Nutritional Data

All plant breeding methods, traditional and modern, have the potential to alter the nutritional value of plants or lead to unexpected or unintended changes in concentrations of various natural toxicants or antinutrients. It is important that all new varieties be evaluated, in order to reduce the likelihood that unexpected changes will produce adverse health effects.

Unintended changes in levels of nutrients can theoretically arise in several ways. Insertion of genetic material could conceivably disrupt or alter the expression of normally expressed plant genes. Expression of the introduced gene – through protein synthesis - might reduce the availability of amino acids used for synthesis of normal plant compounds. Production of normal plant compounds might also be affected if the expressed protein diverted substrates from other important metabolic pathways. Finally, either the expressed protein or altered levels of other proteins might have antinutritional effects. These possible concerns are related to the randomness of DNA insertion. However, changes in gene expression can also occur when traditional breeding methods are used; such changes may be less frequent in transgenic plants since only a limited number of genes are transferred during the genetic modification.

Food safety assessment should consider the potential for any change in nutritional composition, especially in key elements that have a significant impact on the diet, as well as the potential for any change in the bioavailability of key nutritional components. Where additional assurance of safety is sought, analytical methods traditionally applied in the evaluation of food constituents such as total protein, fat, ash, fibre and micronutrients may need to be augmented with additional analyses to identify unexpected effects.

Because of the potential for broad changes in nutrient levels and interactions with other nutrients and unexpected effects, it may be necessary in certain instances to undertake feeding studies in animals to determine outcomes that result from changes in nutrient profiles and nutrient bioavailability. Nutritional modifications which are within the normal range of nutrient variation might require a less extensive evaluation than those outside normal ranges.

For genetically engineered plants that were not developed to have intentionally altered nutritional value, the aim of the nutritional evaluation is to demonstrate that there has been no unintentional changes in the levels of key nutrients, natural toxicants or antinutrients, or the bioavailability of nutrients. In this case, food substitution using products from the genetically engineered plant should not adversely affect the health or nutritional status of the consumer. Implications for the population as a whole and for specific subgroups (e.g., children and the elderly) should be considered.
Nutrition data were obtained from analyses of glyphosate-tolerant and control soybeans (parental variety A5403) grown at nine field locations in 1992. These sites were chosen to be representative of the wide geographical area in which soybeans are grown. In addition, a four-site field test with limited analytical evaluations was performed in 1993. As the emphasis of these analyses was to examine any effects of the introduced gene and protein, the test material was derived from soybeans that had not been treated with glyphosate herbicide.

Although many of the analyses were performed on soybean seed, several soy protein products were also manufactured from GTS 40-3-2 for additional testing. Toasted meal was chosen because it is the main soybean protein product used in animal feed, defatted meal (flour) was prepared because it is the starting material for a large number of soybean products used in food, and protein concentrate from defatted meal was also evaluated because of its food use. In addition, crude lecithin and refined, bleached deodorized oil were manufactured.

Proximate Analysis

Compositional (proximate) analyses were performed on soybean seeds derived from GTS 40-3-2 and the parental non-transgenic control line, A5403. The concentrations of carbohydrate, protein, fat, moisture, fibre, and ash, expressed on a dry-weight basis, were measured according to published procedures of the Association of Official Analytical Chemists (AOAC).

Methods
Ash: Volatile organic matter was driven off when the sample was ignited at 550°C in an electric furnace. The residue was quantitated gravimetrically and calculated to determine percent ash (AOAC method 923.03, 1990). Using a 3 g sample, the lowest confidence level of this method was 0.2%.

Carbohydrates: Carbohydrates were calculated by difference using the fresh weight-derived data and the following equation (USDA Agricultural Handbook No. 8, 1975):


Crude Fibre: Crude fibre is the loss on ignition of dried residue remaining after digestion of the samples with 1.25% sulfuric acid and 1.25% sodium hydroxide solutions under specific conditions (AOAC method 7.066-7.070, 1984). Using a 2 g sample, the lowest confidence level of this method was 0.2%.

Fat: The fat was extracted using ether and hexane. The extract was washed with a dilute alkali solution and filtered through a sodium sulfate column. The remaining extract was evaporated, dried and weighed (AOAC methods 920.39C). Using a 2 g sample, the lowest confidence level of this method was 0.1% fat.

