Trends in Analytical Chemistry 52 (2013) 2–15

Contents lists available at ScienceDirect

Trends in Analytical Chemistry

journal homepage: www.elsevier.com/locate/trac

Review

Foodomics strategies for the analysis of transgenic foods

Alberto Valdés, Carolina Simó, Clara Ibáñez, Virginia García-Cañas ⇑
Laboratory of Foodomics, CIAL (CSIC), Nicolás Cabrera 9, 28049 Madrid, Spain

a r t i c l e i n f o a b s t r a c t
Keywords: Advanced tools in transcriptomics, proteomics, and metabolomics in combination with bioinformatics
Foodomics and chemometrics may provide chemical compositional data that may be helpful in studying any effects
Gene-expression profiling
(intended or not) derived from genetic transformation. In this review, we critically discuss the use of the
Genetically-modified organism (GMO)
main omics technologies to characterize genetically-modified organisms at the transcriptome, proteome
Metabolite profiling
Metabolomics and metabolome levels.
Protein profiling Ó 2013 Elsevier Ltd. All rights reserved.
Proteomics
Substantial equivalence
Transcriptomics
Transgenic food

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. The role of foodomics in GMO analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3. Gene-expression profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4. Protein profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5. Metabolic profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5.1. NMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
5.2. GC-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.3. LC-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.4. CE-MS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.5. FT-ICR-MS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.6. Metabolomic multi-platform studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6. Cross-omics platforms and data-integration studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
7. Conclusions and future outlook. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Abbreviations: 2-DGE, Two-dimensional gel electrophoresis; CE, Capillary electrophoresis; DIGE, Differential in-gel electrophoresis; EI, Electron impact; ESI, Electrospray
ionization; FID, Flame-ionization detection; FT-ICR, Fourier transform ion-cyclotron resonance; GC, Gas chromatography; GM, Genetically modified; GMO, Genetically-
modified organism; HC, Hierarchical clustering; ICP, Inductively-coupled plasma; IT, Ion trap; iTRAQ, Isobaric tags for relative and absolute quantitation; LC, Liquid
chromatography; MALDI, Matrix-assisted laser desorption/ionization; MS, Mass spectrometry; PCA, Principal component analysis; PLS-DA, Partial least squares-discriminant
analysis; qPCR, Quantitative polymerase-chain reaction; TOF, Time of flight.
⇑ Corresponding author. Tel.: +34 910 017 950; Fax: +34 910 017 905.
E-mail address: [email protected] (V. García-Cañas).

0165-9936/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.trac.2013.05.023
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 3

1. Introduction ments for individual crops [9]. This analytical approach has been
reported to cover more than 95% of the GMO composition [10].
The adoption of genetic engineering (or recombinant DNA tech- Although this targeted strategy has enabled identification of unin-
nology) has been considered the fastest growing trend in the his- tended effects in some cases [11], its application to compare the
tory of agriculture, representing one of the foremost compositional differences between GMOs and their comparators
technological advances of the past decades in modern biotechnol- within the substantial equivalence framework has raised several
ogy. The organism derived from recombinant DNA technology is concerns. Specifically, it has been pointed out that this targeted
commonly termed genetically-modified organism (GMO), whereas strategy is biased, and that some unforeseen, unintended effects
food products containing or derived from GMOs are typically re- that could result directly or indirectly from the genetic transforma-
ferred to as transgenic foods. The rapid development of recombi- tion may escape detection [12].
nant DNA technology in agriculture has prompted the production
of GM crops that include benefits in agronomic productivity and
industrial processing. In 2012, the global area of approved GMOs 2. The role of foodomics in GMO analysis
was 170 million hectares in 30 countries, representing almost a
100-fold increase since the first commercialized GM crop in 1996 The bias mentioned and the uncertainties associated with tar-
[1]. Since then, the total accumulated land area cultivated with get-based analysis in comparative compositional evaluation of
transgenic crops has remarkably increased to 1.5 billion hectares. GMOs have motivated development and integration of new, more
Among the most important traits present in authorized GMOs, tol- powerful analytical approaches to deal with the complexity of this
erance to herbicide and resistance to insects are prevalent world- problem and to improve the chances of detecting unintended ef-
wide. Nevertheless, a number of novel traits, such as different fects. As an alternative approach to targeted analysis, the European
micronutrient content, faster ripening, improved feed value, high Food Safety Agency (EFSA) has recommended development and
levels of antioxidants, and tolerance to drought, might be in the use of profiling technologies able to broaden the coverage in com-
pipeline for commercialization in coming years. parative analyses of GMOs [13]. In a recent report, a panel of ex-
Since this development and despite its significant economic po- perts on risk assessment and management suggested the
tential, commercialization of GMOs has become highly controver- application of profiling specifically in those cases where the most
sial, within not only the public sector but also the scientific scientifically valid isogenic and conventional comparator would
community. Regarding safety issues of GMOs, the presence of po- not grow, or not grow as well, under the relevant stress condition
tential unintended changes resulting from genetic transformation [14]. Nevertheless, the profiling of all compounds found in a GMO
represents one of the central issues for debate. Unintended effects (or in any other living organism) in a single analysis is currently
go beyond the primary expected effects of genetic modification, unfeasible. Consequently, multiple analytical techniques are com-
and represent statistically significant differences in a phenotype bined to provide analytical coverage of genes, proteins, and meta-
compared with an appropriate phenotype control [2]. Unintended bolites. In this context, foodomics, recently defined as ‘‘a new
changes may have different origins {e.g., unexpected mutations in- discipline that studies the food and nutrition domains through the
duced by the genetic transformation or during tissue-culture application of advanced omics technologies in order to improve con-
stages of GMO development [3], and secondary effects of gene sumers’ well-being and confidence’’, can play a relevant role in the
expression [4]}. As they are unpredictable, unintended effects, investigation of GMOs [15–17]. Thus, foodomics can provide valu-
regardless of their origin, are considered a significant source of able information about molecular profiling, based on genomics,
uncertainty that might have an impact on human health and/or transcriptomics, proteomics and/or metabolomics analysis, which
the environment [5]. These unexpected modifications would prob- could be essential for GMO characterization and traceability
ably be observed if the changes result in a distinct phenotype, (Fig. 1). In addition to these potential applications, foodomics can
including compositional alterations. Thorough characterization of also be helpful in developing new transgenic organisms, offering
the plant at the molecular level would therefore facilitate detection unprecedented opportunities to study the molecular mechanisms
and description of potential unintended effects in GMOs. leading to a particular phenotype or the mechanisms operating
After several years of intense debate, strict regulations on differ- in important cellular processes, such as the response to different
ent aspects of GMOs, including risk assessment, marketing, label- stresses [18].
ing, and traceability have been established in the European Recently, certain criticisms were made about the usefulness of
Union (EU) and other countries. A leading accepted strategy, molecular profiling for GMO risk assessment [19]. Despite the pub-
termed substantial equivalence, proposes to start risk assessment lication of guidelines describing the minimum standards necessary
of GMOs by comparing them with traditional varieties. Substantial to ensure that microarray and proteomics experiments can be
equivalence is based on the assumption that commercialized tradi- properly interpreted and independently verified [20,21], the lack
tional crops have been consumed for decades and have gained a of standardized, validated procedures that prevent their routine
history of safe use; consequently, they can be used as comparators use in laboratories is still one of the main arguments against pro-
for the safety assessment of new GMOs derived from established filing. Another significant issue of omics techniques is linked to
plant varieties [6,7]. Compositional equivalence between GM crops the poor predictive capacity of the profiles obtained for safety eval-
and conventional (non-GM) comparators is considered to provide uation. Although molecular profiling can effectively measure rela-
an equal or increased assurance of the safety of foods derived from tive differences in compounds between two varieties with high
GM plants. However, this comparative safety assessment is only a sensitivity, the biological relevance of such differences cannot be
starting point within the hazard-identification stage and, depend- determined without previous knowledge of the natural variability
ing on the outcomes in this phase, additional testing may be re- of the crop composition [22]. Furthermore, in most cases where the
quired in further stages of the safety-assessment process [8]. In bandwidth variation of a compound is known, it remains difficult
comparative safety assessment, comparisons are typically per- to decipher the biological meaning of the detected differences in
formed on the basis of targeted analysis of predefined compounds terms of food safety. With respect to the natural variation of the
that include macronutrients and micronutrients, antinutrients, and chemical composition of a crop, it should be evaluated in a wide
natural toxins, following recommendations in the Organization of range of situations (plants grown in different climates, seasons,
Economic Cooperation and Development (OECD) consensus docu- locations, and under different farming practices) to make this
4 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15

Fig. 1. Ideal foodomics platform to analyze genetically-modified organisms (GMOs).

overview as complete as possible [23]. In addition, it has been ar- as a profiling technology over other technologies is that a number
gued that profiling approaches accumulate false positives, which of bioinformatics tools are already available for biological interpre-
would provide superfluous data that may distract from the detec- tation of the results. As an example, differentially-expressed genes
tion of relevant biological information [10]. Despite this open de- can be linked and mapped to relevant biological processes and
bate around the usefulness of omics techniques within the pathways by means of exploratory functional enrichment analysis.
framework for safety assessment of GMOs, a number of reports This strategy opens the possibility of studying functional differ-
demonstrating the suitability and the applicability of different pro- ences between samples to improve prediction of potential unin-
filing approaches for comparative analysis of GMOs suggest good tended effects in GMOs.
acceptance of these fast-evolving techniques by the scientific com- Due to extensive optimization and standardization of instru-
munity [24,25]. It is interesting to note that the majority of the ments and protocols, gene-expression microarray has become a
published omics studies report some differences that can be linked leading analytical technology in transcriptomics studies. The
to genetic transformation. However, a finding that is consistently DNA-microarray technique is based on hybridization of specific
observed in omics profiling studies is that the differences between nucleic acids and it can be used to measure the relative quantities
conventional varieties are in general more pronounced than the of specific mRNAs in two or more samples for thousands of genes
divergence observed between GM and non-GM crops. This obser- simultaneously. Microarrays are collections of oligonucleotides or
vation can also extend to the differences found when the same probes, representing thousands of genes, attached to a substrate,
variety is cultured in different locations and/or seasons. usually a glass slide, at predefined locations within a grid pattern.
Typically, the total RNA fraction, isolated from the GMO under
study, is retrotranscribed to cDNA, which is further labeled with
3. Gene-expression profiling a detectable marker, and allowed to hybridize with individual
probes attached to the glass slide of the microchip. After the
The identification and the quantification of mRNA species under hybridization stage, the slides are washed and imaged using a con-
different conditions have long been of interest to biologists. A pro- focal laser scanner to capture fluorescent signals obtained from the
filing-based analysis of gene expression provides the opportunity sequences bound to their specific probes. Images are converted to
for the global study of the transcriptome. Nowadays, transcrip- raw data, and subsequently transformed into gene-expression data
tome-profiling-based analytical approaches are considered power- using dedicated image-processing software, normalization algo-
ful tools for the broad analysis of signaling pathways, metabolic rithms and statistical tools. Despite the great amounts of highly
routes and networks in a given biological system [26]. In the past informative, useful data generated by microarray, these data suffer
decade, gene-expression profiling has advanced significantly to- from some limitations, including:
wards more sensitive, automatic and high-throughput techniques.
The latest developments in high-throughput technologies, includ-  high cost, which requires significant investment in equipment
ing DNA microarray and next-generation sequencing, have pro- and consumables;
vided an important impulse to expand genomics and  a large number of false discoveries (consequence of the multiple
transcriptomics [27]. One of the benefits of using transcriptomics hypothesis testing);
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 5

Table 1
Gene-expression-profiling studies in genetically-modified organisms (GMOs) using microarray technique

GM crop Tissue Donor species Genetic modification Phenotype Ref.

