determining juice quality in relation to hlb infection · week 0.605 0.136 0.551 -0.442 0.627 0.044...
TRANSCRIPT
Determining Juice Quality in Relation to HLB Infection
M Irey, G Thelwell, H Hou, D VanStrijp, P Sun – SGC/USSC
L Baldwin, A Plotto, W Zhao, S Raithore, J Bai, J Manthey – USDA ARS
Quick Update on the Southern Gardens GMO TrialsM Irey, R Kress, G Thelwell, J Kurimai, J Snively, R Terra, B Ingram
Determining Juice Quality in Relation to HLB Infection
M Irey, G Thelwell, H Hou, D VanStrijp, P Sun – SGC/USSC
L Baldwin, A Plotto, W Zhao, S Raithore, J Bai, J Manthey – USDA ARS
What do we know (or what do we think we know…)?
Juice Quality in Relation to HLB
• What we know……
– Off flavors are worse in Hamlin
than in Valencia
• Subtle but there
• More so early in the season
– The levels of the bitter
compounds nomilin and limonin
are related to perceived off-
flavors
• Most likely related to, but not directly
responsible for the off flavors
– Brix tends to go down as a result
of HLB infection
– Trend towards higher acid and
lower ratio
• What we don’t know…..
– What compound(s) cause(s) the
off flavors
• As a result, we don’t know what to
measure
– For what we do measure, we
don’t know the critical thresholds
that parallel the perceived off
flavors
• Complex matrix
• Metabolites change during season
– The ranges overlap
4
5Baldwin/Plotto
Managing Quality
• With respect to HLB, off flavors are correlated to the amount of
symptomatic fruit in the mix
– More symptomatic = more off flavors
• Controlled studies have shown that it is possible to blend out the
off-flavors (25:75 symptomatic to healthy, above that, trained
panelists can tell a difference)
– But where do you get the healthy fruit?
• All groves are infected to some degree, many groves up to 100% infected
• We need a way to measure the potential to produce off flavors
– It would be nice to have a real time assay
• If we had a way to measure the potential we may be able to
manage the harvest or processing to minimize the negative quality
effects of HLB
6
Possible Parameters to Measure as Indicators
• Fruit
size
• Limonin and Nomilin
1
• PCR on orange juice – Patent pending test to measure DNA in
orange juice
Use of qPCR to predict quality of
orange juice affected by HLB
Elizabeth Baldwin, Wei Zhao, Jinhe Bai, Anne Plotto,
John Manthey, Smita Raithore and Mike Irey
US Horticultural Research Laboratory , Ft. Pierce, FL
Corr Li/HLB Corr LJ/HLB
-0.650221308 -0.788697173
Corr Li/Bitter Corr LJ/Bitter
-0.665778254 -0.679466371
Corr Li/metallic Corr LJ/metallic
-0.679974119 -0.704863099
Corr LI/sweet Corr LJ/sweet
0.61662541 0.712966687
Corr LI/sour Coor LJ/sour
-0.194126855 -0.097221665
CorrLI/stale Corr LJ stale
-0.495466167 -0.688194752
Corr grape/LI Corr grape/LJ
-0.605449742 -0.621003765
fruity non cit/LI Fruity non cit/LJ
0.565460735 0.681861145
Li\orange LJ/orange
0.527810667 0.705207427
Li /green LJ/green
-0.54042317 -0.693573863
Li/orange peel LJ/orange peel
-0.548996016 -0.438095888
Li/oxid oil LJ/Oxid oil
-0.645452635 -0.724286668
Li/umami LJ/umami
-0.645510022 -0.771602072
Li/body LJ/body
0.445981827 0.595169395
Li/tingly LJ/tingly
-0.600287878 -0.555725211
Li/astringent LJ/astringent
-0.625057855 -0.606409874
Li/burning LJ/burning
-0.450891717 -0.679466371
Li/aft bitter LJ/aft bitter
-0.665778254 -0.679466371
Li/aft astring LJ/after astring
-0.639675817 -0.640706474
Li/aft burn LJ/aft burn
-0.47504825 -0.425505407
Li/Sum aft LJ/sum aft
-0.648777106 -0.638515573
LI/sum pos LJ/sum pos
0.588213854 0.727153998
Li/sum neg LJ sum neg
-0.696179869 -0.791868269
Corr LI/sour
Ct: Bigger number = better quality
Smaller number = lower quality
So what did we do for 2014-2015 Crop?