Moisture: The sample was dried to a constant weight in a vacuum oven at 133°C (approximately 2 hours) (AOAC method 44-15A, 1987). The moisture loss was determined gravimetrically.

Protein: Protein and other organic nitrogen in the sample were converted to ammonium sulfate by digesting the sample with sulfuric acid containing a potassium sulfate/titanium dioxide/cupric sulfate catalyst mixture. The acid digest was made alkaline, and the ammonia was distilled and titrated with standard acid. The percent nitrogen was determined and converted to protein using the factor 6.25 (AOAC method 988.05, 1990). Using a 1 g sample, the lowest confidence level of this method was 0.1% protein (0.02% nitrogen).

Results
Compositional analyses of protein, fat, fibre, ash, and carbohydrate of GTS 40-3-2 and control soybean seeds obtained from nine field trial sites in 1992 and four trial sites in 1993 are presented in Figures 8.2 and 8.3, respectively. For each of the components measured, there were no statistically significant differences between GTS 40-3-2 and control soybeans, and with the exception of total carbohydrate, the measured values were within the range reported in the scientific literature. For the nine-site study, the mean GTS 40-3-2 seed carbohydrate content was 37.1% dry weight, compared to a literature high of 34%. This difference was not judged as significant from a safety perspective as the mean carbohydrate concentration measured in control soybeans harvested from the same sites was 38.1% dry weight.

Similar analyses performed on samples of toasted (Fig. 8.4) and non-toasted meal, and protein concentrate prepared from GTS 40-3-2 and control non-transgenic soybeans did not reveal any appreciable differences in the levels of macronutrients.


Methods
Seed samples were subjected to acid hydrolysis using 6N HCl, then adjusted to pH 2.2 and the individual amino acids were quantitated using an automated amino acid analyzer equipped with post-column ninhydrin derivatization and colorimetric detection (Moore & Stein 1954).

Results
For the 18 amino acids measured, there were no statistically significant differences in the levels of any amino acid, including aromatic amino acids, between GTS 40-3-2 seeds and control non-transgenic soybean seeds (Fig. 8.5).

The shikimate pathway plays a central role in plant metabolism and it has been estimated that about one-fifth of the carbon fixed by plants is subsequently channelled through this pathway (Haslam 1993). The lack of any difference in the levels of aromatic amino acids between transgenic GTS soybean seeds and non-transgenic seeds is supported by the fact that all available evidence suggests that EPSPS is not a rate-limiting step in the shikimate pathway, but that regulation of this pathway occurs at the first step in the conversion of erythrose 4-phosphate to 2-keto-3-deoxy-D-arabinoheptulosonate 7-phospate (DAHP) by DAPH synthase (Weiss & Edwards 1980). Increased EPSPS activity would not, therefore, be expected to increase the levels of aromatic compounds in plants, and it has been observed that plant cells expressing 40-times more EPSPS than wild-type cultures do not overproduce aromatic amino acids (Smart et al. 1985).

Fatty Acid Composition

Methods
Samples of soybean seed or refined soybean oil were extracted with chloroform/methanol, saponified with alcoholic potassium hydroxide, and the free fatty acids were then extracted with hexane, washed with water and dried with sodium sulfate. Fatty acids were esterified with methanol, using boron trifluoride as a catalyst, taken up in heptane and subjected to gas chromatographic analysis (AOAC method 983.23 1990). The percent abundance of individual fatty acid methyl esters was calculated relative to the total amount of fatty acid methyl esters present. The lowest confidence level of this method was 0.1% of an individual fatty acid methyl ester.

Results
The relative abundances of individual fatty acids were determined for samples of soybean seed and refined, bleached, deodorized oil derived from GTS 40-3-2 and control non-transgenic soybeans (Fig. 8.6). There was only one statistically significant difference in the seed fatty acid composition between GTS 40-3-2 and control soybeans; this was for C22:0 fatty acids, which represent less than 0.6% of the total fatty acid fraction. All values, even those for C22:0 from seeds, were within the normal range of values for each respective fatty acid as reported in the literature.

