Bradyrhizobium elkanii nod regulon: insights through genomic analysis

Abstract

A successful symbiotic relationship between soybean [Glycine max (L.) Merr.] and Bradyrhizobium species requires expression of the bacterial structural nod genes that encode for the synthesis of lipochitooligosaccharide nodulation signal molecules, known as Nod factors (NFs). Bradyrhizobium diazoefficiens USDA 110 possesses a wide nodulation gene repertoire that allows NF assembly and modification, with transcription of the nodYABCSUIJnolMNOnodZ operon depending upon specific activators, i.e., products of regulatory nod genes that are responsive to signaling molecules such as flavonoid compounds exuded by host plant roots. Central to this regulatory circuit of nod gene expression are NodD proteins, members of the LysR-type regulator family. In this study, publicly available Bradyrhizobium elkanii sequenced genomes were compared with the closely related B. diazoefficiens USDA 110 reference genome to determine the similarities between those genomes, especially with regards to the nod operon and nod regulon. Bioinformatics analyses revealed a correlation between functional mechanisms and key elements that play an essential role in the regulation of nod gene expression. These analyses also revealed new genomic features that had not been clearly explored before, some of which were unique for some B. elkanii genomes.

Introduction

Soybean symbiotic partners mainly belong to the genus Bradyrhizobium, initially proposed as a group of slow-growing, alkaline-producing root nodule nitrogen-fixing bacteria (Somasegaran and Hoben, 1994). Genetic and physiological information, including biochemical profile, DNA homology, and phylogenomics, is essential for clarifying the differences among isolates and for supporting the taxonomic classification of over 25 species in the genus Bradyrhizobium (Jordan, 1982; Rivas et al., 2009). Among the species of this genus, considerable research efforts have focused on B. japonicum and B. elkanii because of their commercial use as a source for inoculant formulations, with many effective long-term programs for elite strain identification and selection being undertaken in different countries (Campos et al., 2001; Kober et al., 2004; Melchiorre et al., 2011).

The characterization of symbiosis itself is not an easy task because of the high degree of specialization involved in this phenomenon, particularly since a restricted group of rhizobial species/strains interacts with only a select range of plant species/varieties and vice versa. Moreover, specificity may occur at distinct stages of the interaction, from the very initial bacterial contact with the plant root to late nodule development and nitrogen fixation, resulting in biological nitrogen fixation that varies according to the host-microsymbiont combination (Schumpp and Deakin, 2010; Wang et al., 2012). In view of this, symbiosis can be conceived as a complex framework promoted by strong evolutionary forces that involves a stringent initial molecular dialogue and signal exchange between the symbiotic partners. The establishment of a successful mutualistic relationship initially requires products of the bacterial structural nod genes that encode for the synthesis of lipochitooligosaccharide (LCO) nodulation signaling molecules, also known as Nod factors (NFs). Subsequently, the recognition of symbiotic NFs by the plant triggers a signaling cascade that ultimately allows bacterial infection and induces de novo organogenesis of the nodule to accommodate the symbiont and further support nitrogen fixation (Tóth and Stacey, 2015). Bradyrhizobium diazoefficiens USDA 110 (formerly B. japonicum USDA 110) has a wide nodulation gene repertoire involved in NF assembly and modification.

In several species of rhizobia nod genes are frequently organized in an operon (nod operon), which suggests that regulation of their expression involves common mechanisms (Denarie et al., 1993). Indeed, transcription of the nodYABCSUIJnolMNO operon in B. diazoefficiens USDA 110 depends upon transcriptional activators, i.e., products of regulatory nod genes responsive to signaling molecules, such as flavonoid compounds, exuded by host plant roots (Luka et al., 1993; Loh and Stacey, 2003). Central to this regulatory circuit of nod gene expression are NodD proteins, members of the LysR-type regulator family. Upon activation by a particular flavonoid ligand NodD proteins can bind to specific DNA motifs upstream of the nod operon, the so called nod boxes, and selectively control the expression of structural nod genes in the early stages of plant-bacteria interaction (Hong et al., 1987; Henikoff et al., 1988; Goethals et al., 1992). Although multiple isoforms of NodD proteins have been identified in distinct rhizobial species, perhaps indicative of a role in expanding the plant host spectrum of these symbionts, only two were found in the B. diazoefficiens USDA 110 genome, namely, NodD₁ and NodD₂, products of the nodD ₁ and nodD ₂ genes, respectively (Gottfert et al., 1989). These two proteins show distinct expression patterns and play different functional roles in regulating the expression of structural nod genes. Active NodD₁, i.e., in the presence of a flavonoid ligand molecule such as genistein in soybean root exudates, operates as a positive transcriptional regulator of the nod operon in B. diazoefficiens. Unique to this organism when compared to other known rhizobial species, NodD₁ is not constitutively expressed; instead, it is induced by flavonoid compounds and shows autoregulation (Banfalvi et al., 1988). Although nodulation requires nod gene induction by flavonoids in most diazotrophs, the efficiency of this process depends on appropriate spatial and temporal expression of these genes. Hence, it is not surprising that, in addition to positive transcriptional regulators, nod genes are also controlled by repressor elements, as is the case for NodD₂ protein, that acts as a negative regulator of the nod operon (Loh and Stacey, 2003).

