On 2017 Jul 24, Richard E Goodman commented:

As one of the senior authors of the Siruguri et al., 2015 publication, with 20 years experience in evaluating the safety of Genetically Engineered (GE or GM) crops, I feel obligated to respond to the statements Dr. Sunil Verma is posting on PubMed COMMONS and now also in Science as an e-letter to the 2016 publication by Priyanka Pulls describing the development of this GE mustard. Dr. Verma's second comment posting here lists his letter in Science. Importantly, neither my comments, nor those of Verma are peer reviewed. We are giving our opinions (which differ markedly as does our experiences). I have written a response to Verma's e-letter in Science and it is available at http://science.sciencemag.org/content/352/6289/1043/tab-e-letters . It addresses the issues of the accepted hazard and risk evaluation of GE crops in India and internationally. Our 2015 publication here describes the assessment looking at the source of the genes, the sequences of proteins and the scientific rational is to evaluation potential risks for those who might be allergic to the protein (Barnase, Barstar or Bar), or to proteins that are highly identical, and could share IgE binding. Dr. Verma did not provide any data that demonstrates we are wrong, or that there are risks from this mustard. Instead in his supplemental information, he compared the sequence of Ani s 9, a minor allergen of a fish parasitic worm, to other sequences in the AllergenOnline.org database. And he implies that cross-reactivity might occur due to associated proteins like the SXP/RAL-2 proteins (Ani s 5 and Ani s 9). However, as noted by Garcia-Mayoral et al., 2014), similar proteins do not exist outside of worms (Nematodes). Ani s 9 has very little sequence similarity to Barnase, as described in our paper. Comparing Ani s 9 in AllergenOnline demonstrates that it is rather unique and unlikely to have cross-reactivity outside of the parasitic worm allergens. This mustard contains Barnase, not Ani s 9. Furthermore, he points to the six amino acid match of Barnase to Ani s 9. But as described in our paper and in my letter in Science, that six amino acid segment matches hundreds of proteins in the NCBI database, without any evidence of cross-reactivity or allergy. Furthermore, there is no evidence that a six amino acid match predicts cross-reactivity. The standard in CODEX is sequences matching >35% identity over 80 amino acids, and such matches are quite conservative (overpredict) both primary and confirmational epitopes (Goodman, 2006, Goodman et al., 2008). CODEX indicates you may do a short sequence match, but must justify the methods. If there are matches of >35% identity over 80, then serum IgE tests would be warranted using sera from at-risk (specifically allergic subjects (Goodman, 2008). In the future there will be improvements in the assessment (Goodman and Tetteh, 2011), however, Dr. Verma has not described any new method or any proof that he has an improvement. Instead, he has proposed hypothetical issues, a letter in Science and he has not posted the letter on facebook. If he thinks there can be improvements, he should do experiments and submit his results to a peer reviewed journal for scientific evaluation. The authors of the Siruguri et al 2015 paper stand by our results that this GM mustard is as safe as the non-GM mustards in use in India today.

References: Garcia-Mayoral MF, Trevino MA et al., (2014). Relationships between IgE/IgG4 epitopes, structure and function in Anisakis simplex Ani s 5, a member of the SXP/RAL-2 protein family. PLOS, Negl Tropical Dis 8(3):e2735. Goodman RE. (2006) Practical and predictive bioinforamtics methods for the identiifcation of potentially cross-reactive protein matches. Mol Nutr Food Res 50:655-660. Goodman RE, Vieths S et al, (2008). Allergenicity assessment of genetically modified crops--what makes sense? Goodman RE. (2008) Performing IgE serum testing due to bioinformatics matches in the allergenicity assessment of GM crops. Food Chem Toxicol 46(Suppl 10):S24-S34. Goodman RE, Tetteh AO. (2011). Suggested improvements for the allergenicity assessment of genetically modified plants used in foods. Curr Allergy Asthma Rep 11(4):317-324.

On 2017 Jul 21, Sunil Verma commented:

Reply to - Comment of Prof. Richard E. Goodman, and Vasanthi Siruguri 2017 Jul 19 2:50 p.m.

Sunil Kumar Verma, Principal Scientist CSIR-Centre for Cellular and Molecular Biology, Hyderabad 500 007, India.

Dear Authors,

I read your elaborated reply with great enthusiasm and with hope to find specific answers to my very specific queries on this specific paper (PMID: 26079618). Instead of replying to my scientific observations on the methodology used in this paper, your reply focus more on defending the release of GM mustard in India, which is not really the central theme of this specific paper in question. The scope of this specific paper was limited to assess the allergenic potential of the transgene Bar, Barnase, and Barstar expressed in Genetically Modified Indian Mustard for heterosis breeding, wherein, the technical methodology which was used in this paper was questionable as specifically explained in my previous comment (supported by 15 pages of supplementary information, please see attachment in previous comment) in detail.

