effect of carbohydrate composition in barley and oat...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Nov. 2009, p. 7006–7016 Vol. 75, No. 220099-2240/09/$12.00 doi:10.1128/AEM.01343-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Effect of Carbohydrate Composition in Barley and Oat Cultivars onMicrobial Ecophysiology and Proliferation of Salmonella enterica in

an In Vitro Model of the Porcine Gastrointestinal Tract�†Robert Pieper,1,2‡§ Jerome Bindelle,1,2,4,5§ Brian Rossnagel,3

Andrew Van Kessel,1 and Pascal Leterme2*Department of Poultry and Animal Science, University of Saskatchewan, 51 Campus Drive, Saskatoon, Saskatchewan S7N 5A8,

Canada1; Prairie Swine Centre Inc., 2105 8th Street East, Saskatoon, Saskatchewan S7H 5N9, Canada2; Department ofPlant Science, University of Saskatchewan, 51 Campus Drive, Saskatoon, Saskatchewan S7N 5A8, Canada3; Animal

Science Unit, Gembloux Agricultural University, Passage des Deportes 2, 5030 Gembloux, Belgium4; andFonds de la Recherche Scientifique-FNRS, Rue d’Egmont 5, 1000 Brussels, Belgium5

Received 9 June 2009/Accepted 17 September 2009

The influence of the carbohydrate (CHO) composition of cereal cultivars on microbial ecophysiology wasstudied using an in vitro model of the porcine gastrointestinal tract. Ten hull-less barley cultivars, six barleycultivars with hulls, six oat cultivars, and six oat groats that differed in �-glucan, nonstarch polysaccharide(NSP), and starch contents and starch type were hydrolyzed enzymatically and incubated for 72 h with pigfeces. Fermentation kinetics were modeled, and microbial compositions and short-chain fatty acid (SCFA)profiles were analyzed using terminal restriction fragment length polymorphism and gas chromatography.Cluster analysis and canonical ordination revealed different effects on fermentation and microbial ecologydepending on the type of CHO and cultivar. First, in cultivars of barley with hulls and oats, the cellulose andinsoluble NSP contents (i) increased Ruminococcus flavefaciens-like and Clostridium xylanolyticum-like phylo-types, (ii) increased acetate production, and (iii) decreased fermentation activity. Second, in hull-less barleycultivars the �-glucan, amylose, amylopectin, crude protein, and soluble NSP contents determined the micro-bial community composition and activity as follows: (i) the amylose contents of the hull-less barley varietiesincreased the butyrate production and the abundance of Clostridium butyricum-like phylotypes, (ii) the �-glu-can content determined the total amounts of SCFA, and (iii) the amylopectin and starch contents affected theabundance of Clostridium ramosum-like phylotypes, members of Clostridium cluster XIVa, and Bacteroides-likebacteria. Finally, the effect of CHO on proliferation of Salmonella enterica in the model was determined.Salmonella cell counts were not affected, but the relative proportion of Salmonella decreased with hull-lessbarley cultivars and increased with oat cultivars as revealed by quantitative PCR. Our results shed light on thecomplex interactions of cereal CHO with intestinal bacterial ecophysiology and the possible impact on hosthealth.

The gastrointestinal tract (GIT) of pigs is colonized by ahighly diverse microbial community, which can be affected byvarious factors, including diet and environmental factors (15,38). Manipulating the composition and metabolic activity ofthe gut microbiota through the diet to improve gut health is anincreasing focus of nutritionists in the postantibiotic era. Dif-ferent strategies, including the use of prebiotics, probiotics,organic acids, or zinc, have been used to manipulate the intes-tinal ecosystem (31, 36). Surprisingly, the fact that compoundsin the basal diet, such as indigestible carbohydrates (CHO) incereal, can also affect the intestinal microbial ecophysiology isoften neglected by nutritionists. Since cereals are a major com-

ponent in the diets of pigs and other monogastric species andsince intestinal bacteria vary in their genetic potential for sub-strate utilization, there is great potential to beneficially manip-ulate microbial ecology in the GIT by choosing cereal cultivarswith specific CHO compositions.

The cereal CHO composition can be markedly different indifferent cultivars (17, 19), and CHO fractions, such as �-glu-cans, could be used as functional food ingredients (6). Forexample, Pieper et al. (32) recently showed that the intestinalmicrobial community composition can be modified using thevariability in the �-glucan content of barley and oat cultivars.The specific mode of action of the other CHO fractions thatare present in these cultivars, such as cellulose, soluble andinsoluble nonstarch polysaccharides (NSP), and starch (e.g.,starch content and amylose/amylopectin ratio), on the micro-bial communities and their activity is still unclear.

In addition to direct effects on intestinal physiology, changesin intestinal microbial composition may enhance or suppressthe growth of specific pathogenic microorganisms by compet-itive exclusion. For example, Salmonella infections are amongthe most frequent and widespread zoonoses in the world, andthere might be opportunities to influence Salmonella coloniza-

* Corresponding author. Mailing address: Prairie Swine Centre Inc.,2105 8th Street East, Saskatoon, SK S7H 5N9, Canada. Phone: (306)667-7445. Fax: (306) 955-2510. E-mail: [email protected].

‡ Present address: Institute of Animal Nutrition, Department ofVeterinary Medicine, Freie Universitat Berlin, Brummerstrasse 34,D-14195 Berlin, Germany.

§ R.P. and J.B. contributed equally to this work.† Supplemental material for this article may be found at http://aem

.asm.org/.� Published ahead of print on 25 September 2009.

7006

on August 26, 2019 by guest

http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

tion via nutritional strategies (30). Studying the interactions ofintestinal bacteria using an in vitro simulation of the porcineGIT could help rapidly screen and evaluate promising strate-gies for reducing Salmonella levels in pigs without the use ofanimal infection models.

The aim of the present study was to determine the influenceof variation in the CHO composition of 10 hull-less barleycultivars, six common barley cultivars, and six oat cultivars andtheir corresponding oat groats (dehulled oats) on large intes-tine fermentation characteristics and microbial communitycomposition. An in vitro model of the porcine GIT describedby Bindelle et al. (5) was employed with or without coinocu-lation with Salmonella enterica, using multivariate canonicalanalysis.

(This study was presented in part at the 11th DigestivePhysiology in Pigs Symposium, Reus, Spain, 19 to 22 May2009.)

MATERIALS AND METHODS

Substrates and chemical characterization. Ten hull-less barley cultivars, sixbarley cultivars with hulls, six oat cultivars, and the groats (dehulled kernels) ofthese oat cultivars were chosen based on typical characteristics, especially the�-glucan and starch contents, as well as the amylose/amylopectin ratio (see TableS1 in the supplemental material). Most of the cultivars were developed andprovided by the Crop Development Centre (CDC) and were grown between2004 and 2006 at the University of Saskatchewan. The panel of cultivars wascompleted with some commercially available cultivars of cereals (McLeod andAC Metcalfe).