Wheat Seed and leaf Triticum aestivum Glu-A1, Glu-D1 Nutritionally enhanced [29]
Rice Seed Oryza sativa OASA1D Free tryptophan accumulation [30]
Rice Seed Hordeum vulgare BCBF1 Control stress-inducible genes [31]
Rice Leaf Aspergillus giganteus afp Fungal resistance [32]
Rice Leaf Bacillus thuringiensis Cry1Ab Insect resistance [33]
Maize Seed B. thuringiensis Cry1Ab Insect resistance [34]
Maize Leaf B. thuringiensis Cry1Ab Insect resistance [35]
Soybean Leaf Agrobacterium tumefaciens CP4 EPSPS Herbicide tolerance [36]
Wheat Seed Aspergillus fumigatus PhyA Endosperm-specific phytase expression [37]
Soybean Leaf A. tumefaciens CP4 EPSPS Herbicide tolerance [38]
Potato Tuber Solanum tuberosum Sus4 Starch metabolism [39]
Rice Root and shoot H. vulgare AlaAT Alanine aminotransferase over-expression [40]

 high background noise, which limits the sensitivity; and, only 1.36% of all analyzed rice transcripts (i.e., 680 transcripts),
 saturation of signals, which reduces dynamic range and the real which, interestingly, were distributed over the rice chromosomes.
quantitative capabilities of this technology [28]. Functional enrichment analysis of the list of altered genes using
the Plant MetGenMAP bioinformatics tool allowed the identifica-
In spite of these constraints, this technology is helpful for the tion of changes in genes directly related to stress/defense re-
identification of differences between two or more transcriptomes. sponses and also to amino-acid metabolism in the GM rice.
Indeed, the commercial availability of microarrays covering a wide Furthermore, transcriptomics results were complemented with
portion of transcriptome in a number of species (e.g., barley, cot- amino-acid profiling analysis of the same samples using isobaric
ton, maize, rice, tomato, and wheat) has enabled their application tags for relative and absolute quantitation (iTRAQ) and liquid chro-
in GMO analysis (Table 1). matography (LC) coupled to mass spectrometry (MS). Changes in
Initial works reporting the use of gene-expression microarray the levels of 10 amino acids and c-amino-n-butyric acid in the
for the study of the differences between the transcriptional profiles
in GM plants and their unmodified counterparts suggested that the
presence of transgenes did not significantly alter gene expression
[29]. Dubouzet et al. employed a combination of microarray anal-
ysis with targeted metabolite analysis to investigate the unex-
pected accumulation of tryptophan and the low levels of
tryptophan derivatives in GM rice overexpressing anthranilate
synthase alpha sub-unit [30].
In rice plants, Batista et al. investigated the extent of unex-
pected transcriptome modifications occurring during genetic engi-
neering versus mutation breeding using the high-density
Affymetrix GeneChip Rice Genome Array, containing probes to
query 51,279 transcripts [31]. The study was performed using four
improved rice varieties (two mutagenized and two transgenic) and
the respective unmodified original genotypes as controls. Results
indicated that the acquisition of a new desired trait or modification
by either procedure caused stress, and, consequently, a broad im-
pact on gene expression. The transcriptomics analysis showed few-
er genetic alterations in GM plants than in mutagenized lines,
although authors could not rule out that these observations may
be specific to the particular rice lines examined in the study.
Using the same commercial microarray platform, Montero et al.
have studied the impact of the transgene, the insertion site and the
associated rearrangements on transcriptomic regulation of GM rice
[32]. The study confirmed differences between genotypes in less
than 0.5% transcripts covered by the microarray. Interestingly,
one-third of the detected differences could be attributed to the
procedure used to obtain GM plants (e.g., in vitro culture, cell dedif-
ferentiation, and plant regeneration), whereas around 15% of the
transcriptomic changes observed in the GM line were linked to
the transformation event (i.e. event specific), associated with the
insertion site and the genome rearrangements occurring during
transformation.
More recently, an insect-resistant rice variety was also investi-
gated for unintended effects using gene-expression-microarray
profiling [33]. This GM rice variety was selected as a model in this Fig. 2. Principal component analysis (PCA) of the sequence-expression data.
Classification of samples using PC1 versus PC2 (a) and PC1 versus PC3 (b).
study due to it having shown some unintended effects, such as in-
Rhombuses correspond to Helen Bt samples; squares to Helen samples; triangles
creased susceptibility to certain rice-specific diseases. Comparative to Beles Sur samples, and crosses to Sancia samples. Open shapes represent control
transcriptomics analysis of the GM rice and a suitable comparator (C) and filled shapes low-N conditions (low-N). (Reprinted from [35], with kind
grown under the same conditions revealed significant changes in permission from Springer Science and Business Media.)
6 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15

transgenic rice plants relative to the conventional counterparts al- component analysis (PCA) of the transcriptional profiles of all 37
lowed some links to be established between the reported unin- sequences suggested that plant variety had the highest impact on
tended effects and the differences at the transcriptional and gene-expression patterns, followed by the nitrogen fertilization,
metabolite level. while the transgenic character had the lowest impact on gene-
In a series of articles, Coll et al. applied the microarray-profiling expression variation (Fig. 2).
technology to study gene expression in insect-resistant GM maize. The transcriptomics study on GM soybean by Cheng et al. in-
The first study reported the comparative study of several varieties cluded two transgenic cultivars, both derived from the same inser-
containing the MON810 event with their near-isogenic counter- tion event, and three conventional cultivars, grown under the same
parts [34]. High-density Affymetrix microarray technology was conditions [36]. In the first part of the study, the five soybean cul-
used to analyze plantlets cultured in vitro under highly-controlled tivars were genotyped to establish the genetic distances between
experimental conditions in order to minimize transcriptome cultivars. Gene-expression data from the analysis of 25 samples
changes due to environmental factors. The analyses indicated dif- were analyzed by PCA and unsupervised hierarchical clustering
ferent expression patterns in terms of number and type of genes (HC). The classification indicated divergences between the oldest
between two MON810 varieties and their comparators. The analy- variety samples (Mandarin cultivar), samples displaying more
sis of a sub-set of 38 gene sequences using real-time quantitative RNA degradation and other 15 samples of non-GM and GM soy-
polymerase chain reaction (qPCR) analysis confirmed the micro- beans. Interestingly, the two GM cultivars did not cluster into a
array results. The analysis of the same genes in other MON810/ separate group from the traditional cultivars, suggesting that the
non-GM maize pairs revealed different expression patterns, sup- insertion of the transgene had only minimal effects on global gene
porting the view that the genetic background of each variety influ- expression. Pairwise comparisons between the GM cultivars and
ences the observed divergence in the transcriptome. Later, the non-GM counterparts revealed a lower number of differen-
transcriptomics profiling was extended to MON810 varieties culti- tially-expressed genes than the number obtained from the com-
vated in a natural environment following common agronomic parison between non-GM cultivars. Statistical analysis confirmed
practices [41]. In this case, the expression patterns of 38 sequences the down-regulation of genes involved in cysteine-protease inhib-
assayed using real-time qPCR suggested that most sequences se- itor and dihydroflavonol-4-reductase activities in both transgenic
lected as differentially expressed in maize plants grown in vitro cultivars. However, it could not be concluded whether these
did not show significant changes in the plants grown in the field. changes were an effect of the insertion event, an effect of the trans-
Next, the effect of different agricultural environments and cul- genic product, or a natural variation of the parent genotype.
tural conditions on the transcriptional differences between
MON810 and non-GM maize plants was assessed [35]. Pair-wise 4. Protein profiling
comparisons of GM/non-GM maize varieties grown under low-
nitrogen and control conditions showed divergences in the range Proteins have a key role in gene function and are involved in a
0.07–0.2% of maize transcriptome. The numbers of genes showing variety of cellular processes including, among others, metabolism
differential expression between the transgenic and non-transgenic and development. With regard to food safety, proteins may also
maize were very low in both control and low-level nitrogen condi- act as toxins, antinutrients, or allergens [42]. Thus, the study of dif-
tions (13 and 23, respectively). The expression of 37 gene se- ferentially-expressed proteins in transgenic food has raised inter-
quences, selected from the complete dataset under analysis, was est in safety assessment. In this context, proteomics, able to
confirmed by real-time qPCR technique with a 71.1% degree of quantify hundreds of proteins in parallel, has become a key tech-
coincidence between microarray and PCR data. Principal nology in comparative studies of the entire proteome of GM plants

Table 2
Protein-profiling studies in genetically-modified organisms (GMOs)

Analytical methodology GM crop Tissue Donor species Genetic modification Phenotype Ref.
2-DGE/MALDI-TOF/TOF MS Maize Seed B. thuringiensis Cry1Ab Insect resistance [45]
Pea Seed P. vulgaris aAI1 Insect resistance [46]
Maize Leaf B. thuringiensis Cry1Ab Insect resistance [47]
2-DGE/LC-ESI-Q/TOF MS Rice Leaf O. sativa CDPK13,CRTintP1 Cold tolerance [48]
2-DGE/LC-ESI-IT MS Maize Seed B. thuringiensis Cry1Ab Insect resistance [49]
Maize Grain B. thuringiensis Cry1Ab Insect resistance [50]
Potato Tuber Aureobasidium pullulans W2 Waxy phenotype [51]
Potato Tuber S. tuberosum Mal1 Changes in cell wall structure [51]
Potato Tuber S. tuberosum SamDC Modified metabolism [51]
2-DGE/MALDI-TOF MS Tomato Seed tomato spotted wild virus (TSWV) TSWV-N Virus resistance [52]
Tomato Leaf cucumber mosaic virus (CMV) ScFv (G4) Virus resistance [53]
Rice Seed B. thuringiensis Cry1Ab, Cry1Ac Insect resistance [54]
Wheat Seed Phalaris coerulescens Trx s Resistance to pre-harvest sprouting [55]
Wheat Seed T. aestivum Bar Herbicide tolerance [56]
Grapevine Leaf Escherichia coli Adh Abiotic stress [57]
Wheat Seed Nicotiana tabacum Rab1 Improved functional properties [58]
2-DGE/MALDI-Q/TOF MS Soybean Seed A. tumefaciens CP4 EPSPS Herbicide tolerance [59]
2D-DIGE/MALDI-Q/TOF MS Soybean Seed A. tumefaciens CP4 EPSPS Herbicide tolerance [60]
2D-DIGE/LC-ESI-Q/TOF MS Pea Seed P. vulgaris aAI1 Insect resistance [61]
2D-DIGE/MALDI-TOF/TOF MS Rice Seed B. thuringiensis Cry1Ac, sck Insect resistance [62]
MALDI-TOF MS Potato Tuber S. tuberosum G1-1 Sprouting delay [63]
LC-ESI-IT MS Maize Grain B. thuringiensis Cry1Ab Insect resistance [64]
LC-ESI-TOF MS Maize Grain B. thuringiensis Cry1Ab Insect resistance [65]
CE-ESI-IT MS/CE-ESI-TOF MS Maize Grain B. thuringiensis Cry1Ab Insect resistance [66]
Soybean Seed A. tumefaciens CP4 EPSPS Herbicide tolerance [67]
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 7

