Development and Validation of a Fluorescent Multiplexed

May 13, 2016 - Conflict of Interest Statement. The authors declare no competing financial interest. ... Fluorescent (bead-based) multiplexed immunoass...
0 downloads 11 Views 727KB Size
Subscriber access provided by UOW Library

Article

Development and Validation of a Fluorescent Multiplexed Immunoassay for Measurement of Transgenic Proteins in Cotton (Gossypium hirsutum). Grant Ramsay Yeaman, Sudakshina Paul, Iryna Nahirna, Yongcheng Wang, Andrew E Deffenbaugh, Zi Lucy Liu, and Kevin Challon Glenn J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.6b01441 • Publication Date (Web): 13 May 2016 Downloaded from http://pubs.acs.org on May 14, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Agricultural and Food Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 35

Journal of Agricultural and Food Chemistry

1

Development and Validation of a Fluorescent Multiplexed Immunoassay for Measurement of

2

Transgenic Proteins in Cotton (Gossypium hirsutum).

3

Grant R. Yeaman*, Sudakshina Paul, Iryna Nahirna, Yongcheng Wang, Andrew E. Deffenbaugh

4

Zi Lucy Liu and Kevin C. Glenn.

5

Monsanto Company, 800 North Lindbergh Boulevard, St. Louis, Missouri 63167

6

*

7

The authors declare no competing financial interest.

8

Keywords: fluorescent multiplexed immunoassay Luminex ELISA cotton Gossypium hirsutum

To whom correspondence should be addressed at [email protected]

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

9

Abstract

10

In order to provide farmers with better and more customized alternatives to improve yields,

11

combining multiple GM traits into a single product (called stacked trait crops) is becoming

12

prevalent. Trait protein expression levels are used to characterize new GM products and

13

establish exposure limits, two important components of safety assessment. Developing a

14

multiplexed immunoassay, capable of measuring all trait proteins in the same sample, allows

15

for higher sample throughput and savings in both time and expense. Fluorescent (bead-based)

16

multiplexed immunoassays (FMI) have gained wide acceptance in mammalian research and in

17

clinical applications. In order to facilitate the measurement of stacked GM traits, we have

18

developed and validated an FMI assay that can measure five different proteins (GUS, NPTII,

19

Cry1Ac, Cry2Ab2 and CP4 EPSPS) present in cotton leaf from a stacked trait product. Expression

20

levels of the five proteins determined by FMI in cotton leaf tissues have been evaluated relative

21

to expression levels determined by ELISAs of the individual proteins and shown to be

22

comparable. The FMI met characterization requirements similar to those used for ELISA.

23

Therefore, it is reasonable to conclude that FMI results are equivalent to those determined by

24

conventional individual ELISAs to measure GM protein expression levels in stacked trait

25

products, but with significantly higher throughput, reduced time and more efficient use of

26

resources.

27

ACS Paragon Plus Environment

Page 2 of 35

Page 3 of 35

Journal of Agricultural and Food Chemistry

28

Introduction

29

Measuring the expression levels of proteins introduced into new genetically modified (GM)

30

crops helps to establish exposure levels for assessing food, feed, and environmental safety as

31

part of the product characterization and safety assessment process. Currently, Enzyme-Linked

32

Immunosorbent Assay (ELISA) is the method of choice for measuring expression levels in GM

33

crops 1-3. Newer GM products are being developed that incorporate several transgenic traits

34

(i.e. “stacking”) providing multiple modes of action that are designed to minimize development

35

of resistance in target insect species and in herbicide protection and combination of these crop

36

protection traits with stress and yield traits, e.g. drought resistance. This facilitates the

37

development of trait solutions in a regional and problem specific manner 4-7. The ISAAA

38

database lists over 200 agricultural stacked products approved since YieldGard® Plus (MON810 x

39

MON863) corn was commercialized in 2003 8. The first cotton stack product, Roundup Ready™

40

Flex™ Bollgard II™ Cotton, was commercialized in 2006 8. In order to continue offering

41

sustainable solutions that meet farmers’ needs for insect protection and herbicide tolerance

42

under changing environmental conditions, future commercial products will contain multiple

43

transgenic traits.

44

More recently, new technologies utilizing both antibody based and mass-spectroscopy (MS)

45

based systems have been developed to determine the levels of proteins in samples 9, 10.

46

Recently, MS based approaches have been used to measure recombinant protein expression

47

levels in a stacked GM product 11. Of the antibody based technologies two platforms, Meso

48

Scale Discovery (MSD) and Fluorescent (bead-based) multiplexed immunoassays (FMI), are

49

regarded as mature technologies capable of measuring multiple proteins in a single sample.

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

50

FMIs may be carried out on a range of flowcytometers, however, Luminex technology (Luminex

51

Corporation, Austin TX) is the most robust and commonly used platform 12, 13. MSD and

52

Luminex platforms have comparable performance characteristics but the Luminex system has,

53

for the purposes of this study, greater flexibility in creating custom multiplexes containing a

54

greater number of analytes 14, 15. FMI has gained broad acceptance in clinical diagnostics and in

55

both academic and pharmaceutical research. The Food and Drug Administration (FDA) assesses

56

the effectiveness of diagnostic assays, like all medical devices, prior to approval for marketing,

57

to ensure that any new diagnostic methods are as useful as currently available technologies 16.

58

Acceptance of FMI based diagnostics in clinical medicine is evidenced by at least 84 diagnostic

59

kits that have gained FDA 510k approvals as of 2014 (Personal communication, C. Martin,

60

Luminex Corporation). Underscoring the technique’s wide spread use in academic and

61

pharmaceutical research, a search of Luminex Corporation’s publication database for FMI based

62

methods resulted in 7436 peer reviewed publications

63

(http://www.luminexcorp.com/publications).

64

With the successful adoption of FMI for clinical applications, and the advent of increasingly

65

complex stacked GM crops that express large numbers of introduced proteins, it is a natural

66

progression to adapt FMI to assess protein expression levels in agricultural biotech products.

67

Fluorescent bead based assays have been used to detect and type plant pathogens, both by

68

nucleic acid based 17-19 and antibody based techniques 20. A triplex FMI assay has been used in

69

processed foods to detect soy, pea and soluble wheat protein adulteration of formula milk 21

70

and another multiplex to detect the presence food allergens 22. Eight studies have been

71

published that directly apply these types of assay to GM crops; seven of these are applied to

ACS Paragon Plus Environment

Page 4 of 35

Page 5 of 35

Journal of Agricultural and Food Chemistry

72

the measurement of nucleic acids specific for transgenes in multiple plant species 23-29. An FMI

73

for the detection of a single transgenic protein, Cry1Ab protein derived from Bacillus

74

thuringiensis (subsp. Kurstaki), in maize has been published 30.