Samples were analyzed to determine the dry matter content (method 967.03,AOAC, 1990), crude protein (CP) content (method 981.10; AOAC, 1990), ashcontent (method 923.03; AOAC, 1990), and ether extract (method 920.29;AOAC, 1990). The starch content, amylose/amylopectin ratio, and water-soluble�-glucan content were analyzed colorimetrically after enzymatic hydrolysis usingstandard procedures (Megazyme Ltd., Ireland). The total NSP (tNSP), solubleNSP (sNSP), and insoluble NSP (iNSP) fractions were determined by gas chro-matography using a Varian Star 3400 gas chromatograph equipped with a 30-mfused silica capillary column and a gas flow rate of 36.15 cm/s after the sampleswere hydrolyzed with 12 M H2SO4 as described by Englyst and Hudson (12).

In vitro hydrolysis and fermentation. In vitro hydrolysis and fermentationwere performed using the procedure described by Bindelle et al. (5). Briefly,cereal samples were hydrolyzed with porcine pepsin (pH 2, 39°C, 2 h) andporcine pancreatin (pH 6.8, 39°C, 4 h), and residues were filtered through a42-�m nylon cloth, washed twice with 96% ethanol and 99.5% acetone, and driedat 60°C. The digestible dry matter (dDM) after hydrolysis was recorded. Resi-dues from different hydrolysis replicates of one cultivar were pooled and incu-bated in an inoculum prepared from fresh feces from three growing pigs thatwere fed a nonmedicated diet; the feces were mixed with a buffer solution (29).Fermentation was performed at 39 � 0.5°C using 200 mg of the hydrolyzedresidues and 30 ml of the inoculum placed in 140-ml glass bottles equipped withrubber stoppers. The experimental scheme was as follows: 28 ingredients, threereplicates, and three blanks (containing only inoculum). The gas pressure in thebottles was regularly recorded for 72 h. After 72 h, the fermentation brothwas centrifuged (12,000 � g, 5 min), and the supernatant was removed foranalysis of short-chain fatty acids (SCFA). The pellet was used for extractionof genomic DNA.

Analysis of SCFA. The supernatant of a centrifuged sample (1 ml) was acid-ified to pH 2.5 with metaphosphoric acid, and an internal standard (crotonic acidsolution [2 mg/ml] in double-distilled H2O) was added. SCFA were analyzed bygas chromatography using a fused silica capillary column (30 m by 320 �m by0.25 �m; ZB-FFAP; Phenomenex, Torrance, CA) in an Agilent 6890 gas chro-matography system equipped with a flame ionization detector (Agilent, Boblin-gen, Germany). Helium at a flow rate of 1.9 ml/min was the carrier gas. The flowrates of hydrogen and air were 35 and 350 ml/min, respectively. The initial oventemperature was 100°C, followed by an increase at a rate of 8°C/min and then afinal temperature of 200°C for 13 min.

DNA extraction and terminal restriction fragment length polymorphism(TRFLP) analysis. Genomic DNA of each sample was extracted using aFastDNA kit (Qiagen, Mississauga, Ontario, Canada) according to the manu-facturer’s instructions.

For analysis of the microbial communities, a partial fragment of the bacterial16S rRNA gene was amplified by PCR using universal forward primer S-D-Bact-0008-a-S-20 (AGA GTT TGA TCM TGG CTC AG) labeled with 6-carboxyf-lourescein and reverse primer S-D-Bact-0926-a-S-20 (CCG TCA ATT CATTTG AGT TT) (23). The PCR mixtures contained 5 �l of 10� incubation buffer,1.5 �l of 50 mM MgCl2, 1.5 �l of each primer (10 �M), 1.5 �l of each de-oxynucleoside triphosphate (10 mM), 0.2 �l of Taq polymerase (5 U/�l), andUV-sterilized Millipore water added until the volume was 50 �l. Each PCR wasperformed with a Thermolyne Amplitron II temperature cycler (Barnstead/Thermolyne, Dubuque, IA) with the program set as follows: 5 min at 95°C,followed by 30 cycles of 95°C for 40 s, 55°C for 40 s, and 72°C for 60 s and thena final extension at 72°C for 10 min. The sizes and yields of PCR products werechecked by electrophoresis in a 1.5% agarose gel after staining with ethidiumbromide (0.5 �g ethidium bromide/ml agarose). Each PCR product was subse-quently extracted from the gel using a Qiagen PCR purification kit (Qiagen,Mississauga, Ontario, Canada) according to the manufacturer’s protocol, and theDNA concentration was measured with a NanoDrop ND-1000 spectrophotom-eter (NanoDrop Technologies, Inc., Wilmington, DE).

For TRFLP analysis, 200 ng of the PCR product was digested at 37°C for 6 husing 15 U of MspI (Fermentas, Burlington, Canada) in 2 �l reaction buffer andenough UV-sterilized Millipore water to bring the volume to 20 �l. Two micro-liters of the digestion solution was subsequently mixed with 9 �l of formamideand 0.5 �l of an internal size standard (ABI GeneScan 600 LIZ size standard)and denatured at 95°C for 5 min, which was followed by immediate cooling on icefor 2 min. Fragment sizes were analyzed using an ABI 3130xl genetic analyzer ingene scan mode and GeneMapper v3.7 software. Fragments that differed by �3bp were considered identical as a binning criterion.

To identify the dominant bacterial species, a small clone library (n � 96) wasconstructed from purified PCR products of pooled DNA isolates using theprimers mentioned above (without 6-carboxyflourescein) and standard cloningprocedures as described in the manufacturer’s protocol (pGEM-T Easy vectorsystem; Promega, Madison, WI). Forward sequences were obtained with an ABI3730 capillary sequencer and aligned using greengenes (10; http://greengenes.lbl.gov/). Sequences of the closest cultured relatives were retrieved and incorpo-rated into the ARB phylogenetic software (24; www.arb-home.de) to assignbacterial species or at least bacterial groups to individual terminal restrictionfragments (TRFs) using the TRFcut tool (34; http://www.mpi-marburg.mpg.de/downloads/). The theoretical fragment sizes were calculated for the enzymeMspI, and fragments whose sizes had �1 bp similarity with TRFs obtained wereconsidered matches.

In vitro fermentation and proliferation of S. enterica. Based on the results ofthe first experiment, 10 barley cultivars and six oat cultivars and groats wereselected for study using a coinoculation approach (fecal inoculum plus Salmo-nella). Citrus pectin (Sigma P-9135) was used as a negative control (7), andtryptone soya broth (TSB) medium was used as the growth substrate for entero-bacteria.

A strain of S. enterica subsp. enterica serotype Typhimurium var. Copenhagenthat was resistant to two antibiotics (novobiocin and nalidixic acid [Novr Nalr]),which was obtained from a grow-finish herd during a survey of western Canadianswine herds by the Western College of Veterinary Medicine (Saskatoon, SK,Canada) and the Alberta Research Council (Edmonton, AB, Canada) was se-lected for this study. Salmonella was usually cultivated aerobically at 37°C ineither TSB or on brilliant green agar plates containing 25 �g/ml of each of thetwo antibiotics.