and their non-modified counterparts [43]. Two conceptually differ- used to avoid the gel-to-gel irreproducibility for comparative pro-
ent strategies can be approached in comparative proteomics. The teomics [60]. In 2-D DIGE, different samples labeled with ultra-
first, referred to as bottom-up, has been widely applied to investi- high-sensitive fluorescent dyes, typically Cy5 and Cy3, are loaded
gate the substantial equivalence and the potential unintended ef- in the same gel. Then, dedicated software is used for comparative
fects in GMOs, while the second, the so-called shotgun, has analysis of gel images, allowing the simultaneous detection of pro-
scarcely been used in GMO analysis. tein spots labeled with the two fluorescent dyes. Using this meth-
Bottom-up proteomics usually involves the application of two- odology, equal areas of each image can be compared more
dimensional gel electrophoresis (2-DGE) followed by image analy- accurately. In the study of the GM soybean proteome, a total of four
sis, and subsequent identification of the differentially-expressed spots were detected as differentially expressed compared with the
proteins by MS. Typically matrix-assisted laser desorption/ioniza- control variety of soybean. The authors correlated the proteomics-
tion coupled to time-of-flight MS (MALDI-TOF-MS) or different profiling results with data obtained from enzymatic determination
variants of LC-MS are used for the identification of individual pro- of superoxide dismutase, catalase, ascorbate peroxidase, and gluta-
teins excised from gel spots. So far, 2-DGE provides the highest thione-reductase activities, suggesting an overall high level of oxi-
protein-resolution capacity with a low cost of instrumentation. dative stress in the transgenic soybean seeds. In addition to this
However, this technique suffers from some technical limitations report, other studies have been conducted on GMOs using 2-D
that prevent the separation of highly hydrophobic proteins with DIGE methodology. For instance, Gong et al. investigated the po-
extreme isoelectric point or high molecular weight. Also, the gel- tential unintended effects caused by insect-resistant transgenic
to-gel variation is one of the major sources of error that makes dif- events in rice using 2-D DIGE (Fig. 3) [62]. Apart from the study
ficult an exact match of spots in the image-analysis process. In of the effects caused by the transgenic modification in indica vari-
spite of these limitations, the 2-DGE technique has proved to be eties, the experimental design allowed evaluation of the effects of
valuable for molecular profiling of GMOs, providing highly-re- conventional genetic breeding and natural genetic variation on the
solved proteomics fingerprints comprising thousands of proteins rice proteome. PCA of the 2-D DIGE results showed all non-GM
[44]. Table 2 lists some representative examples of the application varieties clearly separated from each other, while the separation
of this methodology for profiling the proteome of GMOs versus between GM lines and their controls was not evident. The prote-
their corresponding non-GM counterparts. omes showing the higher degree of divergence were those
First reports on the proteomic profiles of insect-resistant GM corresponding to indica and japonica cultivars, followed by the
MON810 maize showed unexpected differences between the GM three indica varieties. Most of the expression changes were linked
variety and its near-isogenic line [45,49]. However, 2-DGE analyses to the variety and not to the transgenic character. A total of 234
also showed a greater number of proteins that were differentially spots, corresponding to differentially-expressed proteins found
regulated by the environmental conditions. Later, Coll et al. inves- across the pairwise comparisons, were successfully identified
tigated the potential unintended effects of MON810 transgene in using MALDI-TOF/TOF-MS. MS results revealed that a significant
two different commercial varieties of maize using bottom-up pro- number of proteins were related with metabolism, in particular,
teomics [50]. In that study, harvested maize grains from plants tricarboxylic-acid cycle and glycolysis; protein folding and modifi-
grown in agricultural fields were subjected to 2-DGE analysis. Sta- cation; and, defense response.
tistical analysis of electrophoretic data indicated a small number of Approaches for the analysis metal-binding proteins (metallo-
differential spots (61.2% analyzed proteins) in each of the two GM/ proteomics) in GMOs have also received attention in recent years.
non-GM comparisons. Furthermore, fold changes were minimal Sussulini et al. applied bottom-up proteomics in combination with
and none of these changes were detected in the two variety pairs synchrotron radiation X-ray fluorescence and atomic absorption
tested. The set of differentially-expressed proteins, identified from spectrometry to study the protein metallome in GM and non-GM
their peptide mass fingerprint using LC-MS/MS, had distinct pre- soybeans [68]. In their work, different patterns of metal ions,
dicted molecular functions and were involved in different biologi- including Ca(II), Cu(II), Fe(II), Mn(II), Ni(II), and Zn(II), were found
cal processes. Hence, the authors could not establish functional
correlations between the proteins differentially expressed in the
GM maize varieties.
A similar analytical approach was used to compare leaf-protein
profiles of four MON810 maize varieties with the four isogenic
varieties grown in environmentally-controlled growth chambers
[47]. In this case and as occurred in the previous study on maize
grains, the differential protein expression observed in leaf extracts
was variety specific, supporting the idea that genetic modification
does not significantly alter global protein expression in the plant.
Similar conclusions have been reported in comparative bottom-
up proteomics studies on tomatoes with a genetically-added resis-
tance to virus and insect attacks, and potatoes showing a delayed
sprouting process and modified cell-wall structure [51–53].
Regarding this type of study, Brandao et al. have highlighted the
importance of optimizing the parameters that influence the com-
parisons of the protein map after different gel runs, including those
parameters involved in image acquisition [59]. Using a strictly-
controlled routine for image analysis of 2-D gels, a maximum of
79% of spot match was achieved when the GM soybean proteome
was compared to the corresponding non-modified soybean line.
The results of this study indicated that the number of protein spots Fig. 3. Representative DIGE image of rice-seed proteins. Internal standard was
labeled with Cy2, and proteins from ZH10 and D68 were labeled with Cy3 and Cy5,
present in the linear pI range 4–7 was larger than that obtained in respectively. Molecular mass (in kiloDaltons) and pI of proteins are on the left and
the pI range 3–10. In another study on GM soybean, the 2-D fluor- at the top of the image, respectively. (Reprinted with permission from [62]. Ó2012,
escence difference gel electrophoresis (2-D DIGE) technique was American Chemical Society.)
8 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15

in proteins randomly selected from 2-DGE separations of GM and zein-proteins fraction from three different GM maize cultivars
non-GM soybean extracts. and their corresponding isogenic lines. Both instruments showed
In addition to the study of potential unintended effects, the similar sensitivity and repeatability, but CE-ESI-TOF-MS provided
investigation of the intended effects induced by the genetic modi- better results with regard to the number of identified proteins.
fication at the protein level can be successfully addressed using Although shotgun proteomics has already been demonstrated
bottom-up proteomics. This strategy has proved to be helpful for to be a suitable strategy for proteome profiling, it has barely been
the study of how the genetic transformation may alter protein used in GMO analysis. In shotgun proteomics, protein digestion is
abundance, structure, or function, and also, the molecular mecha- performed without any pre-fractionation or separation of the pro-
nisms leading to a particular phenotypic characteristic. A remark- teome. The resulting peptides are separated with LC or CE followed
able example of this application has been reported by Guo et al. by MS analysis to provide rapid, automatic identification of pro-
in the study of the resistance to pre-harvest sprouting observed teins in the sample. Based on this concept, a profiling CE-ESI-
in transgenic wheat seeds [55]. The authors applied 2-DGE and TOF-MS method was developed for the investigation of unintended
MALDI-TOF-MS techniques to decipher proteome differences be- effects in GM soybeans [67]. In this work, protein digestion was
tween wild-type and transgenic wheat seeds expressing a heterol- performed without any prefractionation or separation of the prote-
ogous version of antisense thioredoxin gene. The electrophoretic ome. During the first stages of method development, several
analysis provided about 600 matched protein spots in control parameters affecting the separation and the detection of peptides
and transgenic samples, but only the expression of 16 proteins were optimized. Following this approach, a total of 151 peptides
was considered as significantly altered. The identification of the ex- were automatically detected for each soybean line, but none of
pressed proteins and their functions was the foundation to propose the profiles showed statistical differences between samples.
a molecular regulation model involving thioredoxin protein in Shotgun proteomics was also combined with iTRAQ to quantify
wheat-seed germination. differences in protein profiles between GM and unmodified rice
The combined use of 2-DGE and MS was also the preferred ana- [70]. The procedure involved the treatment of four different di-
lytical strategy to study the mechanisms involved in the response gested samples with four independent isobaric reagents, designed
of GMOs under a variety of stresses – abiotic (e.g., chemicals, to react with all primary amines of protein hydrolyzates. Treated
drought, and salinity) and biotic (e.g., pathogens, and parasites) samples were subsequently pooled and analyzed by tandem MS.
[48]. The proteomics study of GMOs might also focus on the tech- The analyses using this analytical technique revealed significant
nological benefits of a particular transformed event. For example, differences between GM and wild-type rice in 103 proteins out
Di Luccia et al. carried out a study on two GM durum wheat culti- of the 1,883 proteins identified in rice-endosperm samples.
vars with modified functional performance of the grain in order to
investigate the technological properties of gluten [58].
Gel-free protein (or peptide) separation techniques, namely LC 5. Metabolic profiling
and capillary electrophoresis (CE), are suitable alternatives to 2-
DGE for bottom-up proteome profiling. These techniques enable Genetic engineering is often used to modify the metabolism for
direct coupling to a mass spectrometer, require smaller amounts optimal production of plant metabolites, which may directly benefit
of the necessary starting material, provide full automation, high- human health and plant growth [71]. A representative example of this
throughput capabilities, and have better reproducibility than 2- is transgenic rice, commonly known as ‘‘golden rice’’, whose genetic
DGE in terms of qualitative and quantitative analysis. For example, modification is being developed to biosynthesize b-carotene in grain
an LC-MS method, based on an ion trap (IT) mass analyzer and [11]. In this context, metabolomics has the potential to play a relevant
electrospray ionization (ESI), was developed to compare the pro- role in GMO analysis, allowing detection of the effects (intended or
tein profiles of transgenic MON810 maize varieties with those ob- not) that might take place as a result of the genetic transformation
tained from their corresponding non-transgenic line [64]. The [9]. However, metabolomics analysis faces several challenges, since
analyses revealed spectral signals that seemed to be very similar metabolites encompass a wide range of chemical species with diver-
between GM and the unmodified lines, whereas several proteins gent physicochemical properties and at very different concentrations.
were found to be characteristic for samples from specific regions. Regarding the technical requirements for comprehensive metabol-
Using a similar strategy, LC-TOF-MS was applied to profile the omics analysis, high sensitivity and resolution are the key parameters
low-molecular weight protein fraction in both GM and non-GM to consider for the selection of an appropriate method [72]. Two ana-
maize [65]. The sample-preparation step consisted of a five-step lytical platforms are currently used for metabolomics analysis: MS
sample fractionation of maize flour with different water-based and nuclear magnetic resonance (NMR). These techniques stand
and ethanol-based buffer solutions. With this procedure, albumin, alone or are combined with separation techniques (typically, LC-
globulin, zein, zein-like glutelin and glutelin fractions could be NMR, gas chromatography (GC)-MS, LC-MS, and CE-MS), can produce
analyzed separately in order to investigate relevant differences in complementary analytical information to attain more extensive
protein profiles that might eventually be useful as markers for metabolome coverage [73]. Compared to NMR, MS is a more sensitive
GMO identification in maize-derived foods. In this case, the globu- technique, and MS coupled to GC, LC, or CE allows higher resolution
lin fraction showed protein profiles with a higher degree of diver- and sensitivity for low-abundance metabolites [74].
gence between GM and conventional maize samples. Typically, metabolic profiling analysis involves the following
Gel-free approaches have also been investigated to study the general steps:
protein metallome in GMOs. For example, Mataveli et al. applied
off-line 2-D LC with inductively-coupled plasma MS (ICP-MS) to (1) metabolite extraction that often has to be adapted on a case-
investigate the protein metallome in GM and non-GM soybeans by-case basis depending on the type of sample and the ana-
[69]. In their report, MS analysis indicated different metal-ion con- lytical platform chosen;
tent among various separated fractions. In addition, 33 proteins (2) sample preparation, which may include partial purification
were characterized using LC-MS/MS. and derivatization steps;
Erny et al. demonstrated the potential of CE coupled to ESI-MS (3) instrumental analysis of samples;
for the proteomics profiling of insect-resistant transgenic maize (4) detection and quantification of metabolite signals in raw
[66]. A comparative study of two different mass analyzers, namely chromatogram/spectrum data to generate a data matrix list-
TOF and IT, was carried out for the analysis of the intact ing metabolites and their intensity data; and,
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 9

Table 3
Metabolite-profiling studies in genetically-modified organisms (GMOs)

Analytical technique GM crop Tissue Donor species Genetic Phenotype Ref.