• Once a week collected random juice samples from the loads
coming to the plant
– Total of 778 samples
• 305 E/M samples
• 455 Valencia samples
• Analyses
– Limonin/Nomilin by HPLC
– Ct value by real time PCR (qPCR)
– Brix, acid, ratio, pieces per box: plant data
– Electronic tongue
– Week of crop (December 1 used as starting point for week calculation)
• Principal Component Analysis
15
Electronic tongue
• 7 Electronic sensors
• Sensors map to sensory
descriptors
– Sourness
– Metallic
– Saltiness
– Umami
– Spiciness
– Sweetness
– Bitterness
• Can be calibrated to standards
(if available)
16
Principal Component Analysis
• Principal component analysis (PCA) is a statistical procedure that uses an
orthogonal transformation to convert a set of observations of possibly
correlated variables into a set of values of linearly uncorrelated variables
called principal components.
17
Principal Component Analysis
• Principal component analysis (PCA) is a technique used to emphasize
variation and bring out strong patterns in a dataset. It's often used to make
data easy to explore and visualize.
18
– Is a variable reduction technique
– Is used when variables are highly correlated
– Reduces the number of observed variables to a smaller number of
principal components which account for most of the variance of the
observed variables
• The first principal component accounts for most of the variance in the data.
• The second component accounts for the second largest amount of variance in the
data and is uncorrelated with the first principal component, and so on.
• In PCA, observed variables are standardized, e.g. mean = 0
Correlation Matrix – E/M
19
Variables SRS GPS STS UMS SPS SWS BRS Acid Brix Ratio Pcs/Box Limonin Nomilin Ct Week
SRS 1 -0.045 0.780 -0.523 0.811 -0.130 0.190 -0.516 0.062 0.447 -0.225 -0.457 -0.466 0.297 0.605
GPS -0.045 1 -0.456 -0.053 -0.093 0.972 0.885 0.118 -0.013 -0.122 0.083 0.002 0.009 -0.142 0.136
STS 0.780 -0.456 1 -0.445 0.895 -0.556 -0.213 -0.277 0.252 0.418 -0.143 -0.452 -0.438 0.382 0.551
UMS -0.523 -0.053 -0.445 1 -0.520 -0.005 -0.110 0.105 -0.229 -0.258 0.044 0.259 0.335 -0.177 -0.442
SPS 0.811 -0.093 0.895 -0.520 1 -0.215 0.163 -0.191 0.307 0.387 -0.068 -0.502 -0.483 0.363 0.627
SWS -0.130 0.972 -0.556 -0.005 -0.215 1 0.808 0.131 -0.064 -0.173 0.098 0.047 0.058 -0.185 0.044
BRS 0.190 0.885 -0.213 -0.110 0.163 0.808 1 0.003 0.046 0.013 -0.004 -0.145 -0.109 0.006 0.233
Acid -0.516 0.118 -0.277 0.105 -0.191 0.131 0.003 1 0.174 -0.646 0.410 0.223 0.147 -0.183 -0.118
Brix 0.062 -0.013 0.252 -0.229 0.307 -0.064 0.046 0.174 1 0.629 0.100 -0.620 -0.409 0.302 0.361
Ratio 0.447 -0.122 0.418 -0.258 0.387 -0.173 0.013 -0.646 0.629 1 -0.241 -0.637 -0.420 0.368 0.359
Pcs/Box -0.225 0.083 -0.143 0.044 -0.068 0.098 -0.004 0.410 0.100 -0.241 1 0.176 0.021 -0.204 -0.066
Limonin -0.457 0.002 -0.452 0.259 -0.502 0.047 -0.145 0.223 -0.620 -0.637 0.176 1 0.707 -0.407 -0.586
Nomilin -0.466 0.009 -0.438 0.335 -0.483 0.058 -0.109 0.147 -0.409 -0.420 0.021 0.707 1 -0.303 -0.499
Ct 0.297 -0.142 0.382 -0.177 0.363 -0.185 0.006 -0.183 0.302 0.368 -0.204 -0.407 -0.303 1 0.287
Week 0.605 0.136 0.551 -0.442 0.627 0.044 0.233 -0.118 0.361 0.359 -0.066 -0.586 -0.499 0.287 1
Values in bold are different from 0 with a significance level alpha=0.05
• Principal component analysis (PCA) is a technique used to emphasize
variation and bring out strong patterns in a dataset. It's often used to make
data easy to explore and visualize.
Principal Component Analysis
• Principal component analysis (PCA) is a technique used to emphasize
variation and bring out strong patterns in a dataset. It's often used to make
data easy to explore and visualize.
