My comments at the EPA's Nov. 28th Scientific Advisory Panel hearing on StarLink follow. I reviewed many of the studies submitted by Aventis -- a case study in "shoddy corporate science." Perhaps this should become our slogan in response to "sound science."
Bill Freese, Policy Analyst
Safer Food - Safer Farms Campaign
Friends of the Earth
PS If anyone wants any of the studies listed in the References section, let me know. Most are available only in hard copy.
Comments of Bill Freese before FIFRA's Scientific Advisory Panel
Assessment of Scientific Information Concerning StarLink Corn
November 28, 2000
At the October 20th, 2000 meeting of the SAP that considered re-registration of Bt crops, the question of substantial equivalence between plant-produced and bacterial-produced Cry proteins was raised with respect to Cry9C. Since most of Aventis' studies have been conducted with bacterial Cry9C, it is obviously of great importance to know whether Cry9C derived from transformed corn plants is in fact substantially equivalent to bacterial Cry9C. If the two differ in any respect that might be considered important, it would cast doubt on the greater part of Aventis' data and perhaps necessitate a new round of studies to resolve any questions that might be raised by the discrepancies.
Aventis submitted a study in 1997 to answer this question (Peferoen, 1997a). The author discovered three differences between the two bacterial-produced Cry9C proteins (E. coli and Cry-minus Bacillus thuringiensis, each transformed with the cry9c gene sequence engineered into the plasmid pGI9CK) and the corn-produced protein. These differences related to: 1) N-terminal amino acid sequence; 2) molecular weight; and 3) glycosylation.
The N-terminal amino acid sequences of the two bacterial Cry9C proteins were identical, while the corn Cry9C differed by 1-4 amino acids at the N-terminal. The difference was attributed to slightly different protease trimming of bacterial versus corn Cry9C.
Corn-produced Cry9C: 73.8 kD
Bt-produced Cry9C: 70.5 kD 3.3 kD or 4.5% lower than corn-produced Cry9C
E. coli-produced Cry9C: 67.9 kD 5.9 kD or 8% lower than corn-produced Cry9C
These molecular weights were determined by SDS PAGE electrophoresis, and the author maintains that "such small differences are within the resolution of the applied technique and indicate that the molecular weight of the corn and the bacterial produced Cry9C proteins is similar." Yet this statement leaves unclear whether the observed 4.5 and 8% differences are real or experimental artifacts. Is SDS PAGE electrophoresis really so imprecise, or is the molecular weight of corn-produced Cry9C in fact 3,000-6,000 daltons greater than the bacterial forms? Could these discrepancies in MW perhaps be due to differences in post-translational processing between corn versus bacterial Cry9C?
One possible difference in post-translational processing is glycosylation. According to the EPA, "glycosylation and other post-translational modifications can be used to discern potential alterations due to plant expression of the introduced protein. Glycosylation of the plant expressed protein is also critical in establishing test substance equivalency when sources other than plant protein are used as the test substance for toxicity testing." [Cry9C Food Allergenicity Assessment Background Document, p. 4].
In the study mentioned above, the author concludes that while both bacterial-produced Cry9C proteins are not glycosylated, the corn Cry9C is "weakly glycosylated." Dr. Noteborn presented results at the October SAP which also indicated glycosylation of plant-produced Cry9C (I believe in tomatoes). This would seem to raise serious questions about the propriety of using bacterial Cry9C extracts rather than the corn-produced protein for the purposes of testing.
Stability to digestion
Although disputed, many scientists consider glycosylation to be a characteristic property of allergenic proteins. One suggested role of carbohydrate moieties in allergenic proteins is protection against peptic hydrolysis. Matsada et al. demonstrated that human IgE antibodies to ovomucoid in egg-allergic subjects react to a carbohydrate-containing domain found in about one-half of ovomucoid molecules, whereas no antibody with reactivity to the carbohydrate-free domain found in the other half of ovomucoid molecules was detected. However, the authors thought it unlikely that the carbohydrate chain itself was involved in the epitope in question, since cleavage of the disulfide bonds of ovomucoid molecules completely abolishes their recognition by human IgE (whether the domain contains carbohydrate or not). When ovomucoid was subjected to peptic hydrolysis, some glycopeptides with immunoreactivity were isolated, whereas peptides lacking carbohydrate had no immunoreactivity. The authors' suggested explanation was that the carbohydrate moieties may facilitate ovomucoid' s IgE reactivity by exerting a protective effect against peptic hydrolysis [discussed in Li-Chan & Nakai, 1989, p. 43].