In the coinoculation approach, the microbial communities in the fermentationbottles were allowed to adapt to the substrate initially for 6 h before inoculationwith Salmonella. The Salmonella strain was then inoculated with a syringe aftergas pressure measurement and gas release to obtain a total concentration of3.20 � 0.2 log CFU/ml. Fermentation was allowed to proceed for 24 h. Samplesof mixed fermentation broth (0.1 ml) were taken at 6, 12, and 24 h and imme-diately plated onto brilliant green agar plates (Nov� Nal�), and the plates wereincubated as described above. Salmonella colonies were counted, and the resultswere expressed in log CFU/ml incubation broth. After 24 h, the fermentationbroth was centrifuged (12,000 � g, 5 min), and the pellet was used for extractionof genomic DNA as described above.

qPCR analysis of bacterial communities. Total genomic DNA was extractedfrom samples in the second experiment as described above. Quantitative real-time PCR (qPCR) was performed using previously described primer sets andannealing temperatures (Table 1). Total bacterial counts and counts for Salmo-

VOL. 75, 2009 CEREAL CARBOHYDRATES AND INTESTINAL MICROBIAL ECOLOGY 7007


http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

nella, enterobacteria, Clostridium cluster XIVa, Clostridium cluster IV, Clostrid-ium cluster I, lactobacilli, and Bacteroides were obtained. Amplification wasperformed using an iQ SYBR green Supermix (Bio-Rad, Guenette, Canada).The amplification conditions were 95°C for 10 min, followed by 40 cycles of 95°Cfor 30 s, 50 to 60°C (depending on the bacterial species [Table 1]) for 40 s, and72°C for 40 s. The amplification was performed using a CFX96 real-time PCRdetection system with a C1000 thermal cycler (Bio-Rad, Guenette, Canada), andthe data were collected at the annealing-extension step. Standard curves weregenerated using serial dilutions of purified genomic DNA of S. enterica forSalmonella quantification. For quantification of the total bacteria, Clostridiumclusters XIVa, IV, and I, enterobacteria, lactobacilli, and Bacteroides, purifiedPCR products that were obtained by standard PCR using the primers shown inTable 1 were used. The detection limit was 102 copies/ml fermentation broth.Melting curves were checked after amplification in order to ensure correctamplification results. The total count results were expressed in log gene copynumber/ml fermentation broth, whereas the values for the other bacterial groupswere expressed in relative numbers compared to total bacteria.

Statistical analysis and calculations. The in vitro digestibility of cereal drymatter after pepsin and pancreatin hydrolysis was calculated. Gas accumulationcurves for fermentation of hydrolyzed cereals were modeled as described byFrance et al. (14):

V � 0

if 0 � t � lag, and

V � Vf �1 � exp�b�t � lag� � c��t � lag��

if t � lag, where V is the gas accumulation, Vf (ml g�1 initial amount of cereal)is the maximum gas volume for t � �, and lag (h) is the lag time before thefermentation starts. The constants b (h�1) and c (h�1/2) determine the fractionalrate of degradation of the substrate (�) (h�1), which is postulated to vary withtime as follows:

� � b �c

2�t

if t � lag.The half-time to asymptotic gas production when V � Vf /2 was indicated by

T1/2.For analysis of bacterial communities, TRFLP profiles were normalized by

calculation of the relative peak area of each individual peak, and only fragmentswith a relative peak area ratio of �1% and fragments larger than 80 bp wereconsidered for a cluster analysis using Pearson correlation and the unweightedpair group method with algorithmic means with Statistica software (version 6.0;Statsoft, Tulsa, OK).

To analyze the complex interactions of cereal carbohydrate fractions on invitro digestibility, fermentation kinetics parameters, SCFA profiles, and relativeamounts of bacterial groups and species, we performed a multivariate analysisusing the CANOCO statistical package, version 4.5 (37). The values for starch,amylose, amylopectin, �-glucan, cellulose, tNSP, sNSP, iNSP, and CP contents in

the cereal cultivars were imported as explanatory variables. Square root-trans-formed values for the relative abundance of TRFs, SCFA production and molarratio, parameters of fermentation kinetics, and dDM values were used as re-sponse variables. Explanatory and response data were used for constrained linearordination analysis (redundancy analysis [RDA]). Whereas unconstrained ordi-nations, such as principal component analysis, are methods that seek one or moregradients representing predictors that best explain response variable composi-tion, in constrained ordinations such as RDA, these predictors are further re-stricted and ordination axes must be generated from linear combinations ofweighted environmental variables. The explanation of response variables withsynthetic variables (ordination axes) can therefore be further defined usingvalues for the explanatory characteristics (37). The significance of the overallordination model and the effect of explanatory variables during developmentof the ordination model were tested using the Monte Carlo permutation test(n � 499).

Finally, statistical analyses of dDM during pepsin-pancreatin hydrolysis, invitro gas production kinetic parameters (lag, T1/2, Vf), total SCFA production,and molar ratios of individual SCFA after 72 h of in vitro fermentation, as wellas qPCR results from the second coinoculation run, were performed by usinganalysis of variance, followed by Tukey’s honestly significant difference test usingSPSS (version 17.0; SPSS, Chicago, IL). P values of �0.05 were consideredsignificant.

Nucleotide sequence accession numbers. Sequences have been deposited inthe GenBank database under accession numbers GQ214260 to GQ214312.

RESULTS

Chemical compositions of cereals. The chemical composi-tions of the cereals used in this study are presented in Table S1in the supplemental material. There was great variation be-tween cereal types and cultivars. The �-glucan content rangedfrom 4.6% (CDC McGwire) to 12.7% (CDC Fibar) in hull-lessbarley cultivars and from 4.1 to 5.9% in common barley culti-vars, whereas slightly lower values were found in the oat cul-tivars (2.9 to 5.1%). The “waxy” hull-less barley cultivars CDCRattan, SR93139, CDC Fibar, SB94917, and HB393 containedlow levels of amylose starch, whereas in the “high-amylase”cultivars SH99250 and SB94893 the amylose accounted for38.9 and 46.1% of the total starch concentration, respectively.The tNSP contents varied from 7.7 to 15.3% for hull-less barleycultivars, from 11.9 to 17.2% for common barley cultivars, andfrom 16.6 to 26.1% for oat cultivars, likely due to the highercontent of cellulose and lignified hulls.

In vitro fermentation parameters. The dDM content afterenzymatic hydrolysis and the fermentation parameters are pre-

TABLE 1. Primers used for qPCR

Target organisms Primer sequences (5�–3�) Ampliconsize (bp)

