and identified using GC-MS(Table 2), and of these 15 were quantified (Table 3). Theremaining 12 identified compounds (Tables 2 and 3) wereconstant throughout the storage period for all milk types andwere therefore not quantified.

Notably, the levels of the five ketones quantified weresignificantly higher in conventional UHT milk (p < 0.001) thanin lactose-hydrolyzed milk in the initial phase of the storageperiod by a factor of 4−5 (Figure 3). All of the quantifiedketones followed a similar trend during storage for all milktypes, and a significant (p < 0.001) increase in theconcentration of ketones was observed during storage (Figure4). However, for 2-pentanone (p < 0.001), 2-heptanone (LHD,p = 0.016; LHI, p = 0.006), and 2-nonanone (LHD, p < 0.001;LHI, p < 0.001) the formation rate significantly increased at ahigher extent for lactose-hydrolyzed milk compared withconventional milk (Figure 4A−C; Supporting Information).In addition, the formation rate of 2-undecanone increasedsignificantly more in LHD than in CONVI (p = 0.029) andLHI (p = 0.023) milks. 2-Nonanone was the only ketone forwhich a difference between direct and indirect heat treatmentcould be observed during storage and a higher formation ratewas seen in LHD (p = 0.0012) milk compared with LHI milk.2-Octanone did not appear before 90 days in lactose-hydrolyzed milk, whereas it was observed already at day 14 inconventional milk. A higher formation rate in the level of 2-octanone was observed in CONVI (p = 0.0224) milk comparedto LHD milk during storage (Figure 4D).With the exception of 2-methylbutanal, no significant change

in the concentration was observed for the quantified aldehydesduring storage (Figure 5). Notably, the concentration of 2-methylbutanal was almost constant in CONVI milk during the9 month storage period, whereas its concentration (Figure 5D;Supporting Information) increased tremendously after 150 daysof storage in LHD (p < 0.001) and LHI (p < 0.001) milk. Inaddition, it was observed that the level of 2-methylbutanalsignificantly increased more in LHI (p < 0.001) milk than inLHD milk from day 200.Heterocycles 2-methylfuran and 2-ethylfuran exhibited a

pattern similar to that of the ketones, and a significant increasein their concentrations was observed for all of the milk typesduring storage (p < 0.001) (Figure 6A,B; SupportingInformation). The increase in concentration of the two furansduring the storage period was significantly higher in CONVI (p< 0.001) milk compared with the lactose-hydrolyzed milk, andthe concentration of 2-ethylfuran increased more in LHI (p =0.012) milk compared with LHD milk. A different pattern wasobserved for 2-pentylfuran (Figure 6C). Whereas a significantincrease in the concentration of 2-pentylfuran was observed inCONVI (p < 0.001) milk during the storage period, theconcentration remained almost constant in LHD and LHI milk.Dimethyl trisulfide (DMTS) increased in LHD and LHI milk

up to 50 days of storage, followed by a decrease to below theinitial concentration. In CONVI milk a higher initialconcentration of DMTS was observed than in LHD and LHImilks; however, the concentration decreased rapidly duringstorage, and a low and constant concentration of approximately0.2 μg/kg milk was reached after approximately 50 days ofstorage (Figure 7). Dimethyl disulfide (DMDS) was constantthroughout the storage period for all milk types (data notshown).

Analysis of Amino Acids by NMR. Four free amino acids,isoleucine, alanine, methionine, and valine (Figure 8), wereidentified and quantified by 1H NMR spectroscopy. The levelsof alanine (LHD, p < 0.001; LHI, p < 0.001), isoleucine (LHD,p = 0.002; LHI, p < 0.001), and valine (LHD, p < 0.001; LHI, p< 0.001) increased in lactose-hydrolyzed milk, whereas theyremained constant in the conventional milk throughout the

Figure 3. HPLC analysis of furosine in conventional indirect UHTmilk (CONVI), lactose-hydrolyzed direct UHT milk (LHD), andlactose-hydrolyzed indirect UHT milk (LHI) during storage.



storage period (Figure 8A,B,D; Supporting Information).

Methionine content was initially significantly higher in

CONVI (p = 0.0011) compared to LHI milk (Figure 8C;

Supporting Information). Throughout the storage period

methionine levels in CONVI and LHD milk decreased to thelevel of LHI milk.In addition to the four amino acids, acetate was also

quantified by 1H NMR spectroscopy (Figure 9; SupportingInformation). The level of acetate significantly increased in

Table 3. Regression Equations, Limits of Detection (LOD), Limits of Quantification (LOQ), and Coefficients of Determination(R2) for Calibration Curves in 1.5% Fat UHT Milk

compound calibration concn (μg/kg) LOD (μg/kg) LOQ (μg/kg) regression eq R2

aldehydes2-methylbutanal 3.06−10.69 0.82 2.50 y = 0.07x 0.993hexanal 2.29−15.97 2.10 6.35 y = 0.046x 0.970nonanal 3.93−19.64 3.66 11.08 y = 0.0308x − 0.0448 0.967furfural 2.27−68.21 0.84 2.53 y = 0.0212x 0.999decanal 4.07−16.23 1.05 3.18 y = 0.0346x 0.997ketones2-pentanone 4.01−64.07 1.72 5.20 y = 0.1474x 0.9972-heptanone 8.04−80.36 3.25 9.84 y = 0.3954x 0.9972-nonanone 7.64−61.10 5.60 16.96 y = 0.1534x 0.9882-octanone 1.61−20.07 0.45 1.35 y = 0.3196x + 0.0718 0.9992-undecanone 16.25−48.75 3.35 10.15 y = 0.0029x 0.995furans2-methylfuran 5.74−28.68 5.09 15.44 y = 0.2506x − 1.4156 0.9692-ethylfuran 4.51−18.06 1.27 3.85 y = 4.711x 0.9962-pentylfuran 5.37−26.86 2.92 8.84 y = 0.5085x − 0.3116 0.991sulfidesDMDS 1.04−10.39 0.58 1.77 y = 2.3148x + 0.8038 0.997DMTS 1.17−35.22 0.56 1.69 y = 0.5219x 0.999

Figure 4. GC-MS analysis of (A) 2-pentanone, (B) 2-heptanone, (C) 2-nonanone, (D) 2-octanone, and (E) 2-undecanone in conventional indirectUHT milk (CONVI), lactose-hydrolyzed direct UHT milk (LHD), and lactose-hydrolyzed indirect UHT milk (LHI) during storage.



lactose-hydrolyzed milk (LHD, p = 0.028; LHI, p = 0.0336)during storage, whereas the increase in conventional milkduring storage was not significant.Analysis of Free Amino Terminals by Fluorescamine

Assay. Slightly higher initial levels of free amino terminals andamino groups of lysine side chains were observed in theCONVI milk compared with lactose-hydrolyzed milk (Figure10). However, the level of free amino terminals, expressed asleucine equivalents, significantly increased in the lactose-hydrolyzed milk (p < 0.001), whereas it remained ratherconstant in the conventional milk during storage (Figure 10;Supporting Information). After 270 days of storage, the level offree amino terminals was about 2 times higher in lactose-hydrolyzed milk than in conventional milk. No differences wereobserved between LHD and LHI milks.

DISCUSSIONIn the present study volatile compounds, Amadori products asdetermined by furosine levels, levels of free amino acids, andlevels of amino terminals, and proteolysis were analyzed inlactose-hydrolyzed and conventional UHT milk to elucidate thedetailed chemical changes during a 9 month storage period(Figure 2).Maillard Reaction Products. An increase in the level of

furosine was observed in all milk types during the 9 monthstorage, which demonstrates a continuous formation of itsprecursors, lactulosyllysine and fructosyllysine (Figures 2 and3). During the first 60 days of storage lactose-hydrolyzed milkwas characterized by a significantly lower furosine level thanconventional milk. This was likely due to the removal of

approximately 40% of lactose prior to the UHT treatment inthe lactose-hydrolyzed milk, resulting in lower levels ofcarbohydrates during the UHT treatment and lower levels ofMaillard reactions in the initial storage period. This indicatesthat furosine precursors were formed during the heat treatment.However, during storage, the formation rate was significantlyhigher in lactose-hydrolyzed milk compared to conventionalmilk. This is consistent with the fact that the monosaccharidespolymerize more progressively19 than disaccharides. Thisfinding reveals that despite lower carbohydrate levels, theMaillard reaction proceeds more rapidly and to a greater extentin lactose-hydrolyzed milk during storage. Previous studies havefollowed the development of furosine in lactose-hydrolyzed andconventional milk for up to 4 months and showed significantincreases in furosine during storage.4,20 Messia et al.investigated the formation of furosine in lactose-hydrolyzedand conventional indirect UHT milk during 4 months ofstorage at 20 °C and found higher values of furosine in theformer.4 The concentrations of furosine found in the lactose-hydrolyzed milk at production and after storage for 4 monthswere 234 and 562 mg/100 g protein, respectively, whereas thecorresponding concentrations found in the conventional milkwere 235 and 310 mg/100 g protein, respectively. These resultsare quantitatively consistent with the results in the presentstudy for the same storage period, temperature, and milk type.However, this study shows that it is important to follow thedevelopment of furosine for longer than 4 months (Figure 3).Tossavainen and Kallionen investigated furosine content inthree milk types: nonhydrolyzed, hydrolyzed, and lactose-reduced-hydrolyzed (similar to the lactose-hydrolyzed milk in

Figure 5. GC-MS analysis of (A) decanal, (B) hexanal, (C) furfural, (D) 2-methylbutanal, and (E) nonanal in conventional indirect UHT milk(CONVI), lactose-hydrolyzed direct UHT milk (LHD), and lactose-hydrolyzed indirect UHT milk (LHI) during storage.



the present study) UHT milk stored for 12 weeks.7

Tossavainen and Kallionen found a 2.5-fold higher level offurosine in the lactose-reduced-hydrolyzed milk and lactose-hydrolyzed milk, respectively, compared with the nonhydro-lyzed milk and concluded that the content of furosine wasdependent on both the amount and type of sugar. In thepresent study it was observed that the content of furosine wasdependent on the amount of sugar the first 90 days and thattype of sugar was important after 90 days of storage. It was also

revealed that the level of furosine in conventional and lactose-hydrolyzed UHT milk is very dependent on the storage period.This has implications for understanding storage-induceddeterioration of milk quality.A very similar formation pattern was observed for the

furosine generation and the Maillard reaction product, 2-methylbutanal (Figure 5D). After initially being observed at alower concentration in LHI and LHD, which is consistent withtotal carbohydrate content, accumulation of 2-methylbutanalwas found to occur more rapidly and to a much greater extentin lactose-hydrolyzed milk. This observation is in goodagreement with the current understanding of the Maillard−Strecker degradation pathway. The volatile 2-methylbutanal is aStrecker degradation aldehyde21 and is formed when isoleucinereacts with dicarbonyl compounds in the milk. The increasedconcentration of 2-methylbutanal in lactose-hydrolyzed com-pared to conventional UHT milk can be ascribed to the type ofsugar (lactose, glucose, and galactose) available in the milk.Kato et al. studied protein polymerization and browning in amodel system consisting of ovalbumin, glucose, and lactose andconcluded that glucose accelerates the protein polymerizationas well as the Amadori degradation.19 Kato et al. also concludedthat the Amadori compounds formed from lactose were lessprone to degradation into aldehyde compounds compared withthe Amadori compounds from glucose. Intriguingly, in thepresent study it could be observed that the free amino acid andprecursor for Strecker degradation, isoleucine, increased duringstorage in lactose-hydrolyzed milk, whereas it remainedconstant in conventional milk (Figure 8A). This finding isprobably related to a higher level of proteolysis in lactase-treated milk and could be due to unwanted side activities of theenzyme preparation. Balagiannis et al. investigated theformation of 2-methylbutanal and observed that the sugarconcentration, the Amadori rearrangement, the rate ofdicarbonyl formation, and the amount of isoleucine were thelimiting steps of its formation.21 The simultaneous increase ofisoleucine, furosine, and 2-methylbutanal in lactose-hydrolyzedmilk after approximately 100−150 days of storage indicates thatthe level of isoleucine and the Amadori rearrangement isimportant for the formation of 2-methylbutanal, which, inaddition to the lower reactivity of lactose with proteins and thenondetectable formation of isoleucine, explains the absence of aformation of 2-methylbutanal in the conventional UHT milk.Furfural is formed in the 3-deoxyosone pathway in the early

Maillard reaction, which is favored by acidic pH. However,furfural may be a precursor of the formation of furan derivativesin the Maillard reaction.22 Berg observed that furfural wasformed in milk in very small concentrations after heating andthat the level of furfural increased with increasing heat load.23

In the present study, furfural was identified and quantified, andit was found that furfural remained almost constant at aconcentration of 0.8−6 μg/kg (8.3−62.4 μmol/kg) throughoutthe storage period. This concentration is higher than theconcentration detected by Berg, who reported 0.68 μmolfurfural/kg in UHT milk.23 This difference is possibly due todifferences in the UHT treatment or differences in thesensitivity of the applied GC-MS method. The results in thepresent study indicate that no formation of furfural is takingplace during storage at room temperature, likely due to thenonpreferred pathway at normal milk pH.Acetate, identified by NMR spectroscopy in the present

study (Figure 9), is a major degradation product of sugars inthe Maillard reaction. In neutral and basic pH, acetate is formed

Figure 6. GC-MS analysis of (A) 2-methylfuran, (B) 2-ethylfuran, and(C) 2-pentylfuran in conventional indirect treated UHT milk(CONVI), lactose-hydrolyzed direct UHT milk (LHD), and lactose-hydrolyzed indirect UHT milk (LHI) during storage.

Figure 7. GC-MS analysis of dimethyl trisulfide in conventionalindirect UHT milk (CONVI), lactose-hydrolyzed direct UHT milk(LHD), and lactose-hydrolyzed indirect UHT milk (LHI) duringstorage.



primarily from C2−C3 β-cleavage of the 1-deoxyosoneisomerization product 1-deoxy-2,4-hexodiulose (2,4-penta-dione).24,25 Acetate is a stable end product of this pathway

and considered as a marker for the 2,3-enolization in theMaillard reaction.24 In addition, acetate can also be formeddirectly from sugar fragmentation.25 In the present study thelevel of acetate increased during storage for all of the milk types,and no significant difference was found between the milk types(Figure 9). The accumulation of acetate in the milk duringstorage indicates that a continuous formation is taking placethroughout storage and that the degradation of the respectiveAmadori compound (fructosyllysine in lactose-hydrolyzed milkand lactulosyllysine in conventional milk) occurs throughoutstorage, leading to a continuous formation of the dicarbonyls,which are included in Strecker degradation in the Maillardreaction (Figure 2). The comparable increase of acetate inlactose-hydrolyzed and conventional milk indicates that thesame level of Amadori compound degradation is present inboth milk types. Kato et al. observed that Amadori compoundsfrom lactose were more difficult to degrade compared with theAmadori compounds from glucose, which could explain thelower level of Strecker aldehydes such as 2-methylbutanal inconventional milk in the present study.19 However, contrary toKato et al., the present study indicates that the 2,3-enolizationfrom the Amadori compounds to 1-deoxyosone is similar inconventional and lactose-hydrolyzed UHT milk and that thehindrance for further degradation to aldehydes is taking placeafter this step in the Maillard reaction cascade (Figure 2).2-Methylfuran and 2-ethylfuran increased significantly during

storage in all milk types (Figure 6), but a larger increase wasseen in CONVI milk compared with LHD and LHI milks. Thefurans 2-methylfuran and 2-ethylfuran are Strecker aldehyde(acetaldehyde and lactaldehyde) products as well as sugar andprotein degradation products.26,27 The higher content of furansin the conventional milk indicates that the formation of furansis favored by lactose, and not glucose and galactose. Thissuggestion is supported by a recent published work byOwczarek-Fendor et al., who investigated furan formation instarch-based systems containing various carbohydrates andproteins, heated at 130 °C for 30 min, at different pH values.28

The study revealed that lactose and lactose−protein systems

Figure 8. NMR spectroscopic analysis of (A) isoleucine, (B) alanine, (C) methionine, and (D) valine in conventional indirect treated UHT milk(CONVI), lactose-hydrolyzed direct UHT milk (LHD), and lactose-hydrolyzed indirect UHT milk (LHI) during storage.

Figure 9. NMR spectroscopic analysis of acetate in conventionalindirect UHT milk (CONVI), lactose-hydrolyzed direct UHT milk(LHD), and lactose-hydrolyzed indirect UHT milk (LHI) duringstorage. Acetate significantly increased for LHD and LHI duringstorage: significance level for LHD (p = 0.0275), LHI (p = 0.0336).