Assuming that glycosylation can in fact protect against peptic hydrolysis in the case of this ovomucoid antigen, the glycosylation of corn-produced Cry9C could perhaps lend it similar protection against peptic digestion. Of the four studies on Cry9C digestibility submitted by Aventis, only the first [Peferoen, 1997b] involved plant-produced Cry9C; the corn Cry9C in this study showed no signs of protein disintegration after 4 hours of exposure to simulated gastric fluid (SGF) with pepsin (0.32% pepsin in gastric buffer, pH = 2.0). While it is true that the Cry9C proteins derived from E. coli and Bt that were also tested in this study proved resistant to degradation as well, the EPA did note differences in the Western blots between the bacterial preparations, on the one hand, and corn Cry9C, on the other, after treatment in SGF: "Interestingly, the three bands from microbially produced Cry9C (either E. coli or B. thuringiensis) were different from the plant source Cry9C." [EPA Data Evaluation Report on Peferoen, 1997b].
The other three studies were all conducted exclusively on bacterial Cry9C preparations. One showed substantial resistance to degradation after 2 hours in pepsin-containing simulated gastric fluid (pH = 2.0) [Noteborn, 1998]. In the two most recent studies, which were conducted at pH 1.5 and 1.2, respectively, the lower pH appeared to reduce protein stability, though it was still stable for at least 30 minutes [Aventis, MRID #45114401 & 45114402]. Addition of pepsin to the SGF increased the rate of degradation. Though Aventis has presented these latter results at low pH as evidence that Cry9C poses no allergenic concerns, EPA notes that gastric pH varies from less than 1.0 up to 3.0, and can be raised significantly by ingestion of food, antacids or medications [EPA Preliminary Assessment, p. 9]. This in itself argues against relying exclusively on the lower pH digestibility studies. The possible effect of glycosylation in protecting against peptic hydrolysis, coupled with strong evidence that plant-produced Cry9C is in fact glycosylated, is another argument against assuming that Cry9C is sufficiently degraded in gastric fluids not to pose allergenicity concerns. It would be interesting to compare digestion of corn-produced Cry9C and bacterial protein in SGF at a range of pH values, with and without pepsin, to test this hypothesis.
Based on the available information, however, it would seem best to accept the conclusion of Dr. Noteborn in the second digestibility study [Noteborn, 1998, p. 22], which also refers to in vivo evidence of resistance to digestion:
"The data on the digestive stability of the Lys mutant Cry9C protein . in simulated gastric and intestinal digestive fluids and in the intestinal system of the rat suggested a probability of the Lys mutant Cry9C protein surviving to be absorbed via the intestinal mucosa during consumption and thus, may [sic] potentially trigger the production of antibodies, including the antigen-specific immunoglobulin E antibodies."
Amino Acid Homology with Known Allergens
One of the main difficulties in evaluating the potential hazards posed by genetically engineered foods is the fact that many of the genes that are introduced via this technology encode for novel proteins that have never before been in the human diet and/or come from sources that have no known history of allergenicity. Cry9C fits both criteria. The tolworthi strain of Bt comes from the Philippines, and there are no Bt sprays based on Cry9C, so it is a novel ingredient in food. As of now, Bt tolworthi is not known to be a food allergen, though as we shall see there is evidence suggesting that other strains of Bt might be allergenic.
In any case, schemes have been set up to assess the allergenic potential of new genetically engineered foods. Appendix 1 presents one widely used decision tree [Metcalfe, et al., 1996]. The approach recommended for transgenes derived from known allergenic sources, such as tree nuts, fish or eggs, is well established, and involves in vitro assays, skin prick tests and double-blind placebo-controlled food challenge tests (left side). However, ever since the soybean spliced with a Brazil nut gene proved to be allergenic to those with Brazil nut allergies, there has been little interest in transgenes from known allergenic sources. Hence this side of the decision tree has little relevance. In fact, it has been reported that: "All of the genes that have been transferred to plants which are currently commercialized fall into the category of 'derived from sources without allergenic history.'" [Consumer & Biotechnology Foundation, 1999, ]
This right-hand side of the decision tree dealing with transgenes from sources that have no history of allergenicity is much more problematic, because there are unlikely to be sera available from allergic subjects to test the novel protein [Lehrer, 1999]. For this reason, such proteins must be assessed by indirect parameters of allergenicity. Cry9C possesses two of these attributes - stability to digestion and heat - which are discussed elsewhere.