Of course, GM Mustard release could be a consequence of this publication, which could be discussed in detail at appropriate forum when safety data and complete dossier which you have cited in your reply above is made available for public review by the concerned authorities (see ref. 1). Until that complete dossier is available for review, your justifications on GM mustard release would be one sided. And for that reason, I would not like to even touch upon the matter beyond the scope of this specific paper and my technical comments which are still unaddressed.

With reference to your first argument that I have misinterpreted recommendations of Indian Council of Medical Research (ICMR, 2008) guidelines on the safety of Genetically Engineered (GE) crops, I would like to highlight that even the ICMR guideline also emphasize on overall structural similarities to predict the allergenic potentials (see page 13-14 in ref. 2), however, in the light of my previous comment and provided data therein, we now know that primary amino acid sequence comparisons are not the best known indicators of structural similarities among proteins, therefore my technical objection remains valid.

My second technical comment was on Barnase-barstar complex and its possible differential immune response compared to individual proteins in free form; this aspect you have left unaddressed in your reply above.

In response to my third technical objection, you have written that AllergenOnline.org database is a peer reviewed database. I understand that it is a peer-reviewed database; however, it's peer-reviewed status does not change the newly revealed fact that this database fails to detect the known allergen 'Ani s 9' (GenBank: ABV55106.1) as allergen using the strategy as was followed to examine the GM Mustard transgenes; thus, the conclusions drawn in this analysis remain erroneous.

My commentary on your paper and flawed conclusion therein has been published in Science (3). Authors are welcome to take this debate further so that a consensus could be reached beyond a reasonable doubt on the conclusion of this specific paper in question and a suitable action could be taken to correct the conclusions of this paper by the way of erratum if appropriate, as an outcome of this post publication review.

References:

(1) Priyanka Pulla, India nears putting GM mustard on the table. Science 27 May 2016: Vol. 352, Issue 6289, pp. 1043. DOI: 10.1126/science.352.6289.1043

(2) Guidelines for the Safety Assessment of Foods Derived from Genetically Engineered Plants. Indian Council of Medical Research. New Delhi. 2008. https://goo.gl/HuAnGC Link accessed on 22/07/2017

(3) Sunil Kumar Verma, RE: Letter on In Depth "New India nears putting GM mustard on the table" Science 10 July 2017. http://science.sciencemag.org/content/352/6289/1043/tab-e-letters Link Accessed on 22/07/2017

On 2017 Jul 19, Richard E Goodman commented:

Sunil Verma, CSIR Hyderabad, India, provided hypothetical comments about our study (PMID 26079618). He suggests our evaluation methods and conclusions were wrong and suggests our conclusions should be changed. We reviewed our publication and Verma’s comments. We conclude that he is misinterpreting the rationale for our study, the recommendations of the CODEX Alimentarius guidelines (2003/2009) and Indian Council of Medical Research (ICMR, 2008) on the safety of Genetically Engineered (GE) crops and our study. The dossier including study was reviewed and approved by the RCGM and GEAC committees of the government of India. Dr. Verma did not present data from studies of the GE plant or the proteins.

Our response: 1) Indian mustard (Brassica juncea) is an important oil seed crop consumed as oil, seed and greens in India. However, production efficiency and plant diseases limits yield. Heterosis breeding often leads to production of agronomically superior hybrid plants.

2) Mustard is usually self-pollinating. Therefore, scientists at the University of Delhi South Campus developed a GE Indian mustard system producing a male sterile plant and a fertility restorer line that allows hybrid production. The male sterile line was created by insertion of a gene that encodes an RNase (Barnase) that is expressed only in the tapetum (outer cell layer of the anther or pollen sac in early flower buds). The DNA construct includes a herbicide tolerant gene (bar). Expression of Barnase is tightly regulated by the construct of DNA as explained in Jagannath et al., 2002, where a long spacer DNA was inserted between the bar gene and promoter and the Barnase gene to ensure it is only expressed in the anther as demonstrated from field trial plants. Barnase is not detectable in seeds or green tissue of the plant (Jagannath et al., 2001). The fertility restoring line was developed by inserting a gene encoding Barstar also using an anther (tapetum) restricted promoter. Barstar binds to Barnase in a very tight protein complex that inhibits the RNase activity of Barnase (Bisht et al., 2004). The restorer line also includes the bar herbicide tolerance gene. Data on the two GE plants and their hybrid offspring were well characterized and data, including a full safety evaluation, was reviewed by the RCGM and GEAC committees of the Indian government with recommendations the plants are as safe as non-GE mustard.