Figure 10. Concentration of free amines in conventional indirectUHT milk (CONVI), lactose-hydrolyzed direct UHT milk (LHD),and lactose-hydrolyzed indirect UHT milk (LHI) determined byfluorescamine assay during storage.



generated higher amounts of furan at pH 6 compared withglucose and fructose systems.Thermal Degradation and Oxidation Products. In

addition to 2-methylbutanal, four other aldehydes werequantified. The nonsystematic variation in aldehydes couldpotentially be ascribed to the low concentration of thesecompounds. During thermal processing fatty acids mayundergo thermal degradation reactions, including isomer-ization, cyclization, and cleavage, and aldehydes are mainlyformed by oxidation of fatty acids;29,30 straight-chain aldehydesare the most important markers for fat oxidation.31 Vazquez-Landaverde et al. observed that aldehydes were not affected byheat to the same extent as ketones.2 Valero et al. found a smallincrease in aldehydes during 120 days of storage.3 Jeon et al.investigated the effect of addition of ascorbic acid, acting asantioxidant, on the formation of aldehydes in UHT milk andfound that the addition of ascorbic acid reduced the formationof aldehydes.32 Osada and Shibamoto investigated theantioxidative activity of volatile extracts from Maillard modelsystems and observed that the combined effect of manycompounds produced in the Maillard reaction possessed anantioxidative effect that decreased the oxidation of hexanal.33

Thus, in the present study, an antioxidant activity from theadvanced Maillard reaction products34 or peptides andproteins,35 formed from proteolysis, may inhibit the formationof aldehydes. This could explain the low level of changeobserved for the aldehydes during storage in all milk types.However, the low level of aldehydes may also indicate a lowlevel of oxidation in the milk.During storage a significant increase in the concentration of

ketones was observed in the milks (Figure 4). Ketones arederived from fatty acids through β-oxidization30,36 and areprimarily formed by thermal degradation.29 The ketones areknown to be temperature-dependent, and increased levels ofketones have been observed in milk exposed to highertemperatures.2 The present study revealed that both the leveland the formation rate of ketones were affected by milk type.Whereas the level of ketones was higher in conventional UHTin the beginning of the storage period, the levels of 2-pentanone, 2-heptanone, and 2-nonanone increased signifi-cantly higher in the lactose-hydrolyzed compared to conven-tional milk. In addition, a higher increase in 2-undecanone wasobserved in LHD milk compared with CONVI milk. Thesefindings indicate that the thermal degradation of the fatty acidsis more pronounced in lactose-hydrolyzed milk than conven-tional UHT milk. However, because the heat treatmentsapplied were similar, the exact mechanism remains unknown.A significant increase in 2-pentylfuran was found in CONVI

milk during storage, whereas this compound was found to beconstant in lactose-hydrolyzed milk (Figure 6C). Overall, 2-pentylfuran is formed from oxidation of unsaturated lipids.27,37

A possible explanation for the increased formation of 2-pentylfuran in the conventional UHT milk compared with thelactose-hydrolyzed milk may be a higher antioxidant activity inthe lactose-hydrolyzed milk due to an increased Maillardreaction34,38 or increased proteolytic activity35 formingantioxidative compounds.The antioxidative compounds maydecrease the oxidation of the unsaturated lipids present in themilk and thereby the furan formation.When β-lactoglobulin is heat-denatured, methionine and

cysteine are exposed to further reactions, such as oxidation andMaillard reactions, and Strecker degradation. The Streckerdegradation of methionine may lead to the formation of

methional, which further oxidizes to DMDS and DMTS,36,39

and in the present study, DMTS apparently was formed after orduring heat treatment in all milk types. In lactose-hydrolyzedmilk the DMTS concentration increased further during storageand then decreased rapidly after approximately 50 days (Figure7), whereas an immediate decrease was observed in theconventional UHT milk. In addition, it was observed thatmethionine decreased in lactose-hydrolyzed milk, which couldindicate that methionine was degraded to DMTS in the initialpart of the storage period. However, for the conventional UHTmilk, methionine first increased followed by a slow decrease,and the decrease could not be correlated to any simultaneousformation of DMTS. Sulfides and sulfur-containing amino acidsare very reactive, and the decrease of these compounds duringstorage is mainly thought to be due to degradations andoxidations.40,41 Because of observations of increased levels offree amino acids or peptides in lactose-hydrolyzed milk duringstorage (Figure 8A,C,D), a similar observation was expectedwith methionine; however, no increase was observed. This maybe explained by the specificity of the proteolytic enzyme, whichmay not cleave methionine or methionine-containing peptides,or rapid degradation of exposed methionine residues, whichwould prevent its detection in the NMR.

Protein Hydrolysis during Storage. In the present study1H NMR spectroscopy was used to determine amino acidsacting as precursors in Maillard reaction and also to determineacetic acid, which can be formed during the Maillard reaction.In addition to methionine and isoleucine already discussed,alanine and valine were quantified in the milk by 1H NMRspectroscopy. This quantification revealed that these aminoacids increased significantly in lactose-hydrolyzed milk duringstorage, whereas they remained constant at a lower level in theconventional milk (Figure 8B,D). In addition, an increase in thelevel of free amino terminals, obtained from the fluorescamineassay, was observed in lactose-hydrolyzed milk compared toconventional UHT milk in Figure 10. These resultsdemonstrate a higher proteolytic activity in lactose-hydrolyzedmilk compared to conventional milk. An increased level of freeamino terminals and amino groups of lysine side chains inlactose-hydrolyzed milk may contribute considerably to thehigher level of Maillard reaction products in the lactose-hydrolyzed milk. An increased Maillard reaction decreaseslysine availability in lactose-hydrolyzed milk, and therefore theprotein nutrition value may be lower in this type of milk. Alower protein availability might pose a challenge when aiming atdecreased malnutrition in countries with a high prevalence oflactose intolerance. Thus, it is of interest to optimize theproduction of lactose-hydrolyzed milk to avoid high levels ofMaillard reactions and thereby maintain a high nutritional valueof the milk. The increased level of free amino acids and aminoterminals (Figures 8 and 9) in lactose-hydrolyzed milk duringstorage indicates that the proteolytic activity is morepronounced than the Maillard reaction activity. Tossavainenand Kallioinen analyzed proteolytic activity in lactose-hydro-lyzed and unhydrolyzed UHT milk directly after productionand after 4 weeks of storage and observed a higher proteolyticactivity in the hydrolyzed UHT milk compared to theunhydrolyzed UHT milk, which is in agreement with thepresent study.10 The higher proteolytic activity is probably dueto the proteolytic side effect of the added lactase enzymepreparation, which gives rise to substrates for the Maillardreaction and thereby an increased risk for milk qualitydeterioration during storage.



Effect of Heat Treatment. In addition to the effect oflactose hydrolysis, the effects of heat treatment were alsoelucidated in the present study as two different types of heattreatments were included: a direct and an indirect heattreatment. However, no differences in detection in furosinecontent or free amino terminals were observed between directand indirect heat-treated lactose-hydrolyzed milks. Higherfurosine levels in conventional indirect UHT milk comparedwith conventional direct UHT milk have been reported inprevious studies.1,42 The levels of 2-nonanone and 2-undecanone increased more in LHD than in LHI milk, whichis contrary to the observation of Perkins et al., who found anincreased total amount of ketones in indirect heat-treatedconventional UHT milk, compared with the direct heat-treatedmilk. The significantly higher increases in 2-methylbutanal and2-ethylfuran during storage in LHI compared with LHD milkcould be explained by the different heat loads from direct andindirect heat treatments.2 To the best of our knowledge, nostudies comparing direct and indirect heat treatments oflactose-hydrolyzed milk, where lactose is removed andhydrolyzed after heat treatment, have been reported.The present study, which included several complementary

analyses, shows that the type of carbohydrate has a greaterimpact on the Maillard reaction, heat load indicators such asfurosine, and total concentration of ketones, compared to theimpact of different heat treatments (direct and indirect UHT).The formation of 2-methylbutanal and furosine was consid-erably higher in lactose-hydrolyzed milk compared withconventional UHT milk after a 9 month storage period. Inaddition, an increase in free amino acids and free aminesincreased tremendously in lactose-hydrolyzed milk, whereasthey remained constant in conventional UHT milk. Residualproteolytic activity from the lactase-enzyme preparation may beresponsible for the higher levels of free amino acids observed inlactose-hydrolyzed UHT milk. This study reveals that lactose-hydrolyzed UHT milk, compared to conventional UHT milk, ismore vulnerable to reactions that can impair the milk quality,primarily the Maillard reaction and proteolysis but, to someextent, also oxidation.

ASSOCIATED CONTENT

*S Supporting InformationOverview of storage and effects on the amount of volatilecompounds in lactose-hydrolyzed direct UHT milk (LHD),lactose-hydrolyzed indirect UHT milk (LHI), and conventionalindirect UHT milk (CONVI). This material is available free ofcharge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author*(H.C.B.) E-mail: [email protected]. Phone: +45871 58353.

Present Addresses(S.N.) University of Toronto, 555 University Avenue,Toronto, Ontario, Canada.(A.S.) O. Kavli AB, Armborstvagen 7, 12444 Alvsjo, Sweden.

FundingThe present study is part of the Ph.D. work of T.J. and wasfinancially supported by Arla Foods Amba, Food FutureInnovation (FFI), Danish Dairy Research Foundation, andAarhus University.

NotesThe authors declare no competing financial interest.

ACKNOWLEDGMENTS

We are thankful for the excellent technical assistance of BirgitteFoged and Hanne S. Møller.

ABBREVIATIONS USED

CONVI, conventional indirect UHT milk; DMDS, dimethyldisulfide; DMTS, dimethyl trisulfide; DSQ, dual-stage quadru-pole; ESL, extended shelf life; HCl, hydrochloric acid; HPLC,high-pressure liquid chromatography; GC-MS, gas chromatog-raphy−mass spectrometry; LHD, lactose-hydrolyzed directUHT milk; LHI, lactose-hydrolyzed indirect UHT milk;LME, linear mixed model; LSMEANS, least-squares means;LOD, limit of detection; LOQ, limit of quantification; NMR,nuclear magnetic resonance; TIC, total ion current; SIM,selected-ion monitoring; UHT, ultra-high temperature; UV,ultraviolet

REFERENCES(1) Datta, N.; Elliott, A. J.; Perkins, M. L.; Deeth, H. C. Aust. J. DairyTechnol. 2002, 57, 211−227.(2) Vazquez-Landaverde, P. A.; Velazquez, G.; Torres, J. A.; Qian, M.C. J. Dairy Sci. 2005, 88, 3764−3772.(3) Valero, E.; Villamiel, M.; Miralles, B.; Sanz, J.; Martinez-Castro, I.Food Chem. 2001, 72, 51−58.(4) Messia, M. C.; Candigliota, T.; Marconi, E. Food Chem. 2007,104, 910−917.(5) Ingram, C. J. E.; Mulcare, C. A.; Itan, Y.; Thomas, M. G.;Swallow, D. M. Hum. Genet. 2009, 124, 579−591.(6) Harju, M.; Kallioinen, H.; Tossavainen, O. Int. Dairy J. 2012, 22,104−109.(7) Tossavainen, O.; Kallioinen, H. Milchwissenschaft 2008, 63, 22−26.(8) Naranjo, G. B.; Pereyra Gonzales, A. S.; Leiva, G. E.; Malec, L. S.Food Chem. 2013, 141, 3790−3795.(9) Kato, Y.; Matsuda, T.; Kato, N.; Watanabe, K.; Nakamura, R. J.Agric. Food Chem. 1986, 34, 351−355.(10) Tossavainen, O.; Kallioinen, H. Milchwissenschaft−Milk Sci. Int.2007, 62, 410−415.(11) Reineccius, G. In Flavor Chemistry and Technology; Taylor &Francis Group: Boca Raton, FL, USA, 2006; pp 108−139.(12) Bylund, G. Dairy Processing Handbook, 2nd ed.; Tetra PakProcessing Systems: Malmo, Sweden, 2003; pp 227−246.(13) Jansson, T.; Jensen, S.; Eggers, N.; Clausen, M. R.; Larsen, L. B.;Ray, C.; Sundgren, A.; Andersen, H. J.; Bertram, H. C. Dairy Sci.Technol. 2014, 94, 311−325.(14) Dolan, J. W. LCGC North Am. 2009, 27, 224−230.(15) Dolan, B. J. W. LCGC North Am. 2009, 27, 306−312.(16) International Organization for Standardization, InternationalDairy Federation. International Standard ISO 18329, 2004.(17) Sundekilde, U. K.; Poulsen, N. A.; Larsen, L. B.; Bertram, H. C.J. Dairy Sci. 2013, 96, 290−299.(18) Dalsgaard, T. K.; Nielsen, J. H.; Larsen, L. B. Mol. Nutr. FoodRes. 2007, 51, 404−414.(19) Kato, Y.; Matsuda, T.; Kato, N.; Nakamura, R. J. Agric. FoodChem. 1988, 806−809.(20) Evangelisti, F.; Calcagno, C.; Nardi, S.; Zunin, P. J. Dairy Res.1999, 66, 237−243.(21) Balagiannis, D. P.; Parker, J. K.; Pyle, D. L.; Desforges, N.;Wedzicha, B. L.; Mottram, D. S. J. Agric. Food Chem. 2009, 57, 9916−9922.(22) Whitfield, F. B.; Mottram, D. S.; Brock, S.; Puckey, D. J.; Salter,L. J. J. Sci. Food Agric. 1988, 42, 261−272.



http://pubs.acs.org

mailto:[email protected]

(23) Berg, H. Reactions of Lactose during Heat Treatment of Milk: AQuantitative Study; Agricultural University: Wageningen, The Nether-lands, 1993; pp 1−176.(24) Davídek, T.; Devaud, S.; Robert, F.; Blank, I. J. Agric. FoodChem. 2006, 54, 6667−6676.(25) Brands, C. M.; van Boekel, M. a. J. Agric. Food Chem. 2001, 49,4667−4675.(26) Limacher, A.; Kerler, J.; Davidek, T.; Schmalzried, F.; Blank, I. J.Agric. Food Chem. 2008, 56, 3639−3647.(27) Perez Locas, C.; Yaylayan, V. A. J. Agric. Food Chem. 2004, 52,6830−6836.(28) Owczarek-Fendor, A.; De Meulenaer, B.; Scholl, G.; Adams, A.;Van Lancker, F.; Eppe, G.; De Pauw, E.; Scippo, M.-L.; De Kimpe, N.Food Chem. 2012, 133, 816−821.(29) Nawar, W. W.; Cancel, I. E.; Fagerson, I. S. J. Dairy Sci. 1962,45, 1172−1177.(30) Hawke, J. C. J. Dairy Res. 1966, 33, 225−243.(31) Hall, G.; Andersson, J.; Lingnert, H.; Olofsson, B. J. Food Qual.1985, 7, 153−190.(32) Jeon, I. J.; Thomas, E. L.; Reineccius, G. A. J. Agric. Food Chem.1978, 26, 1183−1188.(33) Osada, Y.; Shibamoto, T. Food Chem. 2006, 98, 522−528.(34) Gu, F.; Kim, J. M.; Hayat, K.; Xia, S.; Feng, B.; Zhang, X. FoodChem. 2009, 117, 48−54.(35) Clausen, M. R.; Skibsted, L. H.; Stagsted, J. J. Agric. Food Chem.2009, 57, 2912−2919.(36) Belitz, H.-D.; Grosch, W.; Schieberle, P. Food Chemistry, 4th ed.;Belitz, H.-D., Schieberle, P., Eds.; Springer: Berlin, Germany, 2001;Vol. 73, pp 158−402.(37) Zhang, Y.; Guo, S.; Liu, Z.; Chang, S. K. C. J. Agric. Food Chem.2012, 60, 7457−7462.(38) Morales, F. J.; Jimenez-perez, S. Food Chem. 2001, 72, 119−125.(39) Pripis-Nicolau, L.; de Revel, G.; Bertrand, A.; Maujean, A. J.Agric. Food Chem. 2000, 48, 3761−3766.(40) Al-Attabi, Z. Volatile Sulphur Compounds in UHT Milk; Schoolof Land, Crop, and Food Sciences, The University of Queensland: St.Lucia, QLD, Australia, 2009; Vol. 49.(41) Min, D. B.; Boff, J. M. Compr. Rev. Food Sci. Food Saf. 2002, 1,58−72.(42) Elliott, A. J.; Dhakal, A.; Datta, N.; Deeth, H. C. Aust. J. DairyTechnol. 2003, 58, 3−10.