In B. diazoefficiens USDA 110, the core regulatory mechanism involving NodD₁ and NodD₂ is extended with additional regulators that act synergistically with NodD proteins to modulate the expression of nod genes. Of these, the roles of NolA, a MerR-type regulator encoded by nolA, and NodVW, the product of nodVW, that form a two-component regulatory system are particularly noteworthy. Initially identified as a soybean genotype-specific nodulation factor, NolA was later shown to be an activator of nodD ₂ involved in negative regulation with NodD₂, with both involved in the feedback and quorum regulation of nod genes (Sadowsky et al., 1991; Garcia et al., 1996; Loh and Stacey, 2003). On the other hand, the NodVW two-component regulatory system provides an alternative flavonoid responsive pathway for nod gene activation, which explains the residual nodulation of soybean plants in NodD₁ mutants (Göttfert et al., 1990; Loh et al., 1997).

As the major source for the soybean inoculant industry, strains of B. japonicum and B. elkanii differ markedly in their physiology and in their competitive fitness (Minamisawa, 1989; Vasilas and Furhmann, 1993). Such differences indicate the need to identify and characterize the genetic nature of specificity in the symbiotic relationship as a crucial step in developing guided strategies to enhance the effectiveness of soybean inoculants, given that biological nitrogen fixation ultimately provides a resource for more sustainable agricultural systems.

While there is a considerable amount of knowledge regarding the genetics and molecular mechanisms of the soybean symbiont B. diazoefficiens, including the complete genome sequence, information on the genetics of B. elkanii is restricted mainly to research focused on specific features. Although genomic data and a few drafts of the genome are available for B. elkanii, comparative genomic analyses of these two species have not been reported. In this study, publicly available B. elkanii genomes were compared with the closely related B. diazoefficiens USDA 110 reference genome (Kaneko et al., 2002) to gain some insights into the mechanisms of nod gene expression in B. elkanii, especially for the nod operon and nod regulon. The results of this analysis should provide a more comprehensive understanding of the molecular dynamics and complexity of mechanisms involved in fine-tuning signal communication between this symbiont and its host plants.

Material and Methods

Strains and genomic data

Genomic data consisting of the complete genome of B. diazoefficiens USDA 110 (reference genome) and draft genomes of B. elkanii strains SEMIA 587, CCBAU 05737, CCBAU 43297, USDA 94, USDA 3254 and USDA 3259 were obtained from the publicly available database of The National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov). Genome features and accession numbers are described in Table 1. The type-strain taxonomy of each organisms genome was confirmed by 16S rRNA sequence analysis using the RDP SeqMatch k-nearest-neighbor (k-NN) classifier (Wang et al., 2007) and checked with Basic Local Alignment Search Tool (BLAST) searches and pairwise global sequence alignments implemented in the EzTaxon server database (Kim et al., 2012).

Table 1. Characteristics of the genomes analysed.

	B. diazoefficiens	B. elkanii	B. elkanii	B. elkanii	B. elkanii	B. elkanii	B. elkanii
	USDA 110	SEMIA 587	CCBAU 05737	CCBAU 43297	USDA 94	USDA 3254	USDA 3259
Geographic location	USA	Brazil	China	China	USA	USA	USA
Number	1*	2,431	751	654	174	87	102
of contigs
Size (Mb)	9.11	8.68	9.77	9.35	9.56	8.98	8.72
Av. read coverage	-	32x	109x	120x	NA**	NA	NA
N50 (kb)	-	6,929	25,010	25,032	138,573	249,311	245,986
G+C (%)	64.1	63.6	63.5	63.8	63.7	63.8	63.9
Number of	3 rRNA	3 rRNA	3 rRNA	3 rRNA	3 rRNA	3 rRNA	3 rRNA
RNA calls	49 tRNA	40 tRNA	47 tRNA	48 tRNA	47 tRNA	47 tRNA	47 tRNA
Number of CDS calls	8648	8236	9396	8928	9122	8565	8310
NCBI BioSample	GCF000011365.1	SAMN02471378	SAMN02469485	SAMN02469464	SAMN02584917	SAMN02441448	SAMN02440775
NCBI BioProject	PRJNA17	PRJNA86995	PRJNA77219	PRJNA77219	PRJNA165317	PRJNA165319	PRJNA162999
NCBI assembly accession	GCA_000011365.1	GCA_000257685.1	GCA_000261505.1	GCA_000261525.1	GCA_000519225.1	GCA_000472765.1	GCA_000473005.1