75

The number of individual genes present in a stack is a direct multiplier of the number of assays

76

needed to give an accurate measure of expression levels of each protein. For example: a typical

77

expression study for cotton with ten transgenic proteins may require measuring 10 proteins, in

78

3 tissue types, at 5 geographic sites, with 4 replicates measured per site. By conventional

79

individual ELISAs this would require a total of 600 individual determinations; whereas 60

80

determinations would be required when using an FMI capable of measuring all ten proteins

81

simultaneously in each tissue sample.

82

The goal of this study was to assess whether an FMI can measure all trait proteins in the same

83

samples with a level of accuracy comparable to the individual ELISAs. An FMI method was used

84

to measure five proteins expressed in a stacked GM cotton product, alongside the ELISA

85

methods for the same five respective proteins. The five proteins of interest (POIs) were present

86

in leaf samples from Genuity® Bollgard II® with Roundup Ready® Flex Cotton (MON15985 x

87

MON 88913). These proteins have been introduced to cotton to achieve a variety of functions:

88

e.g. insect control (Cry1Ac and Cry2Ab2) and herbicide tolerance (CP4 EPSPS); therefore, not

89

surprisingly; they differ in their levels of expression and physicochemical properties. The

90

technical challenges in applying FMI to the detection of GM traits in plant tissue samples are

91

addressed; demonstrating that the development of an FMI assay is suitable for use in protein

92

expression studies to characterize stacked trait products.

93

Materials and Methods

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

94

Test Samples

95

The FMI was developed to measure proteins of interest (POIs) present in leaf samples from

96

Genuity® Bollgard II® with Roundup Ready® Flex Cotton (MON15985 x MON88913). Bollgard II®

97

Cotton (MON15985) contains the Cry1Ac and Cry2Ab2 proteins from B. thuringiensis subsp.

98

kurstaki which provides protection to Lepidopteran cotton pests and two selectable marker

99

proteins, β-glucuronidase (GUS) and neomycin phosphotransferase II (NPTII) proteins. Roundup

100

Ready® Flex Cotton (MON88913) contains the 5-enolpyruvyl-shikimate- 3-phosphate synthase

101

protein isolated from Agrobacterium sp. strain CP4 (CP4 EPSPS) that confers tolerance to

102

glyphosate. Bollgard II® Cotton was crossed with Roundup Ready® Flex Cotton by traditional

103

breeding to produce the stacked trait Genuity® Bollgard II® with Roundup Ready® Flex Cotton.

104

Tissue samples used for the development, characterization and validation experiments were

105

produced in a greenhouse (samples were not exposed to herbicides). Ten leaf samples

106

containing the five POIs and ten conventional Cocker 130 cotton leaf samples negative for the

107

POIs were used during assay development through validation. Cocker 130 leaf was used to

108

produce the negative quality control (QC-) matrix. Direct comparisons of FMI and ELISA derived

109

expression levels were carried out using the same extracts from the positive tissues from the

110

greenhouse production. FMI expression levels were determined using two separate stacks,

111

Bollgard II® Cotton (containing NPTII, GUS, Cry1Ac and Cry2Ab2) and Roundup Ready® Flex

112

Cotton (containing CP4 EPSPS), from a 2013 field trial conducted in the USA. Four replicate leaf

113

samples were collected from each of 5 sites located in Arizona, Texas, Louisiana, North Carolina

114

and Georgia. Expression levels detected by FMI were compared to expression levels obtained

115

by ELISA of samples taken from two separate stacks, Bollgard II® Cotton and Roundup Ready®

ACS Paragon Plus Environment

Page 6 of 35

Page 7 of 35

Journal of Agricultural and Food Chemistry

116

Flex Cotton, from two field sites grown in Australia during the 2011-2012 growing season. The

117

comparison of two different sites for ELISA and FMI was chosen so that a comparison of the

118

expression levels using the two techniques was made between independent studies conducted

119

under GLP conditions.

120

Sample Extraction

121

All samples were extracted at a buffer to tissue ratio of 50:1 (v:w) in Tris borate buffer, pH 7.8

122

using three ⅛” beads in a Genogrinder® (SPEX SamplePrep, NJ) for 3.5 minutes at 1500 RPM.

123

Extracts were clarified by centrifugation, aliquoted and frozen at -80°C until use.

124

Standard Proteins and Controls

125

Certified protein standards for each of the analytes were used to construct standard curves and

126

positive quality controls (QCs) and for spiking into negative tissues for determination of matrix

127

effects. Negative QCs (QC-) were prepared from Coker 130 leaf by extraction and dilution to 10

128

fold (minimum matrix dilution), then aliquoted and frozen at -80°C until use. Two positive QCs

129

were used: a QC+ Spike was prepared by spiking known amounts of each of the protein

130

standards into the QC- matrix prior to the final 10X matrix dilution step, then aliquoted and

131

frozen at -80°C until use. The second positive QC (QC+ Biol.) was prepared from a sample

132

transgenic for all of the POIs that was extracted, diluted, aliquoted and frozen as described

133

above.

134

Preparation of xMAP beads

135

Capture antibodies used for coupling to beads were: anti-GUS Rabbit polyclonal IgG, anti-

136

Cry1Ac murine mIgG2b, anti-Cry2Ab2 murine mIgG2a, anti-CP4 EPSPS murine mIgG2a and anti-

137

NPTII rabbit polyclonal IgG. These were the same antibodies used for plate coating in the

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

138

corresponding sandwich ELISAs. xMAP beads were coupled with the appropriate capture

139

antibodies using Antibody Coupling (AbC) kit according to the manufacturer’s instructions

140

(Luminex Corp, Austin TX, Cat # 40-50016). The following bead sectors were assigned: 7-NPTII,

141

8-Cry2Ab2, 9-GUS, 12-Cry1Ac, 20-CP4 EPSPS. Coupling was confirmed, following the

142

manufacturers recommendation, by titrating beads against multiplexed RPE-labeled anti-

143

species IgG specific antibodies (RPE-labeled goat anti-rabbit IgG [SLBD3857], RPE-labeled Rabbit

144

anti-goat IgG [028K4761] and goat anti-mouse IgG [O95K6156] all obtained from Sigma, St

145

Louis, MO). Titrations indicated that optimal and reproducible substitution rates were obtained

146

at a concentration of 5µg of antibody/1x106 beads and these conditions were used for all

147

subsequent bead labeling. Coupling was highly reproducible between batches of labeled beads

148

labeled at three different times over a 9 to 10 month period (data not shown).

149

Detection antibodies used in the Luminex assay

150

Anti-GUS rabbit polyclonal IgG, anti-Cry1Ac goat polyclonal IgG, anti-Cry2Ab2 goat polyclonal

151

IgG, anti-CP4 EPSPS goat polyclonal IgG and anti-NPTII goat polyclonal IgG were biotinylated and

152

used as detection antibodies in the multiplexed assay. These biotinylated antibodies are the

153

same as those used in the validated ELISAs for these proteins, except for CP4 EPSPS where the

154

ELISA uses a direct HRP conjugated monoclonal detection antibody and for GUS where a new

155

antibody production was used. Streptavidin-R phycoerythrin (Thermo, Cat# 21627) was used as

156

the detection reagent in all cases.