Lecithin, which is a phosphatide removed from crude soybean oil, is used as a natural emulsifier, lubricant, and stabilizing agent (Waggle & Kolar, 1979). In addition to analysis of the free fatty acid profile of refined, bleached, deodorized soybean oil prepared from GTS 40-3-2 and non-transgenic soybeans, these oil samples were used to prepare crude lecithin fractions that were analyzed for phosphatide composition (phosphatidyl ethanolamine, phostidic acid, phosphatidyl inositol, phosphatidyl choline) (AOAC method Ja 7b-91). The relative abundance of each of these phosphatide components was comparable between crude lecithin fractions prepared from GTS 40-3-3 soybean oil and control non-transgenic soybean oil (Fig. 8.7).


The profiles of seed storage proteins extracted from GTS 40-3-2 and control non-transgenic soybean seeds were compared by sodium dodecylsulfate (SDS) polyacrylamide gel electrophoresis (PAGE). There were no discernable differences between transgenic and control soybeans (Fig. 8.8), which indicates that the gross protein compositions of GTS 40-3-2 seeds are not materially different from that of the control soybeans.



Soybean is naturally a source of several compounds that have been associated with antinutritive effects. These include protease inhibitors, such as soybean trypsin inhibitor, lectins (e.g., soybean hemagglutinin), isoflavones, and phytate, which complexes with inorganic phosphorous in seed but can also sequester other metallic ions such as iron, calcium, zinc, and magnesium, rendering these elements nutritionally unavailable. The levels of these antinutrient factors were determined in samples of GTS 40-3-2 soybean seed, as well as toasted soybean meal used for livestock feed, and compared with the levels found in the parental non-transgenic soybean line.

Trypsin Inhibitors

The antinutritive effect of trypsin inhibitors in unheated soybean products has been the subject of much research (Rackis et al, 1986). The destruction of trypsin inhibitors and consequent elimination of hypertrophic pancreas effects is an important step in the processing of raw soybeans into products with excellent protein quality (Anderson et al. 1979).

Trypsin inhibitory activity was measured on alkaline (pH 9.5 - 9.8) extracts of raw soybean seed, or toasted meal, by incubation with a known concentration of trypsin, followed by the addition of benzoyl-D-arginine-p-nitroanilide (BAPNA). Measurements of the absorbance at 410 nm were taken after 10 minutes of reaction. Uninhibited trypsin catalyzes the hydrolysis of BAPNA, forming a yellow-coloured p-nitroaniline. One trypsin unit was defined as an increase equal to 0.01 absorbance units at 410 after 10 minutes per 10 ml reaction volume. The lowest confidence level of this method was 1 trypsin inhibitor unit (TIU) / mg sample, using a 1 g sample.

In comparing extracts of raw soybean seeds from GTS 40-3-2 and non-transgenic control lines (Fig. 8.9), there were no statistically significant differences in trypsin inhibitor activity. The normal processing of soybean meal to produce toasted meal results in a greater than 90% elimination of trypsin inhibitor activity from both GTS 40-3-2 and control material (Fig. 8.9).

Lectin Analysis

Plant lectins are a class of proteins with specific binding affinities for carbohydrate containing glycoproteins that are usually present in plant cell walls and the plasma membrane of cells. The binding of lectins to cell surface glycoproteins may cause agglutination, mitosis, or other biochemical changes in the cell. The ingestion of lectins, such as soybean hemagglutinin, has been associated with a range of antinutritive effects and some disease pathologies. Soybean lectin has been quoted as being responsible for about 25% of the growth inhibition attributable to the ingestion of raw soybean meal by rats (Leiner 1953), although it has since been concluded by some that soybean agglutinin does not play any major role as a determinant of the nutritional quality of soybean protein (Leiner 1980). Other authors still believe that circumstantial evidence exists that soybean lectin may make an appreciable contribution to observed growth inhibition caused by dietary exposure to uncooked soybean meal (Pusztai 1989).

The levels of soybean lectin in raw and toasted soybean meal were estimated by measuring the hemagglutination activity of various extracts against rabbit red blood cells (Leiner, 1955; Klurfeld & Dritchevski, 1987). There were no statistically significant differences in the lectin activity between GTS 40-3-2 and control non-transgenic soybeans. The level of hemagglutination activity in raw soybean meal was less than 7 hemagglutination units (HU) / mg protein and essentially undetectable in samples of toasted meal (Fig. 8.10). A comparison of the hemagglutinin activity observed for raw meal in these tests with previously published values of 60-426 HU / mg protein was not informative due to the variability in red cell lots. The sensitivity of the assay was established in positive control tests with purified soybean lectin, in which values of 461-541 HU / mg protein were measured.