3) The safety evaluation followed the Indian Council of Medical Research guidelines (2008), which are consistent with the international guideline (CODEX Alimentarius, 2003 and 2009). One primary concern is whether the protein expressed by the novel genes encode allergens or proteins of moderate sequence identity (e.g. >35% identity over 80 amino acids) to an allergen so that allergenic cross-reactivity might be expected. The guidelines are referenced in the paper and are publically available. The methods and databases are cited. There is no evidence that proteins of lower sequence identity (including the FAO/WHO 2001 suggestion of 6 consecutive amino acid matches) are sufficient to demonstrate potential risks of allergy.

4) The assessment for potential risks of allergenicity of GE crops have been extensively addressed (e.g. Goodman et al., 2008). The risks of food allergy recognized by food regulators around the world are that foods must be labeled with the source of ingredients so that people who have allergies to allergenic sources (e.g. peanuts, specific tree nuts, milk, wheat, eggs), can avoid ingesting foods that cause their allergies. Governments have adopted specific food labeling laws for all foods and hold producers accountable. The standard is the same for GE crops. When Pioneer Hibred was developing a nutritionally enhanced soybean by adding a gene from Brazil nut, Dr. Taylor at the University of Nebraska conduct tests since the gene was from an allergenic tree nut. The results demonstrated that the protein was an allergen that would put Brazil nut allergic subjects at risk (Nordlee et al., 1996) That GE plant never entered food production. There are no published scientific reports of a GE protein in an approved crop causing allergic reactions in humans.

5) The methods suggested by Verma have not been demonstrates to predict allergenicity. A number of publications demonstrate that six-amino acid matches are common and not predictive (e.g., Hileman et al., 2002). Proteins with less than 50% overall identity are unlikely to be cross-reactive. Similar three-D shapes do not predict allergenicity. The AllergenOnline.org database is a peer reviewed database that includes Ani s 9, and there is no significant math. The database is updated every year and is accepted by international governments (Goodman et al., 2016).

6) A similar GE mustard (canola) was produced PGS in Europe, now owned by Bayer CropScience. The canola include the same proteins, expressed in similar amounts and patterns. Seven events have been approved for production and consumption in Canada, the US, and Australia. Food and feed products from those events are approved for import and consumption in Japan, China, South Africa, Korea and the European Union. The food and feed safety evaluation summaries and details are available on the CERA-GMC.org/GmCropDatabase Result website. First approvals were in 1995 and there is no evidence of harm even though 95% of canola in North America is GE.

7) The standard for food safety of GE crops in all countries that approve GE crops is that the GE plant does not pose any greater risk of allergy or toxicity than the similar non-GE crop. This GE Indian mustard developed at South Delhi campus meets those standards.

We stand by our conclusions that the male-sterile and restorer Indian mustard events and progeny are as safe as the non-GM counter-parts. There is no data that suggests otherwise. Statements by Dr. Verma are hypothetical, not based on facts. This mustard meets the same standards used to evaluate GE crops approved in the US, Canada, Australia-New Zealand, Japan and other countries.

Richard E. Goodman, and Vasanthi Siruguri

References: 1.Siruguri V, Bharatraj DK, Vankudavath RN, Rao Mendu VV, Gupta V, Goodman RE. 2015. Evaluation of Bar, Barnase and Barstar recombinant proteins expressed in genetically engineered Brassica juncea (Indian mustard). For potential risks of food allergy using bioinformatics and literature searches. Food Chem Toxicol 83:93-102. 2.Jagannath A, Bandyopadhyay P, et al. 2001. The use of a spacer DNA fragment insulates the tissue-specific expression of a cytotoxic gene (barnase) and allows high-frequency generation of transgenic male sterile lines in Brassica juncea L. Mol Breeding. 8:11-23. 3.Jagannath A, Arumugam N, et al., 2002. Development of transgenic barstar lines and identification of a male sterile (barnase)/restorer (barstar) combination for heterosis breeding in Indian oilseed mustard (Brassica juncea).Curr Sci 82:46-52. 4.CODEX Alimentarius Guidelines 2003, 2009. 5.ICMR Guidelines. 2008. 6.Bisht NC, Jagannath A et al., 2004. A two gene-two promoter system for enhanced expression of a restorer gene (barstar) and development of improved fertility restorer lines for hybrid seed production in crop plants. Mol Breeding 14:129-144. 7.Hileman RE, Silvanovich A et al., 2002. Bioinformatic methods for allergenicity assessment using a comprehensive allergen database. Int Arch Allergy Immunol 128:280-291. 8.Goodman RE, Ebisawa M, et al. 2016. AllergenOnline: a peer-reviewed, curated allergen database to assess novel food proteins for potential cross-reactivity. Mol Nutr Food Res 60:1183-1198. 9.Nordlee JA, Taylor SL, Townsend JA, Thomas LA, Bush RK. 1996. Idnetification of a Brazil-nut allergen in transgenic soybeans. N Engl J Med. 334:688-692. 10.Goodman RE, Vieths S, et al. 2008. Allergenicity assessment of genetically modified crops—what makes sense? Nat Biotechnol 26:73-81.