157

Standard Multiplexed Immunoassay Method

158

Antibody coupled xMAP beads specific to each protein were diluted together in 1 × PBS/1% BSA

159

(w/v) [PBS-BSA] to a final concentration of 100 beads/µl. The appropriate wells in 96 well

ACS Paragon Plus Environment

Page 8 of 35

Page 9 of 35

Journal of Agricultural and Food Chemistry

160

plates (FLUOTRACTM 200 96W Medium Bind microplate, Greiner bio-one, Frickenhausen,

161

Germany) were loaded with 50 µl of diluted beads. Fifty microliters of standard protein

162

dilutions, QCs, and tissue samples were added to corresponding wells and incubated for at least

163

30 minutes at room temperature while shaking at 400 rpm. Plates were washed 3 times using a

164

magnetic plate washer (ELx405, BioTek Instruments Inc., Winooski VT) with

165

1 × PBS/0.05 % Tween 20 (v/v) [PBST]. A cocktail of biotinylated secondary antibodies (each

166

antibody at a 1:2000 dilution) was prepared in PBS-BSA containing mouse IgG (1 mg/ml) and

167

added at 50 µl/well and incubated as described above. Plates were washed as above, prior to

168

the addition of 50 µl/well of Streptavidin-RPE (4µg/ml; Thermo Scientific, Rockford IL) and

169

incubated as described above. After washing, the beads were resuspended by adding 75 µl of

170

PBS-BSA buffer and shaking at 400 rpm for at least 10 minutes. Plates were then read on a

171

Luminex FlexMap 3D® system running xPONENT® Software version 4.2. Quantification of each

172

of the proteins was accomplished by interpolation from each of the protein standard curves

173

using Milliplex™ Analyst software version 5.1 to analyze the output .CSV file.

174

ELISA Protein Detection

175

All ELISAs were carried out in a similar manner using validated protocols, except for the already

176

noted differences in antibodies and in that the ELISAs use a range of extraction buffers, sample

177

buffers and reagent diluents.

178

Validation

179

The following parameters were determined during Validation in three independent

180

experiments by three different operators: 1. Extraction efficiency was established by repeated

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

181

extraction of three positive leaf samples by three operators in three independent experiments.

182

The POI levels were determined by FMI in each extract and the amount extracted in the first

183

extract was expressed as a percentage of the total to give an estimate of extraction efficiency.

184

2. Matrix effects were determined by spiking known amounts of standard proteins into

185

negative tissue matrix and comparing the recovery (as measured by FMI) to recoveries obtained

186

from spiking standard proteins into assay buffer. 3. Dilutional parallelism was established by

187

measuring the recovery by FMI of POI from positive leaf samples at different dilutions. Data

188

used in the evaluation of extraction efficiency, dilutional parallelism, matrix effects and assay

189

precision were the interpolated concentrations obtained from Milliplex™ Analyst (version 5.1)

190

for each experiment. Calculations were carried out using Excel spreadsheet templates

191

(Microsoft Office Excel v2007). All data and calculation spreadsheets underwent rigorous

192

quality control procedures. Precision Performance was carried out using a modification of EP5-

193

A2, the CLIA approved guidelines 31 over three days by three operators with one run per

194

operator/day. Two independently prepared standard curves and QC sets were included in each

195

run for a total of 9 runs and 18 sets of standards and QCs. Five negative leaf extracts were

196

included in each run and used to determine the LOD for each protein. The inter-assay and intra-

197

assay precision was calculated using 16 standard curves/QCs (8 plates) – one run was removed

198

from the calculation as an outlier. The dose dependent precision profile was calculated using

199

the remaining 16 independently prepared standard curves with each standard curve point and

200

the positive QCs interpolated through both curves in the run. The resulting back-calculated

201

concentrations of each standard and QC were used to calculate precision. The CV’s for each

202

standard concentration were averaged resulting in the final precision profile value for the

ACS Paragon Plus Environment

Page 10 of 35

Page 11 of 35

Journal of Agricultural and Food Chemistry

203

method. Acceptable QC ranges were determined as the mean back calculated concentration ±

204

3 standard deviations (s.d.). Precision Data visualization, curve fitting for curve comparisons

205

and comparison of means was carried out using GraphPad Prism version 6.04 for Windows

206

(GraphPad Software, La Jolla California USA, www.graphpad.com.). Standard curves were fitted

207

to a 5-parameter logistic model using the least squares method and a 1/Y2 weighting. Linear

208

regression analysis was carried out using the GraphPad Prism default options and the logged

209

expression values (all curves shown had an R2 > 0.98). Comparison of means from ELISA and

210

FMI expression studies was carried out using GraphPad Prism to run unpaired T-tests with

211

Welsh’s correction, p ≤ 0.05 was considered as significant.

212

The limits of quantification of each standard curve, in each run, were determined by the

213

Milliplex Analyst software. The best-fit feature of Milliplex™ Analyst was used when fitting

214

curves. In practice this is always a five parameter logistic fit using either the net MFI (median

215

fluorescent intensity with background signal subtracted) or the log10 of the net MFI. Using a

216

combination of the confidence interval and the precision error from each curve fit, Milliplex

217

Analyst determines a minDC and a maxDC value; these values are the limits beyond which

218

interpolated values would be statistically unreliable (analogous to the 25% CV value typically

219

used in ELISA). These values are a property of the individual standard curve and vary with the

220

fitting for each curve; an approach that has been used previously for Luminex based assays32.

221

These values are a property of the individual standard curve and do not take into account any

222

differences caused by the presence of matrix in test samples. Therefore, a tissue specific lower

223

limit of detection (LOD) was also determined by taking the mean interpolated values of multiple

224

negative control tissues (5 tissues were evaluated on 9 runs and interpolated through 2

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

225

standard curves for each run i.e. a total of 18 determinations for each tissue for a total of 90

226

determinations). Tissue LODs were set as the mean interpolated concentrations plus three

227

times the standard deviations. The effective lower limits of quantification (LLOQs) for the assay

228

were then taken as the higher of the minDC or the tissue LOD for each analyte. In practice the

229

tissue LOD was nearly always found to be higher than the minDC.