107

Paper 3

Paper 3

Chemical and proteolysis-derived changes during long-term

storage of lactose-hydrolyzed UHT milk

Jansson, T., Jensen, H.B., Sundekilde, U.K., Clausen, M.R., Eggers, N., Larsen,

L.B., Colin, R., Andersen, H.J., Bertram, H.C.

Submitted to Journal of Agricultural and Food Chemistry

This document is confidential and is proprietary to the American Chemical Society and its authors. Do not copy or disclose without written permission. If you have received this item in error, notify the sender and delete all copies.

Chemical and proteolysis-derived changes during long-term

storage of lactose-hydrolyzed UHT milk

Journal: Journal of Agricultural and Food Chemistry

Manuscript ID: jf-2014-04104q

Manuscript Type: Article

Date Submitted by the Author: 27-Aug-2014

Complete List of Authors: Jansson, Therese; Aarhus University, Dept. Food Science Jensen , Hanne; Aarhus University, Department of Food Science - Milk and Egg Science

Sundekilde, Ulrik; Aarhus University, Dept. of Food Science Clausen, Morten; University of Aarhus, Department of Food Science Eggers, Nina; Aarhus University, Dept. Food Science Larsen, Lotte; Ministry of Food, Agric. & Fisheries, Dept of Animal Product Quality; Aarhus University, Department of Food Science - Milk and Egg Science Ray, Colin; Arla Foods Strategic Innovation Centre, Andersen, Henrik; Arla Foods Ingredients, Bertram, Hanne; University of Aarhus, Department of Food Science; University of Aarhus, Department of Food Science

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Chemical and proteolysis-derived changes during long-term storage of lactose-hydrolyzed UHT milk

Therese Jansson1, Hanne B. Jensen2, Ulrik K. Sundekilde1, Morten R. Clausen1, Nina Eggers1, Lotte

B. Larsen2, Colin Ray3, Henrik J. Andersen4, Hanne C. Bertram1*

1Department of Food Science, Aarhus University, Kirstinebjergvej 10, DK-5792 Aarslev, Denmark.

2Department of Food science, Aarhus University, Blichers Allé 20, DK-8830 Tjele, Denmark.

3Arla Foods Strategic Innovation Centre, Lindhagensgatan 126, 10546 Stockholm, Sweden.

4 Arla Foods Ingredients, Sønderhøj 10-12, DK-8260 Viby J, Denmark.

*Corresponding author. Email: [email protected], +45 871 58353

Page 1 of 33



2

Abstract 1

Proteolytic activity in milk may release bitter-tasting peptides and generate free amino terminals 2

that react with carbohydrates which initiate the Maillard reaction. Ultra high temperature (UHT) 3

heat treatment inactivates the majority of proteolytic enzymes in milk, however, in lactose-4

hydrolyzed milk a β-galactosidase preparation is applied to the milk after heat treatment. This 5

enzyme preparation has proteolytic side-activities that may induce quality deterioration of long-6

term stored milk. In the present study proteolysis, glycation and volatile compounds formation were 7

investigated in conventional (100% lactose), filtered (60% lactose) and lactose-hydrolyzed (<1% 8

lactose) UHT milk using reverse phase-high pressure liquid chromatography-mass spectrometry, 9

proton nuclear magnetic resonance, and gas chromatography-mass spectrometry. Proteolysis was 10

observed in all milk types, however, the degree of proteolysis was significantly higher in the 11

lactose-hydrolyzed milk compared to the conventional and filtered milk. The proteins most prone to 12

proteolysis were β-CN and αs1-CN, which was clearly hydrolyzed after approximately 90 days of 13

storage in the lactose-hydrolyzed milk. 14

15

Keywords: Lactose-free milk, proteolysis, heat treatment, lactose hydrolysis, Maillard reaction, 16

Shelf-life 17

18

Page 2 of 33



3

Introduction 19

During storage of ultra-high temperature (UHT) heated milk, the Maillard reaction, oxidation and 20

proteolysis proceeds and changes the chemical composition of the milk. It has been shown that 21

lactose-hydrolyzed milk manufacturing favors initiation of the Maillard reaction compared to 22

conventional UHT milk.1–3 This fact can be ascribed to a higher reactivity of glucose and galactose 23

with the proteins in the milk (mainly lysine) as compared with lactose.4,5 Moreover, a higher 24

proteolytic activity has been observed in lactose-hydrolyzed milk,6,7 which through the formation of 25

peptides and amino acids likewise favors Maillard reactions and add to flavor-active compounds 26

and thereby contribute to the formation of a stale aroma and flavor.8–11 Heat treatment decreases the 27

proteolysis in milk due to denaturation of the proteolytic enzymes. However, during long-term 28

storage, proteolytic activity may deteriorate the milk even in products that have been exposed to a 29

UHT treatment.12 The activity of plasmin may cause bitterness, gelation and precipitation during 30

long-term storage which decreases the shelf-life of UHT milk.13 In lactose-hydrolyzed milk β-31

galactosidase is added before or after the heat treatment of the milk to hydrolyze the lactose. The β-32

galactosidase enzyme preparations may contain undesired proteolytic side-activities especially if the 33

enzyme is added after heat processing of the milk, since the enzymes are not inactivated and the 34

milk is stored for several months at room temperature. Few studies have investigated the proteolytic 35

activity in lactose-hydrolyzed UHT and only for up to 3 months of storage.6,7,14,15 Consequently, a 36

detailed understanding of the chemical changes in lactose-hydrolyzed milk during long-term storage 37

is lacking. The aim of the present study was to get a further understanding of the chemical changes 38

taking place in enzymatically lactose-hydrolyzed UHT milk and conventional UHT milk during a 9-39

months storage period. This was achieved through an elucidation of the concomitant proteolysis, 40

glycation and volatile compounds formation by using a combination of reverse liquid 41

chromatography-mass spectrometry (RP-HPLC MS), proton nuclear magnetic resonance (1H NMR) 42

spectroscopy, and gas chromatography-mass spectrometry (GCMS). 43

Page 3 of 33



4

Material and methods 44

Chemical references 45

Bis-Tris, D2O containing 0.025% sodium trimethylsilyl-[2,2,3,3-2H4]-1-propionate (TSP), 1,4-46

dithioerythritol, Guanidine hydrochloride, Sodium-citrate dehydrate, Trifluoroacetic acid were 47

obtained from Sigma-Aldrich Inc. (Steinheim, Germany). Acetonitril was purchased from Mikrolab 48

RA 1016 (Rathburn Chemicals, Ltd. Walkerburn, UK) 49

Milk samples 50

Four types of milk (conventional indirect UHT, lactose-hydrolyzed direct UHT, lactose-hydrolyzed 51

indirect UHT, and filtered (lactose-reduced) indirect UHT) were produced at a commercial Danish 52

dairy plant (Arla Foods, Esbjerg, Denmark) (Table 1). All the milk were UHT heated using indirect 53

heat treatment where a tubular heat exchanger was used to heat the milk, followed by slow cooling 54

(140-145 °C for 2-3 sec.).16 An additional lactose-hydrolyzed milk batch was heat-treated using 55

direct heat treatment where steam was injected into the milk, followed by flash cooling (140-145 °C 56

for 2-3 sec.).16 Ultra and nano-filtration were applied to reduce the lactose content to approximately 57

60% in the lactose-hydrolyzed and the filtered milk before heat treatment. The remaining lactose 58

was enzymatically hydrolyzed after heat treatment, by addition of β-galactosidase (Maxilact LAG 59

5000, DSM, Netherlands). All milk samples were packaged aseptically in 1 Liter Tetra Brik® 60

cartons. 61

62

The milk samples were stored in an air-conditioned room at 22 °C. After 14, 21, 28, 45, 60, 90, 120, 63

150, 180, 210, 240, 270 days of storage one carton of each milk type was opened for analyses. Milk 64

samples for 1H NMR spectroscopic analysis were filtered using Amicon Ultra 0.5 mL 10 kDa 65

(Millipore, MA) spin filters at 10,000 g for 15 min and subsequently frozen at -80 °C in 2 mL tubes. 66

Milk samples for RP-HPLC MS analysis were skimmed by centrifugation (Sorvall® RC 5B plus, 67

Page 4 of 33



5

Thermo, Waltham, Massachusetts, US) at 12800 g for 30 minutes, 4 °C, and subsequently frozen at 68

-80 °C in 2 mL tubes. 69

Aroma profiling by dynamic headspace sampling and GCMS 70

Dynamic headspace sampling and gas chromatography-mass spectrometry (DHS GCMS) was used 71

to extract and analyze the volatile compounds in the fresh, unfrozen milk according to the procedure 72

described by Jansson et al. 2014.17 Adsorbent traps (Tenax TA, 200 g Tenax TA, 35/60 mesh 73

(Markes International Limited, LIantrisant, UK)) were used to isolate the volatile compounds which 74

were then desorbed (250 °C, 15 min) onto a cold trap (U-TIIGCP, Markes International Limited) 75

using a thermal desorption system (ultra-UNITY™, Markes International Limited). Subsequently, 76

the volatiles were injected into a gas chromatograph (Finnigan trace GC ultra, Thermo, Waltham, 77

Massachusetts, US) and separated using a CP-Wax 52CB column (50 m x 0.25 mm, film thickness 78

0.25 µm, (Varian, Palo Alto, CA)). The separated volatile compounds were detected using a single 79

quadrupole mass spectrometer (Finnigan Trace dual-stage quadrupole (DSQ), Thermo). The mass 80

spectrometer scanned over a mass to charge (m/z) range of 45-650 (4.46 scans/sec) in TIC mode. In 81

total 28 compounds were identified using the NIST mass spectral library (NIST MS search 2.0, 82

2011) and external standards. Four of these: dimethyldisulfide (DMDS), dimethyltrisulfide 83

(DMTS), 2-ethylfuran, 2-pentylfuran) were quantified using external standard curves as previously 84

described.1 85

Metabolite profiling by 1H NMR spectroscopy 86

Prior to proton nuclear magnetic resonance (1H NMR) spectroscopy, the milk filtrates were thawed 87

in randomized order and analyzed as described previously by Jansson et al 20141: Four hundred µL 88

milk was mixed with 200 µL D2O containing 0.025% TSP as an internal standard. The samples 89

were analyzed on a Bruker Avance III 600 spectrometer (Bruker Biospin, Germany) equipped with 90

a 5 mm TXI probe. From a single 90° pulse experiment with a relaxation delay of 5 s standard one-91

dimensional spectra were obtained. The water peak was suppressed by irradiation during the 92

Page 5 of 33



6

relaxation delay, and a total of 64 scans were collected into 32K data points scanning with a spectral 93

width of 12.15 ppm. The data were multiplied by a 0.3 Hz line-broadening function and 94

subsequently Fourier transformed. The TSP signal at 0 ppm was used as a reference, and all the 1H 95

spectra were manually baseline and phase corrected using Topspin 3.2™ (Bruker Biospin). The 96

sugar regions at 5.29-5.22, 5.0-4.49 and 4.11-3.32 ppm and the region with citrate at 2.74-2.47 ppm 97

were removed from the data to facilitate the interpretation of the other metabolites. Metabolite 98

concentrations were calculated for alanine, isoleucine, leucine, lysine, methionine, tryptophan, 99

tyrosine, valine, formate lactose, glucose and galactose, using Chenomx NMR suite 7.7 (Chenomx, 100

Edmonton, Canada). 101

Protein profiling and determination of glycation by RP-HPLC MS 102

The RP-HPLC MS procedure used in the present study was modified according to Jensen et al. 103

(2012)18 and the procedures for sample preparation, settings for the HPLC, ESI source, and the 104

mass selective detector were as described by Frederiksen et al. (2011).19 Proteins were separated by 105

reversed phase chromatography using an Agilent 1100 system (Agilent Technologies, Santa Clara, 106

CA) with a Jupiter C4 column (250 mm x 2 mm, 5µM particle size, 300 Å pores; Phenomenex, 107

Torrance, CA) operated at 40 °C and a G1315A diode array detector with Ultra Violet (UV) 108

detection at 214 nm coupled to a mass selective detector for identification and relative 109

quantification. Average molecular masses (Mw) of the milk proteins were obtained using the 110

deconvolution algorithm of the ChemStation software (rev.B.04.01 SP [650], Agilent 111

Technologies). All milk samples were analyzed in technical duplicates. The relative protein content 112

of the intact milk proteins was calculated as the integrated area of each intact protein normalized 113

with total integrated peak area in each LC-chromatogram. ‘Total CN’ comprised the sum of αS1-114

CN, αS2-CN, β-CN, and κ-CN. 115

Page 6 of 33



7

Statistical analysis 116

Prior to multivariate data analysis 1H NMR spectroscopic data were mean-centered and Pareto-117

scaled. Principal component analysis (PCA) was applied to the centered and scaled data to elucidate 118

differences between the milk types. A partial least square (PLS) regression model was used to 119

predict the storage time, and a corresponding variable important in projection (VIP) plot was 120

applied to identify the metabolites correlating strongest with the storage-induced changes. The PLS 121

model was cross-validated (CV) using segmented CV with nine splits, where it was ensured that 122

each sample replicate was in the same split. The root mean square error of cross validation 123

(RMSECV) was used as a measure of model performance. PCA was performed using SIMCA P+ 124

13.0 (Umetrics, Umeå, Sweden) and PLS was performed using PLS Toolbox 6.71 (Eigenvector 125

Research Inc. Wenatchee, WA) in MATLAB 7.11 (Mathworks, Natick, MA). 126

To test the effect of milk type and storage time on the quantified proteins, amino acids, formate, 127

carbohydrates and the Strecker degradation compounds linear mixed models (lme), least square 128

means (lsmeans), and Tukeys honest significant differences test (α = 0.05) were calculated in R 129

(version 3.0.2, R Foundation for statistical Computing, Vienna, Austria). 130

Page 7 of 33



8

Results 131

Chemical changes in UHT-heated milk stored at 22 °C were followed over a period of nine months. 132

1H NMR spectroscopy was applied to follow proteolysis through the detection of free amino acids. 133

In total, nine free amino acids could be quantified (Table 2) and most of them increased 134

significantly in the lactose-hydrolyzed milk during storage (Figure 1), while they remained 135

constant or were not detected (phenylalanine, tryptophan and tyrosine) in the filtered and 136

conventional UHT milk (Figure 1). Only exceptions were methionine which was constant 137

throughout the storage period and lysine which increased in all milk types. 138

139

In order to investigate which proteins that gave rise to the increased level of amino acids in the 140

lactose-hydrolyzed UHT milk, the intact proteins in the milk were analyzed using RP-HPLC MS. 141

Proteolysis was observed for all proteins in all milk types during storage. However, the level of 142

proteolysis was significantly higher in the lactose-hydrolyzed UHT milk compared to the filtered 143

and conventional UHT milk (Table 2). The relative amount of total intact CN was reduced by 56% 144

for lactose-hydrolyzed indirect UHT-heated milk and by 61% for lactose-hydrolyzed direct UHT-145

heated milk during the storage period, as compared to 15% for conventional and filtered indirect 146

UHT-heated milk (Table 2). The two most extensively hydrolyzed proteins during storage were β-147

CN and αS1-CN (Figure 2), which were reduced by approximately 80% and 45%, respectively 148

(Table 2), in lactose-hydrolyzed milk compared to a more moderate reduction for conventional and 149

filtered milk. 150

κ-CN and β-LG decreased in all milk types, (Table 2), while αs2-CN only decreased in the filtered 151

and lactose-hydrolyzed UHT milk. 152

The identified amino acids and their involvement in Strecker degradation product formation is 153

summarized in Table 3. Five putative Strecker degradation products were identified (Table 2) 154

however no clear connection between the volatile compounds and the increased levels of free amino 155

Page 8 of 33



9

acids could be observed. Likewise no clear effects of lactose hydrolysis, filtration or heat treatment 156

on volatile compound formation were revealed (Table 2). 157

Using RP-HPLC MS we detected glycated proteins in all milk samples at all storage times as N-158

linked glycans with mass addition of 162 Da for one hexose and 324 Da for lactose, or multiple 159

hexoses. Furthermore hydrophilic glycoproteins were identified as the first peak eluting from the 160

