Index | Search | Home | Table of Contents

Gresshoff, P.M. 1990. The importance of biological nitrogen fixation to new crop development. p. 113-119. In: J. Janick and J.E. Simon (eds.), Advances in new crops. Timber Press, Portland, OR.

The Importance of Biological Nitrogen Fixation to New Crop Development

Peter M. Gresshoff

  6. Table 1
  7. Fig. 1
  8. Fig. 2


The element nitrogen is highly abundant in Earth's atmosphere and is a major component of dietary, proteins (as incorporated in amino acids). Plant growth is directly influenced by the availability of reduced nitrogen, leading to the long accepted practice of manuring, fertilizer application, or rotational crop practices (Gresshoff and Rao 1986).

Most plants have the biochemical "machinery" to assimilate nitrogen in the form of nitrate (NOW, Numerous microbes in the soil facilitate the conversion of dead organisms or excrement to this readily usable anion. Some plants, of which most legume families are the most commonly known and used, possess an additional mechanism of nitrogen gathering which involves the reactions commonly termed Biological Nitrogen Fixation. In this process N2 gas is cleaved by the metallo-enzyme nitrogenase (itself made up of two components, the iron protein and the molybdenum-iron protein), linked to hydrogen atoms to yield ammonia (NH3-). This in turn is easily assimilated, just like the product of nitrate reduction, to form the building blocks for intermediate metabolism (purine, pyrimidine, alkaloid, and amino acid biosynthesis).

The reactions of nitrogen fixation are limited to the prokaryotic kingdom as nitrogenase genes have only, been found there. The reaction, in general, requires an anaerobic (lacking O2 or micro-aerobic (low in O2 concentration) environment, although exceptions to this rule are found in bacteria that have evolved special O2 protection mechanisms.

In legumes the root nodule is induced after infection with the soil bacterium Rhizobium or, Bradyrhizobium. In the last decade the advent of techniques in molecular genetics has allowed a detailed definition of bacterial genes causatively involved in nodulation and nitrogen fixation (see Rolfe and Gresshoff 1988 for review).

Legumes like pea, clover, lentils, acacias, and soybean, facilitate a physiological milieu supportive of bacterial nitrogen fixation through the development of a specialized structure called the nodule. Fig. 1 depicts a schematic diagram of the nodulation steps as they are thought to occur in soybean. In short, bacteria in the rhizosphere receive isoflavone signals from plant exudates, which convert the gene product of the nodD gene to an active form, which now is a transcriptional activator of several nodulation and host range genes. Specific subepidermal cells, about 1 cm behind the growing root tip of the main and lateral roots, respond to a yet undefined bacterial factor, which causes cell divisions without cell expansion in these target cells. The cell division cluster leads to the induction of infection of a developing root hair just above. Infection proceeds by means of an infection thread, which penetrates the cortical cell division focus, ramifying into several branches as it develops. Concurrently, cells in the pericycle start to divide, leading to a second growth center involved in nodule growth. The two division foci amalgamate leading to a visible nodule. As cells start to expand, infection threads penetrate the young cell walls and initiate bacterial release. Bacteria are physiologically transformed to become bacteroids, in which the enzyme systems required for nitrogen fixation are expressed. Meanwhile specific nodule function genes are induced, resulting in a fine-tuned symbiotic relationship. The nitrogenase reaction, while biochemically oxygen sensitive, requires commonly the oxygen molecule to generate chemical energy (ATP) needed to cleave the triple bond of N2 gas. The nodule is capable of energy supply (sucrose translocation and breakdown), O2 transport and regulation, ammonia assimilation and restriction of potential parasitic growth of the microbe in plant tissue (Mellor and Werner 1989).

Biological nitrogen fixation, as relevant to agricultural production is not restricted to legumes. Significant nitrogen input into the global nitrogen cycle stems from the Azolla-Anabaena aquatic symbiosis (Plazinski 1989); free living nitrogen fixation in Azotobacter, Azospirillum or Pseudomonas species (Krotzky and Werner 1987); and the Frankia (Actinomycetes)—non-legume nodulation (Simonet et al. 1989).