modification
NMR Maize Grain B. thuringiensis Cry1Ab Insect resistance [76]
Maize Grain B. thuringiensis Cry1Ab Insect resistance [77]
Pea Leaf Streptomyces Bar Herbicide tolerance [78]
hygroscopicus
Wheat Seed T. aestivum Glu-A1, Glu-D1 Nutritionally enhanced [79]
Lettuce Leaf E. coli asn A Growth enhanced [80,81]
Potato Tuber A. pullulans W2 Waxy phenotype [82]
GC-EI-Q MS Potato Tuber S. tuberosum AGPase, StcPGM, Altered starch [83–
StpPGM composition 85]
Tomato Leaf and Arabidopsis thaliana AtHXK1 Altered carbohydrate [86]
fruit metabolism
Rice Seed B. thuringiensis Cry1Ac, sck Insect resistance [87]
Rice Seed O. sativa RCH10, RAC22, b- Antifungal activity [88]
Glu, B-RIP
Raspberries Fruit Raspberry bushy Virus movement Virus resistance [89]
dwarf virus protein
Cucumber Fruit Thaumatococcus Thaumatin-II Sweet flavor [90]
daniellii
Soybean Seed A. tumefaciens CP4 EPSPS Herbicide tolerance [91]
Maize Grain B. thuringiensis Cry1Ab Insect resistance [92]
Maize Grain B. thuringiensis Cry1Ab Insect resistance [93]
Maize Grain A. tumefaciens CP4 EPSPS Herbicide tolerance [93]
Soybean Leaf and N. tabacum ASA2 Nutritionally enhanced [94]
seed
GC-EI-TOF MS Cucumber Fruit T. daniellii Thaumatin-II Sweet flavor [90]
Potato Tuber Cynara scolymus 1-SST, 1-FFT Inulin synthesis [95]
LC-ESI-Q MS Potato Tuber C. scolymus 1-SST, 1-FFT Inulin synthesis [95]
Rice Leaf Zea mays C1, R-S Flavonoid production [96]
Tomato Fruit Vitis vinifera Stilbene synthase Resveratrol synthesis [97]
Rice Seed E. coli glgC-TM Nutritionally enhanced [98]
Rice Seed N. tabacum ASA2 Nutritionally enhanced [99]
LC-ESI-Q/TOF MS Rice Seed B. thuringiensis Cry1Ac, sck Insect Resistance [100]
CE-ESI-TOF MS Maize Grain B. thuringiensis Cry1Ab Insect resistance [101]
Soybean Seed A. tumefaciens CP4 EPSPS Herbicide tolerance [102]
Soybean Seed A. tumefaciens CP4 EPSPS Herbicide tolerance [103]
CE-ESI-Q MS Rice Seed O. sativa YK1 Stress tolerance [104]
FT-ICR MS Maize Grain B. thuringiensis Cry1Ab Insect resistance [105]
Rice Leaf and O. sativa YK1 Stress tolerance [106]
calli
Multiplatform: GC-EI-Q MS and LC-ESI-IT MS Grapevine Leaf E. coli Adh Abiotic stress [107]
Multiplatform: GC-EI-Q MS and LC-ESI-Q MS Papaya Pulp and Carica papaya rep Virus resistance [108]
leaf
Multiplatform: GC-EI-TOF MS, LC-ESI-Q/TOF MS and Tomato Fruit Richadella dulcifica Miraculin Sweet flavor [109]
CE-ESI-Q/TOF MS

(5) statistical analysis of metabolite profiles [75]. maize variety and of the corresponding traditional variety were
investigated using 1-D and 2-D NMR techniques [77]. About 40
Analysis of the vast amount of data obtained in these studies is water-soluble metabolites in the maize-seed extracts were de-
also challenging. Consequently, the application of chemometrics tected. Some of them were identified for the first time in the 1H
and bioinformatics has become essential to exploit the benefits NMR spectrum of the maize-seed extracts. Some quantitative dif-
of metabolic profiling fully to discover significant differences ferences found between transgenic and non-transgenic samples al-
among different plant genotypes. Table 3 summarizes representa- lowed discrimination of the genotypes by PCA.
tive examples of metabolite-profiling studies on GMOs. A comparative NMR metabolomics study was also carried out
on samples from pea plants (six different transformation events),
5.1. NMR a null segregant control from which the transgene has been lost,
and a parental line [78]. Multivariate analysis on the basis of PCA
NMR provides qualitative and quantitative data for many and linear discriminant analysis failed to provide an acceptable
metabolites in a given sample (usually the more abundant constit- classification of samples. Moreover, statistical analysis revealed
uents). Manetti et al. investigated the potential of NMR to identify similar results, suggesting that the null segregant group was signif-
and to classify maize seeds derived from GM and non-GM maize icantly different from the wild-type, indicating that factors other
using multivariate techniques [76]. The analyses revealed correla- than the presence of the transgene had a large effect on the metab-
tions among the different metabolites, and discriminant metabo- olite profile.
lites (choline, asparagine, histidine, and trigonelline), which were Baker et al. carried out a metabolomics study, based on the
lower in transgenic samples than in controls. In a later report, application of NMR, to investigate transgenic wheat varieties over-
the metabolic profiles of seeds from an insect-resistant transgenic expressing high-molecular-weight sub-units of glutenin grown in
10 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15

replicate field experiments on two different locations for four years GC-MS was also demonstrated to be a valuable tool for profiling
[79]. In order to gain more information on the metabolic composi- aroma compounds in transgenic fruits and vegetables. For
tion of the samples, the levels of amino acids were also analyzed in example, Malowicki et al. investigated the volatile fraction of GM
the same samples by GC-MS. The NMR technique was useful in raspberries with added resistance to virus attack using GC-MS
identifying some differences in the levels of disaccharides maltose [89]. In their comparative study, volatile compounds were ex-
and sucrose between parental and transgenic lines. Similarly, dif- tracted from the red raspberries with a stir-sorptive bar. Using
ferent accumulation of amino acids, in particular aspartate and selective-mass ions to avoid any interference between co-eluted
glutamine, was found in plants grown at different sites. However, compounds, MS quantification was performed with quantification
comparative studies of the null transformant with the control line, curves. The study of 30 selected compounds belonging to various
and of the transgenic lines with their respective controls, sug- chemical classes did not reveal significant differences between
gested that the transformation process and the expression of the the transgenic variety and the unmodified counterpart.
transgenes did not have a significant, reproducible impact on the In another study on the profiles of aroma compounds of four
grain metabolome. lines of GM cucumber using GC-MS, solid-phase microextraction
In a series of reports, Sobolev et al. employed NMR to study dif- provided better results in terms of number of extracted volatile
ferent metabolic aspects of transgenic lettuce with enhanced compounds than microdistillation [90]. The analysis of cucumber
growth properties [80,81]. Statistical analysis of NMR spectra re- extracts using GC-EI-Q-MS and GC-EI-TOF-MS suggested that the
vealed an increase in content (up to 30 times) of short-chain inulin significant differences found between GM and non-GM varieties
oligosaccharides in transgenic lettuce leaves, and this difference were quantitative rather than qualitative, regardless of the MS-
was independent of the plant-growth stage. This finding appeared based technique used.
as an unexpected effect because the transgene aims to modify the Other reports paid attention to the metabolite-extraction proce-
asparagine content, together with the nitrogen status, rather than dure. For example, Bernal et al. combined GC-MS with several
the carbohydrate content. selective extraction methods, including supercritical fluids or
accelerated solvents, to analyze amino acids and fatty acids in
GM soybean and maize [91,92].
5.2. GC-MS Frank et al. recently applied GC-MS for metabolite profiling in
insect-resistant and herbicide-tolerant GM maize varieties involv-
GC-MS is one of the most popular profiling techniques to study ing a metabolite-extraction scheme designed to obtain four frac-
the metabolome of GMOs. The high separation efficiency and the tions containing: major lipids; minor lipids; sugars, sugar
reproducibility of GC-MS analysis combined with the stable ioniza- alcohols and acids; and, amino acids and amines [93]. Fractions ob-
tion and fragmentation achieved by electron-impact (EI) ionization tained from transgenic and control (near-isogenic) grains from
make this coupling well suited to the analysis of complex samples. plants grown at distinct locations and in different seasons were
GC-MS is able to determine the levels of primary metabolites, such independently analyzed using GC-EI-Q-MS. Multivariate statistical
as amino acids, organic acids, and sugars, by employing chemical analysis revealed compounds from the two polar factions as the
derivatization. major contributors to the variation observed between samples
In a series of reports, Roessner et al. used GC-MS to characterize grown in different locations and seasons. However, none of the
metabolite profiles of transgenic potato tubers with modified sugar fractions indicated statistically separation of GM maize from the
or starch metabolism. In the first study, GC-MS was applied to ana- near-isogenic variety.
lyze polar metabolites in extracts obtained from potato tubers [83].
Sample preparation involved methoximation and silylation to vol- 5.3. LC-MS
atilize various classes of compounds. This methodology in combi-
nation with MS libraries enabled the identification of 77 out of LC-MS has proved to be useful for the metabolic profiling of GM
150 compounds detected by GC-MS, providing valuable informa- crops. Among the main benefits of LC-MS are its wide dynamic
tion associated with the altered metabolic pathways and unex- range and reproducible quantitative analysis. LC-MS is able to sep-
pected changes in the levels of some compounds in the GM arate and to analyze extremely complex samples, and it shows
potatoes. high versatility for metabolite analysis. LC-MS is frequently used
In a different study, the GC-MS analysis of transgenic potato tu- to profile polar/non-volatile, large and thermolabile compounds,
bers with altered sucrose catabolism revealed a massive elevation demonstrating good performance on profiling secondary metabo-
of amino-acid levels [84]. Further studies on GM potato and tomato lites and complex lipids. For example, LC-MS was applied to the
varieties demonstrated the suitability of data-mining tools (e.g., investigation of flavonoids in GM wheat and rice [96], and stilbenes
PCA and HC) in combination with GC-MS to discover differences in GM tomato [97].
that allow the discrimination of the transgenic varieties from the Although reversed-phase is the most frequent mode used in LC-
respective non-modified lines [85,86]. After these pioneer works, MS metabolite profiling in GMO analysis, other suitable modes
a number of GC-MS studies for metabolic profiling were published have been demonstrated to be useful. For example, Nagai et al.
(see Table 3). used a HILIC phase with LC-ESI-MS/MS for the separation of the
More recently, Zhou et al. investigated the unintended effects in major carbon metabolites of transgenic rice [98].
insect-resistant transgenic rice by the combined use of GC-flame- In a recent report, an LC-MS method was developed to explore
ionization detection (FID) and GC-MS [87]. However, in this study, the unintended effects of the genetic transformation in three vari-
GC-MS was exclusively used to identify certain important com- eties of insect-resistant transgenic rice [100]. Metabolite profiles
pounds after GC-FID profiling. were obtained using HPLC-ESI-QTOF-MS, and raw data were pro-
In another study on GM rice, GC-MS data were complemented cessed and aligned using MassHunter Qualitative Analysis soft-
with the results obtained from near-infrared reflectance, ICP atom- ware prior to multivariate statistical analysis. PCA and partial
ic emission spectroscopy, and LC to investigate the compositional least squares-discriminant analysis (PLS-DA) enabled the correct
differences between three transgenic varieties expressing anti- classification of samples. Metabolite signals having a significant
fungal proteins [88]. Multivariate analysis of the results revealed influence on the classification of the different groups were identi-
several changes in some nutrients, such as protein, amino acids, fied by targeted MS/MS analysis. Using this methodology, 15
fatty acids, and vitamins, which were variety specific. metabolites were found to have significantly contributed to the
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 11

