As mentioned earlier, a protein complex called nitrogenase (composed of enzymes with metal co-factors) makes nitrogen fixation possible in plants. The first one is dinitrogenase and the second one is dinitrogenase reductase [33]. According to the active site co-factor binding metal, there exist three types of dinitrogenase in nature. (a) Molybdenum (Mo) nitrogenase; it is most abundant and carries the most significance in the nitrogen-fixing bacterial and archaeal niche and the alternative vanadium (V) and iron-only (Fe) nitrogenases [34]. The molybdenum-dependent dinitrogenase is formed by nifD and nifK gene products and dinitrogenase reductase is a homodimer of the nifH gene product [30,35]. It is well documented that molybdenum nitrogenase is produced in all diazotrophs in nature, while some produce vanadium or iron nitrogenase in addition to Mo-nitrogenase [36,37]. The rhizobium bacteria residing in nodules fix atmospheric nitrogen gas to NH₃, which plants can assimilate via glutamine synthase to form glutamine. In response, the bacteria derive plant carbohydrates, mainly as malate for food and an energy source for nitrogen fixation. Nodules are very complex structures, containing several processes which operate and interact at distinct levels. The process of nodule formation requires a coordinated exchange of signals between the two symbiotic partners [38]. Bacteria had their symbiotic genes first characterized by transposon mutagenesis; this achieved the definition of over 50 nodulation genes (Nod and Nol) in bacteria, and about the same number controlling nitrogen fixation; thus many nod- and fix-bacterial strains exist in many species of rhizobia. Legume–rhizobium symbiosis starts with molecular signaling between the two partners. Early nodulation gene cascades in legumes. Plants release signals such as flavonoids (e.g., the flavone 7,4 dihydroxyflavone and the isoflavone genistein) which are picked up by compatible bacteria in the rhizosphere [39,40] leading to the production of Nod factors (NF) which trigger early events in the nodulation process [41,42]. This triggers the downstream gene cascade including those involved in nucleoporin, cation channels, calcium spiking, early nodule expression, and cytokine signaling leading to cortical and pericyclic cell divisions, and concomitant bacterial infection. Rhizobia are entrapped by root hair curling after the Nod factor has been perceived, which results in initiating the formation of an infection thread (a tubular structure). This infection thread facilitates the penetration of root hair cells and adjacent cortical cells [43]. Cell divisions in cortical and pericycle occur simultaneously resulting in the formation of the nodule primordium. Bacterial cell division facilitates the rhizobial traveling through the infection thread and is eventually freed into the induced nodule primordium cells [44,45]. As nodules mature with time, bacteria are enclosed within the symbiosome membrane, resultant from an inverted plasma membrane of plant origin. In this encapsulated chamber, the bacteria experience a micro-aerobic environment (lower oxygen concentration) and differentiate into bacteroids, fixing diffused nitrogen gas using their nitrogenase enzyme complex [46,47]. Depending on whether or not the meristem remains active for the life of the nodule, two main types of nodules are formed on the various legume species, (i) indeterminate or (ii) determinate. In the case of determinate nodules, nodular meristematic activity is terminated early and is usually initiated sub-epidermally in the outer cortex, thus giving rise to spherical nodules [48]. In indeterminate nodules, the inner cortex undergoes cell division (anticlinally) followed by periclinal divisions in the pericycle. Here, cylindrical nodules are formed due to more persistent meristems [49,50].

The understanding of the genetic basis of this relationship is of paramount importance and essential for the optimization of nitrogen acquisition rates in legumes themselves. Bacterial nif genes are well known to encode the components of the nitrogenase enzyme complex. nifH, nifD, and nifK genes encode the structural subunit of di-nitrogenase reductase and the 2 subunits of di-nitrogenase, respectively. Many rhizobial genes have been fully sequenced, for instance, Mesorhizobium loti, Sinorhizobium meliloti, and Bradyrhizobium japonicum [51,52,53]. These proteins have similar sequences and common structures and functions in many diazotrophs, for instance, Azotobacter vinelandii, Herbaspirillum seropedicae, Pseudomonas stutzeri, and Bradyrhizobium japonicum [54,55,56,57]. Furthermore, genetic and biochemical analyses revealed that many additional nif genes, including nifE, nifN, nifX, nifQ, nif W, nifV, nifA, nifB, nifZ, and nifS, play roles in the regulation of nif genes and maturation processes of electron transport and FeMo-cofactor biosynthesis and assembly [58,59]. In addition, the fixABCX genes first identified in Rhizobium meliloti [60,61] and subsequently in other diazotrophs were reported to encode a membrane complex participating in electron transfer to nitrogenase [62]. The degree of specificity between legumes and rhizobia varies. The Nod factors produced by Rhizobium etli and Rhizobium loti produce identical Nod factors; however, they have distinct host ranges (Phaseolus spp. and Lotus spp., respectively) [63]. Moreover, different rhizobia nodulating the same plant may excrete completely different Nod factors. For instance, Rhizobium tropici and R. etli produce different Nod factors (sulfated and acetylfucosylated, respectively), but both are known to nodulate Proteus vulgaris [64]. More examples include Bradyrhizobium elkanii and Bradyrhizobium japonicum, which have a number of mutual hosts, but their Nod factors differ considerably [65].

Several studies have identified QTL associated with traits related to biological N fixation (Table 1 [66,67,68,69]). The QTL markers can be used in marker-assisted selection for breeding plants with better nitrogen fixation attributes. A QTL for the total ureides (acyl derivatives of urea) was identified on chromosome 17 in soybean which explained 13.26% phenotypic variation [70]. Li and the team [71] cloned a candidate gene associated with a major QTL in soybean for increasing nodule size and named it INCREASING NODULE SIZE1 (GmINS1). The overexpression of GmINS1 increased the N content and the biomass of the soybean plant due to an increase in number, biomass, the abundance of infection cells, and nitrogenase activity of large nodules [71]. The result was the opposite when GmINS1 was suppressed by RNA interference [71].