Like any bacteria-plant interactions, the rhizobium–legume symbiosis is a complex and delicate process, which involves the cooperation of multiple cellular functions. Metabolic modeling of S. meliloti suggests that chromid genes are more actively involved in rhizosphere fitness than in bulk soils, while chromosome has a similar contribution to fitness in two niches ^[1][2][116,117]. A large number of R. leguminosarum genes are differentially expressed in rhizospheres of pea, alfalfa and sugar beet ^[3][118]. Some host-specific genes are related to C metabolism, and many are located on the non-symbiosis accessory plasmids ^[3][118]. These findings indicate that non-symbiosis genes are extensively involved in rhizosphere fitness. By using the transposon insertion sequencing method, about 600 genes of R. leguminosarum are identified as fitness genes during lifestyle adaptations from rhizosphere to symbiosis with pea plants ^[4][115]. Comparative transcriptomics independently demonstrates that hundreds of rhizobial genes are differentially expressed in nodules compared to free-living cells ^{[5][6][7][8][9]}[114,119,120,121,122]. Metabolic modeling suggests global coordination of carbon and nitrogen allocation in bacteroids ^[10][123]. The broad-host-range strain S. fredii NGR234 shows considerable differences in transcriptomes of bacteroids in Leucaena leucocephala and Vigna unguiculata ^[8][121]. Key nitrogen fixation genes on the symbiosis plasmid of S. fredii strains in Glycine nodules are characterized by high connectivity in both intra- and inter-replicon co-expression analyses ^[5][114], which allow further identification of chromosomal core znu and accessory mdt operons involved in host-specific symbiotic adaptation ^[5][114]. However, functional characterizations post genome-wide surveys are usually limited ^[4][5][114,115].

Based on scattered genetic and molecular biological evidence, a general picture of extensive recruitment of core and accessory functions, in addition to key nodulation and nitrogen fixation genes, in optimizing symbiotic efficiency has been proposed earlier ^[11][12][16,17]. The list of core and accessory functions contributing to rhizobial fitness, from rhizosphere to rhizoplane to nodules, is ever-increasing (above 900 genes) and usually in a strain–host-dependent manner ^[13][52], e.g., motility and chemotaxis ^[14][124], surface polysaccharides ^{[15][16][17][18][19]}[125,126,127,128,129], outer membrane vesicles ^[20][130], quorum sensing ^{[21][22][23][24][25]}[131,132,133,134,135], T1SS, T3SS, T4SS and T6SS ^{[26][27][28][29][30][31][32]}[136,137,138,139,140,141,142], dicarboxylate transport ^[33][143], poly-3-hydroxybutyrate ^[34][35][36][144,145,146], transporters of branched amino acids ^[37][38][39][147,148,149], uptake of ions (phosphorus ^[40][41][42][150,151,152], potassium ^[43][44][153,154], molybdenum ^[45][155], iron ^[46][47][156,157], sulfur ^[45][155], zinc ^[48][94] and manganese ^[49][158]), nitrate reduction ^[50][40], NO modulation ^[51][52][159,160], oxygen limitation responses ^{[50][53][54][55]}[40,161,162,163], cell cycle ^[56][57][58][164,165,166], peptide importers ^{[59][60][61][62][63]}[167,168,169,170,171], regulatory non-coding RNAs (e.g., globally acting trans-small RNA AbcR1/2 and those fragments derived from transfer RNA) ^[64][65][172,173] and carbon–nitrogen metabolism coordination by nitrogen-related phosphotransferase system ^[43][66][67][153,174,175]. These efforts indicate that the successful integration of key nodulation and nitrogen fixation circuits in various bacterial recipients involves systematic and dynamic coordination with other functions during the establishment of symbiosis.

Although all key nitrogen-fixing genes are directly activated by NifA in Proteobacteria ^[68][11], a constitutive expression of nitrogenase should be avoided under fluctuating conditions. The nifA gene can be transcriptionally activated by different upstream regulators, e.g., the FixL-FixJ two-component system in S. meliloti ^[69][176], the FixL-FixJ-FixK cascade in A. caulinodans ^[70][71][177,178], the redox responsive regulator RegR and its upstream kinases in B. diazoefficiens ^[54][72][73][162,179,180]. Nitrogenase is O₂ sensitive and microaerobic fitness machineries seem to be relatively conserved in test rhizobia. The fixNOQP operon, encoding the cbb3 terminal oxidase, is transcriptionally activated by the FixL-FixJ-FixK cascade in S. meliloti, A. caulinodans and B. diazoefficiens ^{[69][70][74][75][76]}[176,177,181,182,183], and the hFixL-FxkR-FixKf-FnrN cascade in R. etli and R. leguminosarum ^[55][77][163,184]. On the other hand, NifA can activate the transcription of fixABCX, encoding an electron bifurcating complex that provides low-potential reducing equivalents for nitrogenase ^[78][185], in S. meliloti, R. etli and R. leguminosarum ^{[79][80][81][82]}[186,187,188,189]. Ferredoxin, likely reduced by FixABCX during nitrogen fixation ^[78][185], is a reductant of nitrogenase and its gene transcription is also activated by NifA in S. meliloti, B. diazoefficiens and R. etli ^[83][84][190,191]. Available transcriptomic evidence in B. diazoefficiens, S. meliloti and R. etli indicates that NifA may regulate more functional genes, e.g., molybdenum transporter, cytochrome P450 proteins, GroES, GroEL and uptake hydrogenase in a strain-dependent manner ^[53][82][83][161,189,190]. Notably, stringent sets of NifA regulon in these three rhizobia (19–67 genes) are on the symbiosis plasmid or symbiosis island with rare exceptions ^[53][82][83][161,189,190]. These findings imply that the integration of the key nitrogen fixation gene circuit in terms of transcriptional activation is strain-dependent, and the transcriptional regulation network of bacterial recipients seems to be limitedly subject to direct interference by NifA.