Bioinformatics analyses

Functional annotation was done with the Rapid Annotation using Subsystem Technology (RAST) server (Aziz et al., 2008), with Glimmer set for gene calling (Salzberg et al., 1998), frameshift correction, backfilling of gaps and automatic fixing errors. Assigned functional features were triple-checked with InterProScan (Zdobnov and Apweiler, 2001) by the signature-recognition method in the InterPro database (Hunter et al., 2009), ScanProsite (de Castro et al., 2006) for protein signature matches in the PROSITE database (Sigrist et al., 2010), and BLASTp against the UniProtKB database (Magrane and Consortium, 2011). An inventory of genes involved in nodulation (structural- and regulatory-nod/nol genes) for each genome in this study is provided in Table S1 ^{(136.6KB, pdf)} . The B. diazoefficiens USDA 110 genome (Kaneko et al., 2002) was used as the reference genome and missing nod/nol genes were searched for in other genomes with BLASTn using homologous nucleotide sequences of the closely-related reference species. Possible frameshift annotation errors in assigned genes were corrected and Open Reading Frames (ORFs) were checked with the Expert Protein Analysis System (ExPASy) translate tool (Gasteiger et al., 2003) by comparing with the respective reference.

The subcellular localization of proteins was predicted using sequence-based tools in a coordinated fashion (Emanuelsson et al., 2007). Initially, SignalP 4.1 software (Petersen et al., 2011) was used to screen amino acid sequences for the presence and location of signal peptide cleavage sites characteristic of secretory proteins and this was followed by the ab initio prediction of non-classical protein secretion, i.e., secretion not triggered by a signal peptide, with SecretomeP 2.0 (Bendtsen et al., 2004). Lipoprotein signal peptides and N-terminal membrane helix prediction was done using LipoP 1.0 (Rahman et al., 2008), while the presence and location of potential twin-arginine translocation signal peptide cleavage sites was verified with TatP 1.0 (Bendtsen et al., 2005). Finally, the prediction of transmembrane topology in proteins was assessed using a combined approach based on a hidden Markov model algorithm implemented in the TMHMM 2.0 server (Krogh et al., 2001) and Phobius (Käll et al., 2004).

All bioinformatics analyses used the default parameters of the respective software.

Phylogenetic and sequence analyses

Multiple sequence alignment of NodD proteins was done by distance estimation using kmer counting and progressive alignment with log-expectation scores, followed by refinement using tree-dependent restricted partitioning implemented by MUSCLE (Edgar, 2004). The algorithm was implemented in the Molecular Evolutionary Genetics Analysis - MEGA 6.0 package (Tamura et al., 2013) and the parameters were set for the Neighbor-Joining clustering method in all interactions, with −2.9 and 0 for gap opening and gap extension penalties, respectively (center specified as 0). Subsequent phylogenetic analysis and tree reconstruction were done using the Neighbor-Joining method in the same package, with the molecular distances of the aligned sequences computed based on p-distance parameters and 1,000 bootstrap replicates and pairwise deletion treatment for gaps. Point accepted mutation (PAM) 250 calculations were used as a substitution matrix model for scoring sequence alignments (Dayhoff and Schwartz, 1978).

Protein structure prediction and alignment

Protein 3D structure prediction was done using the SWISS-MODEL web server (Biasini et al., 2014) based on evolutionarily-related structures, amino acid sequences and protein structure homologies. The technique uses hidden Markov model-sensitive searches run against the SWISS-MODEL template library (SMTL) to generate a structural model of the protein of interest. A QMEAN potential to assess model quality is then generated with an independent accuracy evaluation by the Continuous Automated Model EvaluatiOn project – CAMEO (Haas et al., 2013) based on target sequences pre-released by the Protein Data Bank (PDB) (Berman et al., 2007). The predicted protein structures were aligned with a combinatorial extension (CE) algorithm (Shindyalov and Bourne, 1998) implemented in the Research Collaboratory for Structural Bioinformatics (RCSB) PDB Protein Comparison Tool at www.rcsb.org (Bernstein et al., 1977; Berman et al., 2007; Goodsell et al., 2015). The parameters were set as 30 for the maximum gap size allowed during aligned fragment pairs (AFP) extension in fragment size m = 8, with gap open and gap extension penalties of 5.0 and 0.5, respectively.