230

Results and Discussion

231

Fluorescent (bead-based) multiplexed immunoassays (FMI)

232

The FMI was developed using the latest generation of Luminex readers, the FLEXMAP 3D, and

233

xMAP bead technology. A comprehensive description of the technology and its applications is

234

given in Dunbar and Hoffmeyer 33. xMAP Technology uses fluorochrome beads as a capture

235

surfaces in a manner analogous to polystyrene wells in ELISA, except that antibodies are

236

covalently linked to xMAP beads rather than passively adsorbed. Each xMAP bead is given a

237

unique identity (sector) by the incorporation of differing amounts of two different

238

fluorochromes. Each distinct bead can be coated with a capture antibody/molecule specific to a

239

particular biological target. Mixing different beads together (plexing) allows for the

240

simultaneous capture of multiple analytes from a single sample. Captured analytes are then

241

detected using a mixture of detection antibodies that are either directly linked to an

242

appropriate fluorochrome (e.g. R-phycoerythrin; RPE) or tagged with biotin followed by

243

Streptavidin-RPE. Incorporation of magnetic moieties into xMAP microspheres simplifies assay

244

washing and handling. One advantage of FMI is broader dynamic ranges, up to 5 log orders,

245

compared to the conventional 1 to 1.5 log orders of conventional ELISA. The practical dynamic

246

ranges for analytes is influenced by the properties of the particular capture/detection antibody

ACS Paragon Plus Environment

Page 12 of 35

Page 13 of 35

Journal of Agricultural and Food Chemistry

247

pairs used. The affinity of the capture and detection antibodies influences the working range of

248

the assay such that high affinity antibodies will tend to give higher sensitivity but narrower

249

dynamic range. Conversely, lower affinity antibodies will yield lower sensitivity but broader

250

dynamic ranges. In most instances, the extended dynamic range of Luminex allows for the assay

251

of analytes, with greatly differing expression levels, present within the same sample and at the

252

same dilution 34.

253

FMI for the measurement of five proteins (GUS, NPTII, Cry1Ac, Cry2Ab2 and CP4 EPSPS),

254

present in the leaf of a cotton stacked product, has been developed. The approach in

255

developing this FMI closely followed the methodologies employed for ELISA development and

256

validation and is consistent with accepted industry guidelines 35.

257

Screening for Cross-reactivity.

258

In multiplexed immunoassays, the potential for cross-reactivity between antibody pairs and/or

259

antibodies and analytes is much greater than that encountered in single analyte ELISAs. When

260

using a mixture of polyclonal rabbit and goat antibodies and murine monoclonal antibodies, any

261

observed interference would most likely be caused by cross-species antibodies present in the

262

polyclonal IgG preparations 12. It is therefore essential that potential cross-reactivities are

263

thoroughly assessed and that the standard curves for individual analytes are not significantly

264

altered by the presence of other assay reagents. Experiments were performed using the

265

following combinations: all capture antibody beads/single analyte/single detection antibody, all

266

capture antibody beads /all analytes/single detection antibody, and all capture antibody beads

267

/all analytes/all detection antibodies. Cross-reactivity experiments were conducted as

268

recommended in the xMAP Cookbook 34, where ≤1% cross-reactivity is regarded as acceptable

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

269

performance. As shown in Table 1, no significant cross-reactivity was evident. Note that in

270

Table 1, the cross-reactivities for individual analytes against themselves (values on the diagonal)

271

and those for the multiplex are not an exact match; this variability is well within the expected

272

interassay variability and does not represent a significant difference in performance between

273

the single and the multiplexed assays. Figure 1 shows standard curves (solid symbols and lines)

274

obtained in the presence of either 1) a single analyte and a single detection antibody (A, C, E, G,

275

I) or 2) of all analytes and a single detection antibody (B, D, F, H, J). Also shown on each graph

276

are the standard curves with the full multiplex (e.g., all capture antibody beads) included (open

277

symbols & dotted line). There were minimal or no differences in curves between single analyte

278

/single antibody and single analyte/ all detection antibodies (e.g., compare solid line A to solid

279

line B) indicating a lack of interference between detection antibodies or between detection and

280

capture antibodies. Nor was there a significant difference when the complete multiplex of all

281

capture antibody beads /all analytes/all detection antibody combinations was present

282

(compare open symbols dotted lines to solid symbols - solid lines on each graph). Where

283

differences are present, they are well within the variability typically seen between

284

independently prepared standard curves under otherwise identical conditions.

285

Establishment of Standard Curves.

286

Demonstrating the reproducibility (precision) and stability (robustness) of the standard curves

287

used to calculate protein concentrations in samples is an essential part of validating an

288

immunoassay. Final concentrations for standard curve points were adjusted to take into

289

account the dynamic range for each antibody pair; i.e. the highest concentration on the

290

standard curve was set to the lowest concentration that gave saturable binding. Analyte

ACS Paragon Plus Environment

Page 14 of 35

Page 15 of 35

Journal of Agricultural and Food Chemistry

291

standards were prepared as a cocktail at the highest standard concentration for each analyte

292

then aliquoted and stored frozen at -80°C until use. Stability testing over 10 months showed

293

that one freeze-thaw cycle did not alter performance of the standard cocktail mixture (data not

294

shown). Eight standard concentrations were prepared by three-fold serial dilution of the stock

295

cocktail plus a ninth “zero” standard point containing no standard proteins. The reproducibility

296

of the standard curves is shown in Figure 2. Each point in each graph is the average of sixteen

297

runs with independently prepared standard curves, and the error bars indicate 95% confidence

298

intervals (graphs A-E). The solid line in each graph is the 5 parameter logistic-fit determined by

299

fitting the curve to all replicates (rather than the average). The tightness of the 95% confidence

300

intervals shows that the standard curves for all five proteins are reproducible. Table 2 shows

301

the interassay variability of the standard curves (see precision profiling in Materials and

302

Methods). Typically for ELISA, a CV of 25% or less for each individual standard curve

303

concentration and for the average for all standard concentrations is regarded as acceptable

304

performance36. Note that concentrations at the extremes of the curve (anchor points) are

305

excluded from the calculation of %CV as these are typically beyond the quantifiable range of

306

the curve. It is reasonable to conclude, therefore, that the FMI assay meets or exceeds the

307

precision criteria for all tested proteins.

308

Quantifiable Ranges for FMI and Comparison to ELISA.

309

The quantifiable range of an analytical assay lies between the upper and lower limits that can

310

be reliably measured; i.e. typically, the limits are the points at which the CVs of the interpolated

311

values exceed 25% 35. The wider the dynamic range, the greater the utility of the assay method,

312

since fewer samples have to be repeated at a different dilution to lie within in the quantifiable

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

313

range. An important difference between ELISA and FMI is that, whereas most ELISAs have a

314

dynamic range of about 1.5 log orders, FMIs can potentially have a dynamic range of over 4 log

315

orders 34.

316

The dynamic ranges for the proteins in the FMI are summarized and compared to the

317

equivalent values from the corresponding ELISAs in Table 3. The LLOQ values of FMI are

318

comparable to those obtained in the ELISAs. Note that the LODs were calculated by similar

319

methods in both the ELISAs and the FMI, except that the LOD for FMI was somewhat more

320

conservative since only those values that could be interpolated through the curves were

321

included in the calculation of the standard deviation, whereas in the ELISA calculation included

322

negative values in the calculation of standard deviations. ULOQs were based on the maxDC

323

values determined by Milliplex Analyst for each curve fit. As expected, the ULOQs of the FMI

324

are greater than those of the corresponding ELISA assays. For example, the ULOQ for the GUS

325

protein on the FMI is 102X greater than that of the corresponding ELISA ULOQ (Table 3).