20
Week of Crop – Early Mids
21
12/9/2014 – 2/5/2015
Early Mids – All weeks
• Initial PCA looked at all
parameters and then re-ran
with only those parameters
that were most important in the
first two components
• Used a threshold of 0 on the
first component and grouped in
to two categories
– Worse (high lim/nom, low Ct)
– Better (low lim/nom, high Ct)
22
High Lim/Nom
Low Ct
“Worse”
Low Lim/Nom
High Ct
“Better”
Why are the “Worse” E/M’s skewed to the front of the crop
• If you had two blocks, one with excessive drop and one without,
which would you pick first?
25
26
December 2, 2014, 3 replications, 3 tree plots, 5 treatments: Healthy tree, Healthy shaken left on tree, healthy shaken on ground, infected shaken on tree, infected shaken on ground
December 2, 2014, 3 replications, 3 tree plots, 5 treatments: Healthy tree, Healthy shaken left on tree, healthy shaken on ground, infected shaken on tree, infected shaken on ground
Healthy
Unshaken Healthy Tree
Healthy Ground
HLB Tree HLB Ground P-value
Orange 4.4 a 4.5 a 3.8 b 3.1 c 2.1 d < 0.0001
Grapefruit 2.9 c 2.5 c 2.5 c 4.4 b 5.9 a < 0.0001
Fruity-non-citrus 1.8 a 1.7 ab 1.4 bc 1.1 c 1.1 c 0.0009
Orange peel 2.3 b 2.1 b 2.0 b 2.5 ab 2.9 a 0.027
Green 2.4 b 2.4 b 2.4 b 2.8 ab 3.1 a 0.052
Stale 2.4 b 2.6 b 2.6 b 3.3 ab 4.1 a 0.009
Oxidized oil 1.6 b 1.6 b 1.5 b 2.2 ab 2.7 a 0.012
Typical HLB 4.1 c 4.0 c 4.3 c 7.6 b 10.2 a <0.0001
Sweetness 5.5 a 5.1 ab 4.6 b 4.1 c 3.2 d <0.0001
Sourness 5.0 ab 4.6 b 4.8 b 5.2 ab 5.6 a 0.04
Umami 2.3 b 2.2 b 2.4 b 2.8 ab 3.4 a 0.006
Bitterness 4.1 c 3.1 c 3.4 c 7.1 b 9.3 a <0.0001
Metallic 2.4 c 2.0 c 2.1 c 3.1 b 4.3 a <0.0001
Body 4.9 a 4.5 ab 4.3 b 4.4 ab 4.7 ab 0.166
Tingling 1.8 bc 1.6 c 1.5 c 2.3 ab 2.7 a 0.0001
Astringent 2.3 b 1.8 b 2.1 b 3.4 a 3.9 a <0.0001
Burning 1.4 b 1.2 b 1.2 b 2.1 a 2.5 a 0.0003
AfterBitter 2.0 c 1.6 c 2.0 c 4.5 b 6.0 a <0.0001
AfterAstringent 1.3 b 1.3 b 1.1 b 2.6 a 3.1 a <0.0001
AfterBurning 0.8 bc 0.6 c 0.8 bc 1.3 ab 1.7 a 0.001
December 2, 2014, 3 replications, 3 tree plots, 5 treatments: Healthy tree, Healthy shaken left on tree, healthy shaken on ground, infected shaken on tree, infected shaken on ground
Plotto/Baldwin
Valencia – All Weeks
30
High Lim/Nom
Low Ct
“Worse”
Low Lim/Nom
High Ct
“Better”
March 2,2015 – May 5, 2015
So how can this approach be used to manage the harvest?
• Used the same basic PCA approach but basically did it in 2-3
weeks where possible
– Where not possible to group, looked at individual weeks
• Used the same basic “Better” / “Worse” threshold approach for the
component that explained the Ct and Lim/Nom variability
– Group 1 = Better
– Group 2 = Worse
• Averaged the “Group” rating by grower
• 3 Categories
– Better (<1.4) = Green
– OK (>1.4 <1.6) = Yellow
– Worse (>1.6) = Red
34
Early/Mids
35
Valencia
36
Combined
• Quality (as defined by the PCA
analysis using Ct and Lim/Nom
as the criteria for grouping)
appears to vary by grower
– Consistent during the season
– Consistent between varieties
• Going forward
– Need to go down another level
(grove) to see if there are
consistent differences there
– Need to go compare a couple of
years to see if the trends are
consistent
37
Conclusions
• Although no one parameter by itself is sufficient to measure
quality, PCA using a combination of wet chemistry, electronic
tongue (as a proxy for sensory), and molecular technologies
appears promising as a means of evaluating quality
– Ct may be a rough indicator
• Quality (as it relates to HLB associated off flavors) appears to vary
by grower
• Harvesting patterns could potentially affect quality
– Incentives are opposite……
38
39
Thank You / Questions?
