Index | Search | Home | Table of Contents

Nielsen, N.C. 1990. Modification of protein content in soybean to improve seed quality. p. 106-113. In: J. Janick and J.E. Simon (eds.), Advances in new crops. Timber Press, Portland, OR.

Modification of Protein Content in Soybean to Improve Seed Quality*

Niels C. Nielsen

  6. Table 1
  7. Fig. 1
  8. Fig. 2
  9. Fig. 3
  10. Fig. 4
  11. Fig. 5


When discussing new crops, it is tempting to consider only plants that either are not domesticated or which are grown in other parts of the world but are not adapted for growth in our cultural system. In these two cases the new crop has a characteristic that makes it economically desirable, and strategies for domesticating and adapting the new crop to our production schemes are of over-riding concern. Often ignored during discussions about new crops, however, is the potential of conventional crops that have been modified so that they manifest unique new properties. In contrast to unadapted crops, cultural practices used with modified conventional crops are already well established, and their acceptance is likely to be influenced more by market demand for the new properties than by cultural factors. Thus, modification of conventional crops to produce new ones can easily be justified, and from an economic standpoint may be more cost effective to produce.

Biotechnology, much of it still emerging, will provide the instrument required to restructure conventional crops so that they have new properties. Two obvious approaches will enable these endeavors. In one, biotechnology will be used to recognize and utilize genetic variability that already exists within the germplasm pool for the crop. In reality, this is no different than practices already applied by plant breeders, but an increased level of precision can be attained by screening for specific properties using the emerging techniques of biotechnology. In the second case, biotechnology can be used to generate genetic variability not previously available in the gene pool. This can be accomplished either by moving unique genes from one plant to another, or by engineering genes such that they confer new properties to the plant. I will use examples of research taking place in my laboratory to illustrate these two approaches.


It is well known that soybeans [Glycine max (L.) Merr.] have a strong legume taste that is frequently described as a rancid, or grassy-beany flavor. These flavors have been a major impediment to the acceptance of full fat soy protein-based foods, and their reduction or elimination should open the way to new markets for edible soy proteins. The off-flavors arise by oxidation of polyunsaturated fatty acids (Dutton et al. 1951), and lipoxygenases have long been implicated in this process (Wolf 1975; Rackis et al. 1979). The lipoxygenases catalyze hydroperoxidation of polyunsaturated fatty acids that contain 1,3 pentadienes. In soybeans, more than 50% of the fatty acids in the oil are either linoleic or linolenic acid, two fatty acids with such pentadiene structures. The hydroperoxy-fatty acids produced by lipoxygenases subsequently break down to yield short chain aldehydes that are considered responsible for off-flavors. Hexanal and hexenal appear to be the principal source of the beany flavor attribute in soybeans. Since the lipoxygenases account for over 1 % of the total seed protein, a high probability exists that off-flavor compounds will be generated when the enzymes and fatty acids come in contact with one another.

Most cultivated soybean cultivars contain three lipoxygenase isozymes that have been named L1, L2, and L3 (Axelrod et al. 1981). They are visualized as a triplet of protein bands of high molecular weight by electrophoresis in SDS polyacrylamide gels (Fig. 1). L3 is the uppermost member of the triplet, L2 is in the middle and Ll is at the bottom. As can also be seen in Fig. 1, null-alleles have been located for each isozyme. The null-allele for Ll was located by Hildebrand and Hymowitz (1982), whereas we reported the existence of the null-alleles for L2 and L3 (Kitamura et al. 1983; Davies and Nielsen 1986). The genes that encode the three enzymes have been assigned the gene symbols Lxl, Lx2 and Lx3, respectively.