Results

nod operon and regulatory genes

Annotation of the B. elkanii genome followed by manual curation revealed a conserved operon structure and organization of the nodKABCSUIJnolOnodZ genes in strains CCBAU 05737, CCBAU 43297, USDA 94, USDA 3254 and USDA 3259 similar to that in the B. diazoefficiens USDA 110 nod operon, with the substitution of nodY for the corresponding nodK and the lack of nolMN genes. Curiously, in B. elkanii SEMIA 587, a distinct pattern for this “canonical” gene organization was observed, with these genes scattered in the chromosome and organized as nodKABCS, a second sparse block containing nodI, nodJ, nolO and nodZ, and then nodU, which was separate from the other genes of this operon. As in the case of nolMN, nolZ was not identified in the genomes of any B. elkanii strain; likewise, nolY was not detected in the genomes of strains USDA 94, USDA 3254, and USDA 3259 (Figure 1).

Figure 1. Genetic map showing organization of the nod/nol genes in B. diazoefficiens USDA 110 (reference genome) and six B. elkanii genomes. Letters indicate the name of each gene in the genome. Gene classes are color coded and transcriptional orientation is indicated by the arrows. Relative gene distances are indicated or follow the scale presented. Gene location within the genome is represented by a straight line below the gene map of the respective chromosome (B. diazoefficiens) or contig (B. elkanii) ID.

Annotation also highlighted the presence of nodD ₁ and nodD ₂ regulatory genes in all six B. elkanii genomes and in the B. diazoefficiens USDA 110 genome. The organization of these two genes in the B. elkanii genomes followed the pattern observed for B. diazoefficiens USDA 110, i.e., they were positioned close to each other and close to the nod operon, although in opposite orientation. Similarly, the presence of nodVW that coded for the two-component regulatory elements was ubiquitous in all six B. elkanii genomes, with its location relative to the nod operon varying according to each genome. Conversely, the nolA gene showed a conserved location among the B. elkanii genomes, close to nodD ₂; the exception was for strain SEMIA 587, in which this gene was located at a position distant from nodD ₂ (Figure 1).

Analysis of nodD promoter regions

Analysis of a 250-bp region upstream of the nodD ₁ ORFs revealed that all six B. elkanii genomes and the B. diazoefficiens USDA 110 genome contained one −10/-35 σ⁷⁰ potential promoter (TTGCTA-N₁₇-TGGTAAAAT) located 46 bp upstream from the nodD ₁ CDS start site. Additionally, all sequences showed two 47-bp nod boxes, one of which corresponded to a consensus nod box sequence with the respective palindromic structure, located 11 bp from the nodD ₁ CDS start site that controls transcription of the nod operon. The second one was a presumptive nod box-like sequence located in the upstream region (84 bp) of the nodD ₁ CDS start site (Figure 2A). Interestingly, the consensus nod box sequence (Box1) overlapped the sequence of the −10/-35 σ⁷⁰ putative nodD ₁ promoter, while the presumptive nod box-like sequence was located just 6 bp upstream to it (Figure 2A).

Figure 2. Alignment of the ~250-bp region upstream from nodD ORFs in Bradyrhizobium genomes. (a) The nodD ₁ nod box sequences are boxed and show a consensus nod box sequence near the nodD ₁ ORF start codon (Box 1) and a presumptive nod box-like sequence (Box 2) upstream, with conserved residues highlighted in grey, and (b) nodD ₂ partial nod box sequence structure (boxed) found in some B. elkanii genomes. Smaller boxes denote the putative nod box motifs (ATC-N₉-GAT) with their respective palindromic structure. Yellow shaded sequences show a −10/-35 σ⁷⁰ potential promoter. Sequences read 5' to 3' from right to left, with the first nucleotide in the sequence representing that immediately prior to ATG from the nodD ₁ (a) or nodD ₂ (b) CDS start site.

On the other hand, analysis of the 250-bp region upstream of nodD ₂ revealed that only the B. elkanii SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 genomes exhibited a −10/-35 σ⁷⁰ potential promoter with the sequence TTGTCG-N₁₃-CTGTAAAAG, along with a partial nod box-like sequence located 152 bp upstream from the putative σ⁷⁰ promoter. In the B. diazoefficiens USDA 110 and B. elkanii USDA 3254 and USDA 3259 genomes, neither of these elements could be identified in the nodD ₂ promoter region, i.e., they apparently had no potential −10/-35 σ⁷⁰ promoters nor a partial (or complete) nod box-like sequence (Figure 2B).