RP-HPLC column. This peak increased significantly during storage in all milk samples, and this 161

increase was highest in indirect heated lactose-hydrolyzed milk (Table 2). Even though protein 162

glycation was observed, the level of free carbohydrates remained constant throughout the storage 163

time. Though, after 240 days of storage a decrease in glucose and galactose content was observed in 164

all lactose-hydrolyzed milk samples (data not shown), which is consistent with a decrease in the 165

concentration of free amino acids, as shown in Figure 1. 166

Finally, formate can be formed from the deoxyosones and may therefore be an indicator of the 167

Maillard reaction. It was found that initial formate concentrations were highest in conventional 168

milk, however during storage similar formation rates were observed for all milk types, except for 169

the filtered milk (Table 2). 170

171

Further, the 1H NMR spectroscopic data were analyzed using an untargeted approach in order to 172

elucidate how storage affected overall metabolite composition. PCA of 1H NMR spectroscopic data 173

showed a separation along PC1 and PC2 between the filtered (lactose-hydrolyzed and filtered UHT 174

milk) and non-filtered (conventional UHT milk) samples, probably due to the dilution occurring in 175

this step. In fresh samples (< 50 days old), no clear separation between lactose-hydrolyzed milk and 176

filtered milk could be observed. In addition, along PC1 a pronounced effect of storage was observed 177

for lactose-hydrolyzed milk, while for the conventional and filtered milk the storage effect was 178

much smaller. A PLS regression model with 1H NMR spectroscopic variables as X and storage time 179

as Y, was generated for each milk type (Figure 4A, 4C, 4E). The models revealed that the storage 180

Page 9 of 33



10

time of lactose-hydrolyzed and filtered milk could be predicted within a variance of 11 days (R2 CV 181

was 0.99 and 0.98 respectively; RMSECV was 11.23 days and 11.45 days respectively), and the 182

storage time of conventional milk could be predicted within a variance of 19 days (R2 CV was 0.95 183

and RMSECV was 18.59 days). In order to find the metabolites responsible for the changes during 184

storage VIP scores were plotted for each PLS model (Figure 4B, 4D, 4F). Choline was significant 185

in all milk types, whereas amino acids (leucine, isoleucine, valine, alanine, phenylalanine), as 186

expected, were much more important in lactose-hydrolyzed milk as compared to the other milk 187

types. In the non-hydrolyzed milk samples hippurate, creatine/creatinine, N-acetyl carbohydrates 188

and acetate were identified as significant. 189

190

Page 10 of 33



11

Discussion 191

Storage stability is of importance in UHT milk to maintain the consumer acceptability of the milk. 192

In the present study, chemical changes that could be detrimental to milk quality were investigated in 193

UHT milk produced with and without lactose hydrolysis during a storage period of nine months. 194

Furthermore, filtered indirectly heated UHT milk was included to follow the effect of the β-195

galactosidase added to the milk. Two different heat treatments (direct and indirect) applied to the 196

lactose-hydrolyzed milk were included to investigate the effect of severity of heat treatment. 197

However, the type of heat treatment did not have any effect on volatile compounds, metabolite 198

profile or the proteolytic activity, and therefore the term lactose-hydrolyzed milk refers to both milk 199

types. 200

Proteolysis has major effect in lactose-hydrolyzed milk 201

1H NMR spectroscopy was used to identify metabolites in the milk, and PCA of 1H NMR 202

spectroscopic data revealed much more severe storage-induced changes in lactose-hydrolyzed milk 203

than in the conventional and filtered UHT milk (Figure 3). This result shows that major chemical 204

changes are not induced by the filtration but rather they can be attributed to the addition of the β-205

galactosidase preparation. The PLS models calculated (Figure 4) could predict the time of storage 206

of each milk with a high precision, as reflected in the low RMSECV values (11-19 days). However, 207

the prediction was considerably better for lactose-hydrolyzed UHT milk than conventional UHT 208

milk which supports the more severe storage-induced changes in the lactose-hydrolyzed UHT milk 209

seen in the PCA. It is highly relevant to investigate these changes in more detail to obtain an 210

increased understanding of the underlying mechanisms. In the present study proteolysis was the 211

major chemical change observed. Interestingly lysine was the only amino acid that increased in all 212

milk types during storage and this could be due to the presence of heat stable endogenous proteases 213

which cleave proteins after lysine and arginine (Lys or Arg in P1 position).20–23 A decrease in lysine 214

could be expected due to its involvement in the Maillard reaction. However, according to the results 215

Page 11 of 33



12

in the present study the binding of carbohydrates to lysine seems to be of minor importance 216

compared to the proteolytic activity. 217

The increased proteolytic activity was further elucidated by RP-HPLC MS by investigating the 218

content of intact protein during storage. Our study showed that the caseins present in the lactose-219

hydrolyzed UHT milk were considerably affected by the proteolytic activity, while in the 220

conventional UHT milk only a moderate proteolytic activity could be observed during storage. This 221

is in agreement with the results from Tossavainen and Kallioinen (2007) who investigated lactose-222

hydrolyzed and conventional UHT milk stored at different temperatures (5, 22, 30 and 45° C) for 12 223

weeks.6 In addition, Tossavainen and Kallioinen (2007) found that the αs1-, αs2- and β-CN were the 224

most extensively degraded proteins, while β-LG, α-LA and κ-casein were less affected.14 The 225

proteolytic activity in the filtered milk did not deviate from the conventional UHT milk, which 226

indicates that no proteolysis originated from the filtration step. The most hydrolyzed proteins in the 227

lactose-hydrolyzed UHT milk were β-CN and αs1-CN, which significantly decreased after 90 days 228

of storage, while only a minor decrease was observed in the conventional and filtered UHT milk 229

(Table 2, Figure 2). This is an important and interesting pattern depicting differences between the 230

lactose-hydrolyzed and conventional UHT milk during storage. Another interesting observation was 231

that the free amino acids increased immediately during storage while the reduction in β-CN and αs1-232

CN was not observed before after 90 days in the lactose-hydrolyzed milk. These findings could 233

indicate that a low degree of hydrolysis of proteins gives rise to an increase in the level of free 234

amino acids but do not affect the proteins sufficiently to be detectable by RP HPLC MS. 235

The decrease in β-CN and αs1-CN after approximately 90 days of storage (Figure 2) seems to relate 236

with the development of an off-flavor, which typically develops after 90 days.24 Casein contains a 237

high amount of hydrophobic amino acids, which are highly correlated to bitterness,25,26 and a 238

proteolysis of casein and especially β-CN and αs1-CN would consequently result in a bitter taste.27–239

30 McKellar et al. (1984) investigated proteolysis and bitterness in UHT milk during 8 months 240

Page 12 of 33



13

storage at ambient temperature and observed a high correlation between storage time, proteolysis, 241

and bitterness.8 In addition, McKellar et al. (1984) found a higher level of proteolysis and an earlier 242

formation of bitter flavor in direct heated UHT milk compared to the indirect heated UHT milk.8 243

The Maillard reaction in the lactose-hydrolyzed milk 244

The Maillard reaction is dependent on the reaction between amines (primarily protein-bound lysine 245

and N-terminals) and reducing carbohydrates, and the type and concentration of these substrates are 246

highly important for the level of Maillard reaction product formation.1,2 In the present study protein 247

glycation, formate and Strecker degradation product formation were investigated in filtered, lactose-248

hydrolyzed and conventional UHT milk. The relative content of the glycoprotein increased during 249

storage for all milk types (Table 2), although the absolute degree of glycation could not be 250

quantified based on the RP-HPLC MS method applied. This is in contrast to the study by Carulli et 251

al. (2011) who observed a higher degree of glycation of α-LA and β-LG in presence of glucose and 252

galactose compared to lactose.5 Likewise, we were not able to detect a significant difference in the 253

degree of formate or in the Strecker degradation product formation analyzed (2-pentylfuran, 2-254

methylfuran, DMDS, DMTS and benzaldehyde) during storage. 255

Strecker aldehydes have been shown to affect the milk aroma31,32 and both lactose hydrolysis and 256

proteolysis were expected to promote the formation of Strecker aldehydes in lactose-hydrolyzed 257

milk as compared to conventional UHT milk. However the lack of significant differences could 258

indicate that the Maillard reaction may not the most important pathway for their formation.33,34 For 259

instance benzaldehyde was observed in all milk types, while its precursor, phenylalanine,35 was 260

only detected in lactose-hydrolyzed milk (Table 2). Valero et al. (2001) observed an increase of 261

benzaldehyde in skim milk during storage.34 However, the present study could not support this 262

observation. 263

Page 13 of 33



14

We previously observed a striking correlation between 2-methylbutanal, isoleucine and furosine 264

formation in lactose-hydrolyzed UHT milk after approximately 90 days of storage.1 This seems to 265

be closely connected to the decrease in intact casein concentrations observed here. Hence, a 266

possible link between proteolysis, glycation and Strecker aldehyde formation seems likely, though 267

we cannot prove causality. 268

269

A final aspect that might affect the quality of lactose-hydrolyzed milk is the dilution that occurs 270

during the filtration step. Due to the filtration lactose-hydrolyzed and filtered UHT milk were heat-271

treated with a 40% lower level of lactose compared to the conventional milk. This was reflected by 272

a lower formate level in fresh lactose-hydrolyzed and filtered milk (Table 2), but did not result in 273

different rates of formate formation during storage. The most significant pathway for formate 274

generation is through α-dicarbonyls generated in the Maillard reaction36,37 but our results indicate 275

that the type of sugar as well as proteolysis are not decisive factors for the formation of formate in 276

milk. 277

The present study shows that the side-effects of the added β-galactosidase are highly relevant for 278

the chemical reactions taking place in the lactose-hydrolyzed milk during storage. Consequently, an 279

alternative enzyme preparation method would be attractive for the production of lactose-hydrolyzed 280

milk in order to decrease the observed problems. A possible alternative could be to add the enzyme 281

before the heat treatment to inactivate all the enzymes. Tossavainen and Kallioinen (2007)6 282

investigated how heat treatment before and after addition of β-galactosidase to lactose-hydrolyzed 283

milk affects the proteolytic activity, but no significant difference was observed between the 284

different enzymatic treatments. In milk hydrolyzed prior to the heat treatment, a higher 285

concentration of β-galactosidase is added compared to the milk hydrolyzed after the heat treatment 286

to obtain the same hydrolytic effect. However, the implication of this is that a higher amount of 287

Page 14 of 33



15

enzyme is active after heat treatment. Another alternative, and possibly a more suitable method 288

would be to avoid the side-activity from the enzyme by applying a more pure enzyme preparation to 289

the milk, and thereby decreasing the proteolysis in lactose-hydrolyzed milk. This calls for further 290

investigations into the commercial β-galactosidase enzymes used during production of lactose-291

hydrolyzed products. 292

293

The generation of free amino acids during storage of lactose-hydrolyzed milk originates most likely 294

from either enzymatic impurity in the used β-galactosidase preparation and/or side-activities of the 295

β-galactosidase. Higher levels of proteolytic activity contribute to the flavor of the milk through 296

formation of flavor-active amino acids/peptides and/or through formation of additional Maillard 297

reaction products, e.g. Strecker aldehydes. The reactivity of glucose and galactose determines the 298

rate of the Maillard reaction, however, in addition, our results indicate that a sufficient level of free 299

amino acids is required in order to accelerate the Maillard reaction. In order to gain an improved 300

understanding of the chemical changes observed in lactose-hydrolyzed milk it would be relevant to 301

focus on the proteins β-CN and αs1- CN, which are most affected by the proteolytic activity, and 302

analysis of the peptide profile of the hydrolyzed proteins would be relevant to identify the enzymes 303

responsible for the proteolysis. In addition, investigations on the enzyme purity and how it affects 304

the chemical changes observed in the lactose-hydrolyzed UHT milk would probably facilitate the 305

development of lactose-hydrolyzed products with a long shelf-life. 306

Abbreviations 307

CN, Casein; CV, cross-validated; Da, Dalton; DMDS, Dimethyl disulfide; DMTS, Dimethyl 308

trisulfide; DSQ, Dual-stage quadrupole; RP HPLC, Reversed phase high pressure-liquid 309

chromatography; GCMS, Gas chromatography-mass spectrometry; LA, Lactalbumin; LG, 310

Lactoglobulin; LME, Linear mixed model; LSMEANS, Least square means; MS, Mass 311

Page 15 of 33



16

spectrometry; NMR, Nuclear magnetic resonance; PC, Principal component; PCA, Principal 312

component analysis; PLS, Partial least square; RMSECV, Root mean square error of cross 313

validation; TIC, Total ion current; TSP, sodium trimethylsilyl-[2,2,3,3-2H4]-1-propionate; UHT, 314

Ultra-high temperature; UV, Ultraviolet; VIP, Variable importance in projection; 315

Page 16 of 33



17

Acknowledgment 316

The authors are thankful for the excellent technical assistance of Birgitte Foged and Hanne S. 317

Møller. The present study is part of the Ph.D. work of Therese Jansson and was financially 318

supported by Arla Foods Amba, Food Future Innovation (FFI), Danish Dairy Research Foundation, 319

and Aarhus University. 320

321

Page 17 of 33



18

References 322

(1) Jansson, T.; Clausen, M. R.; Sundekilde, U. K.; Eggers, N.; Nyegaard, S.; Larsen, L. B.; Ray, 323

C.; Sundgren, A.; Andersen, H. J.; Bertram, H. C. Lactose-hydrolyzed milk is more prone to 324

chemical changes than conventional ultra-high-temperature (UHT) milk. J. Agric. Food 325

Chem. 2014, 62, 7886–7896. 326

(2) Messia, M. C.; Candigliota, T.; Marconi, E. Assessment of quality and technological 327

characterization of lactose-hydrolyzed milk. Food Chem. 2007, 104, 910–917. 328

(3) Tossavainen, O.; Kallioinen, H. Effect of lactose-hydrolysis on furosine and available lysine 329

in UHT skim milk. Milchwissenschaft 2008, 63, 22–26. 330

(4) Kato, Y.; Matsuda, T.; Kato, N.; Nakamura, R. Browning and protein polymerization 331

induced by amino-carbonyl reaction of ovalbumin with glucose and galactose. J. Agric. Food 332

Chem. 1988, 806–809. 333

(5) Carulli, S.; Calvano, C. D.; Palmisano, F.; Pischetsrieder, M. MALDI-TOF MS 334

characterization of glycation products of whey proteins in a glucose/galactose model system 335

and lactose-free milk. J. Agric. Food Chem. 2011, 59, 1793–1803. 336

(6) Tossavainen, O.; Kallioinen, H. Proteolytic changes in lactose hydrolyzed UHT milks during 337

storage. Milchwissenschaft-Milk Sci. Int. 2007, 62, 410–415. 338

(7) Evangelisti, F.; Calcagno, C.; Nardi, S.; Zunin, P. Deterioration of protein fraction by 339

Maillard reaction in dietetic milks. J. Dairy Res. 1999, 66, 237–243. 340

Page 18 of 33



19

(8) McKellar, R. C.; Froehlich, D. A.; Butler, G.; Cholette, H.; Campbell, C. The effect of 341

uncooled storage on proteolysis bitterness and apparent viscosity in ultra high temperature 342

milk. Can. Inst. Food Sci. Technol. J. 1984, 17, 14–17. 343

(9) Van Boekel, M. A J. S. Formation of flavor compounds in the Maillard reaction. Biotechnol. 344

Adv. 2006, 24, 230–233. 345

(10) Rerkrai, S.; Jeon, I. J.; Bassette, R. Effect of various direct ultra-high temperature heat 346

treatments on flavor of commercially prepared milks. J. Dairy Sci. 1987, 70, 2046–2054. 347

(11) Kokkinidou, S.; Peterson, D. G. Control of Maillard-type off-flavor development in 348

ultrahigh-temperature-processed bovine milk by phenolic chemistry. J. Agric. Food Chem. 349

2014, 62, 8023–8033. 350

(12) Enright, E.; Bland, A. P.; Needs, E. C.; Kelly, A. L. Proteolysis and physicochemical 351

changes in milk on storage as affected by UHT treatment, plasmin activity and KIO3 352

addition. Int. Dairy J. 1999, 9, 581–591. 353

(13) Rauh, V. M.; Johansen, L. B.; Ipsen, R.; Paulsson, M.; Larsen, L. B.; Hammershøj, M. 354

Plasmin activity in UHT milk: Relationship between proteolysis, age gelation, and bitterness. 355