326

Extraction Efficiency, Dilutional Parallelism and Matrix Effects.

327

Establishment of acceptable levels of performance for extraction efficiency, dilutional

328

parallelism and a lack of matrix effects, in a tissue based quantitative assay are essential in

329

determining that an assay is capable of measuring protein levels with sufficient accuracy for

330

expression studies36. In conventional ELISAs, where each protein is considered in isolation, it is

331

relatively easy to optimize extraction conditions and assay buffers to maximize extraction

332

efficiency and dilutional parallelism and minimize matrix effects. In multiplexed assays, the best

333

overall extraction conditions for proteins with disparate physicochemical properties must be

334

established and it is unlikely that these conditions will be optimal for them all. Therefore, it is

ACS Paragon Plus Environment

Page 16 of 35

Page 17 of 35

Journal of Agricultural and Food Chemistry

335

important to establish the optimal overall extraction conditions and to determine whether or

336

not these conditions give acceptable performance for each of the proteins to be measured.

337

Extraction efficiency was established in a similar experiment to that used for the conventional

338

ELISAs by repeatedly extracting the same tissue sample and monitoring the cumulative amount

339

of extracted protein until greater than 90% of the available analyte is recovered. The extraction

340

efficiency experiments for the FMI differed from ELISA in that the extraction volumes used were

341

1 ml, rather than the customary 10 ml volume used for conventional ELISAs. Small volume

342

extraction for FMI yields significant gains in efficiency by extracting samples in a format that

343

was readily transferable to a 96 well format. In the FMI the relative amount of carryover

344

(residual liquid that cannot be recovered) from one extraction to the next is greater: i.e. 50 µl of

345

carryover in the FMI experiment represents 5%, whereas the same volume in the ELISA

346

experiment represents 0.5% carryover. Therefore, it is reasonable to expect that the typical

347

90% threshold used in ELISA is not possible for FMI. Consensus indicates that extraction

348

efficiency of greater than 70% is acceptable 36. For FMI, the extraction efficiencies determined

349

for NPT II, Cry2Ab2, GUS, Cry1Ac, and CP4 EPSPS were 95%, 62%, 88%, 86%, and 76%,

350

respectively. Therefore, almost all of the analytes met the 70% criteria and, accepting the

351

caveats stated above, are similar to those for ELISA. The one exception was that the extraction

352

efficiency for Cry2Ab2 was determined to be 62%. It is likely that further assessment of the

353

extraction efficiency of Cry2Ab2 in the FMI assays could yield results closer to 70% for two

354

reasons: 1) no differences in expression levels were observed when comparing FMI to ELISA

355

using aliquots from independent samples. 2) Cry2Ab2 performance parameters for dilutional

356

parallelism and matrix effects fell within acceptance criteria (Figures 3A and 3B). For the

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

357

purposes of the present study, however, it was not deemed critical to conduct the additional

358

studies needed to better understand the observed extraction efficiency for this one protein out

359

of the five.

360

Dilutional parallelism was carried out using samples from six separate leaf specimens

361

expressing the proteins from both the Bollgard II® and Roundup Ready® Flex Cotton traits.

362

During initial experiments it was found that a 10-fold dilution was optimal for leaf extracts for

363

detecting low expressing proteins - NPT II and Cry1Ac. However, CP4 EPSPS expression levels

364

were well above the ULOQ at these dilutions and CP4 EPSPS had to be assayed by a separate

365

FMI procedure in which the samples were diluted (usually) 800-fold. Dilutional parallelism was

366

assessed at 10, 15, 20 and 25-fold dilutions for all proteins except CP4 EPSPS which was

367

assessed at 400, 600, 800 and 1000-fold dilutions. Figure 3A shows that the recommended 70-

368

130% target range for dilutional parallelism was met for all analytes in the FMI 35.

369

Matrix effects were determined using four different levels of standard proteins spiked into leaf

370

matrix from conventional (i.e. non GM) cotton plants. The highest spike level (Spike #1) for each

371

protein was 80% of the third highest standard curve concentration and spikes 2, 3, and 4 were

372

3-fold serial dilutions of spike #1. The mean recovery for all analytes meets the target range of

373

70 to 130% recovery; except for the highest CP4 EPSPS spike concentration and the lowest

374

Cry1Ac spike concentration. Since it was determined that leaf CP4 EPSPS levels were higher

375

than the quantifiable range of the curve at 10X dilution, and had to be run at an 800X dilution,

376

the matrix results were acceptable since CP4 EPSPS cannot be determined at this matrix

377

concentration and will always be run at a high enough dilution to be beyond any matrix effects.

378

The recovery for Cry1Ac at the lowest spike level was 137%, which is above the target range of

ACS Paragon Plus Environment

Page 18 of 35

Page 19 of 35

Journal of Agricultural and Food Chemistry

379

70 – 130%. Cry1Ac spike level 4 is 0.395 ng/ml and is below the determined LOD for leaf (0.581

380

ng/ml); this point was excluded from assessment of the matrix effects experiment. Therefore,

381

the target range and the target criteria for acceptable matrix effects on the five tested protein

382

analytes were met.

383

Assessing FMI and ELISA Equivalence

384

Multiple reports in mammalian systems have compared conventional ELISAs to their

385

corresponding multiplex assays (Luminex and MSD). FMIs have been extensively applied to

386

profiling panels of related proteins from different areas of biology, examples include: cell

387

signaling pathways37, immune response pathways 38, transplantation biology 39, pathogenesis of

388

specific diseases 38, 40 and cytokine and chemokine pathways 10, 41-45. Over recent years,

389

commercial multiplex kits capable of measuring greater than 30 POIs have become available 46.

390

Studies comparing Luminex multiplex assays for the measurement of soluble cytokines showed

391

a high degree of variability in sample concentrations obtained by using kits from different

392

manufacturers 47, 48. However, when multiplex assay results were compared to the results

393

obtained using conventional ELISA kits from the same manufacturer (usually employing the

394

same capture/detection antibodies) the reported levels showed close agreement 47, 48.

395

Differences in protein standards supplied by manufacturers also contribute to variability 49. In

396

fact, the variability seen between Luminex multiplex assays are probably no greater than those

397

seen between conventional ELISA kits from different manufacturers 50. In the present study,

398

since we were using the same standard proteins and nearly the same antibody pairs, close

399

agreement was expected.

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

400

In this study, equivalency of the two assays systems was assessed by measuring the same

401

sample extracts with both systems and by comparing expression levels obtained in leaf samples

402

from two independent field trials, where one was measured by ELISA and the other by FMI.

403

Comparison of the same sample extracts measured on both systems gave a high degree of

404

linear correlation and concordant expression levels for the five proteins (Figure 4). The

405

differences in antibodies between ELISA and FMI appears to have had inconsequential effects

406

on the results.