Annealingtemp (°C) Reference

All bacteria CGGYCCAGACTCCTACGGG 200 60 21TTACCGCGGCTGCTGGCAC

Clostridium cluster XIVa AAATGACGGTACCTGACTAA 440 50 28CTTTGAGTTTCATTCTTGCGAA

Clostridium cluster IV GCACAAGCAGTGGAGT 239 50 27CTTCCTCCGTTTTGTCAA

Clostridium cluster I TACCHRAGGAGGAAGCCAC 346 63 35GTTCTTCCTAATCTCTACGCAT

Bacteroidetes CTTCCTCCGTTTTGTCAA 212 60 27GRCCTTCCTCTCAGAACCC

Lactobacilli GCAGCAGTAGGGAATCTTCCA 346 55 41GCATTYCACCGCTACACATG

Enterobacteria CCTACTTCTTTTGCAACCCACTC 364 60 8ATGGCTGTCGTCAGCTCGT

Salmonella AGCCAACCATTGCTAAATTGGCGCA 430 60 1GGTAGAAATTCCCAGCGGGTACTG

7008 PIEPER ET AL. APPL. ENVIRON. MICROBIOL.


http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

sented in summarized form for the different cereal types inTable 2. Values for individual cereals and a statistical compar-ison of the means for cultivars are presented in Table S2 in thesupplemental material. To illustrate the variability, the meanvalues, standard deviations, and ranges are presented for eachtype of grain. The in vitro dDM was different for different typesof grain and cultivars and was partially related to the amylose/amylopectin ratio in hull-less barley cultivars. Common barleyand oat cultivars had similar values with less variability,whereas the highest digestibility was found for oat groats (up to89.7%). The fermentation data also showed that there wasconsiderable variation within and between types of grain. Forexample, the lag time was greater with high-amylose hull-lessbarley cultivars. Most differences were observed for the finalgas volume, which was expressed per g of original nonhydro-lyzed cereal. The starch type and �-glucan content had appar-ent effects on gas production. The values were highest forhigh-amylose barley cultivars SB94893 and SH99250, followedby the high-�-glucan cultivars. The total SCFA production washighest with hull-less barley cultivars and oat groats but rangedfrom 374 mg/g substrate (SB90300) to 535 mg/g substrate(CDC Fibar) for hull-less barley cultivars and from 365 mg/gsubstrate (HiFi) to 459 mg/g substrate (CDC ProFi) for oatgroats. The lowest values (except for the CDC SO-I value)were found for oat cultivars, whereas the values for the com-mon barley cultivars were intermediate. The high-hull-contentoat cultivars had higher molar ratios of acetate (59.7 to 71.2%)and lower values for butyrate (7.6 to 12.0%) than the othervarieties. The highest values for propionate were obtained withoat groats (24.6 and 26.6% for Morgan and CDC ProFi, re-spectively). The levels of branched-chain fatty acids (BCFA) asindicators of protein breakdown were found to be lowest withoat cultivars (0.7 and 4.2% for CDC SO-I and CDC ProFi,respectively), whereas higher values were observed with hull-less barley cultivars (5.3 and 6.1% for SB90354 and SB90300,respectively) and oat groats (5.2 and 5.7% for Morgan andCDC ProFi/CDC Sol-Fi, respectively).

Bacterial community composition. The cluster analysis basedon the TRFs (Fig. 1) revealed distinct bacterial profiles dependingon the cereal type (except for CDC Clyde and SB90300). Twomain clusters were formed by hull-containing common barley andoat cultivars and by hull-less barley cultivars and oat groats. CDCSO-I clustered separately from all other cultivars. However, forcultivars of hull-less barley, there was very high variability andthere were small subclusters compared to the clusters of verysimilar common barley and oat cultivars.

To assign bacterial phylotypes to individual TRFs, an insilico TRF cut tool was implemented in ARB phylogeneticsoftware. The results (Fig. 2) revealed that cereal type andcultivar influenced mainly the abundance of members of clos-tridial clusters IV and XVIa but also clostridial clusters I andXVIII and members of the genus Bacteroides. For example, thelevels of Ruminococcus flavefaciens-like (TRF20) and Clostrid-ium xylanolyticum-like (TRF16) phylotypes were enhancedwith high-cellulose hull-containing cereals, whereas the levelsof other bacteria, belonging to clostridial clusters XIVa and I,were enhanced with hull-less barley cultivars (data not shown).Unfortunately, TRF7, TRF8, TRF9, TRF10, TRF19, andTRF21 could not be identified by this approach. However, they

TA

BL

E2.

Invitro

dDM

,fermentation

kineticparam

eters,andbacterialm

etaboliteprofiles

ofthe

fourcereals

Cerealtype

n

Ferm

entationparam

eters%

oftotalSC

FA

TotalSC

FA

concn(m

g/g)dD

M(%

)L

agtim

e(h)

T1/2

(h)V

f (mlg

�1

�dryw

t�)A

cetatePropionate

Butyrate

BC

FA

Hull-lessbarley

1064.7

�10.4

(45.0–75.3) a1.8

�1.3

(0.9–4.2)9.4

�0.7

(8.3–10.5)77

�25

(47–124)53.3

�0.5

(52.5–54.4)23.2

�0.9

(22.0–24.5)16.3

�0.7

(15.1–17.2)5.7

�0.7

(5.3–6.1)478

�51

(374–535)

Hulledbarley

666.4

�5.9

(55.1–71.1)1.9

�0.8

(1.0–2.3)10.5

�0.2

(10.2–10.8)62

�15

(48–88)56.8

�0.5

(55.9–57.4)21.6

�0.7

(20.3–22.1)15.8

�0.7

(14.9–16.9)4.4

�0.7

(4.1–5.0)378

�23

(354–418)

Oats

664.0

�3.5

(59.3–68.9)1.1

�1.2

(0.4–3.6)18.8

�6.4

(11.6–30.0)34

�17

(25–67)66.2

�4.0

(59.7–71.2)20.4

�1.5

(19.1–23.3)10.2

�1.5

(7.6–12.0)2.5

�1.5

(0.7–4.2)198

�94

(131–380)O

atgroats

687.9

�1.1

(86.3–89.7)1.3

�0.2

(1.1–1.4)8.6

�0.5

(8.0–9.1)22

�2

(19–25)55.3

�1.1

(53.8–56.7)25.7

�0.7

(24.6–26.6)12.9

�0.4

(12.3–13.3)5.5

�0.4

(5.2–5.7)414

�33

(365–459)

aT

hedata

arethe

means

�standard

deviations(ranges).



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

were included in the analyses since they represented �1% ofthe total bacterial communities.

Multivariate canonical analysis. Results of the RDA for theinteraction between cereal nutrient composition and digestibil-ity and fermentation responses for all cereals and for barley,oat, and hull-less barley cultivars separately are presented inFig. 3a to d. The RDA of the four grain types together (n � 28)(Fig. 3a) revealed the major influence of the cellulose (P �0.002), �-glucan (P � 0.006), and CP contents (P � 0.030) anda trend for the amylose content (P � 0.058) in the overallordination model. Cellulose, iNSP, and tNSP were highly cor-related with the first ordinational axis as a consequence of theclustering of the oat cultivars with high hull and cellulosecontents along this axis, whereas the other axis of the modelcovered only 12.1% of the variance. The following is an exam-ple of how to read and interpret the data in the RDA graphs.The acetate molar ratio, T1/2, and R. flavefaciens-like and C.xylanolyticum-like phylotypes (TRF16 and TRF20) werepositively correlated with cellulose, iNSP, and tNSP, as in-dicated by the small angle between the arrows for thesevariables (�90°). An angle of �90° would indicate a nega-tive correlation (i.e., between amylopectin and cellulose inFig. 2a; r � �0.56). Compared to cellulolytic materials, the

sNSP and CP values were positively correlated with thepropionate molar ratio and members of clostridial clusterXIVa (TRF12). Formation of BCFA was associated withClostridium ramosum-like bacteria (TRF17), amylopectin,and starch contents. The concentrations of �-glucan wereassociated with high fermentation activity (Vf), largeamounts of SCFA, a high butyrate molar ratio, and membersof clostridial cluster XIVa (TRF11).