The isoflavones genistein, daidzein, and coumestrol are naturally present in soybeans and their ingestion has been linked to a number of biochemical effects in mammalian species, including estrogenic and hypocholesterolemic activities (Wang et al. 1990; Murphy 1982). They have also been reported to contribute to deleterious effects on livestock animals fed soybean meal (Setchell et al. 1987).

The bound and free forms of daidzein and genistein were determined in samples of raw and toasted soybean meal by high pressure liquid chromatography (HPLC) separation (Pettersson & Kiessling, 1984). Sample extracts, and extracts following acid hydrolysis to liberate bound isoflavones, were analyzed to calculate the concentrations of free and total isoflavones, respectively. Concentrations of bound isoflavones were calculated as the difference of these two values.

No statistically significant differences in the levels of any isoflavones measured in either raw or toasted soybean meal were detected between GTS 40-3-2 and non-transgenic control soybeans (Fig. 8.11). The large variability observed in values determined for seeds harvested from different field trial sites was attributed to the effect of environmental variability on the formation of these compounds in plants.

Stachyose, Raffinose, and Phytate Analysis of Soybean Meal

The low molecular weight carbohydrates, stachyose and raffinose, are primarily responsible for flatus activity, which is a well known characteristic of soybean products (Rackis 1976). Phytic acid (phytate) is a hexaphosphoric acid derivative of inositol, and exists mainly in soybean seeds as an insoluble, non-nutritionally available calcium-magnesium-potassium complex (Mohamed et al. 1991). Phytate is not broken down in monogastric animals (e.g., poultry, fish, swine) and is the main reason that livestock feeds for these animals must be supplemented with additional phosphorus and other minerals, or with phytase enzyme to degrade phytate.

The levels of stachyose and raffinose in extracts prepared from toasted soybean meal were determined by HPLC (Dunmire & Otto 1979). Phytic acid was extracted with dilute HCl and separated from inorganic phosphates by anion exchange chromatography (Ellis & Morris 1983). Bound phytate was eluted with NaCl solution and digested with a mixture of sulfuric and nitric acid to liberate free phosphate, which was quantitated spectrophotometrically following reaction with ammonium molybdate and sulfonic acid. Values were converted to phytic acid based on molecular weight equivalence and the lowest confidence level of the assay was 0.028% phytic acid based on a 2 g sample.

There were no statistically significant differences in the respective levels of stachyose, raffinose, or phytate measured in samples of toasted meal prepared from GTS 40-3-2 or non-transgenic control soybeans (Fig. 8.12).

Nutrient Bioavailability - Confirmatory Animal Feeding Studies

In order to establish that the genetic modification resulting in GTS 40-3-2 did not adversely affect the wholesomeness (ability to support typical growth and well-being) of soybean products, animal feeding studies were performed with laboratory rats, broiler chickens, catfish, and dairy cows. Both processed and unprocessed soybean meal was tested on rats because the majority of soybeans used for human food and animal feed are processed by heat treatment, and because rats serve as a surrogate for wild mammals that may eat soybeans in the field. Poultry consume about 49% of the soybeans fed to farm animals and were the subject of a six-week growth study, and dairy cows were included in a four week study since ruminants are normally fed raw soybeans as a source of protein. The catfish study was included since soybean meal is used in diets for commercial aquaculture. Lastly, unprocessed soybean meal was fed for 5 days to bobwhite quail, since birds may feed on soybeans left in the field after harvest.

Methods
Rat Four-Week Feeding Study: Eight week old male and female Charles River CD rats were fed rodent chow containing either processed or unprocessed soybean meal from GTS 40-3-2 or control non-transgenic soybeans for four weeks, ad libitum, at substitution levels of 24.8% or up to 10%, respectively. Feed consumption and body weight were measured at weekly intervals, and rats were observed twice daily for mortality and adverse clinical signs. At the end of the study, all test animals were sacrificed and necropsied. Liver, testes, and kidneys were weighed and approximately 40 tissues were collected and saved from each animal. Dunnett’s multiple range comparison test (two-tailed) was used to compare inlife body weights, cumulative body weight gain and food consumption for test and control groups. Terminal body weights, absolute organ weights, and organ/body weight ratios were evaluated by decision-tree statistical analysis procedures to detect group differences and analyze for trends.