J. Agric. Food Chem. 2014, 62, 6852–6860. 356

(14) Tossavainen, O.; Kallioinen, H. Furosine formation and proteolytic changes in carbohydrate 357

reduced UHT milks. Milchwissenschaft 2008, 63, 254–258. 358

(15) Mittal, S. B.; Newell, G.; Hourigan, J. A.; Zadow, J. G. The effect of protease 359

contaminations in lactase in the flavour of lactose-hydrolyzed milks. Aust. J. dairy Technol. 360

1991, 46, 46–47. 361

Page 19 of 33



20

(16) Tetra Pak. In:Dairy processing handbook; Tetra Pak, Ed.; 2nd ed.; Tetra Pak processing 362

systems AB: Malmö, Sweden, 2003; pp. 227–246. 363

(17) Jansson, T.; Jensen, S.; Eggers, N.; Clausen, M. R.; Larsen, L. B.; Ray, C.; Sundgren, A.; 364

Andersen, H. J.; Bertram, H. C. Volatile component profiles of conventional and lactose-365

hydrolyzed UHT milk- a dynamic headspace gas chromatography-mass spectrometry study. 366

Dairy Sci. Technol. 2014, 94, 311–325. 367

(18) Jensen, H. B.; Holland, J. W.; Poulsen, N. a; Larsen, L. B. Milk protein genetic variants and 368

isoforms identified in bovine milk representing extremes in coagulation properties. J. Dairy 369

Sci. 2012, 95, 2891–2903. 370

(19) Frederiksen, P. D.; Andersen, K. K.; Hammershøj, M.; Poulsen, H. D.; Sørensen, J.; Bakman, 371

M.; Qvist, K. B.; Larsen, L. B. Composition and effect of blending of noncoagulating, poorly 372

coagulating, and well-coagulating bovine milk from individual Danish Holstein cows. J. 373

Dairy Sci. 2011, 94, 4787–4799. 374

(20) Sørhaug, T.; Stepaniak, L. Psychrotrophs and their enzymes in milk and dairy products: 375

Quality aspects. Trends food Sci. Technol. 1997, 8, 35–41. 376

(21) Hinz, K.; Larsen, L. B.; Wellnitz, O.; Bruckmaier, R. M.; Kelly, A L. Proteolytic and 377

proteomic changes in milk at quarter level following infusion with Escherichia coli 378

lipopolysaccharide. J. Dairy Sci. 2012, 95, 1655–1666. 379

(22) Wedholm, A.; Møller, H. S.; Lindmark-Månsson, H.; Rasmussen, M. D.; Andrén, A.; Larsen, 380

L. B. Identification of peptides in milk as a result of proteolysis at different levels of somatic 381

cell counts using LC MALDI MS/MS detection. J. Dairy Res. 2008, 75, 76–83. 382

Page 20 of 33



21

(23) Crudden, A.; Fox, F.; Kelly, A. L. Factors affecting the hydrolytic action of plasmin in milk. 383

Int. Dairy J. 2005, 15, 305–313. 384

(24) Jensen, S.; Jansson, T.; Eggers, N.; Clausen, M. R.; Jensen, H. B.; Larsen, L. B.; Ray, C.; 385

Sundgren, A.; Andersen, H. J.; Bertram, H. C. Storage-induced changes in the sensory 386

characteristics and volatiles of conventional and lactose-hydrolyzed UHT processed milk. 387

Food Res. Int. 2014. Submitted. 388

(25) Guigoz, Y.; Solms, J. Bitter peptides, occurrence and structure. Chem. Senses 1976, 2, 71–389

84. 390

(26) Ney, K. H. Bitterness of peptides: Amino acid composition and chain length. ACS Symp. Ser. 391

1979, 115, 149–173. 392

(27) Kilara, A.; Panyam, D. Peptides from milk proteins and their properties. Crit. Rev. Food Sci. 393

Nutr. 2003, 43, 607–633. 394

(28) Gomez, B. M. J.; Garde, S.; Gaya, P.; Medina, M.; Nun, M. Relationship between level of 395

hydrophobic peptides and bitterness in cheese made from pasteurized and raw milk. J. Dairy 396

Res. 1997, 64, 289–297. 397

(29) Harwalkar, V. R.; Cholette, H.; McKellar, R. C.; Emmons, D. B. Relation between 398

proteolysis and astringent off-flavor in milk. J. Dairy Sci. 1993, 76, 2521–2527. 399

(30) Karametsi, K.; Kokkinidou, S.; Ronningen, I.; Peterson, D. G. Identification of bitter 400

peptides in aged cheddar cheese. J. Agric. Food Chem. 2014, 62, 8034–8041. 401

Page 21 of 33



22

(31) Pripis-Nicolau, L.; de Revel, G.; Bertrand, A; Maujean, A. Formation of flavor components 402

by the reaction of amino acid and carbonyl compounds in mild conditions. J. Agric. Food 403

Chem. 2000, 48, 3761–3766. 404

(32) Hofmann, T.; Schieberle, P. Formation of aroma-active Strecker-aldehydes by a direct 405

oxidative degradation of Amadori compounds. J. Agric. Food Chem. 2000, 48, 4301–4305. 406

(33) Hougaard, A. B.; Vestergaard, J. S.; Varming, C.; Bredie, W. L. P.; Ipsen, R. H. Composition 407

of volatile compounds in bovine milk heat treated by instant infusion pasteurisation and their 408

correlation to sensory analysis. Int. J. Dairy Technol. 2011, 64, 34–44. 409

(34) Valero, E.; Villamiel, M.; Miralles, B.; Sanz, J.; Martinéz-Castro, I. Changes in flavour and 410

volatile components during storage of whole and skimmed UHT milk. Food Chem. 2001, 72, 411

51–58. 412

(35) Adamiec, J.; Rössner, J.; Velíšek, J. Minor Strecker degradation products of phenylalanine 413

and phenylglycine. Eur. Food Res. Technol. 2001, 212, 135–140. 414

(36) Smuda, M.; Glomb, M. A. Fragmentation pathways during Maillard-induced carbohydrate 415

degradation. J. Agric. Food Chem. 2013, 61, 10198–10208. 416

(37) Brands, C. M.; van Boekel, M. A. J. S., Reaction of monosaccharides during heating of 417

sugar-casein systems: building of a reaction network model. J. Agric. Food Chem. 2001, 49, 418

4667–4675. 419

420

421

Page 22 of 33



23

Figure captions 422

423

Figure 1. Concentration of isoleucine in conventional, lactose-hydrolyzed and filtered UHT milk, 424

quantified by 1H NMR spectroscopy. 425

Figure 2. Relative quantification of αs1-casein and β-casein in conventional, lactose-hydrolyzed and 426

filtered UHT milk as function of storage. 427

Figure 3. PCA score plot of 1H NMR spectroscopic data of conventional, filtered and lactose-428

hydrolyzed UHT milk during storage. Samples are colored according to days of storage. The first 429

principal component explains 43.2% of the variation and the second principal component explains 430

20.1% of the variation. 431

Figure 4. PLS regression plots from 1H NMR data spectroscopic with predicted versus actual and 432

corresponding VIP plot for model based on A, B) conventional UHT milk (n=33), R2 CV:0.95, 433

RMSECV: 18.59 days, C, D) lactose-hydrolyzed UHT milk (n=65), R2 CV:0.98, RMSECV: 11.23 434

days, and E, F) filtered UHT milk (n=23), R2 CV:0.98, RMSECV: 11.45 days. The numbers 435

indicate the metabolites important for the regression; 1,2,3) Leucine, isoleucine, valine, 4) Alanine, 436

5,6) Lysine, Leucine, 7) Acetate, 8) N-acetyl carbohydrates, 9,10) Methionine, 11) Not identified 437

12,13) Creatine/creatinine, choline, 14,15,16,17) Hippurate, tyrosine, tryptophan, phenylalanine. 438

439

440

441

Page 23 of 33



24

Tables 442

Table 1. Overview of milk types, heat treatment and lactose level included in the study. All 443

milk contained 1.5% w/w fat and 3.5% w/w protein. 444

Milk

Enzymatic

hydrolysis

Heat

treatment

Filtration

Level of lactose

% of original

level

Amount of

lactose % w/w

Conventional

indirect UHT

- Indirect - 100 5

Lactose- hydrolyzed

direct UHT

+ Direct + <1 0.01

Lactose- hydrolyzed

indirect UHT

+ Indirect + <1 0.01

Filtered

indirect UHT

- Indirect + 60 3

445

Page 24 of 33



25

Table 2. Overview of the chemical changes in carbohydrates, proteins and Strecker 446

degradation compounds during storage of conventional, lactose-hydrolyzed and filtered UHT 447

milk. 448

Changes during storage

Chemical

compound

Conventional

indirect UHT

Lactose-

hydrolyzed

direct UHT

Lactose-

hydrolyzed

indirect UHT

Filtered

indirect UHT

Method

Lactose* NS

(54.18-54.31)a

NS

(0.278-0.214)b

NS

(0.249-0.207)b

NS

(27.30-39.19)c

1H NMR

Galactose* NS

(0.356-0.407)b

NS

(34.53-27.94)a

NS

(24.56-27.88)a

NS

(0.208-0.275)b

1H NMR

Glucose* ND NS

(41.43-32.66)a

NS

(29.17-32.84)a

NS

(0.136-0.148)b

1H NMR

Formate* (↑) p=0.025

(0.02-0.034)a

(↑) p=0.001

(0.007-0.022)a

(↑) p=0.001

(0.008-0.02)a

NS

(0.009-0.019)a

1H NMR

Alanine* NS

(0.019-0.018)b

(↑) p<0.001

(0.018-0.026)a

(↑) p<0.001

(0.015-0.027)a

NS

(0.016-0.016)b

1H NMR

Isoleucine* NS

(0.003-0.003)b

(↑) p=0.007

(0.007-0.129)a

(↑) p<0.001

(0.067-0.154)a

NS

(0.004-0.003)b

1H NMR

Leucine* NS

(0.001-0.001)b

(↑) p=0.015

(0.003-0.033)a

(↑) p=0.003

(0.003-0.033)a

NS

(0.001-0.006)b

1H NMR

Lysine* NS

(0.007-0.016)a

(↑) p=0.012

(0.007-0.018)a

(↑) p=0.026

(0.007-0.023)a

NS

(0.005-0.021)a

1H NMR

Methionine* NS

(0.004-0.007)a

NS

(0.002-0.005)a

NS

(0.003-0.003)a

NS

(0.002-0.004)a

1H NMR

Phenylalanine* ND (↑) p=0.006 (↑) p=0.001 ND 1H NMR

Page 25 of 33



26

(0.002-0.007)a (0.001-0.009)a

Tryptophan* ND (↑) p=0.063

(0.002-0.027)a

(↑) p=0.023

(0.002-0.036)a ND 1H NMR

Tyrosine* ND (↑) p=0.002

(0.002-0.005)a

(↑) p<0.001

(0.002-0.006)a ND 1H NMR

Valine* NS

(0.006-0.006)b

(↑) p=0.007

(0.012-0.118)a

(↑) p=0.001

(0.011-0.132)a

NS

(0.007-0.005)b

1H NMR

Total CN# (↓) p<0.001

(65.3-54.9)b

(↓) p<0.001

(64.1-24.9)a

(↓) p<0.001

(64.4-28.5)a

(↓) p<0.001

(65.6-55.7)b HPLC MS

Glycoprotein# (↑) p=0.002

(3.3-4.8)b

(↑) p<0.001

(2.3-4.3)b

(↑) p<0.001

(3.1-6.5)a

(↑) p=0.001

(3.0-4.9)b HPLC MS

κ-CN# (↓) p<0.001

(2.3-1.0)b

(↓) p<0.001

(3.4-0.6)a

(↓) p<0.001

(3.6-0.6)a

(↓) p<0.001

(2.7-1.3)b HPLC MS

αs1-CN# NS

(25.5-21.5)b

(↓) p=0.001

(22.8-12.6)a

(↓) p<0.001

(24.3-14.1)a

NS

(25.4-21.0)b HPLC MS

αs2-CN# NS

(3.5-2.6)b

(↓) p=0.004

(4.1-2.0)a

(↓) p=0.004

(3.7-1.7)a

(↓) p=0.003

(4.3-2.2)a HPLC MS

Β-CN# (↓) p=0.001

(30.8-25.0)b

(↓) p<0.001

(31.5-5.4)a

(↓) p<0.001

(29.6-5.6)a

(↓) p<0.003

(30.3-26.3)b HPLC MS

β-LG# (↓) p<0.001

(3.8-1.5)b

(↓) p<0.001

(4.8-0.7)a

(↓) p<0.001

(5.6-1.3)a

(↓) p<0.001

(4.5-2.5)b HPLC MS

Benzaldehyde¤ NS

(2.31-3.38)a

NS

(1.88-1.76)a

NS

(2.28-2.87)a

NS

(2.36-2.05)a

GC-MS

DMDS¤ NS

(0.49-0.11)a

NS

(0.44-0.12)a

NS

(0.60-0.07)a

NS

(0.27-0.49)a GC-MS

DMTS¤ NS

(1.75-0.17)a

NS

(1.42-0.21)a

NS NS

(1.36-0.24)a GC-MS

Page 26 of 33



27

(1.89-0.20)a

2-ethylfuran¤ (↑) p=0.005

(0.64-0.77)a

NS

(0.60-0.79)a

NS

(0.59-0.68)a

(↑) p=0.002

(0.62-0.76)a GC-MS

2-pentylfuran¤ NS ((↑)p=0.053)

(0.99-1.07)a

NS

(0.86-1.04)a

NS

(0.88-0.94)a

NS

(1.04-1.07)a GC-MS

2-methylfuran¤ ND ND ND ND GC-MS

* Average start and end concentration, [mM] 449

# Average start and end concentration, % intact protein 450

¤ Average start and end concentration, [µg/kg] 451

a,b,c Indicate a significant difference in concentration between the milk types at storage day 270, p<0.05 452

ND Not detected 453

NS Not significant 454

(↑) Indicates a significant increase during storage 455

(↓) Indicates a significant decrease during storage 456

457

458

Page 27 of 33



28

Table 3. Amino acids identified using 1H NMR spectroscopy and their involvement in 459

formation of Strecker degradation aldehydes identified using GCMS 460

Amino acid Strecker degradation product Identified with GCMS

Isoleucine 2-methylbutanal

Leucine 3-methylbutanal

Valine Methylpropanal

Alanine Acetaldehyde2-methylfuran,

2-pentylfuran X

Lysine

Methionine MethionalDimethyldisulfide

and dimethyltrisulfide X

Phenylalanine Benzaldehyde, 2-methylfuran X

Tyrosine

Tryptophan

461

462

Page 28 of 33



29

Figures 463

464

465

Figure 1 466

467

468

Page 29 of 33



30

469

470

Figure 2. 471

472

473

474

475

476

Page 30 of 33



31

477

478

479

480

481

482

Figure 3. 483

484

Page 31 of 33



32

485

486

Figure 4. 487

488

489

490

491

Page 32 of 33



33

TOC 492

493

494

495

496

497

Page 33 of 33



143

Paper 4

Paper 4

Storage-induced changes in the sensory characteristics and

volatiles of conventional and lactose-hydrolyzed UHT processed

milk

Jensen S., Jansson, T., Eggers, N., Clausen, M.R., Jensen, H.B., Larsen, L.B.,

Colin, R., Sundgren, A., Andersen, H.J., Bertram, H.C.

Submitted to Food Research International

Elsevier Editorial System(tm) for Food Research International Manuscript Draft Manuscript Number: Title: Storage-induced changes in the sensory characteristics and volatiles of conventional and lactose-hydrolyzed UHT processed milk Article Type: Research Article Keywords: Bitter taste; Lactose-hydrolyzed milk; UHT; Sensory descriptive analysis; GC-MS; Stale flavor Corresponding Author: Dr H.C. Bertram, Corresponding Author's Institution: First Author: Sidsel Jensen Order of Authors: Sidsel Jensen; Therese Jansson; Nina Eggers; Morten R Clausen; Hanne B Jensen; Lotte B Larsen; Colin Ray; Anja Sundgren; Henrik J Andersen; H.C. Bertram Abstract: Storage-induced changes are known to be more prominent in lactose hydrolyzed (LH) milk compared to conventional milk. Therefore the present study aimed at identifying off-flavors resembling from formation of volatiles during storage of ultra high temperature treated (UHT) LH milk and conventional UHT milk. Further, the influence of heat processing, indirect or direct, on UHT LH milk was also examined. Storage-induced changes in sensory attributes, volatiles and primary amines were investigated during a four months period. Conventional UHT milk (with 5 % lactose) processed using indirect heat treatment (CONVI) and two types of UHT lactose-hydrolyzed (LH) milk (with less than 0.01% lactose) produced using either direct heat treatment (LHD) or indirect heat treatment (LHI) were represented in the study. Sensory descriptive analysis showed that fresh samples of CONVI, LHI and LHD differed in sensory properties and the samples could be differentiated according to boiled and stale aroma as well as color saturation. Differentiation of the fresh samples based on the volatile GC-MS profile was not achievable. During storage, samples developed differently with respect to sensory characteristics, volatiles and the amount of primary amines. Partial least squares models (PLS1) including only methyl ketones and aldehydes showed that 2-butanone, 2-pentanone, 2-heptanone, 2-nonanone, heptanal, octanal and nonanal predicted stale flavor. Bitter taste, on the other hand, correlated with the amount of primary amines (Pearson's correlation, r2=0.71). This finding indicates that storage-induced changes in sensory characteristics and chemical composition depend on both differences in lactose content and the applied heat processing.