Quick Update on the Southern Gardens GMO TrialsM Irey, R Kress, G Thelwell, J Kurimai, J Snively, R Terra, B Ingram
Upon the discovery of HLB, Southern Gardens decided to:
• Use the opinions of the best experts available
– 3 pronged approach
• Open our diagnostic lab to the industry at no cost
– Now partially funded by CRDF
• Work with Federal, State and private researchers to develop
solutions
• Establish our own projects to develop a long term solution to HLB
– Multiple partners
– Multiple projects
– Multiple approaches
41
42
Southern Gardens Research
• Texas A & M University
–Disease resistant plants
• Integrated Plant Genetics
–Disease resistant plants
• Cornell University
– Insect resistant plants
• AgroMed LLC
– Identification of synthetic resistance genes
• University of Florida
–Gene delivery system
• USDA
–Screening of potential genes, juice quality
Transgenic Trees
• Stable integration-long term
• Multiple projects
– Antimicrobial peptides
• Integrated Plant Genetics
–Gene from a bacteriophage
• Agromed
–Synthetics
•Texas A & M
–Spinach defensins (multiple)
– Resistance to ACP
• Cornell
–Hirsutellin
–Lectins
43
Transgenic Trees
• Ideally, looking for immunity
• Process involves screening many lines (200+)
• Testing
– Greenhouse
– Field
• Field testing is done under USDA-APHIS-BRS permits
– 3 years
– Defined conditions
• EPA and FDA not involved in permitting (unless EUP is required)
44
Non-transgenic Rohde Red Val
Transgenic Rohde Red Val
2 2
5 5
7 7
Texas Project
• Spinach defensins
– Has been a learning curve
• Generations 1, 2, 3, 4, 5, …..
–Multiple genes
–With and without signal peptides
–Different codon optimizations schemes
–Corrected sequences
–Alone or stacked
– All trees created in Texas at Texas A & M, Weslaco by Dr. Erik Mirkov
• Hamlin, Valencia, Lemons, Mexican Lime, Grapefruit, rootstocks
53
Texas Project
• FDACS Regulatory Process
– Since trees were created in Texas, they were subject to CGIP
• Issues:
–Time delay
–Overloading of the system
» Would not be a good use of FDACS resources given that the majority of the lines would not make it
(200+ lines and more still coming)
– Over time and multiple iterations, a system was developed that worked
• Protected the Florida industry
• Allowed us to move material to Florida
54
So where are we with the deregulation process -EUP
• EUP- allows for greater than 10ac of trials
• For the transgenic tree defensin project, we have an EUP that
allows:
– 400 ac in Florida
– 200 ac in Texas
– Two defensins
– 4 Plant lines
• 2 Sweet Oranges
• 1 Grapefruit
• 1 Lemon
55
Temporary Tolerance Exemption
• Temporary tolerances are granted by the EPA to allow holders of
an EUP to use the products that come off the trials in the EUP
• A petition for a temporary tolerance or an exemption must be
submitted with the application for the EUP (if not a crop destruct)
• In the case of spinach defensins in transgenic trees, a tolerance
exemption was granted
– Based on the data that was submitted, there were no toxicity issues
– Already in the food chain
56
Going forward…..
• We have many more lines being tested and it is likely that we may
have lines better than those in the current EUP
• The EUP process is:
– Tedious
– Costly
– Takes time
– Is restrictive
• Specific lines
• Specific constructs
57
Going forward…..
• Have a new EUP amendment submission
– Multiple events (i.e. not restricted to 4 lines) as long as they were restricted to
certain constructs
• First step towards a “by construct” approach
• This will be necessary going forward
• In the process of:
– EUP for spinach defensin using viral vector
– Temporary tolerance for 3 spinach defensins in CTV
– Temporary tolerance for all spinach defensins
58
Other efforts
• Global approval
– Have started the process to determine the landscape for approval of a
transgenic tree-based commercialization of spinach defensins
• 5 key markets in addition to the US
–Canada
–Japan
–Korea
– Israel
–EU
59
Other efforts
• Also working on:
– Consumer acceptance
• Proactively to improve general acceptance
–Consumers
–Retailers
• Legal preparation for the eventual law suits
– Freedom to operate
– Stewardship issues
• “Defensins plus”
• Second mode of action
60
61
Thank you (again..) / Questions