Despite the large genetic complexity of the host plant, analyses have ascertained several genes of eukaryotic origin necessary for nodulation and nitrogen fixation. Plant genes are defined either by mutagenesis or molecular probes indicating specific nodule expression or amplification (see Gresshoff and Delves 19861; Nap and Bissehng 1989 for review).

When determining the utility and feasibility of new crops in agriculture (whether these are developed by selection from natural habitats, introgressive plant breeding, somatic hybridization, polyploidization or mutagenesis) the essential requirement of all plants for a reduced nitrogen source needs to be considered.

Rarely is this done as evidenced by the description of important characteristics of new alfalfa or soybean registered lines. Plant breeding has focused predominantly on the aspect of yield increases, either by direct selection or disease resistance breeding. Modern sustainable agriculture requires viewing the entire equation, involving input costs and indirect cost to the environment. Thus the advances of the Green Revolution were strongly based on the use of high nitrate fertilization. Little emphasis was given to increase the application of legumes in rotational agricultural practices. Furthermore little plant breeding was done to improve the nitrogen fixing abilities of existing crop legumes or to understand the complexities of the symbiosis. Discussions on new crops need to recognize the advantage of the legume (as well as the actinorhizal plant) symbiosis and include a strategy leading away from the environmental contaminant nitrate.

The essence of this communication is to up-date the reader to the present level of understanding of symbiotic nitrogen fixation as shown by legumes plants in association with Rhizobium or Bradyrhizobium. Special emphasis will be given to an old crop, namely Glycine max (L.) Merr. (soybean) which, when given a set of new genetic characters, may be considered a new crop.


Soybean is grown in many areas of the globe in both tropical and temperate climates (i.e. from Brazil to Southern Canada and Poland). It can yield, on average, between 3 to 5 t per hectare, given good soil and water conditions. Soybean seed is rich in protein (about 40%) and oil (about 20%). The oil is used as a vegetable oil. Soybean meal is used as a protein additive to both human (tufutti, soyaroni, artificial meat) and animal feeds. The United States, Brazil and Argentina are major world suppliers. The European Community (EC) imports 1 to 2 billion dollars worth of soybean products yearly. A complete review of soybeans is found in Wilcox (1987).

The classical uses of soybean in Asia (e.g. bean curd, tofu) have extended to the Western world because of increased health conscientiousness. Additional expansion may arise as industrialized nations recognize that widely-planted basic crop plants can produce renewable industrial resources. The current improvement strategy is progressively shifting towards reducing inputs as compared to increasing output. Yield, water use efficiency and salt tolerance are controlled by several gene systems, and therefore it is unlikely that these characters can be improved easily by new gene manipulation techniques. However, aspects that involve "input costs" such as fertilizer use, application and transport, pesticide and herbicide costs, show the potential for genetic manipulation due to the fact that few genes (often one) can effect these characters. The recent analysis of the plant's contribution to symbiotic relationships has shown that generic manipulation to increase nodulation, nitrogen fixation, and indirectly yield is possible.


Using induced mutagenesis (Carroll et al. 1985a, b, 1986) we have isolated soybean mutants that either fail to nodulate (resistance to Bradyrhizobium) or show increased amounts of nodulation (termed supernodulation). The supernodulation mutant gene also conferred the ability to nodulate and fix nitrogen in the presence of nitrate and hence were given the label nts' (nitrate tolerant symbiosis).

The original isolation, the physiological, biochemical and agronomic characterization are published (Carroll et al. 1985a, b, 1986, 1988; Day et al. 1986, 1987; Delves et al. 1986, 1987a, b, 1988; Gresshoff et al. 1985 1988a, b; Mathews et al. 1987, Mathews 1987; Alva et al. 1988; Olsson et al. 1989, Schuller et al. 1988a, b).