The RDA of barley cultivars (n � 16) (Fig. 3b) highlightedthe strong influence of �-glucan (P � 0.002), tNSP (P � 0.002),and amylopectin (P � 0.002) on the model. The �-glucan andCP contents were positively correlated with the production ofSCFA and BCFA. As observed in the analysis of the four typesof grain together, the acetate molar ratio, the half-time toasymptote (T1/2), and the R. flavefaciens- and C. xylanolyticum-like phylotypes (TRF16 and TRF20) were correlated positivelywith the cellulose, tNSP, and iNSP contents but negatively withthe �-glucan content of the barley cultivars. There was also apositive correlation of amylopectin with the propionate molarratio and the abundance of Bacteroides/Cytophaga-like phylo-types (TRF6), whereas the amylopectin content was negativelycorrelated with the butyrate molar ratio with the barley culti-vars. However, the butyrate molar ratio was positively corre-lated with the lag time, Vf, and Clostridium butyricum-likephylotypes (TRF23).

The RDA of only oat cultivars and oat groats together (n �12) (Fig. 3c) highlighted the strong impact of the cellulosecontent of the oat cultivars on the ordination model (P �0.002). Similar to findings for the overall model, the cellulose,iNSP, and tNSP contents were associated with an elevatedacetate molar ratio, abundant of R. flavefaciens- and C. xylano-lyticum-like phylotypes (TRF16 and TRF20), and fermentationcharacteristics. In contrast, the oat groats were associated withmost of the other response variables.

Finally, an RDA was performed for only hull-less barleyvarieties (n � 10) (Fig. 3d) since they displayed the greatestvariation in nutrient composition between cultivars. As re-vealed by the analysis of all barley cultivars together, the amy-lopectin content was positively correlated with the propionatemolar ratio, dDM, production of BCFA, and Bacteroides/Cy-tophaga-like phylotypes (TRF6). In contrast, the �-glucan con-tent was positively correlated with C. ramosum-like (TRF17)phylotypes, other members of clostridial clusters IV and XIVa(TRF11 and TRF12), and the production of SCFA, whereas itwas negatively correlated with R. flavefaciens-like and C. xyl-anolyticum-like phylotypes (TRF16 and TRF20). Higher molarratios of butyrate were also associated with C. butyricum-likebacteria (TRF23) and positively correlated with the amyloseconcentration. Other bacterial phylotypes reacted differentiallyand showed no clear association with cereal factors.

Impact of cereal cultivars on Salmonella proliferation. Afterinoculation of the S. enterica strain in the second experiment,no significant changes were observed in the numbers of Nalr

Novr S. enterica cells during fermentation, suggesting thatCHO had no effect (data not shown). However, qPCR resultsobtained using genomic DNA extracts after 24 h revealedsignificant (P � 0.05) differences for the relative amount ofSalmonella per total bacterial 16S rRNA gene copy in hull-lessbarley cultivars CDC McGwire, CDC Fibar, SH99250, andSB94893 (Table 3). The values for these four cereal cultivars

FIG. 1. Cluster analysis based on relative TRF abundance in bac-terial TRFLP profiles after 72 h of fermentation of 10 hull-less barleycultivars (light gray), six common barley cultivars (dark gray), six oatcultivars, and six oat groats (black). The cluster was constructed basedon Pearson correlation and the unweighted pair group method withalgorithmic means using Statistica software. The scale bar indicates therelative similarity of profiles.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

were similar to the value for the negative control pectin. Incontrast, the relative proportion was highest in the blank andwith the oat cultivars. The relative contribution of enterobac-teria and lactobacilli was generally low, whereas Clostridiumclusters XIVa, IV, and I and Bacteroides dominated the bac-terial communities. Overall, 77% of the total bacterial com-munities were detected with the qPCR approach used (range,61% [TSB] to 93% [pectin]). Similar to the TRFLP results,there were significant differences depending on the grain typewith Clostridium clusters IV and I. Almost no differences wereobserved for cluster XIVa, whereas Bacteroides showed noclear response to grain type or cultivar.

DISCUSSION

Although the results presented here were obtained with anin vitro model of the porcine GIT, they confirm the greatpotential for manipulating intestinal microbial compositionthrough use of cereal sources containing carbohydrates withprebiotic properties. Due to the use of the in vitro fermenta-tion model, the results are limited to the porcine large intes-tine, but as an improvement compared to other in vitro models,we simulated the nutrient digestion in the upper GIT using apepsin-pancreatin pretreatment, aiming to obtain the indigest-ible fraction of the ingredients, which is likely to undergofermentation in the large intestine in vivo (5). The results ofboth experiments described here support the hypothesis thatthere is a complex interaction between CHO and the microbialecosystem. It is known that contrasting sources of CHO canaffect fermentation characteristics in vitro, causing differencesin the lag time, the slope of gas production curves, and the final

gas volume (2, 4, 5). Furthermore, it has been shown thatcontrasting CHO sources affect the profiles of the fermenta-tion end products (2). Similar to these results, in the presentstudy, differences in fermentation characteristics and bacterialmetabolites were observed between cereal types (hull-less bar-ley, common barley, oat, and oat groats). However, the effectsof the different CHO compositions of cultivars of the samecereal species or type on these parameters in vitro are stillunknown. The results indicate that, especially within the hull-less barley cultivars, there was considerable variation of theseparameters depending on the cultivar. Parameters also par-tially overlapped with other cereal types. These results wereconfirmed by cluster analysis of TRFLP profiles, which showedvery little variation between common barley types and a closerelationship to high-cellulose-content oat cultivars, whereasgreater differences where found for hull-less barley cultivarsand oat groats. This suggests that fibrous materials in the hulls(cellulose, lignin, iNSP) have a dominant effect on the fermen-tation patterns. The fact that CDC SO-I was not included inthe oat-common barley cluster could be explained by its dif-ferential fermentation behavior, starting with a high level ofactivity toward the end of the fermentation (see Table S2 in thesupplemental material). This cultivar is characterized by a lowlignin content, suggesting that there is increased celluloseavailable for fermentation. The great variability with hull-lessbarley cultivars confirms the complex interaction of CHO inthese cultivars, suggesting that other carbohydrate fractions,such as iNSP or sNSP, or the amylose/amylopectin ratio of thestarch fraction had a strong influence on bacterial communitycomposition in vitro.

FIG. 2. Neighbor-joining tree showing the phylogenetic relationships of partial 16S rRNA gene sequences obtained in this study. TRF peaknumbers from the TRFLP analysis are indicated after the species names and refer to the RDA in Fig. 3. Bacillus subtilis was used to root the tree.The bar indicates a calculated evolutionary distance of 10%. Cl., Clostridium.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

FIG. 3. RDA of the effect of nutrient composition (solid arrows) on response variables (dashed arrows), including in vitro fermentation characteristics(dDM, T1/2, Vf, lag time [lag t]), dominant bacterial phylotypes (TRF [see species in Fig. 2]), and bacterial metabolite molar ratios (SCFA, BCFA, acetate,propionate, butyrate). RDA was performed for all types of cereal (n � 28) (a), for only barley (n � 16) (b), for only oat and oat groats (n � 12) (c), and foronly hull-less barley (n � 10) (d). In the RDA, the length, direction, and angle between arrows are direct measures of correlations between variables or betweenvariables and canonical axes (e.g., � � 0° and r � 1; � � 90° and r � 0; and � � 180° and r � �1). The percentages on the x and y axes (e.g., 67.6% and 12.1%in panel a) indicate the proportions of the variability of the data that are described with the corresponding canonical axes in the model. The significance of theoverall model and the effect of nutrients were tested using the Monte Carlo permutation test (n � 499).