On 2017 Jul 09, Sunil Verma commented:

Comment on - Evaluation of Bar, Barnase, and Barstar recombinant proteins expressed in genetically engineered Brassica juncea (Indian mustard) for potential risks of food allergy using bioinformatics and literature searches

Sunil Kumar Verma, Principal Scientist CSIR-Centre for Cellular and Molecular Biology, Hyderabad 500 007, India.

In this study, the authors have tested the allergenic potential of the transgene Bar, Barnase, and Barstar expressed in Genetically Modified Indian Mustard for heterosis breeding. To this end, the authors have done the primary amino acid sequence comparisons of these proteins with the primary amino acid sequences of the known allergens listed in Allergenonline.org and NCBI Entrez protein database until January 2015 and 9 March 2015, respectively. Based on these bioinformatics comparisons authors concluded that the Bar, Barnase and Barstar proteins are unlikely to present any significant risk of food allergy to consumers. The authors also recommended not to perform any human serum IgE testing to further evaluate possible binding to the Bar, Barnase or Barstar proteins.

I hereby propose that the above conclusions drawn by the authors in this study are incorrect and require a major revision.

The main criteria used by the authors in these bioinformatics comparisons was the primary amino acid sequence homology searches of the proteins in question with that of the primary amino acid sequences of the potential allergen listed in above databases. All the hits with less than 50% primary amino acid sequence identities for full length proteins and less than 35% identity in the sliding window 80 amino acid segments of each proteins were ignored; the argument was that these matches could not have led to significant structural similarities among the proteins in question, therefore can be ignored.

Several independent studies have shown that in many cases, even though the primary amino acid sequence similarity between two proteins / domains are very less (<20%), but the tertiary structures of the proteins may be highly similar. One classical example of this is high structural similarity between N terminal half of the Krit-B41 domain with that of the RA domain of RalGDS (1RAX:A) with an r.m.s. deviation of 2.9A for 80 aligned positions; despite a very low homology in their primary amino acid sequences (sequence identity =8.7%). [1, S1] It is notable that both RalGDS and Krit-1 interact with Rap1A through the RA and B41 domains, respectively [2, 3], and so the talin [4]. Thus, the high primary amino acid sequence similarity between two proteins may though infer greater chances of structural homology between these proteins; however, low primary amino acid sequence similarity does not necessarily infer that proteins in question will necessarily have higher structural dissimilarities.

Since it is the conformationally determined structure of the proteins/epitopes which finally decide immunogenicity and allergenicity - and not just the primary amino acid sequences; the conclusion drawn in this study based on merely the primary amino acid sequence comparisons are scientifically inappropriate.

Secondly, in real scenario, both the Barnase and Barstar proteins are expressed simultaneously and these two proteins remain in a complex and not as individual proteins in plant [5, 6]. It is not unlikely that structure of a specific protein in complex may be different than that of the structure of the same individual protein in free form. Also, there may be the possibilities of formation/exposure of new epitope(s) surfaces, particularly as we know now that there are several antibodies known that recognize just the native proteins and some may indeed require complex assembly.

Thus, these conformationally determined epitopes that are recognized in the complex but not the free protein of interest may be reveled in differential screening between a protein and a complex form of the same protein. The conformationally determined epitopes could then be compared for structural homology with the epitopes in known allergens to determine the allergenic potential of two proteins in complex; such studies however, were not conducted in this paper; and the fact that Barnase and Barstar remain in complex and not in free form, was completely ignored throughout the study.

Finally, I found that the overall implication of the Allergenonline.org database itself on correctly predicting the allergenic potential of a new antigen was also questionable.<br> To test this, I assumed that 'Ani s 9' (which is a very well known allergen from SXP/RAL-2 protein family) [7] is a new putative allergen and that this group of proteins are not yet listed in the database; and asked whether or not one can predict if 'Ani s 9' is a potential food allergen using the strategy as was used in this study for Barnase, Barstar and Bar transgenic proteins. The full length primary amino acid sequence comparison of 'Ani s 9' (GenBank: ABV55106.1) using default parameter i.e 'E' value cut off = 1 identified 7 hits (excluding the hits with its own sequences) with 'tropomyosin' allergen from various organisms and 'AAEL002761-PC ' allergen from Aedes aegypti, respectively; however, none of the hits was with significant similarity cut off (>50%). Thus, this bioinformatics search criteria wrongly predicted that the 'Ani s 9' is not a potential food allergen. [S2]

The another criteria i.e. greater than 35% identity in the sliding window of 80 amino acid segment also did not produce any hit at all (other than self hits, which were excluded as explained above), indicating that this criteria also failed to identify 'Ani s 9' as potential food allergen. [S3] The third criteria i.e. 8 continuous amino acid segment search also did not identify any hit with any of the allergen in the database.[S4]

Thus, the bioinformatics search as used in this study following any of the criteria defined could not identify 'Ani s 9' as a potential food allergen. This confirms that the criteria used in this study by authors could easily give false negative results.

