Collapse of the “Life Science” Corporations

In recent months, major pharmaceutical companies have been divesting the agricultural components of their corporations. Originally, these companies wished to build enormous “Life Science” corporations. It was their hope that the use of innovative technologies implemented in the pharmaceutical industry would produce “major cost savings and a host of promising new products” if applied to agriculture¹. But Pharmacia, Upjohn, Novartis, and AstraZeneca have either sold or “spun off” their respective agricultural businesses since the beginning of the fourth quarter this year¹. Aventis, in the news recently for the fiasco surrounding its Starlink corn, is the last to sell off its Crop Sciences division. We will look specifically at Aventis’ Starlink corn to prove that, while a host of excuses have been proposed for failure of the Life Science model, one cannot deny that the short-comings of the previously mentioned technology, originally anticipated to be so helpful, played a part as well.

Currently, there are nine different stages that a new product like Starlink corn must pass. The NIH, USDA, EPA, and FDA all must approve the product². The most intense screening of Starlink corn occurred at the EPA level because of the product’s nature. Starlink corn is modified to contain the Cry9C protein, isolated from Bacillus thuringiensis (Bt). Cry9C is a pesticide that, like all Bt proteins, targets the gut of insects, while claiming to be harmless to humans³. Of the above agencies, the EPA is given jurisdiction over pesticide-containing products⁴. Thus before the product is approved for humans, the EPA must determine what it calls the “Human Health Assessment,” which includes a host of factors: toxicology, allergenicity, and effects on the endocrine system, just to name a few⁵. To prove our point, we will look specifically at allergenicity.

The ability to measure allergenicity of any protein is at the forefront of this debate. As a panelist notes in a recent discussion “[T]here are no animal models that are appropriate and effective for assessing allergenicity [in humans]”⁶. As another panelist notes, “It's very difficult to assess something that's going to become an allergen. We just don't know until it does. People are very individualistic”⁷. While Aventis acknowledges the validity of both of these points, they are quick to point out that “The absence of an animal model test system does not mean that other scientific tools cannot be used to predict human response”⁸. The nature of a food allergy makes it much more difficult to screen for than one to a drug. In the case of foods, there is no large-scale antibody test that would be worthwhile because, as Aventis also acknowledges, antibodies are not involved in one of the most common form of food allergies: food intolerances⁸. Many organizations attempted to implement some of the techniques we have discussed in class thus far to overcome the shortcomings of the formerly mentioned procedures.

A measure of allergenicity is currently attained through two main laboratory procedures. First, the protein is screened for resistance to degradation by acid, heat, and proteases; characteristics which all allergens tend to possess. The second major component is a search for homology to known toxins or allergens. This is where our techniques are implemented. The EPA first reported on Cry9C’s “Amino Acid Homology Comparison”⁹. In this study, Cry9C’s 626 amino acid sequence was compared to all sequences in the PIR, SWISS Prot, and HIV AA databases. The comparison was carried out using the FAST program, specifically in which matches were scored as +1, mismatches and gaps scored –1, and gap size was weighted by adding (0.05 X “gap size”) to the penalty. The results of the study showed that Cry9C had homology with 300 sequences, the most significant of which were to other delta-endotoxins in the Bt protein family. Thus no unexplained homology was found. Yet even if it were, it would be difficult to classify because the agency “has not yet determined how to judge amino acid sequence homology for risk assessment purposes”⁹.

A similar study was performed at a later date in an attempt to screen the Cry9C protein more closely¹⁰. In this study, a DNA sequence of Cry9C was analyzed as a stepwise series of “overlapping eight amino acids”using the FAST program to screen the SWISS protein database. The rationale was based on the idea that “the optimal number of amino acids needed to elicit an immunological response appears to be between 8 and 12”¹¹. But again, these results only showed homology with other Bt crystal proteins. And the EPA was again unable to classify this protein through this technique, since risk assessment cannot accurately be judged by sequence comparison.

The EPA restricted the use of Starlink to that of animal feed, deciding that the possibility of an allergic reaction in a human is too high if directly consumed. Yet some of our most innovative and promising techniques were unable to detect this. Corporations and the EPA had hoped that these tests would lead to a better understanding of allergenicity. However, these tests are flawed. They are dependent on databases that are limited in size. And even if the protein has a significant homology to a sequence, that sequence may or may not be known to elicit an allergic response in humans. This problem exists, on top of the fact that the EPA openly admits it has no idea how to relate the amino acid sequence to the protein’s relative risk of allergenicity.