Denmark, July 2014

Dear Editor

Enclosed please find the manuscript entitled “Storage-induced changes in the sensory characteristics and volatiles of

conventional and lactose-hydrolyzed UHT processed milk”.

The authors believe that the manuscript is suitable for publication in Food Research International. The manuscript

concerns the investigation of off flavor development in lactose hydrolyzed milk. Off flavor development was

measured by sensory descriptive analysis, GC-MS and fluorescamine assay. The aim of the study was to elucidate the

influence of enzymatic lactose hydrolysis, heat processing and storage on sensory characteristics on milk stored for

up to 4 months. Changes in the sensory characteristics were related to changes in volatiles with a special focus on

the relationship between methyl ketones and aldehydes and stale flavor. Further, the relationship between the

degree of proteolysis and bitter taste was examined. The study showed that changes in sensory characteristics and

volatiles were dependent on both milk type and differences in heat processing which is valuable knowledge in the

process optimization of lactose hydrolyzed milk and for a future development of new lactose hydrolyzed products.

We believe that the topic fits well with the aims and scope of Food Research International.

We hope that you will consider the manuscript for publication and we are looking forward to be hearing from you.

On behalf of the authors,

Sidsel Jensen and Hanne Christine Bertram

Cover Letter

Highlights

Sensory characteristics of stored lactose-hydrolyzed and conventional milk differ

Off-flavor development in lactose-hydrolyzed milk depends on UHT processing method

Development of volatiles depends on milk type and UHT processing method

The bitter taste of milk correlates with the content of primary amines

*Highlights (for review)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

1

Storage-induced changes in the sensory characteristics and volatiles of conventional and lactose-hydrolyzed UHT processed milk

Sidsel Jensena,Therese Janssona, Nina Eggersa, Morten R. Clausena, Hanne B. Jensenb, Lotte B.

Larsenb, Colin Rayc, Anja Sundgrenc, Henrik J. Andersend , Hanne C. Bertrama*

aDepartment of Food Science, Research center Aarslev, Aarhus University, Kirstinebjergvej 10,

DK-5792 Aarslev, Denmark. E-mails: [email protected], [email protected],

[email protected], [email protected].

bDepartment of Food Science, Research center Foulum, Aarhus University, Blichers Allé 20, DK-

8830 Tjele, Denmark. E-mails: [email protected], [email protected].

cArla Foods Strategic Innovation Centre, Lindhagensgatan 126, 10546 Stockholm, Sweden, and

Roerdrumvej 2, DK-8220 Brabrand, Denmark. E-mail: [email protected]

dArla Foods Ingredients, Sønderhøj 10-12, DK-8260 Viby J, Denmark. E-mail:

[email protected]

*Corresponding author. Tlf: +45 87158353, Email: [email protected]

*ManuscriptClick here to view linked References

http://ees.elsevier.com/foodres/viewRCResults.aspx?pdf=1&docID=20099&rev=0&fileID=341534&msid=455BF8C6-A234-4F62-803D-23D16D7D3930

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

2

Abstract Storage-induced changes are known to be more prominent in lactose hydrolyzed (LH) milk compared to

conventional milk. Therefore the present study aimed at identifying off-flavors resembling from formation

of volatiles during storage of ultra high temperature treated (UHT) LH milk and conventional UHT milk.

Further, the influence of heat processing, indirect or direct, on UHT LH milk was also examined. Storage-

induced changes in sensory attributes, volatiles and primary amines were investigated during a four

months period. Conventional UHT milk (with 5 % lactose) processed using indirect heat treatment (CONVI)

and two types of UHT lactose-hydrolyzed (LH) milk (with less than 0.01% lactose) produced using either

direct heat treatment (LHD) or indirect heat treatment (LHI) were represented in the study. Sensory

descriptive analysis showed that fresh samples of CONVI, LHI and LHD differed in sensory properties and

the samples could be differentiated according to boiled and stale aroma as well as color saturation.

Differentiation of the fresh samples based on the volatile GC-MS profile was not achievable. During storage,

samples developed differently with respect to sensory characteristics, volatiles and the amount of primary

amines. Partial least squares models (PLS1) including only methyl ketones and aldehydes showed that 2-

butanone, 2-pentanone, 2-heptanone, 2-nonanone, heptanal, octanal and nonanal predicted stale flavor.

Bitter taste, on the other hand, correlated with the amount of primary amines (Pearson’s correlation,

r2=0.71). This finding indicates that storage-induced changes in sensory characteristics and chemical

composition depend on both differences in lactose content and the applied heat processing.

Keywords: Bitter taste; Lactose-hydrolyzed milk; UHT; Sensory descriptive analysis; GC-

MS; Stale flavor

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

3

Chemical compounds

2-butanone (PubChem CID: 6569); 2-pentanone (PubChem CID: 7895); 2-heptanone (PubChem CID: 8051);

2-nonanone (PubChem CID: 13187); heptanal (PubChem CID: 8130); octanal (PubChem CID: 454); nonanal

(PubChem CID: 31289)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

4

1. Introduction

The prevalence of adult-type hypolactasia is widespread throughout the world (Itan, Jones, Ingram,

Swallow, & Thomas, 2010). Historically, hypolactasia frequency is lowest in Northwestern Europe but the

frequency is increasing probably due to an increase in immigration of citizens with adult-type hypolactasia

(Almon, Engfeldt, Tysk, Sjöström, & Nilsson, 2007). Consequently, the sale of lactose-free dairy products

tripled in Europe from 2007 to 2011 and it is also expected to double in the period 2012 to 2016 (Astley,

2012). However, according to a recent global market report the most significant opportunities exist in

markets that have a prevalence of lactose intolerance and where dairy consumption is rising, such as in Asia

and Latin America (Zenith International, 2012).

Traditionally, lactose-hydrolyzed (LH) milk is sold as UHT-treated products, and they have been claimed to

have a shorter shelf-life due to limited chemical stability compared to conventional UHT milk (Messia,

Candigliota, & Marconi, 2007). The chemical instability is expected to result in perceptible alterations of the

product reducing the overall product quality (Lawless, 1995).

Unaltered sensory characteristics of lactose-hydrolyzed UHT milk during storage can only be achieved, if

the chemical and physical processes from which the off-flavors originate are prevented. Vallejo-Cordoba

and Nakai (1994) found that the shelf-life of pasteurized milk as determined by sensory evaluation was

better predicted by analysis of volatiles than by bacterial count. In UHT-treated milk, the severity of heating

applied to the milk and the concentration of dissolved oxygen in the milk immediately after processing

depends on the type of UHT processing and are the two most important factors contributing to off-flavor

formation (Datta, Elliott, Perkins, & Deeth, 2002). The most commonly applied UHT processing is indirect

heat processing where the milk and the heating source are physically separated at all times. Due to a

shorter heating and cooling time, direct UHT processing, where steam is injected or infused directly into the

milk stream, is more gentle, but also a more costly method compared to indirect UHT processing (Elliott,

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

5

Dhakal, Datta, & Deeth, 2003; Elliott, Datta, Amenu, & Deeth, 2005). The level of dissolved oxygen in direct

UHT processed milk is also lower than in milk exposed to indirect UHT processing (Datta et al., 2002).

Recently, numerous papers encompassing sensory analysis of conventional UHT milk have been reported

(Jokar & Karbassi, 2011; Lorenzen et al., 2011; Ochi et al., 2010). However, only a few of these sensory

studies addresses lactose-hydrolyzed milk (Adhikari, Dooley, Chambers, & Bhumiratana, 2010; Tossavainen,

2008). Tossavainen (2008) investigated sensory differences between conventional and lactose-hydrolyzed

UHT milk stored for 10 weeks at 22°C and found no perceptual difference between the two milk types. The

study revealed that the first quality defects occurred after 6 weeks of storage and encompassed cooked,

heated and burned flavor followed by both an atypical and a bitter taste. In contrast, Adhikari et al. (2010)

found a higher intensity of sweet taste, chalkiness and cooked and processed flavors in fresh samples of

lactose-free ultra-pasteurized milk as compared with pasteurized conventional milk.

In total, more than 400 volatiles have been associated with milk. During storage at room temperature

proteins, lipids and sugars are involved in different chemical reactions in UHT-processed milk, and these

chemical changes are reported to change the sensory characteristics (Singh, Ruhil, Jain, Patel, & Patil, 2009;

Valero, Villamiel, Miralles, Sanz, & Martínez-castro, 2001). The Maillard reaction, which is a heat-induced

condensation reaction involving the free aldehyde group of reducing sugars and free amines in proteins and

peptides, occurs during the UHT treatment of milk and continues to progress during storage (Kato,

Matsuda, Kato, & Nakamura, 1988; Pellegrino, De Noni, & Resmini, 1995). More Maillard reaction products

are formed during storage of LH milk compared to conventional milk (Jansson et al., 2014b; Messia et al.,

2007). Formation of stored, stale and rancid flavors have been found to take place during storage of UHT

milk and this has been assigned to increased concentrations of methyl ketones and aldehydes, which

originate from the ongoing Maillard reaction and lipid oxidation (Jeon, Thomas, & Reineccius, 1978; Rerkrai,

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

6

Jeon, & Bassette, 1987). Proteolysis in UHT milk leads to exposure of free amines and this may likewise

increase formation of Maillard reactions products and thus facilitate the formation of stored, stale and

rancid flavors. Furthermore, bitter peptides formed during proteolysis may contribute to an increased

bitterness. McKellar, Froehlich, Butler, Cholette, & Campbell (1984) found a correlation between the

degree of proteolysis and bitter taste in UHT milk.

In order to obtain a better understanding of how storage-induced changes differ in conventional and LH

milk, the objective of the present study was to elucidate the influence of enzymatic lactose hydrolysis, heat

processing and storage on sensory characteristics on milk stored for up to four months. Changes in the

sensory characteristics were related to changes in volatiles with a special focus on the relationship between

methyl ketones and aldehydes and stale flavor. Further, the relationship between the degree of proteolysis

and bitter taste was examined.

2. Materials and methods

2.1 Chemical standards

Decane (99 %), tetradecane (99 %), dodecane (99 %), 2-methylbutanal (95 %), hexanal (98 %), octanal (99

%), nonanal (95 %), decanal (98 %), undecanal (97 %), , 2-butanone (99 %), 2-pentanone (99 %), 2-

hexanone (99.5 %), 2-heptanone (98 %), 2-octanone (98 %), 2-nonanone (99 %), 2-undecanone (99 %), 6-

methyl-5-heptene-2-one (99 %), 6,10-dimethyl-5,4-undecadien-2-one (97 %), dimethyl disulfide (98 %),

dimethyl trisulfide (98 %), 4-methyl-1-pentanol (97 %), toluene (99.5 %), styrene (99 %), 2-pentylfuran (97

%), caffeine (Ph Eur) and calcium glycerophosphate hydrate (Ph Eur) were all purchased from Sigma-Aldrich

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

7

Inc. (Stenheim, Germany). Dimethyl sulfoxide (99 %) was purchased from Serva electrophoresis GmHB

(Heidelberg, Germany).

2.2 Milk samples

Three types of milk were produced at a commercial Danish dairy plant (Arla Foods amba, Esbjerg,

Denmark). Conventional UHT milk (with 5 % lactose) processed using indirect heat treatment (CONVI) and

two types of UHT lactose-hydrolyzed (LH) milk (with less than 0.01 % lactose) produced using either direct

heat treatment (LHD) or indirect heat treatment (LHI) (Table 1). In LHD and LHI products the reduction in

lactose was obtained by a combination of ultra-filtration (to approximately 60 % of the original

concentration) prior to heat processing and enzymatic hydrolysis by addition of β-galactosidase after UHT

processing.

LHD milk was produced by steam injection directly into the milk stream allowing the milk to instantly reach

a temperature of 140-145 °C. This temperature was kept for a few seconds and subsequently the milk was

cooled by flash cooling. LHI and CONVI milk were produced using a tube heat exchanger where the milk

was heated to a temperature of 140-145°C for a few seconds and after that cooled by a slow cooling

process (Tetra Pak, 2003, Chapter 9). Subsequently the milk was tapped and packed aseptically in 1-L

packages. Milk was produced at three different occasions: September 2012, October 2012 and February

2013 and analyzed simultaneously in February 2013 resulting in fresh milk and milk stored for three and

four months at the date of analysis. During storage, the milk was kept in an air-conditioned room at 22°C.

2.3 Sensory descriptive analysis

The milk used for sensory descriptive analysis (DA) was stored in a refrigerator approximately 24 hours

prior to analysis. Two hours before the sensory session approximately 20 mL milk was poured into

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

8

transparent plastic glasses (total volume 30 mL) from ABENA (Aabenraa, Denmark) labeled with a random

three-digit number. The glasses were sealed with a frosted white plastic lid and placed in a thermostat

controlled heat cabinet from Termaks (Bergen, Norway) adjusted to a temperature of 15 °C for 1-2 hours.

Samples were removed from the heat cabinet and placed at room temperature 15 min before the sensory

session.

An established and trained sensory panel was used for DA. The panel consisted of ten assessors (eight

females/two males, aged from 29 to 59) with one to seven years of experience in evaluation of food

products and all with prior experience in DA of milk. During two training sessions (2 hours each) assessors

were presented for reference samples deliberately designed to represent sensory attributes relevant for

stored UHT milk (sulfurous, cabbage-like, cardboard-like, stale, bitter, chalky, sweet/caramel-like, rice

pudding-like and popcorn-like) together with extreme samples from the actual test set (LHD and CONVI, 0

months and LHI and CONVI 4 months). Initially, the sensory panel agreed upon a list of 12 sensory

attributes divided into four main categories: appearance, aroma (nasal perception), flavor/taste (retro nasal

perception/oral perception) and mouthfeel (Figure 1A-L), and afterwards assessors evaluated a set of

extreme samples in the sensory booths. After each training session, assessors received feedback on their

performance with the aim of improving and standardizing the panel’s discriminating power.

The DA was performed in a sensory evaluation laboratory fulfilling requirements settled by the American

Society for Testing and Materials (ASTM, 1986) and the analysis was conducted in agreement with

specifications provided by the International Organization for Standardization (ISO, 1993). To test the

robustness of the DA the fresh sample of LHD was included twice in the analysis. Samples were analyzed in

quadruples over two test days (two replicates each day).

2.4 Analysis of volatiles

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

9

Sampling and analysis of volatiles was analyzed by dynamic headspace gas chromatography mass

spectrometry (GC-MS) performed according to the optimized method described previously by Jansson et al.

(2014a). Three replicates of each type of milk from the same 1-L package were analyzed at the same days

as the sensory analysis. Triplicates of CONVI stored for four months from two different packages were

encompassed in the analysis to verify the reproducibility of the method. An MS database (NIST MS search

version 2.0, 2011) was used to identify 27 volatiles. In addition, all of the volatiles were verified by

comparison of mass spectral data and retention times of authentic standards. Twelve calibrants (2-

methylfuran (6.6 min), 2-methylbutanal (7.4 min), 2-ethylfuran (8.2 min), 2-pentanone (10.7 min), dimethyl

disulphide (DMDS) (14.9 min), hexanal (19.2 min), 2-heptanone (31.9 min), 2-pentylfuran (36.0 min), 2-

octanone (36.8 min), dimethyl trisulphide (DMTS) (40.1 min), 2-nonanone (41.9 min) and nonanal (42.0

min)) were used for quantification of volatiles belonging to the same class and/or compounds eluting in the

same retention time window. Standard curves, made in duplicates, were prepared using ultra-filtered UHT

milk and sampled and analyzed in a manner resembling the procedure used for analysis of the actual milk

samples. The chromatogram of the pure milk was subtracted from the recorded chromatograms. All

concentrations were corrected in accordance to the response factor of an internal standard (4-methyl-1-

pentanol). Limit of detection (LOD) was calculated from the standard error of the Y-intercept and the slope

of the linear regression (Dolan, 2009).