The only strategy that could have identified 'Ani s 9' as possible food allergen was a '6 continuous amino acid segment search, which could have identified its match with Allergen 'Lol p 5' for the 6-aa segment 'ANAPPA'. [S5]

This criteria however, was not used in current study to predict the allergenic potential of Bar, Barnase and Barstar. If this specific criteria was used, Barnase transgenic protein also could have given a potential hit with Allergen Ber e 2 and Ani s 9 for the 6 continuous amino acid patch LFSTAA, and WVASKG, respectively [S6]; hence, the conclusion of this paper could have been different.

In view of the above, I conclude that the criteria implemented in this study were not sufficient to exclude the possibility of the transgenic protein Bar, Barnase and Barstar being a possible allergen; therefore the conclusion drawn by authors that "the above transgenic proteins are unlikely to present any significant risk of food allergy to consumers" is not beyond a reasonable doubt, and hence need an appropriate correction by the way of erratum.

Further, as discussed above, the Barnase and Barstar proteins are expressed simultaneously in final plant and they remain in a tight complex (i.e. barnase-barstar complex) and not as free form. The current study has not even touched upon the barnase-barstar complex; therefore, until the systematic studies on this complex is conducted and concluded, it is not appropriate to give a 'safe' tag to these transgenic proteins. This is particularly important since the conclusion drawn from this study was one of the major evidence which was used by the Indian regulatory authorities to recently give a safety clearance to the genetically engineered Brassica juncea (Indian Mustard) for commercial cultivation in India. [8, 9]

Ref & Suppl Information

On 2017 Jul 09, Sunil Verma commented:

Comment on - Evaluation of Bar, Barnase, and Barstar recombinant proteins expressed in genetically engineered Brassica juncea (Indian mustard) for potential risks of food allergy using bioinformatics and literature searches

Sunil Kumar Verma, Principal Scientist CSIR-Centre for Cellular and Molecular Biology, Hyderabad 500 007, India.

In this study, the authors have tested the allergenic potential of the transgene Bar, Barnase, and Barstar expressed in Genetically Modified Indian Mustard for heterosis breeding. To this end, the authors have done the primary amino acid sequence comparisons of these proteins with the primary amino acid sequences of the known allergens listed in Allergenonline.org and NCBI Entrez protein database until January 2015 and 9 March 2015, respectively. Based on these bioinformatics comparisons authors concluded that the Bar, Barnase and Barstar proteins are unlikely to present any significant risk of food allergy to consumers. The authors also recommended not to perform any human serum IgE testing to further evaluate possible binding to the Bar, Barnase or Barstar proteins.

I hereby propose that the above conclusions drawn by the authors in this study are incorrect and require a major revision.

The main criteria used by the authors in these bioinformatics comparisons was the primary amino acid sequence homology searches of the proteins in question with that of the primary amino acid sequences of the potential allergen listed in above databases. All the hits with less than 50% primary amino acid sequence identities for full length proteins and less than 35% identity in the sliding window 80 amino acid segments of each proteins were ignored; the argument was that these matches could not have led to significant structural similarities among the proteins in question, therefore can be ignored.

Several independent studies have shown that in many cases, even though the primary amino acid sequence similarity between two proteins / domains are very less (<20%), but the tertiary structures of the proteins may be highly similar. One classical example of this is high structural similarity between N terminal half of the Krit-B41 domain with that of the RA domain of RalGDS (1RAX:A) with an r.m.s. deviation of 2.9A for 80 aligned positions; despite a very low homology in their primary amino acid sequences (sequence identity =8.7%). [1, S1] It is notable that both RalGDS and Krit-1 interact with Rap1A through the RA and B41 domains, respectively [2, 3], and so the talin [4]. Thus, the high primary amino acid sequence similarity between two proteins may though infer greater chances of structural homology between these proteins; however, low primary amino acid sequence similarity does not necessarily infer that proteins in question will necessarily have higher structural dissimilarities.

Since it is the conformationally determined structure of the proteins/epitopes which finally decide immunogenicity and allergenicity - and not just the primary amino acid sequences; the conclusion drawn in this study based on merely the primary amino acid sequence comparisons are scientifically inappropriate.

Secondly, in real scenario, both the Barnase and Barstar proteins are expressed simultaneously and these two proteins remain in a complex and not as individual proteins in plant [5, 6]. It is not unlikely that structure of a specific protein in complex may be different than that of the structure of the same individual protein in free form. Also, there may be the possibilities of formation/exposure of new epitope(s) surfaces, particularly as we know now that there are several antibodies known that recognize just the native proteins and some may indeed require complex assembly.