407

Since the intent is to use FMI for expression studies of field samples, the more important test of

408

comparability is between results obtained by ELISA and FMI from independent field trials. Two

409

independent field trials grown in geographically distinct regions during the 2013 growth season

410

were compared. In the first trial, grown in Australia, leaf expression levels in Bollgard II® Cotton

411

and Roundup Ready® Flex Cotton were measured by conventional ELISAs. In the second, grown

412

in the USA, leaf expression levels in Bollgard II® Cotton and Roundup Ready® Flex Cotton were

413

measured by FMI. As shown in Figure 5, Expression levels were in close agreement with no

414

statistically significant differences in mean expression levels found by t-test for NPTII, Cry2Ab2,

415

Cry1Ac or CP4 EPSPS. Mean GUS expression levels were statistically significantly different

416

(Figure 5, Graph C and legend). It is unclear why GUS expression levels are different since most

417

of the assay parameters were similar between the two assays. This discrepancy may arise from

418

using a different goat polyclonal detection antibody compared to the ELISA. Since different sets

419

of field samples were assayed by the two different analytical methods (ELISA and FMI), it is

420

possible that the difference reflects natural variability in expression levels under the different

421

environmental conditions.

ACS Paragon Plus Environment

Page 20 of 35

Page 21 of 35

Journal of Agricultural and Food Chemistry

422

The development and validation of the FMI for five proteins expressed in Genuity® Bollgard II®

423

with Roundup Ready® Flex Cotton show that the characteristics and performance of this

424

multiplexed assay system is comparable to that of conventional ELISAs: No evidence of

425

interference was observed between the five analytes and the antibodies. The precision of the

426

FMI standard curves was comparable to, or exceeded, those determined for the ELISAs. The

427

dynamic range of the FMI was greater than that of the corresponding ELISAs, a result fully

428

predictable for FMI-based assay technology. A single extraction method was found to be

429

suitable for all five proteins. Dilutional parallelism and matrix effects also met typical

430

immunoassay acceptance criteria. The only limitation found in this study was that CP4 EPSPS

431

could not be measured with the same sample dilutions as used to measure the other four

432

proteins. This limitation may be overcome in the future by employing different antibody pairs

433

for CP4 EPSPS 34. Alternatively, splitting the multiplex into smaller sets, according to expression

434

levels, can be achieved with little loss of efficiency. Expression levels generated by FMI fit well

435

within the expression variability observed among different fields trials by conventional ELISA. It

436

is reasonable to conclude that FMI provides comparable results to the corresponding

437

conventional ELISAs while providing the benefits of simultaneously determining expression

438

levels of multiple proteins expressed by stacked trait GM products.

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

References 1. Engvall, E.; Perlmann, P., Enzyme-linked immunosorbent assay (ELISA). Quantitative assay of Immunoglobulin G. Immunochemistry 1971, 8, 871-874. 2. Lipton, C. R.; Dautlick, J. X.; Grothaus, C. D.; Hunst, P. L.; Magin, K. M.; Mihaliak, C. A.; Rubio, F. M.; Stave, J. W., Guidelines for the validation and use of immunoassays for determination of introduced proteins in biotechnology enhanced crops and derived food ingredients. Food Agric. Immunol. 2000, 12, 153-164. 3. Grothaus, G. D.; Bandla, M.; Currier, T.; Giroux, R.; Jenkins, G. R.; Lipp, M.; Shan, G. M.; Stave, J. W.; Pantella, V., Immunoassay as an analytical tool in agricultural biotechnology. J. AOAC Int. 2006, 89, 913-928. 4. Aldemita, R. R.; Reano, I. M. E.; Solis, R. O.; Hautea, R. A., Trends in global approvals of biotech crops (1992-2014). GM crops & food 2015, 6, 150-66. 5. Castiglioni, P.; Warner, D.; Bensen, R. J.; Anstrom, D. C.; Harrison, J.; Stoecker, M.; Abad, M.; Kumar, G.; Salvador, S.; D'Ordine, R.; Navarro, S.; Back, S.; Fernandes, M.; Targolli, J.; Dasgupta, S.; Bonin, C.; Luethy, M. H.; Heard, J. E., Bacterial RNA chaperones confer abiotic stress tolerance in plants and improved grain yield in maize under water-limited conditions. Plant Physiol. 2008, 147, 446-455. 6. Hernandez-Rodriguez, C. S.; Hernandez-Martinez, P.; Van Rie, J.; Escriche, B.; Ferre, J., Shared Midgut Binding Sites for Cry1A. 105, Cry1Aa, Cry1Ab, Cry1Ac and Cry1Fa Proteins from Bacillus thuringiensis in Two Important Corn Pests, Ostrinia nubilalis and Spodoptera frugiperda. PLos One 2013, 8, 1-7. 7. Edmeades, G. O. Progress in Achieving and Delivering Drought Tolerance in Maize - An Update. http://www.isaaa.org/resources/publications/briefs/44/specialfeature/Progress%20in%20Achieving%20 and%20Delivering%20Drought%20Tolerance%20in%20Maize.pdf (07/27/15), 8. ISAAA's GM Approval Database. http://www.isaaa.org/gmapprovaldatabase/ (07/27/2015), 9. Dudal, S.; Baltrukonis, D.; Crisino, R.; Goyal, M. J.; Joyce, A.; Osterlund, K.; Smeraglia, J.; Taniguchi, Y.; Yang, J. H., Assay Formats: Recommendation for Best Practices and Harmonization from the Global Bioanalysis Consortium Harmonization Team. AAPS J. 2014, 16, 194-205. 10. Stenken, J. A.; Poschenrieder, A. J., Bioanalytical chemistry of cytokines - A review. Anal. Chim. Acta 2015, 853, 95-115. 11. Hill, R. C.; Oman, T. J.; Shan, G. M.; Schafer, B.; Eble, J.; Chen, C., Development and Validation of a Multiplexed Protein Quantitation Assay for the Determination of Three Recombinant Proteins in Soybean Tissues by Liquid Chromatography with Tandem Mass Spectrometry. J. Agric. Food. Chem. 2015, 63, 7450-7461. 12. Elshal, M. F.; McCoy, J. P., Multiplex bead array assays: Performance evaluation and comparison of sensitivity to ELISA. Methods 2006, 38, 317-323. 13. Richens, J. L.; Urbanowicz, R. A.; Metcalf, R.; Corne, J.; O'Shea, P.; Fairclough, L., Quantitative Validation and Comparison of Multiplex Cytokine Kits. J. Biomol. Screening 2010, 15, 562-568. 14. Chowdhury, F.; Williams, A.; Johnson, P., Validation and comparison of two multiplex technologies, Luminex (R) and Mesoscale Discovery, for human cytokine profiling. J Immunol Methods 2009, 340, 55-64. 15. Fu, Q.; Zhu, J.; Van Eyk, J. E., Comparison of Multiplex Immunoassay Platforms. Clin. Chem. (Washington, DC, U. S.) 2010, 56, 314-318. 16. Premarket Approval (PMA). http://www.fda.gov/Medicaldevices/Deviceregulationandguidance/Howtomarketyourdevice/Premarket submissions/Premarketapprovalpma/Default.Htm (07/23/2015), 17. Dumonceaux, T. J.; Green, M.; Hammond, C.; Perez, E.; Olivier, C., Molecular Diagnostic Tools for Detection and Differentiation of Phytoplasmas Based on Chaperonin-60 Reveal Differences in Host Plant Infection Patterns. PLos One 2014, 9, 1-21.