separation of the groups harvested at different times and locations. electrophoretic mobilities and m/z values provided by CE-TOF MS
Furthermore, the transgenic rice grains were clearly separated were very helpful for confirming the identity of various isomeric
from the unmodified rice grains after discriminant analysis. In this compounds that could not be unequivocally identified by FT-ICR-
case, another 15 metabolites were found to be responsible for this MS, such as isomeric compounds.
separation. However, the fold changes were comparatively lower
than those obtained from the comparison of different sowing times
5.6. Metabolomic multi-platform studies
and locations, suggesting that the effects of the environmental fac-
tors were greater than those derived from the genetic
The combined use of more than one analytical platform has im-
transformation.
proved the description of the metabolome status of GMOs. For
example, the combined use of LC-ESI-IT-MS and GC-MS for the
5.4. CE-MS
comparative analysis of GM grapevine varieties with the unmodi-
fied counterpart allowed the detection of differences in some phe-
A wide range of analytes, including ionic and polar thermolabile
nolic compounds and volatile secondary metabolites that belong to
compounds, can be profiled by CE-MS, which is considered com-
the classes of monoterpenes, C12-norisoprenoids, and shikimates
plementary to LC-MS and GC-MS. The main characteristics of CE-
[107].
MS are high efficiency, analysis speed and resolution, moreover
GC-MS and LC-MS were also applied to investigate the compo-
requiring little sample pretreatment. However, only moderate sen-
sitional differences between two transgenic ringspot virus-resis-
sitivity is frequently achieved due to the minute sample volumes
tant papaya varieties and the non-modified counterparts [108].
injected in CE-MS [110,111].
In this case, a pre-fractionation strategy was adopted with subse-
The applicability of CE-MS to metabolic profiling of GMOs has
quent parallel analysis for selective and sensitive analysis of com-
already been demonstrated (Table 3). For example, CE-ESI-TOF-
plex extracts obtained from 80 papaya samples. Thus, profiles of
MS in combination with multivariate statistical analysis proved
volatile organic compounds and sugar/polyals were obtained using
to be helpful in the detection of the statistically significant differ-
GC-MS, whereas profiles of organic acids, carotenoids and alkaloids
ences in metabolic profile of varieties of conventional and insect-
were obtained using LC-MS. When the compositional profiles were
resistant GM maize [101]. The analytical procedure indicated some
analyzed by PCA, data-points from transgenic and unmodified pa-
relevant differences in the levels of L-carnitine and stachydrine be-
paya genotypes formed a tight cluster, showing great similarity in
tween conventional and GM maize.
composition, while papaya samples from different harvesting
Similarly, a novel CE-ESI-TOF MS methodology was developed
times were separated.
for the comparative analysis of metabolic profiles of transgenic
In another report, Kusano et al. proposed a novel strategy based
herbicide-tolerant soybean and its corresponding unmodified
on multiplatform analysis to assess the substantial equivalence be-
parental line [102]. In that study, over 45 different metabolites,
tween GMOs and their unmodified counterparts [109]. In their
including carboxylic acids, isoflavones, and amino acids were ten-
study, two GM tomato varieties overexpressing miraculin glyco-
tatively identified. Differences in metabolic profiles of both lines
protein and a panel of traditional tomato cultivars were selected
were more evident in the concentration of three free amino acids,
to prove the proposed methodology. More specifically, data from
namely proline, histidine and asparagine, while a metabolite tenta-
GC-TOF-MS, LC-QTOF-MS and CE-TOF-MS were summarized in
tively identified as 4-hydroxi-L-threonine disappeared in the trans-
single consensus datasets for further multivariate analysis. The
genic soybean compared to its parental non-transgenic line.
combination of the three platforms allowed the statistical analysis
Another CE-ESI-TOF MS method, based on the addition of mod-
of datasets containing over 175 unique tentatively identified
ified cyclodextrins to the background electrolyte [112,113], was
metabolites and more than 1,400 peaks with no or imprecise
applied to the analysis of herbicide-tolerant transgenic soybean
metabolite annotation. Of the annotated metabolites, 56 were de-
[103]. The results revealed some quantitative divergences in the
tected on more than one platform. Using this analytical set-up, the
chiral amino-acid profile between the GM soybean and the
metabolite coverage achieved was 85% of the chemical diversity
untransformed genotype.
found in the LycoCyc database. Results showed that >92% of the
tested peaks in the transgenic lines deviated less from the control
5.5. FT-ICR-MS
line than the accepted limit estimated using the reference panel of
traditional cultivars. Next, the application of orthogonal PLS-DA to
Ultra-high resolution mass spectrometers, such as Fourier
look for changes attributable to genetic modification indicated
transform ion-cyclotron (FT-ICR)-MS, provide accurate-mass mea-
overall proximity between the GM lines and the control greater
surements for the determination of the elemental compositions of
than observed between ripening stages and traditional cultivars
metabolites, which facilitates annotation procedures for unknown
(Fig. 4).
compounds [114]. However, the number of reports on metabol-
omic studies with FT-ICR-MS is small compared with that for other
MS-based platforms due to the equipment cost, the difficulties in 6. Cross-omics platforms and data-integration studies
hardware handling and the extremely large amount of data
generated. As omics techniques have become more accessible, more re-
In spite of these limitations, León et al. demonstrated its poten- ports based on the use of different high-throughput omics plat-
tial for metabolic profiling in GMO analysis [105]. In their study, forms for GMO analysis have been published. The rationale
FT-ICR-MS was also used in combination with CE-TOF-MS for the behind the application of different high-throughput techniques
metabolic profiling of six varieties of maize, three transgenic in- for analyzing GMOs relies on the view that recording as much
sect-resistant lines and their corresponding near-isogenic lines. information as possible will increase the opportunities to identify
The spectral data obtained in both positive and negative ESI modes potential unintended effects in a GMO under specific conditions.
with FT-ICR-MS were uploaded into MassTRIX server in order to In addition, the possibility of analyzing diverse types of biomole-
identify maize-specific metabolites annotated in the KEGG data- cule implies that inter-relationships between the different levels
base [115]. In a subsequent step, multivariate analysis of data sug- of information can be explored. In studies where several omics
gested the most discriminative masses that contribute to technologies were applied, the general trend was to perform statis-
differentiate the GM samples from the non-GM controls. The tical analysis on each independent omics dataset. For instance,
12 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15

Fig. 4. Evaluating the substantial equivalence of miraculin overexpressing tomatoes grown on hydroponic culture solution. (d) Parallel coordinates plot of the predictive
components from the orthogonal PLS-DA model. Each biological sample is drawn as a line that connects its positions on each of the components. Each dimension describes a
unique aspect of the genotype-correlated variance among the metabolite profiles. All genotypes, except 7C and Moneymaker, are separated on at least one axis. Percentages
indicate the ratio of total variance explained by the corresponding dimension. (e) Result from Sammon’s multi-dimensional scaling of distances computed using the six
predictive orthogonal PLS-DA components shown in (d). (Reprinted from [109].)

Scossa et al. investigated the consequence, at the transcriptome In spite of these remarkable studies, in which several advanced
and proteome level, of GM bread-wheat overexpression of a low- profiling techniques were used in parallel to characterize chemical
molecular-weight glutenin sub-unit [116]. In their work, results composition of GMOs, there is a lack of studies reporting on the
from microarray experiments were in good agreement with 2- comparison over the different omics datasets that allow establishing
DGE data, revealing that glutenin overexpression coincided with correlations. This might, in part, be due to the necessity for novel and
a massive down-regulation of other classes of storage proteins. suitable integrative strategies that enable the high level of complex-
Their findings opened new insight into the compensatory re- ity in the different omics areas to be tackled [119]. In this context,
sponses during wheat-seed maturation. appropriate holistic systems biology is depicted as a promising strat-
Global transcriptome and metabolome-profiling analyses were egy to integrate, to summarize and to visualize data from different
also performed for the study of transgenic barley cultivars exhibit- omics domains [120]. Although there is no consensus on defining
ing different disease-resistance and nutritional traits [117]. Profil- systems biology, most of the reports published recently on this topic
ing analysis was performed on barley plants (GM and non-GM agree on that a system-level approach attempts to consider all the
genotypes) grown in field with and without amendment of soil components of a system and that the interactions and the properties
with mycorrhizal fungi. Interestingly, gene-expression-microarray are connected with functions performed by the system via a compu-
data were not as sensitive as metabolome analysis with LC-ESI-MS tational model. As opposed to the traditional reductionist approach,
for revealing minor differences in three of the four varieties in re- systems biology aims to study biological processes using network
sponse to treatment of mycorrhizal fungi. Moreover, consistent modeling with an integrated approach. However, implementation
with other profiling studies, statistical analysis of omics data from of systems biology remains limited in part by the availability of
both platforms indicated differences between the unmodified cul- the suitable data sets.
tivars more significant than the divergences found between GM Omics techniques should be able to detect and to quantify all
and non-GM barley plants. components of the system for the practical application of a sys-
In another report, Barros et al. extended the omics coverage to tem-level approach. Furthermore, data sets should be reproducible
transcriptome, proteome and metabolome analysis of two trans- and acquired at high-throughput, then analysis of the system un-
genic insect-resistant and herbicide-tolerant maize varieties der study can be performed in a consistent, systematic manner
[118]. In the same study, the influences of genotype, year of har- [121]. Also, before data-integration procedures can be practically
vest, agricultural practices and location on the molecular profiles applied to GMO analysis, much effort will have to be made to de-
were also evaluated. Gene-expression microarray and 2-DGE anal- velop effective integrative statistical approaches and appropriate
ysis were applied for transcriptomics and proteomics analyses, computational frameworks for describing molecular systems and
respectively, whereas two metabolomics platforms, namely 1H- connecting omics databases [122,123].
NMR and GC-MS, were used for metabolome profiling of maize-
grain samples. Evaluation of the effect of individual variables for
the factor genotype (i.e. GM and non-GM maize) was systemati- 7. Conclusions and future outlook
cally performed with one-way ANOVA analysis of individual omics
datasets. PCA indicated that growing seasons and locations had Numerous articles in the recent literature illustrate the poten-
stronger overall influences in the three levels of information of tial of omics technologies for molecular profiling analysis of GMOs.
the three maize genotypes than the genetic modification. Minor The latest technological advances in transcriptomics, proteomics
changes registered between non-GM and GM maize varieties in- and metabolomics, combined with suitable chemometrics tools
cluded the down-regulation of maize-allergen Zea m14 in both (including multivariate and univariate statistical analysis), provide
transgenic maize lines, the accumulation of glucose and fructose the opportunity to explore and to decipher compositional differ-
in the insect-resistant maize, and the decreased level of c-tocoph- ences in samples as complex as GMOs. Although molecular profil-
erol and inositol in the herbicide-tolerant line. ing is not currently applied in any stage of the safety assessment of
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 13