The other parameter recommended for evaluation of novel proteins is amino acid homology to known allergens. Typically, a computer search is undertaken for any sequence of eight amino acids in the novel protein that matches any 8-AA sequence in known allergens stored in various databases kept for this purpose. This is based on the notion that an epitope has an optimal length of 8-12 amino acids [Astwood, et al., in Food Allergy, p. 76]. However, there are several serious limitations to this procedure, as even its supporters concede. 1) Most obviously, we simply don't know the amino acid sequences of many allergens, so they are unavailable for purposes of comparison [SAP, 2000, p. 10]; 2) Focusing on optimal length is inappropriate if one's goal is to catch all potentially significant homologies with known epitopes. Instead, one should choose a number of AA residues that represents a lower limit on linear epitope size. T.P. King, for instance, reports that "the minimal size of a continuous epitope is about 6 amino acid residues" [King, 1994, p. 5]. The minimal epitope size of the major peanut allergen is 6 amino acids, while the major codfish allergen is reported to have IgE binding regions that consist of 2 tetra-peptides separated by a non-binding 6-residue spacer [Consumer & Biotechnology Foundation, p. 23]; 3) Perhaps most importantly, the simplistic 8-AA linear epitope search completely misses conformational and discontinuous epitopes, which are determined by the protein's native, 3-dimensional conformation rather than its linear sequence.
Conformational and discontinuous epitopes
Metcalfe et al. (1996, p. 169) point to the birch pollen allergen, Bet v 3, as an example of a conformational epitope that would not be detected by the recommended 8-AA search. IgE binding of patient sera to Bet v 3 depends on calcium-regulated conformational changes; Bet v 3 is apparently not recognized by IgE in the absence of calcium, the presence of which is required to change the shape of the epitope to permit IgE binding.
According to T.P. King, speaking of B-cell epitopes: "The majority of antibodies formed on immunization with allergens or any other globular proteins are directed toward the discontinuous epitopes. This is true in allergic patients or experimental animals and it is irrespective of antibody isotypes." [King, 1994, p. 4]. Dr. King goes on to note that ". continuous epitopes have fewer amino acid residues than the discontinuous epitopes. This may explain why antibodies for discontinuous epitopes have higher affinities than those for continuous epitopes." [Ibid, p. 5]. In short, discontinuous epitopes are more important in the allergic process and bind IgE antibodies more strongly than continuous sequences.
If this is correct, the simplistic 8-AA homology search could be missing - not just a few - but the majority of epitopes in allergens that are responsible for causing allergies. This view is echoed in the June 29th, 2000 SAP report: "An 8 amino acid sequence homology may be too simplistic, it ignores conformational epitopes." Metcalfe et al. (1996, p. 169) seem to concur: "this approach is limited in that it cannot identify discontinuous conformational epitopes that depend on the tertiary structure of the allergen." Even strong supporters of the 8-AA homology search concede its weakness. "The situation is more complex and less well defined for conformational epitopes, which means that sequence-based comparisons should be used only as indicators of potential linear epitopes" [Astwood, et al., p. 76].
Everyone admits the problem, but few propose solutions. The only specific suggestion I have seen for improving the amino acid homology search procedure comes from a European workshop that brought together consumer, biotech industry and public health representatives, and was partially sponsored by Unilever. The report that issued from that workshop called for reducing the level of matches to four amino acid residues [Consumer and Biotechnology Foundation, 1999, p. 23]. Scientific experts from the European Commission question the relevance of purely linear epitope searches, but state that if they are used, "it is necessary to improve databases, and to develop algorithms and software to compare structures and not only sequences." [Scientific Steering Committee, 2000, p. 9].
In the case of Cry9C, only 8-AA homology searches were conducted to check for possible matches to the linear epitopes of known allergens. Given the broad consensus on the failure of this procedure to detect conformational and/or discontinuous epitopes, and equal agreement on the importance of these epitopes in the allergic process, it is clearly possible that the native Cry9C protein harbors allergenic potential based on potential structural (rather than linear) homology to known allergenic proteins. Coupled with Cry9C's stability to digestion and heat, and its novelty in the human diet, this potential argues against continuing to expose the population to this protein.
Are allergenic proteins always the abundant ones?