The nolA promoter regions (250 bp) of all genomes in this study exhibited a −10/-35 σ⁷⁰ potential promoter sequence (TTGAAT-N₁₆-TTGTAGGCT), except for B. elkanii SEMIA 587. Additionally, apart from B. elkanii SEMIA 587, the nolA promoter regions displayed a high degree of sequence similarity among genomes (Figure S1A ^{(233.9KB, pdf)} ). Potential −10/-35 σ⁷⁰ promoter sequences for the nodVW regulatory gene were found in all genomes, although there were differences in the structure and location of these promoters. In contrast to nolA, the promoter regions of the nodVW genes differed considerably in sequence conservation among genomes (Figure S1B ^{(233.9KB, pdf)} ). A search for the nod box in the nolA and nodVW promoter regions revealed the absence of this regulatory motif in the 250-bp upstream gene sequences in all genomes (Figure S1A,B ^{(233.9KB, pdf)} ).

Sequence analysis of regulatory Nod proteins

All the B. elkanii genomes displayed NodD₁ ORFs of 314 amino acids, in agreement with the size of the NodD₁ protein of B. diazoefficiens USDA 110. In contrast, there was discrete variation in the size of the NodD₂ ORFs among genomes, with ORFs of 330 amino acids in B. diazoefficiens USDA 110, 331 amino acids in B. elkanii strains SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94, and 329 residues in B. elkanii strains USDA 3254 and USDA 3259 (Table S1 ^{(136.6KB, pdf)} ).

NodD₁ and NodD₂ showed a relatively high degree of conservation, although some dissimilarity among organisms ultimately clustered each B. elkanii protein into one of two groups (Figure S2 ^{(149.4KB, pdf)} ). Indeed, a PAM250 matrix showed that B. elkanii strains SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 shared 100% global sequence similarity among themselves for both NodD₁ and NodD₂; consequently, each protein set for these strains grouped together. The same situation occurred with USDA 3254 and USDA 3259, although sequence similarity decreased by 7% for NodD₁ and by 12% for NodD₂ compared to their respective homologs from other strains within the same species. NodD regulator proteins from the B. elkanii strains were still conserved when compared to B. diazoefficiens USDA 110 orthologs, especially NodD₁, which exhibited at least 93% sequence similarity, while for NodD₂ no less than 73% similarity was observed (Figure S3A-B ^{(240.2KB, pdf)} ). Additionally, conservation plots for NodD₁ and NodD₂ proteins from the B. elkanii genomes revealed lower conservation at the carboxy-terminus region of the proteins, especially for NodD₂, when compared to the reference genome of B. diazoefficiens USDA 110 (Figure S4A,B ^{(158.8KB, pdf)} ).

All the sequences analysed belonged to the LysR family of transcriptional regulators, and exhibited a 57-amino acid helix-turn-helix (HTH) LysR-type domain located at position 6-63. These domains also contained a 20-amino acid HTH DNA-binding motif from residues 23-42 in the amino-terminus region of NodD₁ and NodD₂ proteins. The NodD₁ DNA-binding motif from B. elkanii strains SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 was highly conserved and showed 100% identity with the sequence LTAAARQINLSQPAMSAAIA (Figure 3A). Curiously, B. elkanii USDA 3254 and USDA 3259 showed 100% identity with the B. diazoefficiens USDA 110 NodD₁ DNA-binding motif LTAAARKINLSQPAMSAAIA, with a single amino acid substitution occurring at position 29, in which glutamine (Q) was replaced by lysine (K) (Figure 3A).

Figure 3. Multiple sequence alignment (MSA) of the B. elkanii NodD HTH LysR-type domain and sequence logo showing conservation of the HTH DNA-binding motif. The HTH DNA-binding sequence conserved among organisms is boxed in red. The sequences are for the (a) NodD₁ and (b) NodD₂ domains.

The B. elkanii NodD₂ DNA-binding motif also split into two groups, with strains SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 showing 100% identity for the sequence LTAAARKINLSQPAMSAAVA, while for strains USDA 3254 and USDA 3259 a serine (S) replaced the threonine (T) at position 24 of NodD₂, resulting in the DNA-binding motif LSAAARKINLSQPAMSAAVA (Figure 3B). A comparison of NodD₂ among Bradyrhizobium species showed that B. diazoefficiens USDA 110 shared higher similarity with B. elkanii strains USDA 3254 and USDA 3259, although an additional amino acid residue divergence at position 25 was observed, with replacement of an alanine (A) in B. elkanii for a serine (S) in B. diazoefficiens, providing the sequence LSSAARKINLSQPAMSAAVA (Figure 3B).

In silico analysis of the subcellular location of NodD₁ and NodD₂ proteins did not predict the presence of any significant ordinary signal peptide cleavage site in their amino acid sequences or any twin-arginine signal peptide cleavage site. Lipoprotein signal peptides and non-classical protein secretory patterns were also not identified. Additional protein topology and signal peptide examination did not identify any transmembrane helices. Finally, calculations using an integrative approach algorithm indicated that NodD₁ and NodD₂ were more likely to be cytoplasmic components rather than membrane-bound proteins.