Our initial objective was to combine the three null-alleles. To accomplish this, we carried out genetic studies to identify the linkage relationships among the three lipoxygenase genes. Our studies showed that each of the three isozymes was encoded by a different gene and that the three genes were distributed among two genetic loci. The genes encoding Ll and L2 were tightly linked to one another, but alleles for L3 segregated independently from those that encode Ll and L2. The tight linkage between Lxl and Lx2 has precluded obtaining a triple mutant that lacks all three isozymes, although double mutants that lack either Ll + L3 or L2 + L3 can be obtained (Davies and Nielsen 1986a). The other important point that emerged from the genetic studies was that a yield reduction was not associated with removal of the lipoxygenase null alleles. To address this question, near isogenic lines were produced by introducing each of the null-alleles into a 'Century genetic background by a backcross breeding technique. Table 1 shows that seed yields of backcrossed lines that contained each of the lipoxygenase null alleles did not differ from one another significantly.

We tested the possibility that the three lipoxygenase isozymes made unequal contributions in the generation of off-flavors. The near isogenic lines were submitted to the Sensory Evaluation Laboratory at Kansas State University where full fat soy flour and soymilk were prepared for flavor assessment. A number of flavor and aroma attributes were rated in double blind experiments. These included the beany, rancid and oily attributes, as well as traits for cereal and milk. Details concerning the measurement of these traits are published elsewhere (Davies et al. 1987), but an example of the data obtained has been presented in Fig. 2. The beany flavor of soymilk prepared from each of the isolines was ranked relative to a lima bean extract. The L2-less isoline resulted in a reduction of the rating of the beany flavor from very strong to mild compared with the 'Century' control. Statistically significant differences between the control and other isolines were not observed. Fig. 3 shows that the opposite response was observed when soymilk samples were rated for milk flavor. In this case soymilk prepared from the L2-less isoline was judged to have more of a dairy-like milk flavor than the soymilk prepared from the 'Century' control. Thus, our data clearly establish that the L2 isozyme make a major contribution of off-flavors associated with soybean products.

The extent to which the reduction of beany off-flavor will be of value in the food industry remains to be determined. However, in cooperation with Central Soya, Inc., Ft. Wayne, Indiana, a marketing experiment is being conducted to evaluate the acceptance of these beans by the Japanese tofu industry. Small amounts of the L2-less beans were shipped to Japan and evaluated during Spring 1988, with favorable results. During winter 1988/89 approximately 150 t of L2-less beans were sent to Japanese tofu manufacturers for more extensive testing. While involvement of the lipoxygenase-less beans in the tofu market seems obvious, their use can be extended to other products as well. For example, soy proteins have traditionally been used as extenders in numerous dairy and meat products. Soymilk is probably the most rapidly expanding market for soy products world-wide. It is anticipated that milder tasting soybeans will expand into these other markets as the volume of lipoxygenase-less seeds increases.

The establishment of soybeans lacking lipoxygenase in the market will not be without difficulty. The null-alleles which condition loss of the enzymes are recessive to the alleles in cultivated crops which contain the enzymes. In the near-term at least, it is likely that the beans will be grown under contract by farmer cooperators, and then maintained in an identity preserved fashion following harvest and during shipment to processors. This implies that mechanisms must be developed to certify that specific cultivars by farmers are genetically pure. Steps such as these that must be made in order to ensure quality control will add expense to the cost of production, and these must be recovered by value added sale of products. In many respects the problems encountered during development of the beans, and the protocols developed for identity preservation of soybean cultivars with the lipoxygenase null-alleles, will be a model for other value added specialty crops.


The second type of approach that can be used to produce new crops involves the direct manipulation of genes whose products confer unique new properties to the plant. Several examples of this type of approach to crop improvement are already being evaluated for commercial applications. For example, when genes for the coat protein of RNA viruses are introduced into plant genomes, they confer protection against secondary infection by similar virus particles (Nelson et al. 1988). Genes that encode insect specific toxins from Bacillus thuringiensis are being used to protect transgenic tomatoes against hornworm attack (Fishhoff et al. 1988). Transgenic tomato and soybean plants have also been described which contain genes for glyphosate herbicide resistance (Fillatti et al. 1987; Hinchee et al. 1988). In a number of laboratories including ours, methods to improve the nutritional quality of seed proteins by genetic engineering are being evaluated. While not as close to commercial application as some of the previously mentioned projects, the approaches we have followed will illustrate some of the problems that must be solved in development of new crops with improved nutritional quality.