ACS Paragon Plus Environment

Page 22 of 35

Page 23 of 35

Journal of Agricultural and Food Chemistry

18. Ishii, H.; Tanoue, J.; Oshima, M.; Chung, W. H.; Nishimura, K.; Yamaguchi, J.; Nemoto, F.; So, K.; Iwama, T.; Yoshimatsu, H.; Shimizu, M.; Kozawa, T., First application of PCR-Luminex system for molecular diagnosis of fungicide resistance and species identification of fungal pathogens. J. Gen. Plant Pathol. 2008, 74, 409-416. 19. van Brunschot, S. L.; Bergervoet, J. H. W.; Pagendam, D. E.; de Weerdt, M.; Geering, A. D. W.; Drenth, A.; van der Vlugt, R. A. A., Development of a Multiplexed Bead-Based Suspension Array for the Detection and Discrimination of Pospiviroid Plant Pathogens. PLos One 2014, 9, 1-12. 20. Peters, J.; Thomas, D.; Boers, E.; de Rijk, T.; Berthiller, F.; Haasnoot, W.; Nielen, M. W. F., Colourencoded paramagnetic microbead-based direct inhibition triplex flow cytometric immunoassay for ochratoxin A, fumonisins and zearalenone in cereals and cereal-based feed. Anal. Bioanal. Chem. 2013, 405, 7783-7794. 21. Haasnoot, W.; du Pre, J. G., Luminex-based triplex immunoassay for the simultaneous detection of soy, pea, and soluble wheat proteins in milk powder. J. Agric. Food. Chem. 2007, 55, 3771-3777. 22. Gomaa, A.; Boye, J., Simultaneous detection of multi-allergens in an incurred food matrix using ELISA, multiplex flow cytometry and liquid chromatography mass spectrometry (LC-MS). Food Chem. 2015, 175, 585-592. 23. Choi, S. H., Hexaplex PCR assay and liquid bead array for detection of stacked genetically modified cotton event 281-24-236x3006-210-23. Anal. Bioanal. Chem. 2011, 401, 647-655. 24. Choi, S. H.; Oh, Y. T.; Kwon, J. Y.; Lee, S. N.; Han, B. D.; Ryu, K. H., Development of Detection System Using Multiplex PCR and Liquid Beadarray for Stacked Genetically Modified Rice Event (LS28xCry1Ac). J. Korean Soc. Appl. Biol. Chem. 2010, 53, 639-646. 25. Fantozzi, A.; Ermolli, M.; Marini, M.; Balla, B.; Querci, M.; Van den Eede, G., Innovative Application of Fluorescent Microsphere Based Assay for Multiple GMO Detection. Food Analytical Methods 2008, 1, 10-17. 26. Han, X. Q.; Wang, H. Y.; Chen, H. J.; Mei, L.; Wu, S. Q.; Jia, G. L.; Cheng, T.; Zhu, S. F.; Lin, X. M., Development and primary application of a fluorescent liquid bead array for the simultaneous identification of multiple genetically modified maize. Biosens. Bioelectron. 2013, 49, 360-366. 27. Kluga, L.; Folloni, S.; Van den Bulcke, M.; Van den Eede, G.; Querci, M., Applicability of the "RealTime PCR-Based Ready-to-Use Multi-Target Analytical System for GMO Detection" in processed maize matrices. Eur. Food Res. Technol. 2012, 234, 109-118. 28. Querci, M.; Foti, N.; Bogni, A.; Kluga, L.; Broll, H.; Van den Eede, G., Real-Time PCR-Based Readyto-Use Multi-Target Analytical System for GMO Detection. Food Analytical Methods 2009, 2, 325-336. 29. Querci, M.; Van den Bulcke, M.; Zel, J.; Van den Eede, G.; Broll, H., New approaches in GMO detection. Anal. Bioanal. Chem. 2010, 396, 1991-2002. 30. Fantozzi, A.; Ermolli, M.; Marini, M.; Scotti, D.; Balla, B.; Querci, M.; Langrell, S. R. H.; Van den Eede, G., First application of a microsphere-based immunoassay to the detection of genetically modified organisms (GMOs): Quantification of Cry1Ab protein in genetically modified maize. J. Agric. Food. Chem. 2007, 55, 1071-1076. 31. Tholen, M. S.; Kallner, A. K.; Kennedy, J. W.; Krouwer, J. S.; Meier, K., Evaluation of Precision Performance of Quantitative Measurement Methods: Approved Guideline. In EP5-A2, 2nd ed.; Clinical and Laboratory Standards Institute: Wayne, PA, 2004; Vol. 24. 32. Defawe, O. D.; Fong, Y. Y.; Vasilyeva, E.; Pickett, M.; Carter, D. K.; Gabriel, E.; Rerks-Ngarm, S.; Nitayaphan, S.; Frahm, N.; McElrath, M. J.; De Rosa, S. C., Optimization and qualification of a multiplex bead array to assess cytokine and chemokine production by vaccine-specific cells. J Immunol Methods 2012, 382, 117-128. 33. Dunbar, S. A.; Hoffmeyer, M. R., Chapter 2.9 - Microsphere-Based Multiplex Immunoassays: Development and Applications Using Luminex® xMAP® Technology. In The Immunoassay Handbook (Fourth Edition), Wild, D., Ed. Elsevier: Oxford, 2013; pp 157-174.