Structural analysis of regulatory Nod proteins

Global protein structure alignment of NodD₁ from the B. elkanii strains revealed high identity and similarity to NodD₁ protein from the reference genome of B. diazoefficiens USDA 110. In particular, the genomes of B. elkanii strains USDA 3254 and USDA 3259 showed 92.62% and 97.65% structural identity and similarity for NodD₁ compared to the orthologous protein from B. diazoefficiens USDA 110, respectively. Although slightly lower, the corresponding values for NodD₁ structural identity and similarity from B. elkanii strains SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 were 91.61% and 95.97%, respectively. Analysis of the NodD₁ structural alignments indicated that the higher identity and similarity of B. elkanii strains USDA 3254 and USDA 3259 with B. diazoefficiens USDA 110 resided mainly in the HTH motif of the DNA-binding domain (Figure S5A-F ^{(163KB, pdf)} ). In contrast, NodD₂ global protein structure alignments for the B. elkanii SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 genomes showed higher identity (75.17%) and similarity (84.56%) with the NodD₂ ortholog from B. diazoefficiens USDA 110 than with the B. elkanii USDA 3254 and USDA 3259 genomes, for which the identity and similarity were no higher than 74.16% and 82.89%, respectively (Figure S6A-F ^{(177.6KB, pdf)} ). Additionally, even though substantial identity and similarity were observed in the structural alignment of NodD₂ between Bradyrhizobium species, the values were lower compared to the structural alignments for NodD₁ proteins.

When the global structural alignment of NodD₁ vs. NodD₂ proteins was compared within each genome, the identity and similarity were still relatively significant. In the B. elkanii USDA 3254 and USDA 3259 genomes for example, the NodD₁ vs. NodD₂ structural alignment showed 70.47% identity and 81.88% similarity, while for the B. elkanii SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 genomes, the identity and similarity reached values of 69.46% and 81.54%, respectively. The B. diazoefficiens USDA 110 reference genome showed the most dissimilar structural alignment of NodD₁ vs. NodD₂ and consequently had the lowest identity (63.76%) and similarity (80.54%) values (Figure S7A-F ^{(175.4KB, pdf)} ). However, considering only the 57-residue protein segment corresponding to the DNA-binding domain containing the HTH motif, the NodD₁ vs. NodD₂ structural alignment values were higher in all genomes, being identical in the B. elkanii USDA 3254 and USDA 3259 genomes, i.e., 100% identity and similarity. The B. elkanii SEMIA 587, CCBAU 05737, CCBAU 43297 and USDA 94 genomes also showed better structural alignment of the amino-terminus region that accommodated the DNA-binding domain for the NodD₁ vs. NodD₂ comparison, with 91.38% identity and 98.28% similarity, compared to B. diazoefficiens USDA 110, which showed values of 81.03% and 93.10% for identity and similarity, respectively (Figure 4).

Figure 4. Structural alignment of the NodD₁ and NodD₂ 57-residue HTH domain within each genome using the Combinatorial Extension (CE) algorithm. The alignments are for B. diazoefficiens USDA 110 (a), B. elkanii SEMIA 587 (b), B. elkanii CCBAU 05737 (c), B. elkanii CCBAU 43297 (d), B. elkanii USDA 94 (e), B. elkanii USDA 3254 (f) and B. elkanii USDA 3259 (g). Orange residues represent NodD₁ superimposed on light-blue residues representing NodD₂.

Discussion

nod operon/regulon in Bradyrhizobium genomes

Lipochitooligosaccharides (LCOs) produced by the action of bacterial nodulation (nod, nol, noe) gene products are key signaling molecules in establishing the Bradyrhizobium-legume symbiosis that triggers the formation of a new organ (the nodule) in which biological nitrogen fixation occurs (Day et al., 2000). Bradyrhizobium diazoefficiens and B. elkanii can nodulate soybean plants. Despite this common ability, detailed genetic and biochemical studies involving different Bradyrhizobium species have shown distinct physiological profiles that ultimately affect their symbiotic behavior, including nodulation capacity, nitrogen fixation efficiency and competitive ability for nodule formation (Kober et al., 2004). Most of the structural and regulatory nod/nol genes with known function seem to be conserved between the B. diazoefficiens USDA 110 and B. elkanii genomes. Moreover, the presence of the structural nodY/KABCSUIJnolOnodZ genes, arranged in an operon in the B. diazoefficiens and most B. elkanii genomes, may indicate that similar patterns of LCO formation can be expected for these organisms, depending primarily upon a particular operon regulation in each genome.