Supernodulation and the related nitrate tolerant symbiosis has been investigated in detail. We now know, that the nts mutants all fall into the same genetic complementation group, being mendelian recessives (Delves et al. 1988). The marginal mutant nts1116 gives intermediate levels of nitrate tolerance, about 3-5 fold increases in nodule number and nodule mass, and up to 30% increase in total grain yield compared to parent cultivar Bragg. The genetics of mutant nts1116 suggests dominance to nts382, but recessiveness to the wild-type allele (Delves et al. 1988). Supernodulation is caused by the diminished synthesis or transport of a shoot-derived inhibitor. The chemical nature of this substance has yet to be determined but seems to involve a small, water-extractable compound (Gresshoff et al. 1988b). Biologically we can state that the shoot derived inhibitor has the following properties; directly involved in nodulation arrest and autoregulation; lowered substantially in nts mutants; facilitates nitrate inhibition of nodulation; synthesized in leaves; controlled by single mendelian gene; induced by early nodulation events.

We have recognized numerous pleiotropic changes in the chemical composition of mutant and wild-type shoot extracts and are presently assaying these for symbiotic activities (see Gresshoff et al. 1988b for partial HPLC separations).

Complementation analysis between non-nodulation (Nod-) mutants nod49, nod139, nod772 (Carroll et al. 1986) and the naturally occurring line rj1 (Lee) showed that mutant alleles nod49, nod772 and rj1 are in the same complementation group (Mathews 1987; Mathews et al. 1989b). Since independent events occurred in the production of mutation rj1 and nod49, it is possible that this locus is a `hot-spot' as three out of the four known nod- mutations in soybean occur within it. Mutation nod772 is slightly leaky in its phenotype compared to nod49 since occasional root hair curling and infection thread formation have been observed (Mathews et al. 1987). Cytological studies on serially sectioned roots of wild-type and mutant plants allowed a closer verification of their phenotypes. Mathews (1987) and Mathews et al. (1989a) showed that mutants nod49, nod772 and line rj1 share phenotypic similarities (consistent with their common genetic background). Each mutant induces a lower level of subepidermal cell divisions per centimeter of root than wild-type (cultivar Bragg) plants and lack (except for some leakiness observed in mutant nod772) root hair curling, infection thread formation and nodule development when inoculated at moderate titres (107-108 cells of Bradyrhizobium japonicum strain USDA 110 per plant).

Occasional nodulation was observed on non-nodulation mutants (including line rj1 plants) which appeared to be dependent on inoculation titre. From 2 to 6 nodules were observed on each plant when inoculated with 1010-1011 bacteria per seedling. These "occasional" nodules exhibited normal morphology and nitrogen fixation. A higher frequency of occasional nodulation was detected close to cotton-wool plugs in Leonard jar assemblies suggesting a niche exists there which may increase bacterial cell number or, alternatively, concentrate bacterial cell products. Occasional nodulation was never found in field grown or sand grown mutant plants. This observation may relate to the original findings with rj1 (also called T201 and nn; Williams and Lynch 1954) in which the addition of sawdust to sand altered the nodulation phenotype. Mathews (1987) was unable to select a bacterial strain which could suppress the Nod- phenotype. Bacteria reisolated from occasional nodules retained the phenotype of original inoculant strains. Rare infections via threads on curled root hairs occurred in Nod- plants inoculated at high titres of bacteria which seemingly eliminates the possibility of an alternative infection route for example via "crack entry." Linkage analysis (Mathews 1987, Mathews et al. 1989b) shows alleles nod49, nod772 and rj1 are not linked to purple flowers and the nts locus.

Mutation nod139 defines a second non-nodulation locus different to nod49, nod772 and rj1. It also segregates as a mendelian recessive but produces a symbiotic phenotype lacking subepidermal cell divisions, root hair curling and nodule formation. Mutant allele nod139 plants interestingly also exhibit occasional nodulation. The new nod- locus is tentatively named rj6 , in continuation of the established nomenclature for symbiotic mutants of soybean (Mathews et al. 1989b), but should perhaps be renamed nod3 (see below).