GMOs, good prospects can be anticipated for this analytical ap- [4] H.A. Kuiper, G.A. Kleter, The scientific basis for risk assessment and regulation
of genetically modified foods, Trends Food Sci. Technol. 14 (2003) 277–293.
proach in research into GMO foods. However, to reinforce the con-
[5] J.R. Ioset, B. Urbaniak, K. Ndjoko-Ioset, J. Wirth, F. Martin, W. Gruissem, K.
fidence in any molecular profiling techniques, standard validation Hostettmann, C. Sautter, Flavonoid profiling among wild type and related GM
protocols will have to be applied in the future. wheat varieties, Plant Mol. Biol. 65 (2007) 645–654.
On the transcriptomics side, microarray is currently the domi- [6] Organization of Economic Cooperation and Development (OECD) Safety
Evaluation of Foods Derived by Modern Biotechnology: Concept and
nant technique. However, RNA-seq methods based on next- Principles; OECD: Paris, France, 1993.
generation sequencing technology will provide novel means for [7] FAO/WHO. Safety Aspects of Genetically Modified Foods of Plant Origin; Joint
transcriptome profiling in GMOs. RNA-seq technology is still FAO/WHO Consultation: Geneva, Switzerland, 2000.
[8] E.J. Kok, J. Keijer, G.A. Kleter, H.A. Kuiper, Comparative safety assessment of
evolving, but it already presents advantages over microarray, such plant-derived foods, Regul. Toxicol. Pharmcol. 50 (2008) 98–113.
as better sensitivity and broader dynamic range, enabling the study [9] L.V.T. Shepherd, J.W. McNicol, R. Razzo, M.A. Taylor, H.V. Davies, Assessing the
of splice variants, and detecting allele-specific differences (poly- potential for unintended effects in genetically modified potatoes perturbed in
metabolic and developmental processes. Targeted analysis of key nutrients
morphisms). However, although RNA-seq methods are gradually and anti-nutrients, Transgenic Res. 15 (2006) 409–425.
displacing microarrays in other research fields, significant [10] B.M. Chassy, Can -omics inform a food safety assessment?, Regul Toxicol.
improvements are still needed in terms of accuracy in quantifica- Pharmcol. 58 (2010) S62–S70.
[11] X. Ye, S. Al-Babili, A. Kloti, J. Zhang, P. Lucca, P. Beyer, I. Potrykus, Engineering
tion [124]. the provitamin A (beta-carotene) biosynthetic pathway into (carotenoid-free)
Regarding proteomics methods, it might be anticipated that rice endosperm, Science 287 (2000) 303–305.
some limitations of the current technologies, especially the resolu- [12] E. Millstone, E. Brunner, S. Mayer, Beyond ‘substantial equivalence’, Nature
401 (1999) 525–526.
tion of peptides, will be improved to increase protein coverage and
[13] European Food Safety Agency, Guidance document of the scientific panel on
to make identification more reliable. We can also expect that genetically modified organisms for the risk assessment of genetically
advancements in proteomic technology with regard to gel-to-gel modified plants and derived food and feed, EFSA Communications
variability will help proteomic profiling to become standard prac- Department, Parma, Italy, 2006.
[14] Ad Hoc Technical Expert Group (AHTEG) (2010). https://www.cbd.int/doc/
tice in GMO analysis. meetings/bs/bsrarm-02/official/bsrarm-02-05-en.pdf. United Nations
Similarly, we expect that recent advances in the continuously Environment Programme Convention for Biodiversity.
evolving metabolomics field will hit the area of GMO analysis in [15] A. Cifuentes, Food analysis and foodomics, J. Chromatogr. A 1216 (2009) 7109.
[16] M. Herrero, V. García-Cañas, C. Simo, A. Cifuentes, Recent advances in the
the near future. Among the most relevant advances, comprehen- application of capillary electromigration methods for food analysis and
sive multidimensional techniques, such as GC  GC or LC  LC Foodomics, Electrophoresis 31 (2010) 205–228.
coupled to MS, have the potential to improve metabolic profiling [17] V. García-Cañas, C. Simó, M. Herrero, E. Ibáñez, A. Cifuentes, Present and
future challenges in food analysis: foodomics, Anal. Chem. 84 (2012) 10150–
of GMOs by achieving enhanced resolution, selectivity and sensi- 10159.
tivity in comparison with conventional separation techniques. [18] M. Herrero, C. Simó, V. García-Cañas, E. Ibáñez, A. Cifuentes, Foodomics: MS-
Other advances that potentially may improve metabolomics of based strategies in modern food science and nutrition, Mass Spectrom. Rev.
31 (2012) 49–69.
GMOs are the use of novel MS interfaces with minimal [19] N. Doerrer, G. Ladics, S. McClain, C. Herouet-Guicheney, L.K. Poulsen, L.
requirements for sample preparation, improvements to both sam- Privalle, H. Stagg, Evaluating biological variation in non-transgenic crops:
ple-preparation strategies and analytical platforms. Furthermore, executive summary from the ILSI Health and Environmental Sciences
Institute workshop, November 16–17, Paris, France, Regul. Toxicol.
the development and the growth of the available metabolite
Pharmacol. 58 (2010) (2009) S2–S7.
databases providing structural information will improve the rate [20] A. Brazma, P. Hingamp, J. Quackenbush, G. Sherlock, P. Spellman, J. Stoeckert,
of analyte identification. W. Aach, C.A. Ansorge, H.C. Ball, T. Causton, P. Gaasterland, C. Glenisson, F.C.P.
The availability of bioinformatics tools that allow faster annota- Holstege, I.F. Kim, V. Markowitz, J.C. Matese, H. Parkinson, A. Robinson, U.
Sarkans, S. Schulze Kremer, J. Stewart, R. Taylor, J. Vilo, M. Vingron, Minimum
tion of the peaks recorded in the profiles and tools able to summa- information about a microarray experiment (MIAME) – toward standards for
rize multi-platform datasets will also be crucial for efficient microarray data, Nat. Genet. 29 (2001) 365–371.
analysis of metabolomics data [75,125]. Although several public [21] C.F. Taylor, N.W. Paton, K.S. Lilley, P.A. Binz, R.K. Julian, A.R. Jones, W. Zhu, R.
Apweiler, R. Aebersold, E.W. Deutsch, M.J. Dunn, A.J. Heck, A. Leitner, M.
databases focused on integrating metabolite information with data Macht, M. Mann, L. Martens, T.A. Neubert, S.D. Patterson, P. Ping, S.L.
on genes and proteins are available, much effort is still needed to Seymour, P. Souda, A. Tsugita, J. Vandekerckhove, T.M. Vondriska, J.P.
link identified plant metabolites to metabolic pathways and other Whitelegge, M.R. Wilkins, I. Xenarios, J.R. Yates, H. Hermjakob, The
minimum information about a proteomics experiment (MIAPE), Nat.
relevant biological information. Such complementary information, Biotechnol. 25 (2007) 887–893.
frequently named meta-data, provides the opportunity to improve [22] A.E. Ricroch, J.B. Berge, M. Kuntz, Evaluation of genetically engineered crops
the biological interpretation of the results and to realize the full using transcriptomic, proteomic, and metabolomic profiling techniques, Plant
Physiol. 155 (2011) 1752–1761.
potential of omics technologies in GMO research. [23] H. Davies, A role for ‘‘omics’’ technologies in food safety assessment, Food
Control 21 (2010) 1601–1610.
[24] V. García-Cañas, C. Simó, C. León, E. Ibáñez, A. Cifuentes, MS-based analytical
Acknowledgments methodologies to characterize genetically modified crops, Mass Spectrom.
Rev. 30 (2011) 396–416.
This work was supported by AGL2011-29857-C03-01 project [25] J.A. Heinemann, B. Kurenbach, D. Quist, Molecular profiling – a tool for
addressing emerging gaps in the comparative risk assessment of GMOs,
(Ministerio de Economía y Competitividad, Spain), and CSD2007- Environ. Int. 37 (2011) 1285–1293.
00063 FUN-C-FOOD (Programa CONSOLIDER, Ministerio de Educa- [26] J.P. van Dijk, C. Leifert, E. Barros, E.J. Kok, Gene expression profiling for food
ción y Ciencia, Spain). A.V. and C.I. thank the Ministerio de Eco- safety assessment: examples in potato and maize, Regul. Toxicol. Pharmacol.
58 (2010) S21–S25.
nomía y Competitividad for their FPI pre-doctoral fellowships. [27] J.H. Malone, B. Oliver, Microarrays, deep sequencing and the true measure of
the transcriptome, BMC Biol. 9 (2011) 34.
[28] D. Dobnik, B. Morisset, K. Gruden, NAIMA as a solution for future GMO
References diagnostics challenges, Anal. Bioanal. Chem. 396 (2010) 2229–2233.
[29] M.M. Baudo, R. Lyons, S. Powers, G.M. Pastori, K.J. Edwards, M.J. Holdsworth,
[1] C. James, Global Status of Commercialized Biotech/GM Crops: 2012, ISAAA P.R. Shewry, Transgenesis has less impact on the transcriptome of wheat
Brief No. 44 (2012). ISAAA: Ithaca, NY. grain than conventional breeding, Plant Biotechnol. J. 4 (2006) 369–380.
[2] F. Cellini, A. Chesson, I. Colquhoun, A. Constable, H.V. Davies, K.H. Engel, [30] J.G. Dubouzet, A. Ishihara, F. Matsuda, H. Miyagawa, H. Iwata, K. Wakasa,
A.M.R. Gatehouse, S. Kärenlampi, E.J. Kok, J.J. Leguay, S. Lehesranta, H.P. Integrated metabolomic and transcriptomic analyses of high-tryptophan rice
Noteborn, J. Pedersen, M. Smith, Unintended effects and their detection in expressing a mutant anthranilate synthase alpha sub-unit, J. Exp. Bot. 58
genetically modified crops, Food Chem. Toxicol. 42 (2004) 1089–1125. (2007) 3309–3321.
[3] A. Rosati, P.G. Bogani, A. Santarlasci, M. Buiatti, Characterisation of 3’ [31] R. Batista, N. Saibo, T. Lourenc, M.M. Oliveira, Microarray analyses reveal that
transgene insertion site and derived mRNAs in MON810 YieldGard maize, plant mutagenesis may induce more transcriptomic changes than transgene
Plant Mol. Biol. 67 (2008) 271–281. insertion, Proc. Natl. Acad. Sci. USA 105 (2008) 3640–3645.
14 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15