Thus, these conformationally determined epitopes that are recognized in the complex but not the free protein of interest may be reveled in differential screening between a protein and a complex form of the same protein. The conformationally determined epitopes could then be compared for structural homology with the epitopes in known allergens to determine the allergenic potential of two proteins in complex; such studies however, were not conducted in this paper; and the fact that Barnase and Barstar remain in complex and not in free form, was completely ignored throughout the study.

Finally, I found that the overall implication of the Allergenonline.org database itself on correctly predicting the allergenic potential of a new antigen was also questionable.<br> To test this, I assumed that 'Ani s 9' (which is a very well known allergen from SXP/RAL-2 protein family) [7] is a new putative allergen and that this group of proteins are not yet listed in the database; and asked whether or not one can predict if 'Ani s 9' is a potential food allergen using the strategy as was used in this study for Barnase, Barstar and Bar transgenic proteins. The full length primary amino acid sequence comparison of 'Ani s 9' (GenBank: ABV55106.1) using default parameter i.e 'E' value cut off = 1 identified 7 hits (excluding the hits with its own sequences) with 'tropomyosin' allergen from various organisms and 'AAEL002761-PC ' allergen from Aedes aegypti, respectively; however, none of the hits was with significant similarity cut off (>50%). Thus, this bioinformatics search criteria wrongly predicted that the 'Ani s 9' is not a potential food allergen. [S2]

The another criteria i.e. greater than 35% identity in the sliding window of 80 amino acid segment also did not produce any hit at all (other than self hits, which were excluded as explained above), indicating that this criteria also failed to identify 'Ani s 9' as potential food allergen. [S3] The third criteria i.e. 8 continuous amino acid segment search also did not identify any hit with any of the allergen in the database.[S4]

Thus, the bioinformatics search as used in this study following any of the criteria defined could not identify 'Ani s 9' as a potential food allergen. This confirms that the criteria used in this study by authors could easily give false negative results.

The only strategy that could have identified 'Ani s 9' as possible food allergen was a '6 continuous amino acid segment search, which could have identified its match with Allergen 'Lol p 5' for the 6-aa segment 'ANAPPA'. [S5]

This criteria however, was not used in current study to predict the allergenic potential of Bar, Barnase and Barstar. If this specific criteria was used, Barnase transgenic protein also could have given a potential hit with Allergen Ber e 2 and Ani s 9 for the 6 continuous amino acid patch LFSTAA, and WVASKG, respectively [S6]; hence, the conclusion of this paper could have been different.

In view of the above, I conclude that the criteria implemented in this study were not sufficient to exclude the possibility of the transgenic protein Bar, Barnase and Barstar being a possible allergen; therefore the conclusion drawn by authors that "the above transgenic proteins are unlikely to present any significant risk of food allergy to consumers" is not beyond a reasonable doubt, and hence need an appropriate correction by the way of erratum.

Further, as discussed above, the Barnase and Barstar proteins are expressed simultaneously in final plant and they remain in a tight complex (i.e. barnase-barstar complex) and not as free form. The current study has not even touched upon the barnase-barstar complex; therefore, until the systematic studies on this complex is conducted and concluded, it is not appropriate to give a 'safe' tag to these transgenic proteins. This is particularly important since the conclusion drawn from this study was one of the major evidence which was used by the Indian regulatory authorities to recently give a safety clearance to the genetically engineered Brassica juncea (Indian Mustard) for commercial cultivation in India. [8, 9]

Ref & Suppl Information

On 2017 Jul 19, Richard E Goodman commented:

Sunil Verma, CSIR Hyderabad, India, provided hypothetical comments about our study (PMID 26079618). He suggests our evaluation methods and conclusions were wrong and suggests our conclusions should be changed. We reviewed our publication and Verma’s comments. We conclude that he is misinterpreting the rationale for our study, the recommendations of the CODEX Alimentarius guidelines (2003/2009) and Indian Council of Medical Research (ICMR, 2008) on the safety of Genetically Engineered (GE) crops and our study. The dossier including study was reviewed and approved by the RCGM and GEAC committees of the government of India. Dr. Verma did not present data from studies of the GE plant or the proteins.

Our response: 1) Indian mustard (Brassica juncea) is an important oil seed crop consumed as oil, seed and greens in India. However, production efficiency and plant diseases limits yield. Heterosis breeding often leads to production of agronomically superior hybrid plants.