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

34. Angeloni, S.; Cordes R., D., S., Garcia, C., Gibson, G. Martin, C., Stone, V., xMAP® Cookbook: A collection of methods and protocols for developing multiplex assays with xMAP Technology. Second ed.; Luminex Corporation: Austin, TX, p 144. 35. Kelley, M.; DeSilva, B., Key elements of bioanalytical method validation for macromolecules. AAPS J. 2007, 9, E156-E163. 36. Shan, G. M., Immunoassays in agricultural biotechnology. 2011; p xii + 350 pp.-xii + 350 pp. 37. Campbell, M.; Lie, W. R.; Zhao, J.; Hayes, D.; Mistry, J.; Kung, H. J.; Luciw, P. A.; Khan, I. H., Multiplex Analysis of Src Family Kinase Signaling by Microbead Suspension Arrays. Assay Drug Dev. Technol. 2010, 8, 488-496. 38. Behnert, A.; Schiffer, M.; Muller-Deile, J.; Beck, L. H.; Mahler, M.; Fritzler, M. J., Antiphospholipase A(2) Receptor Autoantibodies: A Comparison of Three Different Immunoassays for the Diagnosis of Idiopathic Membranous Nephropathy. J. Immunol. Res. 2014, 2014, 1-5. 39. Lachmann, N.; Todorova, K.; Schulze, H.; Schonemann, C., Luminex (R) and Its Applications for Solid Organ Transplantation, Hematopoietic Stem Cell Transplantation, and Transfusion. Transfus. Med. Hemoth. 2013, 40, 182-189. 40. Kofoed, K.; Schneider, U. V.; Scheel, T.; Andersen, O.; Eugen-Olsen, J., Development and validation of a multiplex add-on assay for sepsis biomarkers using xMAP technology. Clinical Chemistry 2006, 52, 1284-1293. 41. Clendenen, T. V.; Arslan, A. A.; Lokshin, A. E.; Idahl, A.; Hallmans, G.; Koenig, K. L.; Marrangoni, A. M.; Nolen, B. M.; Ohlson, N.; Zeleniuch-Jacquotte, A.; Lundin, E., Temporal reliability of cytokines and growth factors in EDTA plasma. BMC Res. Notes 2010, 3, 1-9. 42. Dossus, L.; Becker, S.; Achaintre, D.; Kaaks, R.; Rinaldi, S., Validity of multiplex-based assays for cytokine measurements in serum and plasma from "non-diseased" subjects: Comparison with ELISA. J Immunol Methods 2009, 350, 125-132. 43. duPont, N. C.; Wang, K.; Wadhwa, P. D.; Culhane, J. F.; Nelson, E. L., Validation and comparison of luminex multiplex cytokine analysis kits with ELISA: Determinations of a panel of nine cytokines in clinical sample culture supernatants. J. Reprod. Immunol. 2005, 66, 175-191. 44. Kellar, K. L.; Douglass, J. P., Multiplexed microsphere-based flow cytometric immunoassays for human cytokines. J Immunol Methods 2003, 279, 277-285. 45. Ray, C. A.; Bowsher, R. R.; Smith, W. C.; Devanarayan, V.; Willey, M. B.; Brandt, J. T.; Dean, R. A., Development, validation, and implementation of a multiplex immunoassay for the simultaneous determination of five cytokines in human serum. J. Pharm. Biomed. Anal. 2005, 36, 1037-1044. 46. Chaturvedi, A. K.; Kemp, T. J.; Pfeiffer, R. M.; Biancotto, A.; Williams, M.; Munuo, S.; Purdue, M. P.; Hsing, A. W.; Pinto, L.; McCoy, J. P.; Hildesheim, A., Evaluation of Multiplexed Cytokine and Inflammation Marker Measurements: a Methodologic Study. Cancer Epidemiol., Biomarkers Prev. 2011, 20, 1902-1911. 47. Khan, S. S.; Smith, M. S.; Reda, D.; Suffredini, A. F.; McCoy, J. P., Multiplex bead array assays for detection of soluble cytokines: Comparisons of sensitivity and quantitative values among kits from multiple manufacturers. Cytometry, Part B 2004, 61B, 35-39. 48. Siawaya, J. F. D.; Roberts, T.; Babb, C.; Black, G.; Golakai, H. J.; Stanley, K.; Bapela, N. B.; Hoal, E.; Parida, S.; van Helden, P.; Walzl, G., An Evaluation of Commercial Fluorescent Bead-Based Luminex Cytokine Assays. PLos One 2008, 3, 1-12. 49. Nechansky, A.; Grunt, S.; Roitt, I. M.; Kircheis, R., Comparison of the Calibration Standards of Three Commercially Available Multiplex Kits for Human Cytokine Measurements to WHO Standards Reveals Striking Differences. Biomarker Insights 2008, 3, 227-235. 50. Ledur, A.; Fitting, C.; David, B.; Hamberger, C.; Cavaillon, J. M., Variable Estimates of Cytokine Levels Produced by Commercial Elisa Kits - Results Using International Cytokine Standards. Journal of Immunological Methods 1995, 186, 171-179.

ACS Paragon Plus Environment

Page 24 of 35

Page 25 of 35

Journal of Agricultural and Food Chemistry

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Figure Captions Figure 1. Standard Curve Performance and Cross-reactivity Screening Under Single and Multiplexed Conditions. Standard curves for NPTII (●), Cry2Ab2 (▲), GUS (■) , Cry1Ac (▼) and CP4 EPSPS (◆) were run with all 5 types of capture beads but single analyte and single detection antibody (A, C, E, G, I) and with all capture beads, all analytes and single detection antibody (B, D, F, H, J). Curve fitting was carried out using GraphPad Prism (v6.04) and a 5 parameter logistic fit model (Robust fit option). Dotted lines and open symbols are the standard curves obtained with the full multiplex (all beads, all analytes and all detection antibodies) for that particular analyte.

Figure2. Standard Curve Reproducibility. Sixteen full multiplex runs were carried out resulting in 16 standard curves per analyte (A: NPTII; B: Cry1Ac; C: Cry2Ab2; D: GUS; E: CP4 EPSPS). Each graph is a plot of the median fluorescence intensity (MFI) as a function of the indicated protein standard concentration. The mean MFI for each standard point is shown () and the error bars represent the 95% confidence of each standard curve point. The solid line shows the curve fitted using a 5 parameter logistic fit calculated using the MFIs from all 16 runs.

Figure 3. Dilutional Parallelism and Matrix Effects. A. dilutional parallelism was carried out using four dilutions of four leaf extracts from analyte positive samples. CP4 EPSPS was run at higher dilutions (bracketed dil. factors) due to expression levels that were above the quantifiable range of the standard curve at the dilutions used for the other POI. Percent recoveries were calculated using the average back calculated concentrations. The grey shading shows the expected performance of 70-130%. B. Matrix effects were determined using four different levels of standard proteins spiked into negative leaf matrix. Key for both graphs: NPTII (●), Cry2Ab2 (▲), GUS (■), Cry1Ac (▼) and CP4 EPSPS (◆).

ACS Paragon Plus Environment

Page 26 of 35

Page 27 of 35

Journal of Agricultural and Food Chemistry

Figure 4. Linearity of FMI and ELISA. Ten analyte samples were extracted as described in the Materials and Methods and then run on either the FMI assay or on the corresponding sandwich ELISA. For ELISAs, sample dilutions were adjusted to correspond with the recommended dilutions for each assay and the results for both assays were back calculated from ng/ml to µg/g fresh weight (fwt). Key: NPTII (●), Cry2Ab2 (▲), GUS (■) , Cry1Ac (▼) and CP4 EPSPS (◆). Linear regression analysis (solid line) indicated a strong linearity and a nearly 1 to 1 concordance between the two assays (Gradient =1.031 ± 0.04, R2 = 0.931, P