An important plank in Aventis' argument that Cry9C poses no allergenic risk to the public is the contention that most allergenic proteins are present at levels of 1 to 40% of the total protein of the allergenic food [Aventis, Updated Safety Assessment, 2000, p. 21]. Though this might be true, most, of course, is not the same as all. And this statement should perhaps be qualified with the adjectives "known" and "major." Unfortunately, the 1-40% rule is often presented as if it does encompass all allergens. One of the strongest proponents of this rule has been Steve Taylor, the University of Nebraska food scientist who is also now serving as an Aventis consultant [Aventis, Updated Safety Assessment, 2000, p. 72]. Thus it is interesting to note that Dr. Taylor once showed more interest in food allergens present in lower quantities:
".we need to be looking at, very carefully, what are the properties of those proteins that exist in low amounts in foods but are still allergens. The 20 kilodalton protein that is problematic for some people with soybean allergy is a protein that's present in infinitesimally low quantities in soybeans. You can't even see the protein band on electrophoresis after purifying the protein, but you can see its reaction to IgE very nicely." [FDA, 1994, p. 145, emphasis added]
Unfortunately, I haven't been able to track down this soybean allergen. But there are other examples of low-level allergens. For instance, Li-Chan and Nakai (1989) report that "a total of 13 egg-white proteins, including minor constituents, have been found to cause production of specific IgE antibodies in individuals allergic to eggs." Bovine serum albumin, a milk allergen, constitutes just 1% of milk protein, while allergenic lectin proteins in soybeans and peanuts represent between 1 and 2% of total plant protein [Astwood]. Does the Panel know of any method for strictly predicting the prevalence or severity of an allergy due to a particular protein from knowledge of the amount of that protein present in a food?
Protein Sensitization and Cross-Reactivity
As even experts in the field will admit, our knowledge of the allergic process is still very primitive. This applies particularly to the process of sensitization. The issue is clouded by conflicting reports on whether extremely low levels of a protein induce sensitization or tolerance, the question of sensitization in utero or through breast milk, the increased susceptibility to sensitization in infants and young children, our extremely inadequate knowledge with respect to cross-reactivity between foods, and the possibility for allergens by one route (inhalant) to also act by another (gastrointestinal). Concerning this latter point, Robert Aalberse reports that the protalins in hazelnuts and hazel pollen share some common amino acid sequences; he offers this as a possible explanation for why people allergic to hazelnut pollen are also allergic to the nuts [FDA, 1994, p. 45).
Given this uncertainty, it would seem wise to demand extremely firm evidence refuting the potential of Cry9C to sensitize people before a decision is made to permit it in foods. It is therefore troubling to note that many statements by Aventis and others who support approval of StarLink in contaminated products argue rather blithely that we simply need not worry about sensitization, since the corn hasn't been around long enough, or in sufficient quantities, to induce it. I will argue elsewhere that Aventis' dietary exposure estimates might be several orders of magnitude off, but what substantiation, for example, is provided for the argument that two years of exposure is insufficient time for sensitization to occur?
Aventis has supplied only two studies that address the issue of sensitization, and both seem to have limited value. The first one looks at the reactivity of sera from suspected corn-sensitive individuals to wild-type and StarLink corn [Lehrer, 1997], while the second [Lehrer, 2000] evaluates the reactivity to StarLink of sera from people with allergies to common foods. The first one was submitted in 1997, before StarLink had been approved for commercial use, so specific IgE reactivity to Cry9C would not be expected, since StarLink had not yet become part of the human diet at that time. And since the tolworthi strain of Bt comes from the Philippines, and has not been used in any Bt sprays, one would not expect significant environmental exposure to Cry9C. Mainly for this reason, EPA rated this study supplemental, but also because no purified Cry9C control was used to establish the background reactivity, which renders the results ambiguous, and because of the failure to identify individual sera in the immunoblot assay, which made it impossible to correlate intensity of staining with percentage reactivity in the RAST assay. [EPA Data Evaluation Review].