It has been recognized for some rime that proteins in seeds from the commonly grown crop species do not contain a nutritionally balanced amino acid content. The seed proteins from cereals are generally deficient in the amino acids lysine and tryptophan, whereas those from legumes are deficient in the sulfur amino acids methionine and cystine. Because of these deficiencies, studies in many laboratories have focused on the genes that encode seed storage proteins (reviewed by Casey et al. 1986; Higgins 1984; Shotwell and Larkins 1989). My laboratory has been interested in glycinin, the more prevalent of the two storage proteins in soybeans. In the remainder of this paper, I will review briefly the rational we have been following and progress we have achieved in modification of the nutritional quality of the glycinins. More complete descriptions can be found elsewhere (Nielsen 1984b, 1989).

Glycinin is found primarily in cotyledons of seeds where it is deposited in organelles called protein bodies. The organization of glycinin and other seed proteins within these organelles (principally beta-conglycinin, but smaller amounts of lipoxygenases, seed lectins and protease inhibitors) is presently not understood. However, x-ray diffraction studies suggest there is a high degree of organization of these proteins within the protein bodies because characteristic diffraction patterns can be obtained using the purified organalles (Coleman et al. 1980). The significance of this observation is that modifications which perturb the ability of storage proteins to assemble properly in protein bodies could have serious detrimental effects on other important agronomic characteristics of the plant such as seed size and yield. Alternatively, the protein may either not be directed to the proper subcellular destination or it may be recognized as a foreign protein and be degraded. Hence, it will be important to understand the consequence that modifications made to storage proteins have on the movement and assembly of the protein complexes within cells of the soybean cotyledon.

Glycinin, as extracted from seeds and purified, is an hexamer with a sedimentation coefficient of about 125 (Derbyshire et al. 1976), and undoubtedly originates from a more complex structure within the protein body. Five different subunits have been purified which account for the majority of the glycinin subunits present in the seed, although other genes that contribute minor subunits may also be present in the genome (Nielsen et al. 1989). The prevalent subunits originate from a family of homologous genes, each of which has been cloned, and whose nucleotide sequences have been determined (Nielsen et al 1989). The mature subunits consist of two peptide chains, both of which originate from the same precursor molecule that is cleaved post-translationally (Turner et al. 1982). The two chains of the mature subunit are joined by a single disulfide bond (Staswick et al. 1984). One chain is derived from NH2-terminal region of proglycinin and has an acidic isoelectric point, while the other arises from the COOH-terminal of the precursor and has a basic isoelectric point. Because the two chains are incorporated as a unit into the complex, and its basic structure originates prior to post-translational cleavage (see below), the pair of peptides rather than individual components is considered a subunit. The five subunits can be divided into two subfamilies (Group-1 and Group-2) based on major differences in the similarity of their amino acid sequences (Nielsen 1985a).

The general scheme followed during the synthesis and assembly of 11S legume proteins such as glycinin has been determined (Chrispeels et al. 1982), and is outlined in Fig. 4. Precursors are synthesized at the direction of polyadenylated mRNA at the surface of rough endoplasmic reticulum (ER). Signal peptides are removed co-translationally as the precursors emerge into the lumen of the ER, and there they assemble into trimers. The pathway for transport of the trimers to the protein bodies is presumed to be through Golgi, as has been demonstrated for the phytohemagglutinin of Phaseolus vulgaris (Chrispeels 1985). Cleavage of the precursors takes place in the vacuoler protein bodies.