Although all the non-nodulation mutants described here are mendelian recessives there is some suggestion that nod49, when heterozygous in the absence of autoregulation (Rolfe and Gresshoff 1988), still affects nodule number (Mathews 1987). This phenotype is possible if the total number of actual infections is reduced. The autoregulation mechanism may thus mask subtle phenotypic differences in early nodule development which occur prior to the developmental bottleneck presented by autoregulation. Precise quantitative analysis of bacterial and plant nodulation genes should not only be done using wild-type plants but also mutant nts plants which lack autoregulation.

The major classes of soybean mutants described here are shown in relation to a broadly defined series of nodulation events in Fig. 2.

Whether the present system of mutant nomenclature is appropriate depends on one's view. The rj prefix refers to the use of Rhizobium japonicum as an inoculant. Today we classify all of these strains as Bradyrhizobium or Rhizobium fredii. Another problem exists in the different symbiotic phenotypes that are pooled under the rj symbol; rj1 (and rj6) result in non-nodulation, while Rj2, Rj3 Rj4 and possibly Rj5 control symbiotic effectiveness (i.e. nitrogen fixation and assimilation) rather than nodule initiation. A further complication comes from the recent suggestion that Rj3 no longer expresses its original phenotype (T. Devine, pers. comm.), making it impossible to prove by complementation that one's newly discovered mutant allele differs from Rj3. Great advances have been made in our understanding of bacterial nodulation genes. The accepted bacterial nomenclature uses the prefix nod (for example; nodA, nodK or nodM) (Appelbaum, 1989) following established generic nomenclature as defined for Escherichia coli.

Mutation of certain bacterial genes can result in the same symbiotic phenotype as certain plant mutants, e.g. mutant nod139 is phenotypically equivalent to a bacterial nodC mutant. It seems more appropriate, then, to label plant mutations in a similar manner. For example, mutation rj1 might be labeled nod1 and mutant nod139 might become nod3. The locus defined by the supernodulation alleles nts382, nts1007, and nts246 would become nod2 as each controls a nodulation step which was genetically defined in our laboratory prior to the nod3 locus (Delves et al. 1988; Carroll et al. 1988). Mutant alleles Rj2, Rj3, Rj4 and Rj5 should become fixl, fix2, fix3 and fix4 respectively, as each alters the nitrogen fixation phenotype. Researchers in Pisum sativum, a classical genetic organism, easily accepted the nomenclature using a three letter prefix followed by a number (see Kneen and LaRue 1984). In view of the variable nomenclature for the rj1 locus (nn, T201 and rj1), it is suggested that future usage should accept a more unified system, tike that proposed here (Table 1).

All non-nodulation mutants of soybean respond to an increased inoculum titre by permitting occasional nodulation. This strongly suggests these mutant plants have a decreased sensitivity to a bacterial signal which is partially compensated when high inoculum titres are applied. The importance of monitoring bacterial titres in symbiotic tests cannot, therefore, be overestimated. Both bacterial and plant mutant phenotypes may be masked if, for example, inoculation titres are too high. Moreover, crucial alterations are probably expressed at all inoculum levels (i.e. in bacterial nodC or plant nts382 mutants) but subtle quantitative changes in phenotype may be missed. A similar precaution was mentioned above regarding the masking of subtle symbiotic alterations by the plant autoregulation mechanism.

We are presently expanding the analysis of the supernodulation and non-nodulation mutants through molecular genetic techniques. Plant exudates and extracts contain flavones and isoflavones, which specifically induce nodulation-essential genes in the Bradyrhizobium (Kosslak et al. 1987). One can monitor this inducing activity by using gene fusions in which the bacterial nodulation gene is linked to the beta-galactosidase gene (lacZ) from Escherichia coli. Thus, a developmental step can be assayed by enzymatic means. Preliminary findings (Mathews et al. 1989c) using a nodC-lacZ fusion in B. japonicum strain USDA110 induced with a dilution series of methanol extracts of 5 day old uninoculated seedling roots of nod49, nts382 and 'Bragg' plants, show that they have similar induction ability. This suggests that the balance between inducing isoflavones and possible antagonists is the same. 'Williams 82', which has a similar nodulation pattern as 'Bragg', has a 10 fold higher inducing ability than 'Bragg' (Mathews et al. 1989c). This suggests that the quantitative level of inducing substances (as assayed by nodC-lacZ fusions) from uninoculated roots has little to do with quantitative aspects of nodulation. Whether this holds true to inoculated seedlings and to exudates rather than extracts is presently investigated. Our investigations on the plant are focusing on comparisons in the macromolecular profiles of supernodulation and non-nodulation mutants compared to wild-type plants. It is now possible to use genetic transformation to evaluate the functional significance of isolated gene sequences in soybean.