2) Mustard is usually self-pollinating. Therefore, scientists at the University of Delhi South Campus developed a GE Indian mustard system producing a male sterile plant and a fertility restorer line that allows hybrid production. The male sterile line was created by insertion of a gene that encodes an RNase (Barnase) that is expressed only in the tapetum (outer cell layer of the anther or pollen sac in early flower buds). The DNA construct includes a herbicide tolerant gene (bar). Expression of Barnase is tightly regulated by the construct of DNA as explained in Jagannath et al., 2002, where a long spacer DNA was inserted between the bar gene and promoter and the Barnase gene to ensure it is only expressed in the anther as demonstrated from field trial plants. Barnase is not detectable in seeds or green tissue of the plant (Jagannath et al., 2001). The fertility restoring line was developed by inserting a gene encoding Barstar also using an anther (tapetum) restricted promoter. Barstar binds to Barnase in a very tight protein complex that inhibits the RNase activity of Barnase (Bisht et al., 2004). The restorer line also includes the bar herbicide tolerance gene. Data on the two GE plants and their hybrid offspring were well characterized and data, including a full safety evaluation, was reviewed by the RCGM and GEAC committees of the Indian government with recommendations the plants are as safe as non-GE mustard.

3) The safety evaluation followed the Indian Council of Medical Research guidelines (2008), which are consistent with the international guideline (CODEX Alimentarius, 2003 and 2009). One primary concern is whether the protein expressed by the novel genes encode allergens or proteins of moderate sequence identity (e.g. >35% identity over 80 amino acids) to an allergen so that allergenic cross-reactivity might be expected. The guidelines are referenced in the paper and are publically available. The methods and databases are cited. There is no evidence that proteins of lower sequence identity (including the FAO/WHO 2001 suggestion of 6 consecutive amino acid matches) are sufficient to demonstrate potential risks of allergy.

4) The assessment for potential risks of allergenicity of GE crops have been extensively addressed (e.g. Goodman et al., 2008). The risks of food allergy recognized by food regulators around the world are that foods must be labeled with the source of ingredients so that people who have allergies to allergenic sources (e.g. peanuts, specific tree nuts, milk, wheat, eggs), can avoid ingesting foods that cause their allergies. Governments have adopted specific food labeling laws for all foods and hold producers accountable. The standard is the same for GE crops. When Pioneer Hibred was developing a nutritionally enhanced soybean by adding a gene from Brazil nut, Dr. Taylor at the University of Nebraska conduct tests since the gene was from an allergenic tree nut. The results demonstrated that the protein was an allergen that would put Brazil nut allergic subjects at risk (Nordlee et al., 1996) That GE plant never entered food production. There are no published scientific reports of a GE protein in an approved crop causing allergic reactions in humans.

5) The methods suggested by Verma have not been demonstrates to predict allergenicity. A number of publications demonstrate that six-amino acid matches are common and not predictive (e.g., Hileman et al., 2002). Proteins with less than 50% overall identity are unlikely to be cross-reactive. Similar three-D shapes do not predict allergenicity. The AllergenOnline.org database is a peer reviewed database that includes Ani s 9, and there is no significant math. The database is updated every year and is accepted by international governments (Goodman et al., 2016).

6) A similar GE mustard (canola) was produced PGS in Europe, now owned by Bayer CropScience. The canola include the same proteins, expressed in similar amounts and patterns. Seven events have been approved for production and consumption in Canada, the US, and Australia. Food and feed products from those events are approved for import and consumption in Japan, China, South Africa, Korea and the European Union. The food and feed safety evaluation summaries and details are available on the CERA-GMC.org/GmCropDatabase Result website. First approvals were in 1995 and there is no evidence of harm even though 95% of canola in North America is GE.

7) The standard for food safety of GE crops in all countries that approve GE crops is that the GE plant does not pose any greater risk of allergy or toxicity than the similar non-GE crop. This GE Indian mustard developed at South Delhi campus meets those standards.

We stand by our conclusions that the male-sterile and restorer Indian mustard events and progeny are as safe as the non-GM counter-parts. There is no data that suggests otherwise. Statements by Dr. Verma are hypothetical, not based on facts. This mustard meets the same standards used to evaluate GE crops approved in the US, Canada, Australia-New Zealand, Japan and other countries.

Richard E. Goodman, and Vasanthi Siruguri

References: 1.Siruguri V, Bharatraj DK, Vankudavath RN, Rao Mendu VV, Gupta V, Goodman RE. 2015. Evaluation of Bar, Barnase and Barstar recombinant proteins expressed in genetically engineered Brassica juncea (Indian mustard). For potential risks of food allergy using bioinformatics and literature searches. Food Chem Toxicol 83:93-102. 2.Jagannath A, Bandyopadhyay P, et al. 2001. The use of a spacer DNA fragment insulates the tissue-specific expression of a cytotoxic gene (barnase) and allows high-frequency generation of transgenic male sterile lines in Brassica juncea L. Mol Breeding. 8:11-23. 3.Jagannath A, Arumugam N, et al., 2002. Development of transgenic barstar lines and identification of a male sterile (barnase)/restorer (barstar) combination for heterosis breeding in Indian oilseed mustard (Brassica juncea).Curr Sci 82:46-52. 4.CODEX Alimentarius Guidelines 2003, 2009. 5.ICMR Guidelines. 2008. 6.Bisht NC, Jagannath A et al., 2004. A two gene-two promoter system for enhanced expression of a restorer gene (barstar) and development of improved fertility restorer lines for hybrid seed production in crop plants. Mol Breeding 14:129-144. 7.Hileman RE, Silvanovich A et al., 2002. Bioinformatic methods for allergenicity assessment using a comprehensive allergen database. Int Arch Allergy Immunol 128:280-291. 8.Goodman RE, Ebisawa M, et al. 2016. AllergenOnline: a peer-reviewed, curated allergen database to assess novel food proteins for potential cross-reactivity. Mol Nutr Food Res 60:1183-1198. 9.Nordlee JA, Taylor SL, Townsend JA, Thomas LA, Bush RK. 1996. Idnetification of a Brazil-nut allergen in transgenic soybeans. N Engl J Med. 334:688-692. 10.Goodman RE, Vieths S, et al. 2008. Allergenicity assessment of genetically modified crops—what makes sense? Nat Biotechnol 26:73-81.