We developed an in vitro system in order to study the assembly of proglycinin (Dickinson et al. 1987). The method is based on transcription-translation technology due to Melton et al. (1984), where transcription from modified cDNAs is accomplished by SP6 polymerase and results in mRNA that encodes all of the authentic subunits except the signal peptide. After purification, the mRNA is translated in rabbit reticulocyte lysates. As is shown in Fig. 5, Group-2 proglycinin subunits produced in this manner are able to self-assemble into trimers in the reticulocyte lysate. The assembly is dependent on both the concentration of the precursors and time. The trimers are equivalent in size to the oligomers that form in endoplasmic reticulum during synthesis glycinin complexes. However, even prolonged incubations of these trimers do not result in the formation of hexamers equivalent to those isolated from seeds. That step is dependent upon proteolytic cleavage of the trimers (Dickinson et al. 1989). Thus, the post-translational modification of proglycinins to form acidic and basic chains is apparently a regulatory step in the assembly of glycinin complexes that prevents their aggregation into an insoluble complex prior to their arrival in protein bodies.

The ability to synthesize and assembly glycinin complexes in vitro is a useful tool for attempts to engineer subunits with an improved nutritional quality. Changes introduced into cDNAs clones that encode proglycinins can be used to produce modified subunit precursors, and these can be tested for deviations in their ability, to assemble into oligomers. The advantage this approach has over transgenic plants whose seeds contain the modified proteins is one of speed of analysis. Weeks rather than years are required to make the evaluations in vitro, and constructions incapable of assembly can be eliminated from consideration. Moreover, regions in the subunits crucial for oligomer formation can be identified using the methodology. Because the three dimensional structure of the subunits has not been determined, the in vitro assembly assay is an important tool that can be used to guide efforts directed toward improvement of the nutritional quality of seed proteins.

The secondary structural preferences of amino acids in glycinin and homologous proteins from seeds of other plant species were studied in an effort to identify regions in the subunits that might be amenable to modifications for improvement of nutritional quality (Argos et al. 1985). Two regions of interest in the subunits were identified. One region was highly conserved among the 11S subunits, and consisted of hydrophobic amino acids that were predicted to be mainly in beta-sheet and turn conformations. It is located in the basic chain, immediately COOH-terminal to the conserved post-translational cleavage site in the precursor. By analogy to similar regions in proteins whose structure is known, we suspect that the highly conserved hydrophobic region is buried within the subunit and makes an important contribution to the overall structure of the molecule. Modifications in this region of the subunit are likely to alter structure considerably and could adversely, affect assembly of the subunits into oligomers. The second domain exhibits a considerable natural size variation among 11S subunits. This region, which we call the hypervariable region (HVR), is located at the COOH-terminal end of the acidic chain, and is adjacent to the post-translational cleavage site in the precursor. The HVR is predicted to contain exclusively alpha-helix and turn secondary structural preferences. By analogy with similar regions in proteins whose structures are known, the HVR is likely to be located at the surface of the glycinin subunits. The HVR varies in size among the five glycinin subunits by nearly 100 amino acids. Because of the natural variability of the HVR and its predicted location at the surface of the subunit, it may, tolerate changes in structure better than more highly conserved regions in the subunits.

To test these notions, appropriate changes were engineered into a cDNA clone that encoded G4-glycinin using site directed mutagenesis. The mutant proglycinin subunits were synthesized and tested for their ability to self-assemble in vitro. As predicted, a deletion in the highly conserved hydrophobic region of the basic chain resulted in mutant proglycinins incapable of assembly into trimers (Dickinson et al. 1987). On the other hand, many structural alterations made in the HVR were tolerated with little or no change in the rate of trimer self-assembly and hexamer re-assembly in vitro (Dickinson 1986). For example, deletion of the entire HVR results in mutant proglycinins that not only assemble, but self-assemble into trimers at a slightly more rapid rate than controls with normal precursors. Insertion of five alternating Arg-Met units also results in mutant subunits capable of assembly at normal or near normal rates. These latter constructions are significant from a nutritional perspective because methionine is considered the limiting essential amino acid in legume seed proteins. These data indicate that the limited types of changes targeted to the HVR thus far do not seem to perturb the assembly process as judged by the in vitro assembly assay. Although the results raise the host that beneficial modification of the storage protein subunits might not be as difficult as imagined, the modifications must be tested in vivo. The recent reports demonstrating that soybeans can be transformed using either disarmed Ti plasmid (Hinchee et al. 1988) or a ballistic method (McCabe et al. 1988) now make such experiments feasible.