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

Despite the great advantage of molecular microbiologicaltools, such as TRFLP, for studying bacterial communities with-out a requirement for cultivation, it has to be noted that thereare some drawbacks, including primer and PCR bias or for-mation of pseudo-TRFs (9, 11). Furthermore, it is difficult toassign bacterial species to individual TRFs since different phy-lotypes could give similar TRFs after restriction digestion. Inthe present study, there were some TRFs that could not beassigned (e.g., TRF7, TRF8, TRF9, TRF10, TRF19, andTRF21), likely due to the size of the clone library. On the otherhand, some TRFs (e.g., R. flavefaciens, C. xylanolyticum, and C.butyricum TRFs) could be assigned to species-like phylotypesdue to their frequent occurrence in the clone library analysis.Furthermore, since samples in the present study were gener-ated based on the same biases, it is possible to compare sam-ples on a relative basis.

To our knowledge, this is the first time that the interactionsof nutrient source and intestinal microbial activity were studiedusing RDA. Using this approach, the effect of the cereal CHOincluded in the grain matrix on the microbial composition andactivity can be visualized. When the data for all cereal typesand varieties together were analyzed (Fig. 3a), high-cellulose-content and iNSP-containing oat and common barley cultivarsshowed slow fermentation and favored cellulolytic, acetate-

producing R. flavefaciens- and C. xylanolyticum-like bacteria(TRF16 and TRF20). These species have specifically adaptedto break down fibrous material (cellulolytic and xylanolytic)during evolutionary coexistence with the host organism (13).They belong to clostridial clusters IV and XIVa, respectively,and are typical colonizers of the distal GIT of monogastricspecies and forestomach of ruminant species (13, 20, 22, 39).Corresponding to the present data, the breakdown of cellulo-lytic materials usually resulted in acetate formation. On theother hand, �-glucans favored C. ramosum-like species (clusterXVIII, TRF17), members of clostridial clusters IV and XIVa(TRF11 and TRF12), and overall SCFA production. In con-trast to these results, isolated �-glucans favored the growth ofClostridium histolyticum-like bacteria and increased the propi-onate molar ratio in another recent in vitro study using humanfecal microbiota (18). This is in contrast to the present findingsand might indicate either that there are differences in thegeneral in vitro model or that isolated �-glucans or wholecereals containing �-glucans in the grain matrix are used (32).RDA of hull-less barley (Fig. 3d) revealed that the high amy-lose content of some hull-less barley cultivars favored the con-tribution of butyrate-producing members of clostridial cluster I(C. butyricum-like phylotypes) but not members of clostridialcluster XIVa, which are commonly involved in butyrate forma-

TABLE 3. qPCR analysis (means) of total bacterial counts (log 16S rRNA gene copy number/ml of fermentation broth) and relativecontributions of six bacterial groups and Salmonella to the overall bacterial community after 24 h of in vitro fermentation of

hydrolyzed cereal varieties, pectin, and TSB using pig feces as the inoculum andwith coinoculation of S. enterica after 6 ha

Cultivar or growth mediumTotal bacterial

counts (logcopies/ml)

% of:

ClostridiumclusterXIVa

Clostridiumcluster IV

Clostridiumcluster I Bacteroides Lactobacilli Enterobacteria

(10�2)Salmonella

(10�4)

Hull-less barley cultivarsSB90354 10.4 A 39.5 A 26.0 CD 11.0 A 7.6 BC 0.2 BC 2.7 A 0.5 CDSB90300 10.1 AB 23.4 B 35.6 BC 12.1 A 4.4 D 0.3 BC 1.3 0.6 CDSR93139 10.1 ABC 28.2 B 30.8 BCD 10.5 A 6.7 BCD 0.2 BC 0.4 C 0.4 CDCDC McGwire 10.3 AB 23.1 B 33.6 BCD 9.3 A 4.2 D 0.2 BC 0.4 C 0.3 DCDC Fibar 10.1 AB 25.7 B 37.2 B 10.4 A 8.7 AB 0.2 BC 0.3 C 0.2 DSH99250 10.3 A 24.2 B 24.8 D 9.9 A 5.3 BCD 0.1 C 0.2 C 0.2 DSB94893 10.2 AB 26.4 B 31.8 BCD 11.4 A 3.6 D 0.2 BC 0.3 C 0.2 D

Common barleycultivarsMcLeod 10.1 AB 24.6 B 24.1 D 10.0 A 10.9 A 0.2 B 0.8 BC 0.9 BCDAC Metcalfe 9.9 BC 25.6 B 40.4 B 14.3 A 5.7 BCD 0.4 B 0.4 C 0.5 CD

Oat cultivarsCDC Dancer 9.9 BC 20.1 B 39.4 B 4.1 B 8.3 ABC 0.3 BC 1.2 ABC 1.6 ABCDC Sol-Fi 9.7 C 24.4 B 53.9 A 5.2 B 6.2 BCD 1.0 A 0.9 BC 0.7 CDCDC SO-I 9.8 C 21.0 B 57.3 A 4.4 B 8.3 AB 0.6 A 0.7 C 1.2 ABC

Oat groatsCDC Dancer groats 10.2 AB 26.7 B 30.5 BCD 2.1 B 5.9 BCD 0.4 BC 1.1 AB 0.8 CDCDC Sol-Fi groats 9.9 BC 28.3 B 37.5 B 3.5 B 7.1 BCD 0.4 BC 0.7 C 0.7 CDCDC SO-I groats 10.1 AB 25.8 B 34.0 BCD 3.2 B 4.6 CD 0.3 BC 0.5 C 0.4 CD

MediaPectin 10.3 A 41.1 A 43.4 A 2.5 B 6.4 BCD 0.2 BC 0.6 C 0.3 DTSB 10.1 AB 21.5 B 30.1 BCD 4.6 B 3.7 D 0.7 A 1.6 ABC 0.7 CDBlank 9.7 CD 19.4 B 48.1 A 10.5 A 11.6 A 0.8 A 2.3 A 2.0 A

SEM 0.042 29.661 43.486 8.054 5.322 0.038 0.009 0.000

a Different letters in a column indicate significant (P � 0.05) differences.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

tion (3, 33). Although butyrate production might be beneficialfor maintaining intestinal health, this result has to be inter-preted with care, since other members of clostridial cluster I,such as Clostridium botulinum, or Clostridium perfringens maybe harmful to the host organism. Finally, the starch and amy-lopectin values were positively correlated with the propionatemolar ratio and Cytophaga/Bacteroides-like bacteria, likely dueto their ability to utilize starch (42).

The differential response of clostridia to the substrates avail-able with hull-less and common barley cultivars, oat cultivars,and oat groats was confirmed by qPCR of dominant bacterialgroups in the second experiment. However, the in vivo contri-bution of these bacterial species might be overestimated withthe current in vitro method as we used a buffered medium. Forexample, bacteria such as Ruminococcus sp. were shown toexhibit less metabolic activity at pH values below 6.3 (16).Furthermore, the level of Bacteroides increases and the level ofRoseburia-like species decreases with a shift from pH 5.5 to pH6.5 (40).