An obvious solution would be to use a larger database, if one currently existed. But databases can only grow as fast as novel proteins are discovered and/or determined to be allergenic. Perhaps a more immediate solution to the problem would be comparing the actual structure of the protein with structures of known allergenic proteins. Pure sequence-homology screening obviously failed to yield the specific evidence we searched for. Yet as a recent article points out, the “data-driven revolution” has progressed beyond sequences to “high-resolution three-dimensional structural analysis of proteins”¹². As we discussed in class, structural comparison can detect almost twice as many relationships between proteins as that of sequence comparison can¹³. We specifically discussed doing a structural comparison using Alignment of Distance Matrices techniques. The residue-residue distance matrices of a particular protein are calculated and then compared to that of other proteins. The result yields “a powerful, flexible method for the detection of spatial similarities in protein structures”¹⁴. Perhaps the implementation of techniques like this in the screening process could benefit the independent Crop Science corporations, as well as the regulatory agencies involved; creating a more precise, efficient, and cost-effective system.

Author: Thomas J. Gioia, MB&B 452a, 12/15/00

References

¹Capell, Kerry. “An About-Face at Aventis.” Business Week. November 27, 2000. pp. 168

²From “Regulation of Agricultural Biotechnology in the United States: How the Process Works.” Available

at http://www.icfcs.org/biotechreg.htm

³From “The Environmental Protection Agency’s White Paper on Bt Plant-pesticide Resistance

Management.” May, 1998. Available at http://www.epa.gov/oppbppd1/biopesticides/white_bt.pdf

⁴From “United States Department of Health and Human Services Food and Drug Administration Public

Meeting: Biotechnology in the Year 2000 and Beyond.” Federal Building, Oakland CA. Dec. 13,

1999. Transcript available at http://www.fda.gov/ohrms/dockets/dockets/99n4282/tr00003.rtf

⁵From the EPA’s “Bt Plant-Pesticides Biopesticides Registration Action Document,” Science Assessment

Section, Sub-Sections IIA and IIB. Available at

http://www.epa.gov/oscpmont/sap/2000/october/brad2_scienceassessment.pdf

⁶Fagan, John, Ph. D., Chairman and Chief Scientific Officer, Genetic ID, at “United States Department of

Health and Human Services Food and Drug Administration Public Meeting: Biotechnology in the

Year 2000 and Beyond.” Transcript page 91. (See above)

⁷Hefle, Susan L., Ph. D., Research Assistant Professor and Co-Director, Food Allergy Research and

Resource Program, University of Nebraska, Lincoln, at “United States Department of Health and

Human Services Food and Drug Administration Public Meeting: Biotechnology in the Year 2000

and Beyond.” Transcript pages 113-114. (See above)

⁸From General Information About Allergenicity, Available at http://www.us.cropscience.aventis.com

⁹From EPA Data Evaluation Report, MRID # 44258109, “Amino Acid Homology Comparison.” January 6,

1998. Available at http://www.epa.gov/oppbppd1/biopesticides/cry9c/der-44258109a.htm

¹⁰From EPA Data Evaluation Report, MRID # 44384404, “Food Allergenicity: Amino Acid Sequence

Homology.” April 8, 1998. Available at http://www.epa.gov/oppbppd1/biopesticides/cry9c/der-44384404a.htm

¹¹From EPA Data Evaluation Report, MRID # 44714001, “Safety Assessment.” July 15, 1999. Available at

http://www.epa.gov/oppbppd1/biopesticides/cry9c/der-44714001a.htm

¹²Wada, Akiyoshi. “Bioinformatics-The Necessity of the Quest for ‘First Principles’ in Life.”

Bioinformatics, Vol. 16, no. 8, 2000. pp. 663-4.

¹³Levitt, Michael and Mark Gerstein. “A Unified Statistical Framework for Sequence Comparison and

Structure Comparison.” Proc. Natl. Acad. Sci USA, Vol. 95, May, 1998. pp. 5913-5920.

¹⁴Holm, Lisa and Chris Sander. “Protein Structure Comparison by Alignment of Distances Matrices.”

Journal of Molecular Biology, Vol. 233, 1993. pp. 123-138.