2.5 Determination of primary amines by fluorescamine assay

The level of primary amines in the milk was determined according to the method of Dalsgaard, Nielsen, &

Larsen (2007). Milk was mixed with an equal volume of 24 % trichloroacetic acid and put on ice for 30 min

in order to accelerate protein precipitation. The precipitated mixture was centrifuged (17,000 g) for 20 min.

Part of the supernatant, 37.5 µL, was added to 1130 µL 0.1M borate buffer and 375 µL of a fluorescamine

solution (0.2 mg/mL fluorescamine in water free acetone). After 18 min, 250 µL of the resulting solution

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

10

was transferred to a 96-well micro plate and fluorescence was measured using a Synergy 2 Multi-Mode

Microplate Reader (Holm & Halby, Denmark) with excitation and emission wavelengths of 400 nm and 485

nm. Quantification was achieved using an external leucine standard curve made in 1 mM HCl (0.5 – 3 mM

leucine) and results were given as leucine equivalents.

2.6 Statistical analysis

Statistical analyses were carried out using the freeware programs PanelCheck v. 1.4.0

(www.PanelCheck.com) and R v. 2.13.1 (R Foundation for Statistical Computing, Vienna, Austria).

PanelCheck was used for assessor evaluations and for sensory product differences tested by Bonferroni

least significant difference (LSD) while R was used on sensory data for a complete analysis of linear mixed

effect models (Kuznetsova, Brockhoff, & Christensen, 2013) with assessor and replicate as random effects.

Analysis of variance (ANOVA) was applied to investigate significant differences of volatiles between milk

samples. Statistical significances were defined at p≤0.05. Compounds with concentrations below LOD were

set to zero. Partial least squares regression (PLS1) was performed using Simca software version 13.0 from

Umetrics (Umeaa, Sweden) in order to predict stale flavor on basis of mean data from the analysis of

selected volatiles. All data was scaled to unit-variance prior to analysis and full cross validation was applied.

Person’s correlation coefficient (r2) was calculated using R to determine the linear relationship between

bitter taste and the content of primary amines.

3. Results and discussion

3.1 Sensory descriptive analysis (DA)

Figure 1 shows the development during storage for each of the twelve attributes included in the DA: boiled

milk aroma (A), popcorn aroma (B), stale aroma (C), color saturation (D), boiled milk flavor (E), popcorn

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

11

flavor (F), stale flavor (G), caramel flavor (H), creamy flavor (I), sweet taste (J), bitter taste (K) and fatty

mouthfeel (L). No significant differences were identified between the two fresh samples of LHD milk

originating from the same 1-L package for any of the tested attributes confirming the robustness of the DA.

Because the two samples were perceived as being identical, the mean intensity score for each repetition of

these two samples was used in the further statistical models.

Fresh samples of milk could be differentiated according to boiled milk aroma (Figure 1A), color saturation

(Figure 1D) and fatty mouthfeel (Figure 1L). Cooked/boiled milk aroma in milk was associated with an

increase in the formation of sulfur-containing compounds most probably arising from the thermally

denaturation of serum proteins (de Wit and Nieuwenhuijse, 2008) and, thus, it was unexpected to find that

fresh LHD, representing the sample with the lowest heat load, had a higher intensity of boiled milk aroma

(Figure 1A) than both CONVI and LHI. However, the result is supported by Ochi et al. (2010) who likewise

found a more intense overall aroma, butter aroma and cooked flavor in direct UHT processed milk as

compared to samples exposed to indirect UHT treatment. In other studies a higher intensity of cooked and

processed flavor and a sweeter taste was found in fresh LH milk compared with fresh samples of

conventional milk (Adhikari et al., 2010; Jokar & Karbassi, 2011) or no observable differences were

recognized (Tossavainen, 2008). After three months of storage, CONVI had a higher degree of color

saturation (Figure 1D) than LHI, a more intense boiled milk flavor (Figure 1E), cream flavor (Figure 1I) and a

more fatty mouthfeel (Figure 1L) compared to LHI and LHD whereas LHI was evaluated to have a more stale

flavor (Figure 1G) compared to CONVI. Color saturation (Figure 1D) was significantly lower in fresh and

three months old LHD than in similar samples of CONVI, while the difference disappeared after 4 months of

storage. This finding indicates that the type of heat processing was decisive for the color saturation of milk

the first three months of storage, possibly due to different effects on the degree of heat-induced protein

denaturation. After four months of storage, differences in boiled aroma (Figure 1A), creamy flavor (Figure

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

12

1I) and fatty mouthfeel (Figure 1L) were identified. LHI was evaluated to have a more intense boiled milk

aroma (Figure 1A) than LHD whereas CONVI was judged to have a higher intensity of creamy flavor (Figure

1I) than both LHI and LHD and to provide a more fatty mouthfeel (Figure 1L) than LHD.

Seven attributes obtained a significant interaction effect between milk type and storage: boiled milk aroma

(p≤0.0001, Figure 1A), stale aroma (p≤0.0001, Figure 1C), color saturation (p≤0.0001, Figure 1D), boiled milk

flavor (P=0.05, Figure 1E), stale flavor (p≤0.0001, Figure 1G), creamy flavor (p≤0.0001, Figure 1I) and bitter

taste (0.05, Figure 1K).

FIGURE 1

Tossavainen (2008) identified an increase in cooked, heated and burned flavor after six weeks of storage as

the first indicator of storage-related quality defects in LH milk. In this study, cooked, heated and burned

flavor were associated with boiled milk aroma (Figure 1A) and flavor (Figure 1E), and LHI and CONVI milk

showed an increase in intensity of boiled milk aroma (Figure 1A) from fresh to 3 months of storage,

whereas a decrease in this attribute was observed for LHD during storage. Boiled milk flavor (Figure 1E) did

not increase in any of the tested milk types during storage. The second indicator of storage-related quality

defects that Toassavainen (2008) observed was a combination of an atypical and a bitter taste. These

quality defects can be associated with stale flavor (Figure 1G) and bitter taste (Figure 1K) in the present

study, which both showed a significant interaction effect between milk type and storage. This could

indicate that development of these two attributes was dependent on the milk type and that a general

conclusion about the development across milk types should be made with caution. Stale flavor originates

from volatiles formed via the Maillard reaction and lipid oxidation (Jeon et al., 1978; Rerkrai et al., 1987),

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

13

both of which are catalyzed by heat. However, no significant effect of direct versus indirect heating could

be observed in the fresh milk, whereas during storage stale flavor increased the first three months of

storage in LHD and then settled. In contrast, LHI and CONVI demonstrated the lowest amount of stale

flavor after three months of storage (Figure 1G).

Bitter taste is known to be associated to proteolytic activity in LH milk (Tossavainen, 2008). In the present

study bitter taste likewise increased in lactose reduce milk (LHD and LHI) while it remained constant in

CONVI. In addition, there was a significant decline in sweet taste for all milk types (Figure 1J) during storage

which could indicate a reduction in free sugars available, e.g. lactose, galactose and glucose, due to ongoing

Maillard reactions during storage (Chávez-Servín, Romeu-Nadal, Castellote, & López-Sabater, 2006) or

alternatively because of a masking effects due to an increase in bitter taste that is known to reduce the

perceived sweet taste (Beck, Jensen, Bjoern, & Kidmose, 2014). The masking effect is more likely because

Jansson et al. (2014b) could not analyze changes in the content of simple sugars in UHT processed LH milk

during 9 months of storage. Fatty mouthfeel (Figure 1L) was the only attribute where a significant effect of

milk type was observed, with CONVI being the milk type with the most profound fatty mouthfeel (Figure

1L). This fact might be due to a smaller milk fat globule size in the LH milk types as a result of an additional

membrane filtration processing, however, this hypothesis needs to be further investigated.

3.2 Relationship between the chemical composition and sensory characteristics

The composition of the volatile components in the milk samples is responsible for the perceived aroma.

Volatiles present in levels above the sensory threshold (Table 2) can be expected to be important

contributors to the aroma of food products, however, also volatiles below the sensory threshold will

possibly influence perceived aroma due to mixture enhancement (Jeon et al., 1978). Only minor differences

in the composition of volatiles between the duplicate CONVI samples stored for four months were found by

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

14

GC-MS analysis and therefore means were used for statistical calculations. Of the 27 compounds verified by

authentic standards, a total of 17 compounds: 2-butanone, 2-methylbutanal, 2-pentanone, decane,

toluene, DMDS, 2-hexanone, hexanal, 2-heptanone, heptanal, 2-octanone, octanal, DMTS, 2-nonanone,

nonanal, tetradecane and decanal, occurred in concentrations equal to or higher than the LOD in one or

more of the tested milk types (Table 2). The presence of these compounds in milk was in agreement with

previous GC-MS studies conducted on milk (Contarini et al., 1997; Jansson et al., 2014a; Valero et al., 2001;

Vazquez-Landaverde, Velazquez, Torres, & Qian, 2005).

Intriguingly, no significant differences between fresh samples of milk were observed for any of the detected

volatiles and thus the GC-MS analysis failed to differentiate and explain why fresh LHD milk was perceived

to have a higher intensity of boiled milk aroma (Figure 1A) compared to CONVI and LHI. After three months

of storage, CONVI had a significantly higher content of 2-butanone and hexanal and a significantly lower

content of DMDS than LHD and LHI and a significantly lower content of 2-methylbutanal, DMTS and

tetradecane than LHD (Table 2). The lower content of 2-methylbutanal in CONVI stored for three months

compared to both types of LH milk was a strong indicator of a more slow progression of the Maillard

reaction in CONVI during storage, considering 2-methylbutanal being a typical Maillard reaction product in

milk (Jansson et al., 2014b). Oxidative degradation of methionine leads to a higher content of DMDS and

DMTS (Ballance, 1961). Consequently, the higher content of dissolved oxygen found in indirect UHT

processed milk compared to direct UHT processed milk (Datta et al., 2002) must be expected to result in

generation of a higher concentration of DMDS and DMTS. However, concurrently a higher level of dissolved

oxygen is reported to cause a more rapid loss of sulfhydryl compounds (Thomas, Burton, Ford, & Perkin,

1975), which may explain why the concentration of DMDS and DMTS cannot be linked to heat processing

method and may also be the reason that no differences in boiled milk aroma were evident between milk

samples stored for three months. After four months of storage, significantly lower contents of toluene,

hexanal, 2-heptanone and 2-nonanone were identified in LHI compared to CONVI and LHD as well as

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

15

significantly lower contents of 2-butanone, 2-pentanone and 2-octanone than CONVI. LHD stored for four

months had a significantly lower concentration of DMDS and DMTS compared to LHI (Table 2).

TABLE 2

Eleven of the identified volatiles exhibited a significant interaction between storage and milk type: 2-

butanone (P=0.007), 2-pentanone (P=0.03), toluene (P=0.0005), DMDS (P=0.0003), 2-hexanone (P=0.006),

hexanal (P=0. 0002), 2-heptanone (P=0.005), 2-octanone (P=0.007), DMTS (P=0.01), 2-nonanone

(P=0.0005), and tetradecane (P=0.0003), supporting the results from the sensory descriptive analysis, which

showed that the changes in the sensory characteristics during storage highly depended on milk type and

heating process.

As LH milk has been found to have a shorter shelf-life than conventional milk (Messia et al., 2007), and as

previous studies have associated storage-related off-flavor in UHT milk with increasing concentrations of

methyl ketones and aldehydes (Jeon et al., 1978; Rerkrai et al., 1987), the development in methyl ketones

and aldehydes during storage were examined in the three types of milk. It was investigated if some of the

methyl ketones and aldehydes were better predictors of the development of stale flavor than others. Stale

flavor was the attribute from the present study assumed to be closest related to the storage related off-

flavor described in the abovementioned studies. For this purpose, PLS1 regression was conducted only

including methyl ketones (six volatiles) and aldehydes (six volatiles) as X-variables and stale flavor as Y-

variable. The optimal number of PC’s in the prediction of stale flavor was four which resulted in a

cumulative cross-validation variance in Y (Q2Y_cum) of 0.81, meaning that the methyl ketones and

aldehydes included in the model were able to predict 81 % of the variance related to stale flavor. The

model explained 94 % of the variance related to methyl ketones and aldehydes (R2X_cum). Figure 2

displays the PLS1 score scatter plot (A) and the corresponding loading plot (B) including volatiles (X-

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

16

variables) and stale flavor (Y-variable) for PC1 and PC2. From the figure it was obvious that fresh samples of

milk and milk stored for three months differ in the content of methyl ketones and aldehydes depending on

milk type. Fresh CONVI and CONVI stored for three months are positioned in the upper and lower left

quadrant of the PLS1 score plot while fresh LHD and LHD stored for three months were positioned in the

lower left and the upper right quadrant. Fresh LHI and LHI stored for three months were positioned in the

lower right and the upper left quadrant. All samples stored for four months were positioned in the right

side of the score plot (Figure 2A). Samples positioned in the lower left quadrant were characterized by a

high concentration of 2-butanone, hexanal and decanal while samples positioned in the upper right

quadrant have a higher concentration of 2-butanone, 2-pentanone, 2-heptanone, 2-nonanone, heptanal,

octanal and nonanal. These compounds were concurrently found to be the best predictors of stale off-

flavor and it appears to be the combination of these compounds, more than the individual volatiles that

gave rise to the stale flavor.

FIGURE 2

Generally, the concentration of ketones increased in CONVI and LHD milk during storage while a decrease

in concentration was seen for LHI milk (Table 2). Three of the six identified aldehydes showed a significant

increase in concentration during storage: 2-methylbutanal (P= 0.002), octanal (P= 0.003) and decanal (P=

0.01) and an interaction effect between milk type and storage was only seen for hexanal. Overall, these

results indicate that the formation of aldehydes during storage was independent of milk type while the

generation or loss of ketones highly depends on the type of milk. Further, both methyl ketones and

aldehydes influenced stale flavor.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

17

A consistent quality defect associated with stored UHT processed milk upon storage is bitter taste (McKellar

et al., 1984; Tossavainen, 2008). In this study, the level of primary amines was significantly correlated with

the intensity of bitter taste (R2 = 0.71, Figure 3). This has previously been observed in conventional milk

(McKellar et al., 1984), but has not previously been elucidated in LH milk underlying the importance of the

results obtained in this study. The level of primary amines increased more progressively in LH compared to

CONVI (Table 3) indicating that proteolysis was found to be more pronounced in LH milk compared to

CONVI. As the level of primary amines and the bitter taste was correlated, it is most probably enhanced

proteolysis that is responsible for the observed development of bitter taste in LH milk. Since there should

be no variation in the level and activity of endogenous enzymes between the investigated milk types we

expect that the increased in proteolysis in LH milk to be the result of proteolytic side-activity from the β-

galactosidase enzyme applied to hydrolyze the lactose in the production of LH milk.

4. Conclusions

Results from the sensory descriptive analysis showed that fresh samples of CONVI, LHD and LHI differed in

the intensity of boiled and stale aroma as well as color saturation whereas a differentiation of the fresh

milk samples based on the volatile profile was not achievable. During storage sensory characteristics and

volatiles developed differently depending on milk type and heat processing. Intriguingly, stale flavor proved

not to be a strictly storage-related off-flavor as the intensity of stale flavor was found to be highest in fresh

CONVI of all tested samples. 2-Butanone, 2-pentanone, 2-heptanone, 2- nonanone, heptanal, octanal and

nonanal were found to be strong predictors of stale flavor in tests only including methyl ketones and

aldehydes. Based on these results, it can be concluded that formation of aldehydes was independent of

milk type while the ketone level highly depended on milk type. Bitter taste could be related to the level of

primary amines which were generally higher in LH milk compared to CONVI and more pronounced in

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

18

samples stored for four months compared to fresh samples and samples stored for three months. The

output from this study is valuable knowledge in the process optimization of LH products and for a future

development of new LH products.