On 2017 Jul 24, Richard E Goodman commented:

As one of the senior authors of the Siruguri et al., 2015 publication, with 20 years experience in evaluating the safety of Genetically Engineered (GE or GM) crops, I feel obligated to respond to the statements Dr. Sunil Verma is posting on PubMed COMMONS and now also in Science as an e-letter to the 2016 publication by Priyanka Pulls describing the development of this GE mustard. Dr. Verma's second comment posting here lists his letter in Science. Importantly, neither my comments, nor those of Verma are peer reviewed. We are giving our opinions (which differ markedly as does our experiences). I have written a response to Verma's e-letter in Science and it is available at http://science.sciencemag.org/content/352/6289/1043/tab-e-letters . It addresses the issues of the accepted hazard and risk evaluation of GE crops in India and internationally. Our 2015 publication here describes the assessment looking at the source of the genes, the sequences of proteins and the scientific rational is to evaluation potential risks for those who might be allergic to the protein (Barnase, Barstar or Bar), or to proteins that are highly identical, and could share IgE binding. Dr. Verma did not provide any data that demonstrates we are wrong, or that there are risks from this mustard. Instead in his supplemental information, he compared the sequence of Ani s 9, a minor allergen of a fish parasitic worm, to other sequences in the AllergenOnline.org database. And he implies that cross-reactivity might occur due to associated proteins like the SXP/RAL-2 proteins (Ani s 5 and Ani s 9). However, as noted by Garcia-Mayoral et al., 2014), similar proteins do not exist outside of worms (Nematodes). Ani s 9 has very little sequence similarity to Barnase, as described in our paper. Comparing Ani s 9 in AllergenOnline demonstrates that it is rather unique and unlikely to have cross-reactivity outside of the parasitic worm allergens. This mustard contains Barnase, not Ani s 9. Furthermore, he points to the six amino acid match of Barnase to Ani s 9. But as described in our paper and in my letter in Science, that six amino acid segment matches hundreds of proteins in the NCBI database, without any evidence of cross-reactivity or allergy. Furthermore, there is no evidence that a six amino acid match predicts cross-reactivity. The standard in CODEX is sequences matching >35% identity over 80 amino acids, and such matches are quite conservative (overpredict) both primary and confirmational epitopes (Goodman, 2006, Goodman et al., 2008). CODEX indicates you may do a short sequence match, but must justify the methods. If there are matches of >35% identity over 80, then serum IgE tests would be warranted using sera from at-risk (specifically allergic subjects (Goodman, 2008). In the future there will be improvements in the assessment (Goodman and Tetteh, 2011), however, Dr. Verma has not described any new method or any proof that he has an improvement. Instead, he has proposed hypothetical issues, a letter in Science and he has not posted the letter on facebook. If he thinks there can be improvements, he should do experiments and submit his results to a peer reviewed journal for scientific evaluation. The authors of the Siruguri et al 2015 paper stand by our results that this GM mustard is as safe as the non-GM mustards in use in India today.

References: Garcia-Mayoral MF, Trevino MA et al., (2014). Relationships between IgE/IgG4 epitopes, structure and function in Anisakis simplex Ani s 5, a member of the SXP/RAL-2 protein family. PLOS, Negl Tropical Dis 8(3):e2735. Goodman RE. (2006) Practical and predictive bioinforamtics methods for the identiifcation of potentially cross-reactive protein matches. Mol Nutr Food Res 50:655-660. Goodman RE, Vieths S et al, (2008). Allergenicity assessment of genetically modified crops--what makes sense? Goodman RE. (2008) Performing IgE serum testing due to bioinformatics matches in the allergenicity assessment of GM crops. Food Chem Toxicol 46(Suppl 10):S24-S34. Goodman RE, Tetteh AO. (2011). Suggested improvements for the allergenicity assessment of genetically modified plants used in foods. Curr Allergy Asthma Rep 11(4):317-324.