Broiler Chicken Six-Week Study: Commercial broiler chicks (White Plymouth Rock x White Cornish; Cobb 500 cockerel x Cobb 500 pullet) were fed test diets containing processed meal from GTS 40-3-2 or the control parental non-transgenic A5403 soybeans, supplemented with corn meal as the only other source of protein. Diets were formulated so as to ensure approximately equal amounts of essential amino acids (methionine, cysteine, lysine, arginine, tryptophan, and threonine), did not contain any medications or growth promoting feed additives, and met the National Research Council requirements for poultry feed. Birds were checked daily for mortality, and any that died on test were removed, weighed and necropsied to determined probable cause of death. Body weights and food consumption were measured, and at the termination of the study, birds were sacrificed and major and minor pectoralis muscles (breast muscles) from the right side were dissected and weighed. Abdominal fat pads were also removed and weighed.

Dairy Cow Four-Week Study: Thirty-six multiparous Holstein dairy cows (93-196 days of lactation) were fed a mixed diet ration (35% alfalfa hay, 17% corn silage, 37% commercial grain mix) containing 10% (w/w dry matter basis) raw soybeans from GTS 40-3-2 or control non-transgenic A5403 soybean lines. This dietary level represented the upper limit for incorporation of raw soybeans into mixed cow diets as fed by dairy farmers, and cows were pre-adapted to high soybean diets prior to the start of the study. Milk samples collected daily during the course of the study were analyzed for lactose, fat, protein, and somatic cells. Total urine and fecal output was collected daily during the last week of the study to determine dry matter digestibility and nitrogen balance.

Catfish Ten-Week Study: Fingerling channel catfish (Ictalurus punctatus), Mississippi Select strain, were maintained for 10 weeks in glass aquaria and reared on a diet containing soybean meal from GTS 40-3-2 or control non-transgenic soybeans at the same substitution levels used commercially (45-47% w/w). All diets were prepared to contain a final protein concentration of 32%. Fish were weighed at the beginning of the study and on weeks 2, 6, and 10, at which times feed consumption was quantified by subtracting the weight of uneaten pellets removed from the bottoms of tanks from the quantity of feed administered. The cumulative feed conversion ratio was estimated at weeks 2, 6, and 10 by dividing the sum of the feed offered to that point by corresponding total weight gain, adjusting for mortalities. At the end of the study, several fish were selected at random and the edible tissue composited and subjected to proximate analysis.

Results
The feed efficiencies (feed conversion ratios) of both GTS 40-3-2 and non-transgenic control soybeans, when used as components of animal feed, were summarized and compared across studies (Table 8.1). The bobwhite quail study was not included in this comparison because of its short duration (5 days). No statistically significant differences in feed efficiencies were observed when GTS 40-3-2 was used as a feed source compared to the parental variety, A5403. These results were consistent with the extensive compositional analyses demonstrating that GTS 40-3-2 was not significantly different from the control soybeans in terms of its nutritional properties.

References

1. Anderson, R.L., Rackis, J.J. & Tallent, W.H. (1979). Biologically active substances in soy products. In: Soy Protein and Human Nutrition. H.L. Wilke, D.T. Hopkins & D.H. Waggle, eds. Academic Press, New York. pp. 209-233.


2. AOAC Method 44-15A. (1987). Moisture-air-oven methods. In AOAC Official Methods of Analysis, Association of Official Analytical Chemists, Arlington, Virginia.


3. AOAC Method 7.066-7.070. (1984). Fiber (crude) in animal feed, ceramic fiber filter method. In AOAC Official Methods of Analysis, Association of Official Analytical Chemists, Arlington, Virginia.


4. AOAC Method 920.39C. (1990). Fat (crude) of ether extract in animal feed. In AOAC Official Methods of Analysis, Association of Official Analytical Chemists, Arlington, Virginia.


5. AOAC Method 923.03. (1990). In Official Methods of Analysis, 15th Edition. Association of Official Analytical Chemists, Arlington, Virginia, modified.


6. AOAC Method 983.23. (1990). Fat in foods, chloroform-methanol extraction. In Official Methods of Analysis, 15th Edition. Association of Official Analytical Chemists, Arlington, Virginia.


7. AOAC Method 988.05. (1990). Protein (crude) in animal feed, CuSO4/TiO2 mixed catalyst Kjeldahl method. In AOAC Official Methods of Analysis, Association of Official Analytical Chemists, Arlington, Virginia.