The more recent study [Lehrer, 2000], though conducted after some exposure to StarLink had occurred in the population, is also of limited value, because it tells us at most that there may not be cross-reactivity between eight common food allergens and Cry9C. Even this is uncertain, however, because a paired t-test shows that differences between the IgE reactivities of these sera to wild-type and transgenic (StarLink) corn might be large enough to constitute a real difference (p = 0.411). In addition, sera donors were only required to meet two out of three inclusion criteria: 1) a positive history of food allergic response to one or more of the test allergens or food allergy compatible with one or more of the test allergens; 2) a positive skin test to the specific test allergen extract; and/or 3) a positive IgE antibody response (> 3% binding to the suspect allergen by RAST). Hence sera from donors without a positive history of food allergic response could have been included in the study (i.e. if they met criteria 2 and 3 but not 1). I believe it is standard practice for such tests to use only sera from individuals with known food allergies [Lehrer, 1999, pp. 3-4]. In any case, according to Dean Metcalfe, "people who've become allergic to food have a very narrow spectrum of reactivity." He notes that 50-60% of the allergic population have allergies to just one food, while 20% are allergic to two, and 20% to three, suggesting that roughly half of food allergic people do not exhibit cross-reactivity to other foods [FDA, 1994, p. 15].
Skin and Serological Tests
A perhaps better means of evaluating the potential allergenicity of Cry9C is offered by an EPA-sponsored study conducted by Bernstein et al. entitled "Immune responses in farm workers after exposure to Bacillus thuringiensis pesticides." The study demonstrated positive skin prick tests to Btk spore extract and the presence of IgE antibodies in 2 of 123 farm workers exposed to Bt sprays containing Cry1Ab and Cry1Ac in the course of their field work. Although the authors feel that "it is unlikely that consumers would develop allergic sensitivity after oral exposure to transgenic foods (e.g. tomatoes, potatoes) that currently contain the gene encoding this protein," they clearly state that "future clinical assessment of this possibility is now feasible because of the availability of reliable Bt skin and serologic reagents developed during the course of this investigation" [Bernstein et al., 1999, p. 581]. Although the authors were unable to link any respiratory symptoms to Bt spray exposure, this could be due to the low levels of Cry proteins found in sprays. Bt crops contain concentrations of delta endotoxins that are one to two orders of magnitude higher than Bt sprays. It should also be noted here that StarLink expresses its delta endotoxin at higher levels than the Cry proteins expressed by other Bt corns. This difference is especially pronounced in the corn kernel, where Cry9C is present at levels 10 to over 400 times the grain concentrations of other Cry proteins [EPA, 2000, IIA5]. Hence corn dust concentrations of Cry9C could be two to three orders of magnitude higher than the level of endotoxin found in Bt sprays.
While far from perfect, it would seem that tests utilizing these skin and serologic agents would at least offer more meaningful results than the studies submitted by Aventis, which test StarLink extract against sera from people with allergies to foods containing allergens that are probably quite different than Cry9C protein. One would expect to find a great deal of homology between the various Cry proteins, and this is in fact demonstrated by the amino acid homology searches conducted by Aventis [MacIntosh, 1997a; Peferoen, 1997c]. The EPA's main objection to this proposal is that the Bt spray preparations used in the skin and sera tests contain numerous bacterial proteins other than the Cry proteins. Hence the observed reactivity might be attributable to these other proteins and not to Cry1Ab and/or Cry1Ac. This is certainly a possibility, but it is no argument against testing purified Cry9C against these sera.
Still better, of course, would be to conduct tests on the sera from farm and millworkers exposed specifically to StarLink corn. One possible pool of such workers would be employees of Garst Seed Company, the largest distributor of StarLink, though since Cry9C expression in pollen is reportedly quite low, a better choice might be millworkers exposed to StarLink grain dust.
Many studies have implicated grain, including corn, dust in allergic reactions, with bakers' asthma being the best known case (Bauer, 1998). And while inhalant allergens do not necessarily elicit allergic reactions via the gastrointestinal route (e.g. bakers' asthma), this is clearly a possibility that needs to be explored in the case of Cry9C should those subject to high occupational exposure prove sensitive to the protein.
Aventis contends that the level of exposure to Cry9C is too low to be of concern even if it is an allergen. This, in a nutshell, is the premise of its involved calculations on the commingling and dilution of StarLink corn from field to food processor to stomach. The only real argument presented to substantiate this view is an analogy - to the peanut allergens Ara h1 and Ara h2.
According to Susan Hefle [Aventis, Updated Safety Assessment, p. 69], "some types of allergic sensitization can occur at very low levels of allergen exposure, particularly inhalant allergy." The June 29th, 2000 SAP [p. 11] report states that "smaller doses [of a protein] are more likely to sensitize the immune system than to induce tolerance." Given the large gaps in our knowledge in this problematic area of sensitization, Aventis' comparison of exposure levels to Cry9C versus peanut allergen to "prove" that it is unlikely to cause sensitization appears to be completely unfounded. In particular, is there any evidence to suggest that the potency of an allergen to elicit an allergic response has anything to do with its "potency" to induce sensitization?