Acknowledgement

The present study is part of the Ph.D. work of Therese Jansson and was financially supported by Arla Foods

amba, Future Food Innovation (FFI), Danish Dairy Research Foundation, and Aarhus University.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

19

References

Adhikari, K., Dooley, L. M., Chambers, E., & Bhumiratana, N. (2010). Sensory characteristics of commercial lactose-free milks manufactured in the United States. LWT - Food Science and Technology, 43(1), 113–118. doi:10.1016/j.lwt.2009.06.017

Almon, R., Engfeldt, P., Tysk, C., Sjöström, M., & Nilsson, T. K. (2007). Prevalence and trends in adult-type hypolactasia in different age cohorts in Central Sweden diagnosed by genotyping for the adult-type hypolactasia-linked LCT -13910C > T mutation. Scandinavian Journal of Gastroenterology, 42(2), 165–70. doi:10.1080/00365520600825257

Astley, M. (2012). Self diagnosed lactose intolerance driving lactose free dairy sales - analyst. Dairyreporter.com. Retrieved July 14, 2014, from http://www.dairyreporter.com/Markets/Self-diagnosed-lactose-intolerance-driving-lactose-free-dairy-sales-analyst

ASTM. (1986). Physical requirements guidelines for sensory evaluation laboratories, American Society for Testing Materials, Philadelphia, PA.

Ballance, P. E. (1961). Production of volatile compounds related to the flavour of foods from the Strecker degradation of DL-methionine. Journal of Science of Food and Agriculture, 12, 532–536.

Beck, T. K., Jensen, S., Bjoern, G. K., & Kidmose, U. (2014). The masking effect of sucrose on perception of bitter compounds in Brassica vegetables. Journal of Sensory Studies, 29(3), 190–200. doi:10.1111/joss.12094

Burdock, G. A. (2010). Fenaroli’s Handbook of Flavor Ingredients (6th ed.). Boca Raton, FL: CRC Press.

Chávez-Servín, J. L., Romeu-Nadal, M., Castellote, A. I., & López-Sabater, M. C. (2006). Evolution of free mono- and di-saccharide content of milk-based formula powder during storage. Food Chemistry, 97(1), 103–108. doi:10.1016/j.foodchem.2005.03.027

Contarini, G., Povolo, M., Leardi, R., Toppino, P. M., Lattiero-caseario, I. S., Lombardo, V., & Salerno, V. B. (1997). Influence of heat treatment on the volatile compounds of milk. Journal of Agricultural and Food Chemistry, 45(8), 3171–3177. doi: 10.1021/jf960849s

Czerny, M., Christlbauer, M., Christlbauer, M., Fischer, A., Granvogl, M., Hammer, M., Hartl, C., Hernandez, N. M., & Schieberle, P. (2008). Re-investigation on odour thresholds of key food aroma compounds and development of an aroma language based on odour qualities of defined aqueous odorant solutions. European Food Research and Technology, 228(2), 265–273. doi:10.1007/s00217-008-0931-x

Dalsgaard, T. K., Nielsen, J. H., & Larsen, L. B. (2007). Proteolysis of milk proteins lactosylated in model systems. Molecular Nutrition & Food Research, 51(4), 404–14. doi:10.1002/mnfr.200600112

Datta, N., Elliott, A. J., Perkins, M. L., & Deeth, H. C. (2002). Ultra-high-temperature (UHT) treatment of milk: comparison of direct and indirect modes of heating. The Australian Journal of Dairy Technology, 57(3), 211–227.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

20

Dolan, J. W. (2009). Calibration Curves, Part II: What are the Limits? LCGC North America, 27(4), 306–312.

Elliott, A. J., Datta, N., Amenu, B., & Deeth, H. C. (2005). Heat-induced and other chemical changes in commercial UHT milks. Journal of Dairy Research, 72(4), 442–446.

Elliott, A. J., Dhakal, A., Datta, N., & Deeth, H. C. (2003). Heat-induced changes in UHT milks - Part 1. Astralian Journal of Dairy Technology, 58(1), 3–10.

Flavornet. (2014). Retrieved July 14, 2014, from http://www.flavornet.org/flavornet.html

ISO. (1993). . ISO 8586-1. General guidance for the selection, training and monitoring of assessors, ISO, Geneva, Switzerland.

Itan, Y., Jones, B. L., Ingram, C. J. E., Swallow, D. M., & Thomas, M. G. (2010). A worldwide correlation of lactase persistence phenotype and genotypes. BMC Evolutionary Biology, 10(36). doi:10.1186/1471-2148-10-36.

Jansson, T., Jensen, S., Eggers, N., Clausen, M. R., Larsen, L. B., Ray, C., Sundgren, A., Andersen, H. J., & Bertram, H. C. (2014a). Volatile component profiles of conventional and lactose-hydrolyzed UHT milk —a dynamic headspace gas chromatography-mass spectrometry study. Dairy Science & Technology, 94(4), 311–325. doi:10.1007/s13594-014-0164-7

Jansson, T., Clausen, M.R., Sundekilde, U. K., Eggers, N., Nyegaard, S., Larsen, L. B., Ray, C., Sundgren, A., Andersen, H. J., & Bertram, H. C. (2014b). Lactose-Hydrolyzed Milk Is More Prone to Chemical Changes during Storage than Conventional Ultrahigh-Temperature (UHT) Milk. Journal of Agricultural and Food Chemistry, in press. doi.org/10.1021/jf501671z.

Jeon, I. J., Thomas, E. L., & Reineccius, G. A. (1978). Production of volatile flavor compounds in ultrahigh-temperature processed milk during aseptic storage. Journal of Agricultural and Food Chemistry, 26(5), 1183–1188. doi:10.1021/jf60219a039

Jokar, A., & Karbassi, A. (2011). In-house production of lactose-hydrolysed milk by beta-galactosidase from Lactobacillus bulgaricus. Journal of Agricultural Science and Technology, 13, 577–584.

Kato, Y., Matsuda, T., Kato, N., & Nakamura, R. (1988). Browning and protein polymerization induced by amino-carbonyl reaction of ovalbumin with glucose and lactose. Journal of Agricultural and Food Chemistry, 36(1973), 806–809. doi: 10.1021/jf00082a034

Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2013). Package “ lmerTest ” (pp. 1–15). Retrieved from http://cran.r-project.org/web/packages/lmerTest/lmerTest.pdf

Lawless, H. (1995). Dimensions of sensory quality: a critique. Food Quality and Preference, 6(3), 191–199. doi: 10.1016/0950-3293(94)00023-O

Lorenzen, P. C., Clawin-Rädecker, I., Einhoff, K., Hammer, P., Hartmann, R., Hoffmann, W., Martin, D., Molkentin, J., Walte, H. G., & Devrese, M. (2011). A survey of the quality of extended shelf life (ESL) milk in relation to HTST and UHT milk. International Journal of Dairy Technology, 64(2), 166–178. doi:10.1111/j.1471-0307.2010.00656.x

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

21

McKellar, R. C., Froehlich, D. A., Butler, G., Cholette, H., & Campbell, C. (1984). The effect of uncooled storage on proteolysis bitterness and apparent viscosity in ultra-high-temperature milk. Canadian Institute of Food Science and Technology, 17(1), 14–17.

Messia, M. C., Candigliota, T., & Marconi, E. (2007). Assessment of quality and technological characterization of lactose-hydrolyzed milk. Food Chemistry, 104(3), 910–917. doi:10.1016/j.foodchem.2006.12.045

Nagata, Y. (2014). Odor measurement review (pp. 118–133). Retrieved from http://www.orea.or.jp/en/PDF/Odor_Measurement_Review,.pdf#page=113

Ochi, H., Sumi, M., Nakata, I., Saito, H., Uozumi, M., & Iwatsuki, K. (2010). Sensometric calibration of sensory characteristics of commercially available milk products with instrumental data. Journal of Dairy Science, 93(5), 1794–806. doi:10.3168/jds.2009-2575

Pellegrino, L., De Noni, I., & Resmini, P. (1995). Coupling of lactulose and furosine indices for quality evaluation of sterilized milk. International Dairy Journal, 5(7), 647–659. doi:10.1016/0958-6946(95)00036-3

Rerkrai, S., Jeon, I. J., & Bassette, R. (1987). Effect of various direct ultra-high temperature heat treatments on flavor of commercially prepared milks. Journal of Dairy Science, 70(10), 2046–2054. doi:10.3168/jds.S0022-0302(87)80252-7

Singh, R. R. B., Ruhil, A. P., Jain, D. K., Patel, A. A., & Patil, G. R. (2009). Prediction of sensory quality of UHT milk – A comparison of kinetic and neural network approaches. Journal of Food Engineering, 92(2), 146–151. doi:10.1016/j.jfoodeng.2008.10.032

Tetra Pak (2003) Dairy processing handbook (2nd ed.) Lund, Sweden: Tetra pak processing systems AB.

Thomas, E. L., Burton, H., Ford, J. E., & Perkin, A. G. (1975). The effect of oxygen content on flavour and chemical changes during aseptic storage of whole milk after ultra-high-temperature processing. Journal of Dairy Research, 42(02), 285–295. doi:10.1017/S0022029900015326.

Tossavainen, O. (2008). Heat-induced changes in lactose hydrolysed milks. Doctoral Dissertation

Valero, E., Villamiel, M., Miralles, B., Sanz, J., & Martínez-Castro, I. (2001). Changes in flavour and volatile components during storage of whole and skimmed UHT milk. Food Chemistry, 72(1), 51–58. doi: 10.1016/S0308-8146(00)00203-X

Vallejo-cordoba, B., & Nakai, S. (1994). Keeping-quality assessment of pasteurized milk by multivariate analysis of dynamic headspace gas chromatographic data .1. Shelf-life prediction by principal component regression. Journal of Agricultural and Food Chemistry, 42, 989–993. doi:10.1021/jf00040a029

Vazquez-Landaverde, P. A., Velazquez, G., Torres, J. A., & Qian, M. C. (2005). Quantitative determination of thermally derived off-flavor compounds in milk using solid-phase microextraction and gas chromatography. Journal of Dairy Science, 88(11), 3764–72. doi:10.3168/jds.S0022-0302(05)73062-9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

22

de Wit, R., & Nieuwenhuijse, H. (2008). Kinetic modelling of the formation of sulphur-containing flavour components during heat-treatment of milk. International Dairy Journal, 18, 539-547.

Zenith International Ltd, Bath, UK (2012). Opportunity for Lactose Free Dairy Report 2012. 115pp.

FIG

UR

E 1

Figu

re 1

: In

ten

sity

of

sen

sory

att

rib

ute

s in

dif

fere

nt

typ

es o

f m

ilk (

LHD

: lac

tose

-hyd

roly

zed

milk

exp

ose

d t

o d

irec

t U

HT

trea

tme

nt;

LH

I: la

cto

se-h

ydro

lyze

d m

ilk e

xpo

sed

to

ind

irec

t U

HT

trea

tmen

t; C

ON

VI:

co

nve

nti

on

al m

ilk e

xpo

sed

to

ind

irec

t U

HT

trea

tmen

t) d

uri

ng

a st

ora

ge p

eri

od

of

fou

r m

on

ths.

Lo

wer

cas

e le

tte

rs in

dic

ate

stat

isti

cal d

iffe

ren

ces

bet

we

en d

iffe

ren

t m

ilk a

t th

e sa

me

sto

rage

tim

e (P

≤0.0

5).

Fig

ure

1

FIGURE 2

Figure 2: (A) Scores and (B) Loadings from PLS1 regression with methyl ketones (6) and aldehydes (6) detected by GC-MS as X-

variable and stale flavor determined from sensory analysis as Y-variable. Different samples of milk (LHD: lactose-hydrolyzed milk

exposed to direct UHT treatment (dark gray); LHI: lactose-hydrolyzed milk exposed to indirect UHT treatment (light grey); CONVI:

conventional milk exposed to indirect UHT treatment (black)) were for up to four months.

Figure 2

FIGURE 3

Figure 3: Correlation plot of the level of primary amines (N-terminals) (X-axis) and bitter taste intensity of three different milk types

(LHD: lactose-hydrolyzed milk exposed to direct UHT treatment (dark grey); LHI: lactose-hydrolyzed milk exposed to indirect UHT

treatment (light grey); CONVI: conventional milk exposed to indirect UHT treatment (black)) stored for up to four months.

Figure 3

TABEL 1

Table 1: Specifications of milk samples

Milk Storage time

[months]

Type of heat treatment

Amount of lactose

[% w/w]

Amount of fat [% w/w]

Amount of protein [% w/w]

Conventional 1a, 3 and 4 Indirect 5 1.5 3.5-3.6

Lactose- hydrolyzed

0b, 3 and 4 Direct 0.01 1.5 3.5-3.6

0b, 3 and 4 Indirect 0.01 1.5 3.5-3.6 aReferred to as fresh, bTested after 1 week of storage

Table 1

TABLE 2

Table 2: Content [ng∙ g-1 of milk, ppb] of individual volatiles in different types of milk (LHD: lactose-hydrolyzed milk exposed to direct UHT treatment; LHI: lactose-hydrolyzed milk exposed to indirect UHT treatment; CONVI: conventional milk exposed to indirect UHT treatment) differing with respect to storage time together with odor detection threshold of each volatile. Different letters indicate significant differences (P ≤ 0.05) between milk samples of equal storage time.

Milk type Odor Threshold

[ppb] CONVI

LHD

LHI

CONVI

LHD

LHI

CONVI

LHD

LHI

Fresh

3 months

4 months

2-butanone Sweet, apricota 440d 43.57

97.02

116.69

144.25 b 20.93 a 28.09 a

181.16 B 147.16 ab 28.66 a

2-methylbutanal Maltb 2b 0.00

0.00

0.00

0.00 a 6.48 b 3.81 ab

2.75

5.96

7.63

2-pentanone Ether, fruita 70a 15.15

13.81

29.58

22.19

18.02

14.13

40.27 b 29.39 ab 20.61 a

Decane Alkanec 620d 7.96

8.01

9.37

7.88

8.66

9.18

10.29

9.33

10.90

Toluene Paintc 330d 7.59

18.06

23.18

10.35

11.02

13.03

26.28 B 29.81 b 11.68 a

DMDS Onion, cabbagec

2d 1.72

1.35

1.28

0.44 a 1.83 b 1.86 b

1.84 ab 0.50 a 2.59 b

2-hexanone Etherc 24d 0.00

3.27

12.41

2.67

6.45

3.47

12.65

7.99

5.72

Hexanal Green, grassyb 2b 0.00

8.56

15.28

10.37 b 1.10 a 1.42 a

15.88 b 18.45 b 5.32 a

2-heptanone Fruit, spice, cinnamon, bananaa

7d 231.31

366.45

767.06

351.04

327.80

283.67

822.70 b 714.75 b 373.28 a

Heptanal Fattya, pungenta

0.2d 0.00

5.33

8.01

0.00

13.12

3.23

10.54

10.39

4.60

2-octanone Floral, fruit, greena

41-62a 0.00

1.98

6.31

0.00

1.31

1.09

7.29 b 5.15 ab 4.10 a

Octanal Citrus, greenb 3b 0.00

2.02

2.47

0.00

0.00

0.00

4.44

2.81

5.57

DMTS Cabbageb 0.01b 2.20

2.05

1.86

0.85 a 3.05 b 1.99 ab

1.27 ab 0.42 a 2.50 b

2-nonanone Fruit, greena 58-82a 12.06

16.79

45.23

18.46

18.01

11.98

42.17 b 34.69 b 16.98 a

Nonanal Citrus, soapb 3b 0.80

2.40

5.45

2.11

6.13

2.30

6.68

4.78

5.72

Tetradecane

1.04

0.04

1.37

0.61 a 2.08 b 0.07 a

0.55

0.89

0.47

Decanal Sweet, wax, floral, citrusa

0.1-6a 0.00

6.41

8.25

3.21

4.54

6.14

11.19

6.42

12.27

aBurdock, 2010, bCzerny et al., 2008, c”Flavornet,” 2014 and dNagata, 2014 , pp. 122-123

Table 2

TABLE 3

Table 3: Level of primary amines (mM) in three different types of milk (LHD: lactose-hydrolyzed milk exposed to direct UHT treatment; LHI: lactose-

hydrolyzed milk exposed to indirect UHT treatment; CONVI: conventional milk exposed to indirect UHT treatment) during a storage period of four

months. Lower case letters indicate statistical differences (P≤0.05) between milk samples.

Category Storage

(months) N-amino terminals (mM)

CONVI Fresh 0.50 e

3 0.53 de

4 0.71 c

LHD Fresh 0.56 de

3 1.00 ab

4 1.01 a

LHI Fresh 0.61 de

3 0.92 b

4 1.02 a

Table 3

chemical changes and off-flavor development in lactose ... · chemical changes and off-flavor...

Documents