[32] M. Montero, A. Coll, A. Nadal, J. Messeguer, M. Pla, Only half the [56] E. Horvath-Szanics, Z. Szabo, T. Janaky, J. Pauk, G. Hajos, Proteomics as
transcriptomic differences between resistant genetically modified and an emergent tool for identification of stress-induced proteins in control
conventional rice are associated with the transgene, Plant Biotechnol. J. 9 and genetically modified wheat lines, Chromatographia 63 (2006)
(2011) 693–702. S143–S147.
[33] Z. Liu, Y. Li, J. Zhao, X. Chen, G. Jian, Y. Peng, F. Qi, Differentially expressed [57] F.X. Sauvage, M. Pradal, P. Chatelet, C. Tesniere, Proteome changes in leaves
genes distributed over chromosomes and implicated in certain biological from grapevine (Vitis vinifera L.) transformed for alcohol dehydrogenase
processes for site insertion genetically modified rice Kemingdao, Int. J. Biol. activity, J. Agric. Food Chem. 55 (2007) 2597–2603.
Sci. 8 (2012) 953–963. [58] A. Di Luccia, C. Lamacchia, C. Fares, L. Padalino, G. Mamone, B. La Gatta, G.
[34] A. Coll, A. Nadal, M. Palaudelmas, J. Messeguer, E. Mele, P. Puigdomenech, M. Gambacorta, M. Faccia, N. Di Fonzo, E. La Notte, A proteomic approach to
Pla, Lack of repeatable differential expression patterns between MON810 and study protein variation in GM durum wheat in relation to technological
comparable commercial varieties of maize, Plant Mol. Biol. 68 (2008) 105– properties of semolina, Ann. Chim. 95 (2005) 405–414.
117. [59] A.R. Brandao, H.S. Barbosa, M.A.Z. Arruda, Image analysis of two-dimensional
[35] A. Coll, A. Nadal, R. Collado, G. Capellades, M. Kubista, J. Messeguer, M. Pla, gel electrophoresis for comparative proteomics of transgenic and non-
Natural variation explains most transcriptomic changes among maize plants transgenic soybean seeds, J. Proteomics 73 (2010) 1433–1440.
of MON810 and comparable non-GM varieties subjected to two N- [60] H.S. Barbosa, S.C.C. Arruda, R.A. Azevedo, M.A.Z. Arruda, New insights on
fertilization farming practices, Plant Mol. Biol. 73 (2010) 349–362. proteomics of transgenic soybean seeds: evaluation of differential
[36] K.C. Cheng, J. Beaulieu, E. Iquira, F.J. Belzile, M.G. Fortin, M.V. Stromvik, Effect expressions of enzymes and proteinsAnal, Bioanal. Chem. 402 (2012) 299–
of transgenes on global gene expression in soybean is within the natural 314.
range of variation of conventional cultivars, J. Agric. Food Chem. 56 (2008) [61] N. Islam, P.M. Campbell, T.J.V. Higgins, H. Hirano, R.J. Akhurst, Transgenic peas
3057–3067. expressing an alpha-amylase inhibitor gene from beans show altered
[37] P.L. Gregersen, H. Brinch-Pedersen, P.B. Holm, A microarray-based expression and modification of endogenous proteins, Electrophoresis 30
comparative analysis of gene expression profiles during grain development (2009) 1863–1868.
in transgenic and wild type wheat, Transgenic Res. 14 (2005) 887–905. [62] C.Y. Gong, Q. Li, H.T. Yu, Z. Wang, T. Wang, Proteomics insight into the
[38] J. Zhu, W.L. Patzoldt, R.T. Shealy, L.O. Vodkin, S.J. Clough, P.J. Tranel, biological safety of transgenic modification of rice as compared with
Transcriptome response to glyphosate in sensitive and resistant soybean, J. conventional genetic breeding and spontaneous genotypic variation, J.
Agric. Food Chem. 56 (2008) 6355–6363. Proteome Res. 11 (2012) 3019–3029.
[39] E. Baroja-Fernández, F.J. Muñoz, M. Montero, E. Etxeberria, M.T. Sesma, M. [63] M. Careri, L. Elviri, A. Mangia, I. Zagnoni, C. Agrimonti, G. Visioli, N. Marmiroli,
Ovecka, A. Bahaji, I. Ezquer, J. Li, S. Prat, J. Pozueta-Romero, Enhancing sucrose Analysis of protein profiles of genetically modified potato tubers by matrix-
synthase activity in transgenic potato (Solanum tuberosum L.) tubers results in assisted laser desorption/ionization time-of-flight mass spectrometry, Rapid
increased levels of starch, ADPglucose and UDPglucose and total yield, Plant Commun. Mass Spectrom. 17 (2003) 479–483.
Cell Physiol. 50 (2009) 1651–1662. [64] M.C. García-López, V. García-Cañas, M.L. Marina, Reversed-phase high-
[40] P.H. Beatty, A.K. Shrawat, R.T. Carroll, T. Zhu, A.G. Good, Transcriptome performance liquid chromatography-electrospray mass spectrometry
analysis of nitrogen-efficient rice overexpressing alanine aminotransferase, profiling of transgenic and non-transgenic maize for cultivar
Plant Biotechnol. J. 7 (2009) 562–576. characterization, J. Chromatogr. A 1216 (2009) 7222–7228.
[41] A. Coll, A. Nadal, R. Collado, G. Capellades, J. Messeguer, E. Mele, M. [65] A. Koc, A. Cañuelo, J.F. Garcia-Reyes, A. Molina-Diaz, M. Trojanowicz, Low-
Palaudelmas, M. Pla, Gene expression profiles of MON810 and comparable molecular weight protein profiling of genetically modified maize using fast
non-GM maize varieties cultured in the field are more similar than are those liquid chromatography electrospray ionization and time-of-flight mass
of conventional lines, Transgenic Res. 18 (2009) 801–808. spectrometry, J. Sep. Sci. 35 (2012) 1447–1461.
[42] A. D’Alessandro, L. Zolla, We are what we eat: food safety and proteomics, J. [66] G.L. Erny, C. León, M.L. Marina, A. Cifuentes, Time of flight versus ion trap MS
Proteome Res. 11 (2012) 26–36. coupled to CE to analyse intact proteins, J. Sep. Sci. 31 (2008) 1810–1818.
[43] C.Y. Gong, T. Wang, Proteomic evaluation of genetically modified crops: [67] C. Simó, E. Domínguez-Vega, M.L. Marina, M.C. García, G. Dinelli, A. Cifuentes,
current status and challenges, Front. Plant Sci. 4 (2013) 41. CE-TOF MS analysis of complex protein hydrolyzates from genetically
[44] M.C. Ruebelt, N.K. Leimgruber, M. Lipp, T.L. Reynolds, M.A. Nemeth, J.D. modified soybeans – a tool for foodomics, Electrophoresis 31 (2010) 1175–
Astwood, K.H. Engel, K.D. Jany, Application of two-dimensional gel 1183.
electrophoresis to interrogate alterations in the proteome of genetically [68] A. Sussulini, G.H.M.F. Souza, M.N. Eberlin, M.A.Z. Arruda, Comparative
modified crops. 1. Assessing analytical validation, J. Agric. Food Chem. 54 metallomics for transgenic and non-transgenic soybeans, J. Anal. At.
(2006) 2154–2161. Spectrom. 22 (2007) 1501–1506.
[45] A.G. Albo, S. Mila, G. Digilio, M. Motto, S. Aime, D. Corpillo, Proteomic analysis [69] L.R.V. Mataveli, M. Fioramonte, F.C. Gozzobc, M.A.Z. Arruda, Improving
of a genetically modified maize flour carrying CRY1AB gene and comparison metallomics information related to transgenic and non-transgenic soybean
to the corresponding wild-type, Maydica 52 (2007) 443–455. seeds using 2D-HPLC-ICP-MS and ESI-MS/MS, Metallomics 4 (2012) 373–378.
[46] H. Chen, G. Bodulovic, P.J. Hall, A. Moore, T.J.V. Higgins, M.A. Djordjevic, B.G. [70] J. Luo, T. Ning, Y. Sun, J. Zhu, Y. Zhu, Q. Lin, D. Yang, Proteomic analysis of rice
Rolfe, Unintended changes in protein expression revealed by proteomic endosperm cells in response to expression of hGM-CSF, J. Proteome Res. 8
analysis of seeds from transgenic pea expressing a bean alpha-amylase (2009) 829–837.
inhibitor gene, Proteomics 9 (2009) 4406–4415. [71] Y. Okazaki, K. Saito, Recent advances of metabolomics in plant biotechnology,
[47] G.M. Balsamo, G.C. Cangahuala-Inocente, J.B. Bertoldo, H. Terenzi, A.C.M. Arisi, Plant Biotechnol. Rep. 6 (2012) 1–15.
Proteomic analysis of four Brazilian MON810 maize varieties and their four [72] S.G. Villas-Boas, S. Mas, M. Akeson, J. Smedsgaard, J. Nielsen, Mass
non-genetically-modified isogenic varieties, J. Agric. Food Chem. 59 (2011) spectrometry in metabolome analysis, Mass Spectrom. Rev. 24 (2005) 613–
11553–11559. 646.
[48] S. Komatsu, G. Yang, M. Khan, H. Onodera, S. Toki, M. Yamaguchi, [73] V. Shulaev, Metabolomics technology and bioinformatics, Brief Bioinform. 7
Overexpression of calcium-dependent protein kinase 13 and calreticulin (2006) 128–139.
interacting protein 1 confers cold tolerance on rice plants, Mol. Genet. [74] H.J. Issaq, Q.N. Van, T.J. Waybright, G.M. Muschik, T.D. Veenstra, Analytical
Genomics 277 (2007) 713–723. and statistical approaches to metabolomics research, J. Sep. Sci. 32 (2009)
[49] L. Zolla, S. Rinalducci, P. Antonioli, P.G. Righetti, Proteomics as a 2183–2199.
complementary tool for identifying unintended side effects occurring in [75] K. Saito, F. Matsuda, Metabolomics for functional genomics, systems biology,
transgenic maize seeds as a result of genetic modification, J. Proteome Res. 7 and biotechnology, Annu. Rev. Plant Biol. 61 (2010) 463–489.
(2008) 1850–1861. [76] C. Manetti, C. Bianchetti, M. Bizzarri, L. Casciani, C. Castro, G. D’Ascenzo, M.
[50] A. Coll, A. Nadal, M. Rossignol, P. Puigdomenech, M. Pla, Proteomic analysis of Delfini, M.E. Di Cocco, A. Lagana, A. Miccheli, M. Motto, F. Conti, NMR-based
MON810 and comparable non-GM maize varieties grown in agricultural metabonomic study of transgenic maize, Phytochemistry 65 (2004) 3187–
fields, Transgenic Res. 20 (2011) 939–949. 3198.
[51] S.J. Lehesranta, H.V. Davies, L.V. Shepherd, N. Nunan, J.W. McNicol, S. Auriola, [77] F. Piccioni, D. Capitani, L. Zolla, L. Mannina, NMR metabolic profiling of
K.M. Koistinen, S. Suomalainen, H.I. Kokko, S.O. Kärenlampi, Comparison of transgenic maize with the Cry1Ab gene, J. Agric. Food Chem. 57 (2009) 6041–
tuber proteomes of potato varieties, landraces, and genetically modified lines, 6049.
Plant Physiol. 138 (2005) 1690–1699. [78] A. Charlton, T. Allnutt, S. Holmes, J. Chisholm, S. Bean, N. Ellis, P. Mullineaux,
[52] D. Corpillo, G. Gardini, A.M. Vaira, M. Basso, S. Aime, G.P. Accotto, M. Fasano, S. Oehlschlager, NMR profiling of transgenic peas, Plant Biotechnol. J. 2 (2004)
Proteomics as a tool to improve investigation of substantial equivalence in 27–35.
genetically modified organisms: the case of a virus-resistant tomato, [79] J.M. Baker, N.D. Hawkins, J.L. Ward, A. Lovegrove, J.A. Napier, P.R. Shewry,
Proteomics 4 (2004) 193–200. M.H. Beale, A metabolomic study of substantial equivalence of field-grown
[53] M. Di Carli, M.E. Villani, G. Renzone, L. Nardi, A. Pasquo, R. Franconi, A. Scaloni, genetically modified wheat, Plant Biotechnol. J. 4 (2006) 381–392.
E. Benvenuto, A. Desiderio, Leaf proteome analysis of transgenic plants [80] A.P. Sobolev, A.L. Segre, D. Giannino, D. Mariotti, C. Nicolodi, E. Brosio, M.E.
expressing antiviral antibodies, J. Proteome Res. 8 (2009) 838–848. Amato, Strong increase of foliar inulin occurs in transgenic lettuce plants
[54] K. Xue, J. Yang, B. Liu, D. Xue, The integrated risk assessment of transgenic rice (Lactuca sativa L.) overexpressing the asparagine synthetase A gene from
Oryza sativa: a comparative proteomics approach, Food Chem. 135 (2012) Escherichia coli, J. Agric. Food Chem. 55 (2007) 10827–10831.
314–318. [81] A.P. Sobolev, G. Testone, F. Santoro, C. Nicolodi, M.A. Iannelli, M.E. Amato, A.
[55] H. Guo, H. Zhang, Y. Li, J. Ren, X. Wang, H. Niu, J. Yin, Identification of changes Ianniello, E. Brosio, D. Giannino, L. Mannina, Quality traits of conventional
in wheat (Triticum aestivum L.) seeds proteome in response to anti-trx s gene, and transgenic lettuce (Lactuca sativa L.) at harvesting by NMR metabolic
PLoS ONE 6 (2011) e22255. profiling, J. Agric. Food Chem. 58 (2010) 6928–6936.
 A. Valdés et al. / Trends in Analytical Chemistry 52 (2013) 2–15 15