The examples described above represent two relatively straightforward means by which a conventional crop might be altered to produce plants with a unique and valuable new properties. It is anticipated that an increasing number of such examples will become evident in the near term as our understanding of gene expression and the rules that govern protein folding become more sophisticated. The next several decades will indeed be exciting ones as new biotechnological strategies are melded with more conventional approaches of plant breeding to generate new crops.


*Summarizes cooperative research between USDA Agriculture Research Service and the Indiana Agricultural Experiment Station. The work was supported in part by grants from the American Soybean Association, Central Soya Inc. and from the USDA Competitive Grants Program.

Table 1. Yield and seed size of lipoxygenase isolines.z

Isoline Yield (bu/acre) Seed size (g/100 seeds)
C1640 53.6 14.5
Century 58.2 17.3
-L1BC5 54.7 15.2
-L2BC3 60.8 16.2
-L3BC5 60.2 17.6
zMeans of two replications. Plants were grown in four 16 ft. rows 24 inches apart with 115 seeds per row. Prior to harvest, two ft. were cut from each end of the 16 ft. rows and then the center two rows were harvested and weighed. There was significant difference in yield (LSD0.05 = 9.7), but the C1640 seeds were significantly smaller than the others (LSD0.05 = 2.24). C1640 is a line with low linolenic acid content described by Wilcox et al. (1983). The pedigrees of -L1BC5, -L2BC3 and -L3BC5 are indicated in Davies and Nielsen (1986b).

Fig. 1. Polyacrylamide gel electrophoresis of extracts from seeds that lack specific lipoxygenase isozymes compared with those from wild type. Cotyledonary tissue from mature seeds was extracted in SDS sample buffer and applied directly to the gel. Migration is towards the anode at the bottom of the figure. Wild type, 'Century'; -LI, Pl 408.251; -L2, Pl 86.023; -L3, Pl 205-085 From Davies and Nielsen (1986).

Fig. 2. Taste panel scores (y axis) for beany off-flavor of soymilk preparations from seed lacking one or two of the lipoxygenase isozymes. The panel ranked beany off-flavor in 10 categories, 1, bland; 2-4, mild; 5-6, moderate; 7-8, strong; and 9-10, very strong intensity. Adapted from Davies et al. (1987).

Fig. 3. Taste panel scores (y axis) for milk flavor of soymilk preparations from seeds lacking one or two of the lipoxygenase isozymes. The panel ranked milk flavor in 10 categories; 1, bland; 2-4, mild; 5-6, moderate; 7-8, strong; and 9-10, very strong intensity. Adapted from Davies et al. (1987).

Fig. 4. Schematic representation of events in the synthesis and assembly of soybean glycinin.

Fig. 5. Time-course for self-assembly of 3H-proglycinin synthesized in vitro. Labeled proglycinin was produced at the direction of clone pSP65/248 as described by Dickinson et al. (1987). Labeled proglycinin was incubated directly in the translation mixture for the indicated times at 37°C before the aliquots were re-moved for sucrose density gradient centrifugation. The position of sedimentation standards in the gradient are indicated at the top of the figure. The data on this figure indicate that there is a time dependent change in the sedimentation profile, as 3S proglychinin monomers are converted 9S trimers equivalent to those formed in endoplasmic reticulum in vitro. Adapted from Dickinson et al. (1987).

Last update August 26, 1997 by aw