An efficient nitrogen fixation process of rhizobia is structurally ensured by nodule organogenesis that is initiated by specific recognition of host symbiotic signals by NodD in most rhizobia. The divergence of NodD and multiple NodD copies facilitate rhizobia efficiently establishing symbiosis under fluctuating conditions and/or exploring more host plants ^{[85][86][87][88][89]}[64,67,68,192,193]. Available studies support a model where a primary NodD binds DNA in the absence of host symbiotic signals while signal–NodD binding enhances DNA bending that allows transcription ^[90][91][194,195]. NodD autoregulates its own transcription in R. leguminosarum bv. trifolii and R. leguminosarum bv. viciae ^[92][93][196,197]. When two or more NodDs are encoded within a genome, these NodDs usually act as a coordination module in a rhizobium- and condition-dependent way. For example, a second NodD copy in some R. leguminosarum bv. trifolii strains enhances nodule colonization competitiveness ^[94][198]. In S. meliloti, NodD1_Sm and NodD2_Sm respond to plant signals and the overexpressed NodD3_Sm can function without flavonoids ^[95][96][62,199]. Among five NodD copies of R. tropici CIAT899, nodD2_Rt expression is induced by osmotic stress, and the engineered overexpression of NodD2_Rt alone is sufficient to replace the other NodD copies ^[97][200]. Mesorhizobium loti R7A has two NodD1_Ml and NodD2_Ml, showing a degree of functional redundancy ^[98][201]. More detailed investigation demonstrates that NodD1_Ml mainly function in infection threads while NodD2_Ml primarily acts in the rhizosphere and within nodules ^[99][202]. This observed division of labor is likely due to their divergence at the signal binding cleft ^[99][202]. Noteworthy, NodD2_Ml activity is negatively affected by NodD1_Ml at the pre-infection stage ^[99][202]. Two NodD copies are also found in other rhizobia, e.g., S. fredii and B. diazoefficiens, where NodD2 can negatively regulate the transcription of nodD1 ^{[89][100][101]}[193,203,204]. The nodD2_Sf mutant shows impaired nodulation on soybean but improved compatibility with Lotus species ^[102][205]. Collectively, these findings suggest a working model where rhizobial NodD and its regulon may be divergently selected in at least two dimensions: 1) from rhizosphere to rhizoplane to infection threads to nodules; 2) different host plants.

Various variables in these niches may interact with the key symbiotic interaction signaling pathway/network. Indeed, NFs are required for the biofilm establishment ^[103][104][206,207], which generally enhances bacterial resilience to various stress factors ^[105][208]. The production of exopolysaccharide (EPS), a common component of biofilm matrix, is negatively regulated by NodD in S. fredii ^[106][107][65,209] while positively regulated by the NodD3-SyrM-SyrA regulatory module in S. meliloti ^[108][210]. This is in line with the fact that EPS is an important symbiotic signal for the host infection of S. meliloti but dispensable in S. fredii symbiosis with test legumes ^[109][110][211,212]. In the presence of symbiotic signal daidzein, S. fredii NGR234 produces the phytohormone indole-3-acetic acid (IAA) in a NodD-dependent manner, and overexpression of NodD2_Sf enhances the transcription of IAA synthesis genes and IAA production ^[111][213]. The positive regulation of IAA production by NodD is also found in R. tropici CIAT 899 ^[88][192]. There is evidence showing that flavonoids induce an NodD-dependent expression of traI that is responsible for the synthesis of short-chain quorum-sensing 3-oxo-C8-HSL in S. fredii ^[104][207]. Independent studies also reveal that the transcription of T3SS and effector coding genes depends on the positive regulator TtsI, while the expression of ttsI can be activated by NodD and plant flavonoids ^[112][214]. In NodD3_Sm over-expressing S. meliloti, more than 200 genes are differentially expressed including upregulated EPS biosynthesis, and downregulated motility and chemotaxis functions ^[113][215]. Among those upregulated genes (above 70 genes), 69% are located on the chromosome and chromid ^[113][215]. Although a systematic survey of direct targets of NodD is not available yet, these examples suggest that NodDs have been evolving to differentially modulate other non-symbiosis functions to improve fitness from rhizosphere to rhizoplane to infection in a strain–host-dependent manner.