Salmonella infections are among the most frequent andwidespread zoonotic diseases in the world. Since there mightbe opportunities to reduce the prevalence of Salmonella usingfeeding strategies, a coinoculation model was developed tostudy the effect of cereal CHO on Salmonella proliferation invitro. There was no reduction in Salmonella counts due todifferent fermentable substrates, indicating that Salmonellawas able to survive in the buffered system and occupy anecological niche. In a recent study of Martin-Pelaez et al. (25),Salmonella counts were significantly reduced with lactulose asa substrate. However, in their studies Martin-Pelaez et al. (25,26) used very high numbers of Salmonella cells (�log 7.0CFU), which would not naturally occur during a normal Sal-monella infection. Callaway et al. (7), using an in vitro simu-lation technique for ruminal fermentation, found that pectincould significantly reduce the prevalence of Salmonella. Thiswas partially confirmed by our results, but only the relativeproportion of Salmonella was reduced with pectin, as was alsothe case with the hull-less barley cultivars CDC McGwire,CDC Fibar, SH99250, and SB94893. It is possible that thepreferential growth of commensal bacteria compared to Sal-monella (decreasing relative proportion of Salmonella) underin vitro batch culture conditions could result in a reducedabsolute number of Salmonella cells in feces under the contin-uous culture conditions that occur in vivo. However, it is notclear to what extent such results could be transferred to in vivoconditions and reduce Salmonella colonization and/or thetransmission among animals.

Conclusions. The current study revealed that differences inCHO composition between cereal cultivars of the same type ofgrain can affect pig intestinal microbial ecophysiology. Further-more, the effects were revealed by multivariate canonical anal-ysis, showing the usefulness of this approach when intestinalmicrobial ecophysiology and nutrient-microbe interactions arestudied. However, irrespective of the grain type, positive cor-relations were found between acetate production, cellulolyticbacteria, and cellulose content, between butyrate productionand amylose and/or �-glucan content, and between propionateproduction and amylopectin content. This shows the typicalecophysiological signatures of CHO fractions, namely, theamylose/amylopection ratio and �-glucan content in the pig

intestinal tract. Differences between cultivars not only result indifferent microbial ecological responses but could also affectthe susceptibility of the host to opportunistic pathogens, suchas S. enterica, due to the relative amounts of CHO in somehull-less barley varieties.

ACKNOWLEDGMENTS

We thank Agriculture & Agri-Food Canada, administered throughthe Agriculture & Food Council of Alberta (ACAAF project AB0446;Nisku AB, Canada), for funding this study. The continuing core sup-port of the Prairie Swine Centre received from Sask Pork, the Mani-toba Pork Council, Alberta Pork, and the Saskatchewan AgricultureDevelopment Fund is gratefully acknowledged. R. Pieper was fundedby a postdoctoral research grant from the German Academic Ex-change Service (DAAD, Germany).

We are grateful to the technical staff of the Prairie Swine Centre andthe Department of Animal and Poultry Science at the University ofSaskatchewan for their assistance.

REFERENCES

1. Aabo, S., O. F. Rasmussen, I. Rossen, P. D. Sørensen, and J. E. Olsen. 1993.Salmonella identification by the polymerase chain reaction. Mol. Cell.Probes 7:171–178.

2. Awati, A., B. A. Williams, M. W. Bosch, Y. C. Li, and M. W. A. Verstegen.2006. Use of the in vitro cumulative gas production technique for pigs: anexamination of alterations in fermentation products and substrate losses atvarious time points. J. Anim. Sci. 84:1110–1118.

3. Barcenilla, A., S. E. Pryde, J. C. Martin, S. H. Duncan, C. S. Stewart, C.Henderson, and H. J. Flint. 2000. Phylogenetic relationships of butyrate-producing bacteria from the human gut. Appl. Environ. Microbiol. 66:1654–1661.

4. Bauer, E., B. A. Williams, C. Voigt, R. Mosenthin, and W. M. A. Verstegen.2001. Microbial activities of faeces from unweaned and adult pigs, in relationto selected fermentable carbohydrates. Anim. Sci. 73:313–322.

5. Bindelle, J., A. Buldgen, C. Boudry, and P. Leterme. 2007. Effect of inoculumand pepsin-pancreatin hydrolysis on fibre fermentation measured by the gasproduction technique in pigs. Anim. Feed Sci. Technol. 132:111–122.

6. Brennan, C. S., and L. J. Cleary. 2005. The potential use of cereal(133,134)-� glucans as functional food ingredients. J. Cereal Sci. 42:1–13.

7. Callaway, T. R., J. A. Carroll, J. D. Arthington, C. Pratt, T. S. Edrington,R. C. Anderson, M. L. Galyean, S. C. Ricke, P. Crandall, and D. J. Nisbet.2008. Citrus products decrease growth of E. coli O157:H7 and Salmonellatyphimurium in pure culture and in fermentation with mixed ruminal micro-organisms in vitro. Food Pathol. Dis. 5:621–627.

8. Castillo, M., S. M. Martin-Oruem, E. G. Manzanilla, I. Badiola, M. Martin,and J. Gasa. 2006. Quantification of total bacteria, enterobacteria and lac-tobacilli populations in pig digesta by real-time PCR. Vet. Microbiol. 114:165–170.

9. Crosby, L. D., and C. S. Criddle. 2003. Understanding bias in microbialcommunity analysis techniques due to rrn operon copy number heterogene-ity. Biotechniques 34:790–798.

10. DeSantis, T. D., P. Hugenholtz, K. Keller, E. L. Brodie, N. Larsen, Y. M.Piceno, R. Phan, and G. L. Andersen. 2006. NAST: a multiple sequencealignment server for comparative analysis of 16S rRNA genes. Nucleic AcidsRes. 34:W394–W399.

11. Egert, M., and M. W. Friedrich. 2003. Formation of pseudo-terminal restric-tion fragments, a PCR-related bias affecting terminal restriction length poly-morphism analysis of microbial community structure. Appl. Environ. Micro-biol. 69:2555–2562.

12. Englyst, H. N., and G. J. Hudson. 1987. Method for routine measurement ofdietary fibre as non-starch polysaccharides. A comparison with gas liquidchromatography. Food Chem. 24:63–76.

13. Flint, H. J., E. A. Bayer, E. T. Rincon, R. Lahmed, and B. A. White. 2008.Polysaccharide utilization by gut bacteria: potential for new insights fromgenomic analysis. Nat. Rev. Microbiol. 6:121–131.

14. France, J., M. S. Dhanoa, M. K. Theodorou, S. J. Lister, D. R. Davies, andD. Isac. 1993. A model to interpret gas accumulation profiles associated within vitro degradation of ruminant feeds. J. Theor. Biol. 163:99–111.

15. Hill, J. E., S. M. Hemmingsen, B. G. Goldade, T. J. Dumonceaux, J. Klassen,R. T. Zijlstra, S. H. Goh, and A. G. Van Kessel. 2005. Comparison of ileummicroflora of pigs fed corn-, wheat-, or barley-based diets by chaperonin-60sequencing and quantitative PCR. Appl. Environ. Microbiol. 71:867–875.

16. Hiltner, P., and B. A. Dehority. 1983. Effect of soluble carbohydrates ondigestion of cellulose by pure cultures of rumen bacteria. Appl. Environ.Microbiol. 46:642–648.