As noted above, inhalant allergens are especially likely to sensitize at very low levels. A complicating factor is the possibility that "inhalation might be important for food allergy" [Robert Aalberse, in FDA, 1994, p. 219]. Aalberse continues: "We discussed a situation where a child develops a food sensitivity without obvious exposure in the food, but when you look in the dust like Martin Chapman did for the mite elements we find that there is as much food allergen in dust as there is mite allergen, in the range of one to 10 micrograms in the gram of dust. These children don't eat the food, perhaps they have some via the breast milk, but you have to keep in mind perhaps the other sensitization route is via inhalation of the dust in the child's environment." Aalberse suggests that egg white and milk protein might induce sensitization via this inhalational mechanism.
Aventis assumes without discussion that the only possible route for induction of sensitization by Cry9C is oral. If inhalational sensitization to a food allergen is indeed a possibility, this constitutes still another reason for conducting sera tests on farm and millworkers who are exposed occupationally to high levels of StarLink corn dust. And since a large percentage of farm or millworkers in this country are also Hispanics, who tend to eat more corn than other ethnic groups, there is the possibility for high levels of exposure by two routes concurrently.
Finally, the SAP that met on October 20, 2000 to consider Bt re-registration in light of human health concerns was asked if it were possible to establish a safety threshold for Cry proteins, in particular Cry9C. Unfortunately, we do not yet have the report from that meeting, but the panelists seemed to agree that it would be impossible to do this based on the available information. Is this Scientific Advisory Panel prepared to set such a safety threshold?
Another important factor impinging on dietary exposure is the stability of Cry9C protein to the high temperatures of food processing. In two studies that addressed this matter, Cry9C was found to be stable at 90* C for two hours [Noteborn, 1998] and 10 minutes [Peferoen, 1997b], respectively. In the former study, some high molecular weight aggregates were observed at 2 hours and afterwards, while in the latter study, the heat treatment appears to have ended after 10 minutes.
In another study [MacIntosh, 1997b], ELISA tests for Cry9C were conducted on processed catfish pellets. No Cry9C protein was detected in either the Bt or control pellets. According to the EPA's data evaluation report, "the study was not conducted in accordance with EPA GLP guidelines, but was, according to the author, designed to meet the spirit of the guidelines." EPA goes on to comment that "no details on the place and time of growth of the corn plants or processing of the kernels into catfish pellets was given. Many procedures/protocols and raw data were not provided." [EPA data evaluation report]. In its preliminary assessment, the EPA comments further that "the limit of detection was not reported in the fish feed study and the study contained very little information on the analytical method used.
The most recent information on Cry9C protein in processed foods comes from "Analysis of taco shells for Cry9C protein" [Aventis, October 23, 2000]. The EPA states that these measurements "used an analytical method that has not been validated." The EPA might be referring to the fact that the taco shells that Aventis tested for protein had only been "found to be positive for the presence of StarLink-derived DNA." [p. 8]. In other words, Aventis did not quantify the amount of StarLink corn in the taco shells it tested for the presence of Cry9C protein. The real-time quantitative polymerase chain reaction procedure employed by Genetic ID, the testing firm that first discovered Cry9C in taco shells for Friends of the Earth, is reportedly capable of detecting "0.01% or lower" levels of GMOs in processed food products (limit of detection). The limit of quantification is reported at 0.01% [Genetic ID's Analytical Methods for Detection and Quantification of Genetically Modified Organisms in Foods and Agricultural Products].
Why didn't Aventis quantify the amount of StarLink corn in the taco shells it tested for Cry9C protein? If real-time quantitative PCR has a limit of detection down to 0.01% or lower, it is possible that the taco shells tested by Aventis contained this low amount of StarLink corn. One would probably not expect to detect Cry9C protein if this in fact were the case. Aventis' negative results are therefore suspect.
Quantification of Cry9C protein in StarLink
The main study upon which Aventis bases its estimates of the level of Cry9C protein in StarLink corn is flawed by contamination of the controls with Cry9C [Shillito, 1999]. Though this might have occurred through contamination during the processing phase of the study, EPA stated that "based upon the data provided, it is not possible to rule out the possibility that there was expression of the Cry9C and PAT proteins in the control corn." After contamination of the original control corn, Aventis replaced it with a new control which was "a different line of corn, grown in a different state under different (unidentified) growth conditions." For this reason, EPA found "the accuracy of these numbers questionable."