The organization of nod genes in clusters/operons is a common and often conserved feature among various rhizobia genomes, many of them with a chromosomal location, as in the case of B. diazoefficiens, thereby facilitating coordination of the expression of these respective genes (Schlaman et al., 1998). Although nod genes in Bradyrhizobium are frequently organized in operons, this is not a fixed pattern, as seen in the B. elkanii SEMIA 587 genome which showed a reduced nodKABCS operon with additional nod/nol genes scattered along the chromosome. Interestingly, separation of the genes from this operon may result in the differential regulation of nod genes, ultimately providing an altered pattern of LCO formation for this strain compared to other B. elkanii genomes, even though this is not mandatory for divergent nod gene arrangement (Vázquez et al., 1991). As in most rhizobia, expression of the nod/nol operon in Bradyrhizobium is dependent on distinct flavonoid compounds produced and secreted by each leguminous host plant, with the quantity and spectrum of these molecules varying according to the age and physiological state of the plant (Phillips and Streit, 1996; Loh and Stacey, 2003; Hassan and Mathesius, 2012).

Regulation of the nod/nol operon in B. diazoefficiens and B. elkanii is essentially under control of the same elements, namely, the nod regulon that consists of nodD ₁, nodD ₂, nolA and nodVW genes, all of them present in the genomes of both species. The pattern of gene organization observed for nodD ₁ and nodD ₂, both of which were invariably located close to each other and in opposite orientation to the nod operon in Bradyrhizobium genomes, reinforced their importance as key components in the core regulatory mechanism of operonic nod gene expression. Indeed, the products of nodD genes are assumed to be fundamental in coordinating structural nod genes in many symbionts, including Bradyrhizobium species (Mulligan and Long, 1989; Honma et al., 1990; Loh and Stacey, 2003; del Cerro et al., 2015). For instance, in B. diazoefficiens products of the nodD ₁ and nodD ₂ regulatory genes orchestrate nodYABCSUIJnolMNOnodZ operon expression by activation or repression, respectively (Loh and Stacey, 2003). Such a regulatory system suggests a common mechanism that may apply for B. elkanii genomes with conserved nod gene organization.

Besides their divergent transcriptional orientation, components of the LysR-type transcriptional regulator (LTTR) family, such as NodD₁ and NodD₂, also exhibit another characteristic feature, namely, autoregulation (Maddocks and Oyston, 2008), which ultimately indicates that expression of the associated operon largely depends on the regulation of these regulatory genes themselves. Since B. diazoefficiens nodD mutants retain a large marginal ability to nodulate soybean plants, alternative transcriptional activators were proposed to take part in this process. Indeed, the identification of a two-component regulatory system consisting of NodV and NodW, products of nodVW, sheds some light on this paradox and provides a suitable explanation. In such a system, the sensor kinase component NodV can detect the environmental stimulus (such as a specific isoflavone) resulting in its autophosphorylation and subsequent signal transduction by transfer of the phosphoryl group to its cognate response regulator protein NodW, which in turn is then able to activate its target nod operon (Loh et al., 1997). Further evidence that Bradyrhizobium nodVW products are essential for efficient nodulation of mung bean [Vigna radiata (L.) Wilezek], cowpea [Vigna unguiculata (L.) Walp.] and siratro [Macroptilium atropurpureum (Moc. & Sessé ex DC.) Urb.], but are not required for soybean, suggests a host-specific role for these proteins, probably through recognition of specific flavonoid inducers produced by the host plants in response to the correlate sensor component NodV (Göttfert et al., 1990; Sanjuan et al., 1994). As shown here, we identified this same system in B. elkanii genomes, suggesting that both species possess an alternative pathway for nod operon activation and possibly a strategy for broadening their respective host ranges.

Whereas successful soybean nodulation depends on timely expression of the structural nod gene at the right place and in suitable amounts, it is reasonable to assume that negative regulation also occurs, in addition to activation. Although this negative control in B. diazoefficiens is mediated by the nodD ₂ product that operates as a nod operon repressor as mentioned before, this process is assisted by the nolA gene product, which establishes an additional level of regulation by inducing nodD ₂ expression under appropriate conditions. In other words, the product of nolA acts as an transcriptional inducer of the repressor nodD ₂ (Garcia et al., 1996). As already mentioned, NolA is a member of the MerR family of transcriptional regulators that activates transcription upon binding to specific DNA motifs and consequently induces DNA binding leading to its appropriate alignment for RNA polymerase positioning and subsequent transcriptional activation (Ansari et al., 1992; Philips et al., 2015). Moreover, nolA has the peculiar feature of encoding for the three proteins NolA₁, NolA₂ and NolA₃ that originate from three in-frame ATG start codons, with NolA₁ controlling the expression of the other two and being involved in the activation of nodD ₂ (Loh et al., 1999). This mechanism is also shared with high homology by the B. elkanii genomes and, in most cases, shows a very similar gene organization and transcription to nodD ₁ and nodD ₂, as previously demonstrated (Dobert et al., 1994). This finding indicates tight nod gene regulation in this species as well.