8. AOAC Method Ja 7b-91 (1993). In Official Methods and Recommended Practices of the Association of Official Analytical Chemists, 4th Edition, Arlington, Virginia.


9. Dunmire, D. & Otto, S. (1979). HPLC determination of sugars in various food products. Journal of the Official Association of Analytical Chemists 62, 176-185.


10. Ellis, R. & Morris, E.R. (1983). Improved ion-exchange phytate method. Cereal Chem. 60, 121-124.


11. Haslam, E. (1993). Shikimic acid: Metabolism and metabolites. John Wiley & Sons, Chichester, England.


12. Klurfeld, D.M. & Kritchevski, D. (1987). Isolation and quantitation of lectins from vegetable oils. Lipids 22, 667-668.


13. Leiner, I.E. (1953). Soyin, a toxic protein from the soybean. I. Inhibition of rat growth. J. Nutr. 49, 527-540.


14. Leiner, I.E. (1955). The photometric determination of the hemagglutinating activity of soyin and crude soybean extracts. Arch. Biochem. Biophys. 54, 223-231.


15. Leiner, I.E. (1980). Anti-nutritional factors as determinants of soybean quality. In: World Soybean Research Conference II: Proceedings. F. Corbin, ed. Westview Press, Boulder, CO. pp. 703-712.


16. Mohamed, A.I., Mebrahtu, T. & Rangappa, M. (1991). Nutrient composition and anti-nutritional factors in selected vegetable soybean (Glycine max L.). Plant Foods Hum. Nutr. 41, 89-100.


17. Moore, S. & Stein, W.H. (1954). Procedures for the chromatographic determination of amino acids on four per cent cross-linked sulfonated polystyrene resins. Journal of Biological Chemistry 211, 893-906.


18. Murphy, P.A. (1982). Phytoestrogen content of processed soybean products. Food Technology 36, 60-64.


19. Pettersson, H. & Kiessling, K.H. (1984). Liquid chromatographic determination of the plant estrogens coumestrol and isoflavones in animal feed. Journal of the Official Association of Analytical Chemists 67, 503-506.


20. Pusztai, A. (1989). Lectins. In: Toxicants of Plant Origin: Volume II, Proteins and Amino Acids. P.R. Cheeke, ed. CRC Press, Boca Raton, FL. pp. 29-71.


21. Rackis, J.J. (1976). Flatulence problems associated with soy products. In: World Soybean Research. L.D. Hill, ed. The Interstate Printers and Publishers, Inc., Danville, IL. pp. 892-903.


22. Rackis, J.J., Wolf, W.J. & Baker, E.C. (1986). Protease inhibitors in plant foods: Content and inactivation. In: Nutritional and Toxicological Significance of Enzyme Inhibitors in Food. M. Friedman, ed. Plenum Press, New York. pp. 299-347.


23. Setchell, K.D.R., Gosselin, S.J., Welsh, M.B., Johnston, J.O., Balistreri, W.F., Kramer, L.W., Dresser, B.L. & Tarr, M.J. (1987). Dietary estrogens - a probable cause of infertility and liver disease in captive cheetahs. Gastroenterology 93, 225-233.


24. Smart, C.C., Johanning, D., Muller, G. & Amrhein, N. (1985). Selective overproduction of 5-enol-pyruvylshikimate acid 3-phosphate synthase in a plant cell culture which tolerates high doses of the herbicide glyphosate. Journal of Biological Chemistry 260, 16338-16346.


25. USDA Agriculture Handbook No. 8. (1975). Composition of Foods. In Agricultural Handbook No. 8. United States Department of Agriculture, Washington, D.C. pp. 159-165.


26. Waggle, D.H. & Kolar, C.W. (1979). Types of soy protein products. In: Soy Protein and Human Nutrition. H.L. Wilke, D.T. Hopkins, and D.H. Waggle, eds. Academic Press, New York. pp. 19-51.


27. Wang, G., Kuan, S.S., Francis, O.J., Ware, G.M. & Carman, A.S. (1990). A simplified HPLC method for the determination of phytoestrogens in soybean and its processed products. J. Agric. Food. Chem. 38, 185-190.


28. Weiss, U. & Edwards, J.M. (1980). Regulation of the shikimate pathway. In: The Biosynthesis of Aromatic Compounds. John Wiley & Sons, New York, pp. 287-301.