Science in Society Archive

GM Protein in Ice Cream

Genetically modified fish antifreeze protein is potentially able to cause inflammation and should not be approved without comprehensive tests. Prof. Joe Cummins, Dr. Mae-Wan Ho and Prof. Malcolm Hooper.

This report has been submitted to the Food Standards Agency to oppose approval of Unilever’s application on behalf of the Independent Science Panel

Unilever is seeking approval of a genetically modified (GM) (FAQ on genetic engineering) ice-structuring protein derived from a polar fish, ocean pout, for use in making ice cream smoother and creamier. The GM protein is produced in transgenic bakers’ yeast. Ice-structuring, or antifreeze protein protects the ocean pout in freezing waters by preventing large ice crystals forming; in ice cream and other frozen food it would have the same effect. Unilever applied to the Food Standards Agency (FSA) UK for approval, and its proposal is now open for public comment [1].  Unilever has sent similar petitions to the United States Food and Drug Administration (FDA) to obtain the Generally Recognized As Safe (GRAS) status for the food additive [2] and to Food Standards Australia New Zealand [3]. Both applications have been approved, which is unfortunate.

The transgenic protein produced in yeast was designated ISP Type III HPLC 12 glyco–ISP.  The preparation tested by Unilever contained peptides from yeast and sugars along with the recombinant protein. Unilever conducted a subchronic feeding test of the preparation on rats by oral gavage (force feeding) for 3 weeks, as well as a battery of genotoxicity tests that proved to be negative.  A series of tests that included those suggested by the World Health Organisation for allergy were carried out, along with tests for reactivity with serum obtained from a few people allergic to fish. The report stressed that the recombinant protein was identical to protein found in edible fish [1], although that kind of statement is generally untrue as will be discussed below.

There is voluminous literature on antifreeze glycoproteins, particularly those from polar fish. There are four main types of glycoproteins each differing significantly from the others. Type III proteins are around 6500 daltons in size, they form a beta-sandwich structure and are found only in ocean pout [4]. Although the antifreeze protein itself is not immunogenic for the ocean pout, there is nevertheless a strong immune response to the micro ice crystals complex with antifreeze protein circulating in the fish’s blood, indicating that the complex functions as conventional antigens for the ocean pout [5].

The GM protein from transgenic yeast is the product of a synthetic approximation of the pout antifreeze protein gene. The code sequence was altered to facilitate production in yeast without altering the amino acid sequence. Multiple copies of the synthetic gene were inserted into the yeast chromosomes to boost the synthesis of the protein [1].

Production of proteins in yeast destined for human consumption or therapy is fraught with the problem of secondary modification of the proteins by glycosylation or other modifications that result in the human (or animal) immune system recognizing the yeast modified proteins as antigens. There has been progress in “humanizing” the glycosylation patterns of proteins produced in yeast [6, 7]. However, there has been no effort to “humanize” the glycosylation pattern of the antifreeze protein produced in the yeast strain used to produce the protein.

Are the cursory studies on allergenicity carried out by Unilever on the GM protein to be used in ice cream adequate to rule out allergy and other immune reactions in the tens of millions of people that will consume the ice cream?

It is worth pointing out that the transgenic protein is already used in ice cream in the USA, Australia and New Zealand, and that ice cream has not been labeled, so any problems resulting from its use may go unrecognized.

We should recall that the transgenic expression of a bean gene in peas turned it into a strong immunogen, resulting in debilitating even fatal lung inflammation in mice. That response was related to the glycosylation pattern of the transgenic protein [8, 9] (“Transgenic pea that made mice ill" SiS 29). Unilever does not appear to have carried out the inflammation tests even though there is every indication from the scientific literature that pouter antifreeze protein is immunologically active.

There is also the question of latency. Some chronic inflammatory diseases emerge gradually, building up from an initial response that is small and clinically variable or insignificant (asymptomatic) [10]. But there is a potential cascade effect that when triggered, will lead to autoimmune effects that could affect any organ. Without long term testing, we could be letting off an immunological time bomb. Tests for inflammatory effects must be done in both young and older animals will full analysis of inflammatory cytokines, antibodies and related molecules. Tests in young animals are particularly important as ice cream is consumed from the earliest age when there are crucial development processes occurring.

In conclusion, contrary to the claims of Unilever, there is no evidence that the transgenic ice- structuring protein is identical to the protein produced in pouter fish. The transgenic protein appears to have the glycosylation pattern of yeast, making that protein a unique antigen. Even though allergenicity was studied in a cursory way, there is clear precedent for studying inflammation comprehensively in the long term in both young and older animals before exposing the European public to the transgenic ice cream.

Article first published 04/07/06


  1. Lewis S.  Application for the approval of ice structuring protein typ III HPLC12 preparation for use in edible ices. Safety and Assurance Centre 2006
  2. US Food and Drug Administration Agency Response Letter GRAS Notice No. GRN000117 2003,
  3. Food Standards Australia New Zealand  Initial Assessment Report  Application A544  Ice Structuring Protein as a Processing Aid for Ice Cream and Edible Ices, 2004,
  4. Harding MM, Anderberg PI and Haymet AD 'Antifreeze' glycoproteins from polar fish. Eur J Biochem. 2003,  270(7), 1381-92.
  5. Verdier JM, Ewart KV, Griffith M and  Hew CL. An immune response to ice crystals in North Atlantic fishes. Eur J Biochem. 1996, 241(3), 740-3.
  6. Wildt S and Gerngross TU. The humanization of N-glycosylation pathways in yeast.  Nat Rev Microbiol. 2005, 3(2), 119-28.
  7. Li H, Sethuraman N, Stadheim TA, Zha D, Prinz B, Ballew N, Bobrowicz P, Choi BK, Cook WJ, Cukan M, Houston-Cummings NR, Davidson R, Gong B, Hamilton SR, Hoopes JP, Jiang Y, Kim N, Mansfield R, Nett JH, Rios S, Strawbridge R, Wildt S and Gerngross TU. Optimization of humanized IgGs in glycoengineered Pichia pastoris. Nat Biotechnol. 2006, 24(2),  210-5.
  8. Prescott VE, Campbell PM, Moore A, Mattes J, Rothenberg ME, Foster PS, Higgins TJ and Hogan SP. Transgenic expression of bean alpha-amylase inhibitor in peas results in altered structure and immunogenicity. J Agric Food Chem. 2005, 53(23), 9023-30.
  9. Ho MW. Transgenic pea that made mice ill. Science in Society 2006, 29, 28-29,
  10. Arbuckle MR, McClain MT, Rubertone MV, Scofield RH, Dennis GJ, James JA, Harley JB. Development of autoantibodies before the clinical onset of systemic lupus erythematosus. New Eng J Med 2003, 349, 1526-33.

Got something to say about this page? Comment

Comment on this article

Comments may be published. All comments are moderated. Name and email details are required.

Email address:
Your comments:
Anti spam question:
How many legs on a tripod?