In addition to NodD, lineage-specific parts regulating nod genes have been identified in a few model rhizobia, e.g., the two-component system NodV-NodW, NwsB and NolA in B. diazoefficiens ^[101][204], SyrM, a NodD homolog, in Sinorhizobium ^{[108][114][115]}[210,216,217], NolR homologs in Sinorhizobium and Rhizobium ^{[116][117][118][119][120]}[218,219,220,221,222]. NodV-NodW positively regulates nodulation genes and is essential for symbiosis with mung bean, cowpea and siratro, but only marginally contributes to nodulation on soybean ^[121][223]. NwsB, a homolog of NodW, is required for both induction and the population density-dependent repression of nodulation genes ^[122][224]. NolA induces nodD2_Bd and consequently represses the transcription of nodulation genes ^[123][225]. SyrM_Sm activates nodD3_Sm transcription and NodD3_Sm induces syrM_Sm, forming a self-amplifying circuit in S. meliloti ^[108][210]. NodD1_Sf induces the transcription of SyrM_Sf that in turn induces NodD2_Sf that represses nodulation genes in S. fredii ^[114][216]. Chromosomal NolR directly targets nodD1_Sm, nodD2_Sm and nodABC_Sm operons in S. meliloti ^{[118][124][125]}[220,226,227] and downregulates nodulation and T3SS genes in S. fredii ^[120][222].

MucR/RosR/Ros/MI (hereafter MucR), a conserved zinc-finger regulator in Alphaproteobacteria, enhances the induction of key nodulation genes by the host signal luteolin in S. meliloti ^[126][228]. The disruption of mucR leads to delayed nodule development in the S. meliloti-alfalfa pair ^[126][228], reduced nodulation competitiveness in R. etli (common bean) ^[127][229], impaired nodulation and nitrogen fixation in R. leguminosarum bv. trifolii (clover) ^[128][230] and deficient nitrogen fixation in broad-host-range S. fredii strains (e.g., increased nodule number on soybean while decreased nodule number on Lotus burttii) ^[129][130][231,232]. Available transcriptomic analyses independently show that MucR is a pleiotropic regulator, e.g., in R. leguminosarum bv. trifolii, S. meliloti, S. fredii and other members of Alphaproteobacteria ^{[126][129][130][131][132][133]}[113,228,231,232,233,234] (Figure 1). MucR autoregulates its transcription and recent ChIP-seq analysis further revealed more than 1350 direct target genes of MucR in the multipartite genome of S. fredii ^[134][111], among which a considerable number of known circuits were identified, e.g., NodD2_Sf ^[89][100][193,203], TtsI, T3SS and its effector NopP ^[135][84], SyrB (negative regulator for SyrM in S. meliloti) ^[136][235], pilus assembly (Cpa), motility (Fla/Fli/Mot/Flg) and chemotaxis (Che/Mcp) ^[14][124], diguanylate cyclases ^[137][236], general stress response (RpoE5 and CspA8) ^[138][237], surface polysaccharides (ExoY and Uxs1) ^[139][140][238,239], carbon–nitrogen metabolism coordination (PtsN3) ^[43][66][153,174], nitric oxide reduction (Nor) ^[50][40], and uptake of potassium (Kdp) ^[43][44][153,154], iron (RirA) ^[46][156] and phosphorus (PhoUB) ^[40][150]. It should be noted that MucR rarely activates its target genes and prefers to bind AT-rich DNA (periodic repeats of “Ts”) that represents a characteristic feature of the symbiosis plasmid, genomic islands and xenogeneic DNA ^[134][111]. More molecular biology evidence shows that MucR shares convergently evolved features of xenogeneic silencers with H-NS, Lsr2, MvaT and Rok from either Gram-positive or Gram-negative bacteria ^[141][112], i.e., oligomerization mediated by N-terminal domain and forming DNA-protein-DNA bridging complex. The importance of these convergent functions in integrating symbiosis/pathogenesis functions is demonstrated by the H-NS from E. coli being interchangeable with MucR from S. fredii in the hemolysis activity assay and symbiotic performance ^[141][112]. Indeed, H-NS and other functional H-NS/MucR or Lsr2/MucR chimeric proteins, show similar ChIP-seq profiles in the multipartite genome of S. fredii and share 286 target genes including those encoding NodD2_Sf, TtsI, T3SS, NopP, RirA, ExoY, Uxs1, TraI, TraR and VisN ^[141][112]. This evidence highlights that the convergent xenogeneic silencer MucR predisposes Alphaproteobacteria to integrate AT-rich symbiosis genes.