Through the genetic definition of major genes that control early steps of nodulation in soybean, it is possible to delineate the molecular mechanisms underlying the observed phenotypic changes. This type of understanding should have significance to the study and optimization of nodulation and nitrogen fixation in other legume crops.

Specifically, the isolation and wide-ranging characterization of supernodulation and non-nodulation mutants of soybean has allowed the recognition of several new phenomena governing the symbiosis.

New agriculturally significant properties may not necessarily have to arise by selection from natural populations, but mutagenesis and directed selection (as it was illustrated above for the nodulation mutants) followed by breeding can yield plants with new qualities. The isolation of hypernodulation mutation like nts1116, with increased grain yield and the acid soil tolerance of mutants nts382 (see Alva et al. 1988), illustrates the quantum jumps in yield improvement can occur.

The induced mutagenesis and selection strategy has helped to delineate several of the plant functions needed in nodulation and nitrogen fixation. This symbiosis contributes substantially to agricultural productivity. This approach could also contribute to the domestication and selection of other desirable new crop species.


Table 1. Nomenclature of genetic loci affecting symbiotic phenotype in soybean.

Category Present allelic nomenclature Proposed nomenclature Symbiotic phenotype
Non-nodulatio n nod49, nod772, rj1 nod1 Hac-, Nod-, Scd+
Non-nodulation nod139 nod3 Hac-, Nod-, Scd-
Supernodulation nts382, nts1007 nod2 Nod+++, Sdi-, Fix+
Supernodulation nts1116 nod4(?) or nod2 Nod++, Sdi-, Fix+
Nitrogen fixation Rj2, Rj3, Rj4, Rj5 fix1, fix2, fix3, fix4 Nod+, Sdi+, Fix-
Abbreviations; Hac = hair curling, Nod or nod = nodulation, Scd = subepidermal cell division, Fix or fix = nitrogen fixation, nts = nitrate tolerant symbiosis, Nod+++ = supernodulation, Nod++ = hypernodulation, Sdi = shoot derived inhibitor, nts1116 may be a leaky allele at the nod2 locus (Delves et al. 1988).

Fig. 1. Ontogeny of nodule initiation in soybean: sequence of events is indicated by letters A to J. Plant roots excrete substances (stage A), which interact with the bacteria (stage B), which produce subepidermal cell division stimulating factors. These interact with specific hypodermal cells near the xylem poles, suggesting a possible positional gradient emanating from the xylem and perhaps the phloem (stage C). The hypodermal (or subepidermal) division focus forms the primary nodule meristem, which potentiates the developing root hair just above it to become a target site for bacterial infection (stage D). Bacteria attach and invade the root hair (stage E) while the primary meristem induces pericycle cell divisions, again near the xylem pole (stage F). Infection threads become clearly visible as the root hair cell grows. The two meristematic foci grow together (stage G), giving rise to a fused cluster of dividing and invaded cell types (stage H). The infection thread ramifies, and the bacteria increase rapidly in number in the cortex. Subsequent dim differentiation of the nodule yields vascular connections and the variety of cell types needed for nodule function (stage J). Reprinted from Rolfe and G resshoff (1988) with permission of publishers.

Fig. 2. Mutational blockage of nodulation development in soybean. Three loci controlling cell division events were characterized genetically (as indicated by the crosses). Resulting phenotypes were non-nodulation and supernodulation (absence of autoregulation). Crosses without genetic labels represent potential sites for mutant isolation (e.g. a root controlled autoregulation affected mutant).

Last update August 26, 1997 by aw