[82] M. Defernez, Y.M. Gunning, A.J. Parr, L.V.T. Shepherd, H.V. Davies, I.J. [104] H. Takahashi, M. Hayashi, F. Goto, S. Sato, T. Soga, T. Nishioka, M. Tomita, M.
Colquhoun, NMR and HPLC-UV profiling of potatoes with genetic Kawai-Yamada, H. Uchimiya, Evaluation of metabolic alteration in transgenic
modifications to metabolic pathways, J. Agric. Food Chem. 52 (2004) 6075– rice overexpressing dihydroflavonol-4-reductase, Ann. Bot. 98 (2006) 819–
6085. 825.
[83] U. Roessner, C. Wagner, J. Kopka, R.N. Trethewney, L. Willmitzer, Technical [105] C. León, I. Rodriguez-Meizoso, M. Lucio, V. García-Cañas, E. Ibañez, P. Schmitt-
advance: simultaneous analysis of metabolites in potato tuber by gas Kopplin, A. Cifuentes, Metabolomics of transgenic maize combining Fourier
chromatography-mass spectrometry, Plant J. 23 (2000) 131–142. transform-ion cyclotron resonance-mass spectrometry, capillary
[84] U. Roessner, L. Willmitzer, A.R. Fernie, High-resolution metabolic electrophoresis-mass spectrometry and pressurized liquid extraction, J.
phenotyping of genetically and environmentally diverse potato tuber Chromatogr. A 1216 (2009) 7314–7323.
systems. Identification of phenocopies, Plant Physiol. 127 (2001) 749–764. [106] H. Takahashi, Y. Hotta, M. Hayashi, M. Kawai-Yamada, S. Komatsu, H.
[85] U. Roessner, A. Luedemann, D. Brust, O. Fiehn, T. Linke, L. Willmitzer, A.R. Uchimiya, High throughput metabolome and proteome analysis of transgenic
Fernie, Metabolic profiling allows comprehensive phenotyping of genetically rice plants (Oryza sativa L.), Plant Biotechnol. 22 (2005) 47–50.
or environmentally modified plant systems, Plant Cell 13 (2001) 11–29. [107] C. Tesniere, L. Torregrosa, M. Pradal, J.M. Souquet, C. Gilles, K. Dos Santos, P.
[86] U. Roessner-Tunali, B. Hegemann, A. Lytovchenko, F. Carrari, C. Bruedigam, D. Chatelet, Z. Gunata, Effects of genetic manipulation of alcohol dehydrogenase
Granot, A.R. Fernie, Metabolic profiling of transgenic tomato plants levels on the response to stress and the synthesis of secondary metabolites in
overexpressing hexokinase reveals that the influence of hexose grapevine leaves, J. Exp. Bot. 57 (2006) 91–99.
phosphorylation diminishes during fruit development, Plant Physiol. 133 [108] Z. Jiao, J.C. Deng, G.K. Li, Z.M. Zhang, Z.W. Cai, Study on the compositional
(2003) 84–99. differences between transgenic and non-transgenic papaya (Carica papaya L.),
[87] J. Zhou, C. Ma, H. Xu, K. Yuan, X. Lu, Z. Zhu, Y. Wu, G. Xu, Metabolic profiling of J. Food Compos. Anal. 23 (2010) 640–647.
transgenic rice with cryIAc and sck genes: an evaluation of unintended effects [109] M. Kusano, H. Redestig, T. Hirai, A. Oikawa, F. Matsuda, A. Fukushima, M.
at metabolic level by using GC-FID and GC-MS, J. Chromatogr. B 877 (2009) Arita, S. Watanabe, M. Yano, K. Hiwasa-Tanase, H. Ezura, K. Saito,
725–732. Covering chemical diversity of genetically-modified tomatoes using
[88] Z. Jiao, X. Si, G. Li, Z. Zhang, X. Xu, Unintended compositional changes in metabolomics for objective substantial equivalence assessment, PLoS ONE
transgenic rice seeds (Oryza sativa L.) studied by spectral and 16 (2011) 16989.
chromatographic analysis coupled with chemometrics methods, J. Agric. [110] A. Giuffrida, A. Contino, G. Maccarrone, M. Messina, V. Cucinotta, Mass
Food Chem. 58 (2010) 1746–1754. spectrometry detection as an innovative and advantageous tool in ligand
[89] S.M.M. Malowicki, R. Martin, M.C. Qian, Comparison of sugar, acids, and exchange capillary electrophoresis, Electrophoresis 32 (2011) 1176–1181.
volatile composition in raspberry bushy dwarf virus-resistant transgenic [111] V. Cucinotta, A. Giuffrida, G. Grasso, G. Maccarrone, A. Mazzaglia, M. Messina,
raspberries and the wild type ‘meeker’ (Rubus idaeus L.), J. Agric. Food Chem. G. Vecchio, Diaminotrehalose-capped b-cyclodextrin, a new member of
56 (2008) 6648–6655. hemispherodextrins: Synthesis, thermodynamic and spectroscopic
[90] R. Zawirska-Wojtasiak, M. Goslinski, J. Gajc-Wolska, S. Mildner-Szkudlarz, characterization and its exploitation in chiral electrokinetic
Aroma evaluation of transgenic, thaumatin II-producing cucumber fruits, J. chromatography, J. Sep Sci. 34 (2011) 70–76.
Food Sci. 74 (2009) C204–C210. [112] A. Giuffrida, R. Caruso, M. Messina, G. Maccarrone, A. Contino, A. Cifuentes, V.
[91] J.L. Bernal, M.J. Nozal, L. Toribio, C. Diego, R. Mayo, R. Maestre, Use of Cucinotta, Chiral separation of amino acids derivatised with fluorescein
supercritical fluid extraction and gas chromatography-mass spectrometry to isothiocyanate by single isomer derivatives 3-monodeoxy-3-monoamino-b-
obtain amino acid profiles from several genetically modified varieties of and c-cyclodextrins: the effect of the cavity size, J. Chromatogr. A 1269
maize and soybean, J. Chromatogr. A 1192 (2008) 266–272. (2012) 360–365.
[92] J.J. Jimenez, J.L. Bernal, M.J. Nozal, L. Toribio, J. Bernal, Profile and relative [113] A. Giuffrida, L. Tabera, R. González, V. Cucinotta, A. Cifuentes, Chiral analysis
concentrations of fatty acids in corn and soybean seeds from transgenic and of amino acids from conventional and transgenic yeasts, J. Chromatogr. B 875
isogenic crops, J. Chromatogr. A 1216 (2009) 7288–7295. (2008) 243–247.
[93] T. Frank, R.M. Röhlig, H.V. Davies, E. Barros, K. Engel, Metabolite profiling of [114] Y. Xu, J.F. Heilier, G. Madalinski, E. Genin, E. Ezan, J.C. Tabet, C. Junot,
maize kernels–genetic modification versus environmental influence, J. Agric. Evaluation of accurate mass and relative isotopic abundance measurements
Food Chem. 60 (2012) 3005–3012. in the LTQ-orbitrap mass spectrometer for further metabolomics database
[94] Y. Inaba, J.E. Brotherton, A. Ulanov, J.M. Widholm, Expression of a feedback building, Anal. Chem. 82 (2010) 5490–5501.
insensitive anthranilate synthase gene from tobacco increases free [115] K. Suhre, P. Schmitt-Kopplin, MassTRIX: mass translator into pathways,
tryptophan in soybean plants, Plant Cell Rep. 26 (2007) 1763–1771. Nucleic Acids Res. 36 (2008) W481–W484.
[95] G.S. Catchpole, M. Beckmann, D.P. Enot, M. Mondhe, B. Zywicki, J. Taylor, N. [116] F. Scossa, D. Laudencia-Chingcuanco, O.D. Anderson, W.H. Vensel, D.
Hardy, A. Smith, R.D. King, D.B. Kell, O. Fiehn, J. Draper, Hierarchical Lafiandra, R. D’Ovidio, S. Masci, Comparative proteomic and transcriptional
metabolomics demonstrates substantial compositional similarity between profiling of a bread wheat cultivar and its derived transgenic line
genetically modified and conventional potato crops, Proc. Natl. Acad. Sci. USA overexpressing a low molecular weight glutenin sub-unit gene in the
102 (2005) 11462–14458. endosperm, Proteomics 8 (2008) 2948–2966.
[96] Y.M. Shin, H.J. Park, S.D. Yim, N.I. Baek, C.H. Lee, G. An, Y.M. Woo, Transgenic [117] K.H. Kogel, L.M. Voll, P. Schafer, C. Jansen, Y. Wu, G. Langen, J. Imani, J.
rice lines expressing maize C1 and R-S regulatory genes produce various Hofmann, A. Schmiedl, S. Sonnewald, D. von Wettstein, R.J. Cook, U.
flavonoids in the endosperm, Plant Biology 4 (2006) 303–315. Sonnewald, Transcriptome and metabolome profiling of field-grown
[97] I. Nicoletti, A. De Rossi, G. Giovinazzo, D. Corradini, Identification and transgenic barley lack induced differences but show cultivar-specific
quantification of stilbenes in fruits of transgenic tomato plants (Lycopersicon variances, Proc. Natl. Acad. Sci. USA 107 (2010) 6198–6203.
esculentum Mill.) by reversed phase HPLC with photodiode array and mass [118] E. Barros, S. Lezar, M.J. Anttonen, J.P. van Dijk, R.M. Rohlig, E.J. Kok, K. Engel,
spectrometry detection, J. Agric. Food Chem. 55 (2007) 3304–3311. Comparison of two GM maize varieties with a near-isogenic non-GM variety
[98] Y.S. Nagai, C. Sakulsinghroj, G.E. Edwards, H. Satoh, T.W. Greene, B. Blakeslee, using transcriptomics, proteomics and metabolomics, Plant Biotechnol. J. 8
T.W. Okita, Control of starch synthesis in cereals: metabolite analysis of (2010) 436–451.
transgenic rice expressing an up-regulated cytoplasmic ADP-glucose [119] L. Martín, A. Anguita, V. Maojo, J. Crespo, Integration of Omics data for cancer
pyrophosphorylase in developing seeds, Plant Cell Physiol. 50 (2009) 635– research, in: W.C.S. Cho (Ed.), An Omics Perspective on Cancer Research,
643. Springer, New York, 2010, pp. 249–266.
[99] F. Matsuda, A. Ishihara, K. Takanashi, K. Morino, H. Miyazawa, K. Wakasa, H. [120] N. Gehlenborg, S.I. O’Donoghue, N.S. Baliga, A. Goesmann, M.A. Hibbs, H.
Miyagawa, Metabolic profiling analysis of genetically modified rice seedlings Kitano, O. Kohlbacher, H. Neuweger, R. Schneider, D. Tenenbaum, A. Gavin,
that overproduce tryptophan reveals the occurrence of its inter-tissue Visualization of omics data for systems biology, Nat. Methods 7 (2010) S56–
translocation, Plant Biotechnol. 27 (2010) 17–27. S68.
[100] Y. Chang, C. Zhao, Z. Zhu, Z. Wu, J. Zhou, Y. Zhao, X. Lu, G. Xu, Metabolic [121] E. Sabidó, N. Selevsek, R. Aebersold, Mass spectrometry-based proteomics for
profiling based on LC/MS to evaluate unintended effects of transgenic rice systems biology, Curr. Opin. Biotechnol. 23 (2012) 1–7.
with cry1Ac and sck genes, Plant Mol. Biol. 78 (2012) 477–487. [122] S.P. Akula, R.N. Miriyala, H. Thota, A.A. Rao, S. Gedela, Techniques for
[101] T. Levandi, C. Leon, M. Kaljurand, V. García-Cañas, A. Cifuentes, Capillary integrating -omics data, Bioinformation 3 (2009) 284–286.
electrophoresis time-of-flight mass spectrometry for comparative [123] A. Conesa, J.M. Prats-Montalbán, S. Tarazona, M.J. Nueda, A. Ferrer, A
metabolomics of transgenic versus conventional maize, Anal. Chem. 80 multiway approach to data integration in systems biology based on
(2008) 6329–6335. Tucker3 and N-PLS, Chemometr. Intell. Lab. 104 (2010) 101–111.
[102] R. Garcia-Villalba, C. León, G. Dinelli, A. Segura-Carretero, A. Fernandez- [124] P.A. McGettigan, Transcriptomics in the RNA-seq era, Curr. Opin. Chem. Biol.
Gutierrez, V. García-Cañas, A. Cifuentes, Comparative metabolomic study of 17 (2013) 4–11.
transgenic versus conventional soybean using capillary electrophoresis-time- [125] H. Redestig, M. Kusano, A. Fukushima, F. Matsuda, K. Saito, M. Arita,
of-flight mass spectrometry, J. Chromatogr. A 1195 (2008) 164–173. Consolidating metabolite identifiers to enable contextual and multi-
[103] A. Giuffrida, C. León, V. García-Cañas, V. Cucinotta, A. Cifuentes, Modified platform metabolomics data analysis, BMC Bioinformatics 11 (2010) 214.
cyclodextrins for fast and sensitive chiral-capillary electrophoresis-mass
spectrometry, Electrophoresis 30 (2009) 1734–1742.