Given the many uncertainties with regard to test substance equivalence and the import of Cry9C glycosylation, the protein's established resistance to digestion and heat, the failure to search for conformational & discontinuous epitopes, the recent discovery of IgE response to Bt sprays and the failure to use available reagents to conduct skin and sera sera testing, the potential for inhalational sensitization, and not least the many and glaring deficiencies in so many of Aventis' studies, the weight of evidence would seem to go against temporary approval of Cry9C protein in contaminated products.
Astwood, et al. "Food Biotechnology and Genetic Engineering," in Food Allergy: Adverse Reactions to Foods and Food Additives, eds. Metcalfe, Sampson & Simon.
Aventis: The Digestibility of the Cry9C Protein by simulated gastric and intestinal fluids. MRiD 45114401.
Aventis: Analysis of taco shells for Cry9C protein, Oct. 23, 2000.
Aventis: Updated Safety Assessment of StarLink Corn containing Cry9C protein.
Aventis: Comparison of the in vitro digestibility based upon pH of the endotoxin Cry9C derived from E. coli and Bacillus thuringiensis. MRID 45114402.
Bernstein et al (1999). "Immune responses in farm workers after exposure to Bacillus thuringiensis pesticides.
Consumer & Biotechnology Foundation (1999) with the European Federation of Asthma and Allergy Associations. Genetically modified foods and allergenicity: safety aspects and consumer information, workshop May 28-29, 1999.
EPA (2000). Bt plant pesticides biopesticides registration action document.
EPA: Cry9C Food Allergenicity Assessment Background Document
EPA: Preliminary Evaluation of Information contained in the October 25, 2000 submission from Aventis CropScience
FDA (1994). Transcript of "Conference on Scientific Issues Related to Potential Allergenicity in Transgenic Food Crops," April 18-19, 1994.
Genetic ID: Analytical Methods for Detection and Quantification of Genetically Modified Organisms in Foods and Agricultural Products
King, T. P. (1994). B Cell Epitopes of Allergens. Presented at FDA, 1994 (see reference).
Lehrer, S. (1997). Investigation of Allergens in Wild-Type and Transgenic Corn, Tulane Univ. School of Medicine.
Lehrer, S. (1999). The potential health risks of genetically modified organisms: How can allergens be assessed and minimized?
Lehrer, S. (2000). Evaluation of IgE Antibody Reactivity of Food Allergic Subjects to StarLink Corn, submitted to EPA Oct. 24, 2000.
Li-Chan, E. and Nakai, S. (1989). Biochemical basis for the properties of egg white. Crit. Rev. Poultry Science 2:21.
MacIntosh (1997a). Amino Acid Sequence Homology Search with the corn expressed truncated Cry9C protein sequence
MacIntosh (1997b). Preparation and characterization of catfish pellets.
Metcalfe, et al. (1996). Assessment of the allergenic potential of foods derived form genetically altered crop plants. Crit. Rev. in Food Science and Nut. 36(S)
Noteborn, H. (1998). Assessment of the Stability to Digestion and Bioavailability of the LYS Mutant Cry9C Protein from Bacillus thuringiensis serovar Tolworthi, AgrEvo.
Peferoen, M. (1997a) Determination of Test Substance Equivalence Between Corn Plant Produced Cry9C and Bacterial Produced Cry9C, Plant Genetic Systems
Peferoen, M. (1997b) In vitro Digestibility and Heat Stability of the Endotoxin Cry9C and Bacillus thuringiensis, Plant Genetic Systems.
Peferoen, M. (1997c). Cry9C Bacillus thuring. insecticidal protein identification of sequence homology with allergenicity by searching protein databanks
SAP 2000. "Food Allergenicity of Cry9C Endotoxin and other Non-digestible proteins, FIFRA Scientific Advisory Panel Report No. 2000-01A, June 29, 2000.
Scientific Steering Committee (2000). Risk assessment in a rapidly evolving field: the case of genetically modified plants (GMP). October 26-27, 2000.
Shillito (1999). Determination of the stability of Pat and Cry9C protein in processed grain of transgenic field corn in fractionated agricultural commodities, AgrEvo.