17. Høltekjolen, A. K., A. K. Uhlen, E. Bråthen, S. Sahlstrøm, and S. H. Knut-sen. 2006. Contents of starch and non-starch polysaccharides in barley vari-eties of different origin. Food Chem. 94:348–358.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

18. Hughes, S. A., P. R. Shewry, G. R. Gibson, B. V. McCleary, and R. A. Rastall.2008. In vitro fermentation of oat and barley derived �-glucans by humanfaecal microbiota. FEMS Microbiol. Ecol. 64:482–493.

19. Izydorczyk, M. S., J. Storsley, D. Labossiere, A. W. MacGregor, and B. G.Rossnagel. 2000. Variation in total and soluble �-glucan content in hullessbarley: effects of thermal, physical, and enzymic treatments. J. Agric. FoodChem. 48:982–989.

20. Julliand, V., A. De Vaux, L. Millet, and G. Fonty. 1999. Identification ofRuminococcus flavefaciens as the predominant cellulolytic bacterial speciesin the equine cecum. Appl. Environ. Microbiol. 65:3738–3741.

21. Lee, D.-H., Y.-G. Zo, and S.-J. Kim. 1996. Nonradioactive method to studygenetic profiles of natural bacterial communities by PCR-single-strand-con-formation polymorphism. Appl. Environ. Microbiol. 62:3112–3120.

22. Leser, T. D., J. Z. Amenuvor, T. K. Jensen, R. H. Lindecrona, M. Boye, andK. Moller. 2002. Culture-independent analysis of gut bacteria: the pig gas-trointestinal tract microbiota revisited. Appl. Environ. Microbiol. 68:673–690.

23. Liu, W. T., T. L. Marsh, H. Cheng, and L. J. Forney. 1997. Characterizationof microbial diversity by determining terminal restriction fragment lengthpolymorphisms of genes encoding 16S rRNA. Appl. Environ. Microbiol.63:4516–4522.

24. Ludwig, W., O. Strunk, R. Westram, L. Richter, H. Meier, Yadhukumar, A.Buchner, T. Lai, S. Steppi, G. Jobb, W. Forster, I. Brettske, S. Gerber, A. W.Ginhart, O. Gross, S. Grumann, S. Hermann, R. Jost, A. Konig, T. Liss, R.Lussmann, M. May, B. Nonhoff, B. Reichel, R. Stehlow, A. Stamatakis, N.Stuckmann, A. Vilbig, M. Lenke, T. Ludwig, A. Bode, and K. H. Schleifer.2004. ARB: a software environment for sequence data. Nucleic Acids Res.32:1363–1371.

25. Martín-Pelaez, S., G. R. Gibson, S. M. Martín-Orue, A. Klinder, R. A.Rastall, R. M. La Ragione, M. J. Woodward, and A. Costabile. 2008. In vitrofermentation of carbohydrates by porcine faecal inocula and their influenceon Salmonella typhimurium growth in batch culture systems. FEMS Micro-biol. Ecol. 66:608–619.

26. Martín-Pelaez, S., A. G. Manzanilla, M. Anguita, M. Fondevila, M. Martín,E. Mateu, and S. M. Martín-Orue. 2009. Different fibrous ingredients andcoarsely ground maize affect hindgut fermentation in the pig in vitro but notSalmonella Typhimurium survival. Anim. Feed Sci. Technol. 153:141–152.

27. Matsuki, T., K. Watanabe, J. Fujimoto, T. Takada, and R. Tanaka. 2004.Use of 16S rRNA gene-targeted group-specific primers for real-time PCRanalysis of predominant bacteria in human feces. Appl. Environ. Microbiol.70:7220–7228.

28. Matsuki, T., K. Watanabe, J. Fujimoto, Y. Miyamoto, T. Takada, K. Matsu-moto, H. Oyaizu, and R. Tanaka. 2002. Development of 16S rRNA-gene-targeted group-specific primers for the detection and identification of pre-dominant bacteria in human feces. Appl. Environ. Microbiol. 68:5445–5451.

29. Menke, K. H., and H. Steingass. 1988. Estimation of the energetic feed value

obtained from chemical analysis and in vitro gas production using rumenfluid. Anim. Res. Dev. 28:7–55.

30. O’Connor, A. M., T. Denagamage, J. M. Sargeant, A. Rajic, and J. McKean.2008. Feeding management practices and feed characteristics associated withSalmonella prevalence in live and slaughtered market-weight finisher swine:a systematic review and summation of evidence from 1950 to 2005. Prev. Vet.Med. 87:213–228.

31. Pettigrew, J. E. 2006. Reduced use of antibiotic growth promoters in dietsfed to weanling pigs: dietary tools, part 1. Anim. Biotechnol. 17:207–215.

32. Pieper, R., R. Jha, B. Rossnagel, A. G. Van Kessel, W. B. Souffrant, and P.Leterme. 2008. Effect of barley and oat cultivars with different carbohydratecompositions on the intestinal bacterial communities in weaned piglets.FEMS Microbiol. Ecol. 66:556–566.

33. Pryde, S. E., S. H. Duncan, G. L. Hold, C. S. Stewart, and H. J. Flint. 2002.The microbiology of butyrate formation in the human colon. FEMS Micro-biol. Lett. 217:133–139.

34. Ricke, P., S. Kolb, and G. Braker. 2005. Application of a newly developedARB software-integrated tool for in silico terminal restriction fragmentlength polymorphism analysis reveals the dominance of a novel pmoA clusterin a forest soil. Appl. Environ. Microbiol. 71:1671–1673.

35. Song, Y., C. Liu, and S. M. Finegold. 2004. Real-time PCR quantitation ofclostridia in feces of autistic children. Appl. Environ. Microbiol. 70:6459–6465.

36. Stein, H. H., and D. Y. Kil. 2006. Reduced use of antibiotic growth promot-ers in diets fed to weanling pigs: dietary tools, part 2. Anim. Biotechnol.17:217–231.

37. Ter Braak, C. J. F., and P. Smilauer. 2002. CANOCO reference manual andCanoDraw for Windows user’s guide: software for canonical communityordination (version 4.5). Microcomputer Power. Ithaca, NY.

38. Thompson, C. L., B. Wang, and A. J. Holmes. 2008. The immediate envi-ronment during postnatal development has long-term impact on gut com-munity structure in pigs. ISME J. 2:739–748.

39. Varel, V. H., and J. T. Yen. 1997. Microbial perspective on fiber utilization byswine. J. Anim. Sci. 75:2715–2722.

40. Walker, A. W., S. H. Duncan, E. C. McWilliam Leitch, M. W. Child, and H. J.Flint. 2005. pH and peptide supply can radically alter bacterial populationsand short-chain fatty acid ratios within microbial communities from thehuman colon. Appl. Environ. Microbiol. 71:3692–3700.

41. Walter, J., C. Hertel, G. W. Tannock, C. M. Lis, K. Munro, and W. P.Hammes. 2001. Detection of Lactobacillus, Leuconostoc, and Weissella spe-cies in human feces by using group-specific PCR primers and denaturinggradient gel electrophoresis. Appl. Environ. Microbiol. 67:2578–2585.

42. Xu, J., M. K. Bjursell, J. Himrod, S. Deng, L. K. Carmichael, H. C. Chiang,L. V. Hooper, and J. I. Gordon. 2003. A genomic view of the human-Bacteroides thetaiotaomicron symbiosis. Science 299:2074–2076.



http://aem.asm

.org/D

ownloaded from

http://aem.asm.org/

effect of carbohydrate composition in barley and oat...

Documents