cis-regulatory nod elements and trans-acting factors

Considering the nature of genomic elements present in the nod operon and especially in the nod regulon of B. diazoefficiens and their marked similarity with those in the B. elkanii genomes, both in terms of structure and organization, we hypothesized that major phenotypic differences in LCO production and secretory patterns in these two species could occur at some transcriptional or post-transcriptional regulatory level. Accordingly, in the presence of the corresponding flavonoid compounds, activated NodD proteins would specifically bind to conserved cis-regulatory elements on bacterial DNA, namely, the nod boxes. These regulatory structures basically consist of a 47-bp conserved region containing the nod box consensus sequence with the palindromic ATC-N₉-GAT motif and are located upstream to the nod operon and control its expression (Rostas et al., 1986; Nieuwkoop et al., 1987; Goethals et al., 1992). The conservation of such cis-regulatory elements and their essentiality in many rhizobia species studied so far strongly suggests that NodD₁ is central in nod operon gene expression in these species, including Bradyrhizobium (Spaink et al., 1987). The nod operon activation by NodD₁ in B. elkanii genomes is no exception to this rule and seems to be under a similar, if not identical, regulatory mechanism since the same genetic features, highly conserved in relation to B. diazoefficiens USDA 110, are observed in this context.

Besides the activation of nod gene expression, an autoregulatory function unique to Bradyrhizobium is assigned to nodD ₁ by its own product NodD₁ upon binding the same isoflavone molecules that activate the nod operon. Due to divergent transcriptional orientation, the nod box sequences are located upstream from the nod operon containing the structural nod genes and the nodD ₁/nodD ₂ regulatory genes. This location of nod boxes in between the nod operon and nodD ₁/nodD ₂ within the genome complicates the analysis of their regulation. Although autoregulation by NodD₁ is also accomplished by a DNA-binding mechanism, this is reportedly achieved by binding to an alternative presumptive nod box-like sequence located upstream to the consensus nod box in B. diazoefficiens USDA 110 (Banfalvi et al., 1988; Wang and Stacey, 1991). Indeed, the conservation of this regulatory sequence in B. elkanii genomes reinforces its proposed function, a conclusion supported by the close proximity of this additional cis-regulatory region to a −10/-35 σ⁷⁰ potential promoter. Despite similarities such as high sequence conservation and location in the genome, a few differences still exist among the presumed nod box-like sequences in distinct Bradyrhizobium species, e.g., the slightly smaller size and some nucleotide divergence in B. elkanii genomes compared to B. diazoefficiens, that ultimately may affect the DNA-binding process itself.

Based on the nodD ₁ and nodD ₂ sequence-homology and their resulting description as members of LysR-type regulators, it seems reasonable to also consider a potential autoregulatory role for NodD₂, akin to what can be observed for NodD₁. Previous screening of Rhizobium japonicum USDA 191 DNA sequences for the presence of a nod box revealed that only the nodD ₁ promoter region showed high sequence homology to other nod boxes, while no extensive similar motifs were identified upstream to nodD ₂, even though several alignments of up to four homologous base pairs were observed (Appelbaum et al., 1988). Although no complete nod box has been found in the promoter region of nodD ₂ in Bradyrhizobium genomes, a remarkably conserved sequence containing only one copy of the palindrome ATC-N₉-GAT, characteristic of a nod box sequence structure, was identified in some B. elkanii genomes. Curiously, the same genomes that contained this palindromic sequence also displayed a potential −10/-35 σ⁷⁰ binding site. Since this relationship has not been reported before, the precise biological significance of this findings is still unclear and may represent another mechanism (in addition to NolA) by which nodD ₂ transcription is regulated in this species; this question deserves further investigation.

The current model for nodD ₂ expression in B. diazoefficiens essentially considers the induction of transcription by NolA upon binding to a putative NolA binding-site present upstream to nodD ₂ (Garcia et al., 1996; Loh and Stacey, 2003). Sequence homology has shown that NolA is a member of the MerR transcriptional regulator protein family that is known to activate suboptimal σ⁷⁰-dependent promoters through protein-dependent DNA distortion that ultimately provides an appropriate alignment of −10/-35 σ⁷⁰ and the correct positioning of RNA polymerase in relation to the respective promoter (Brown et al., 2003). Surprisingly, as shown here, preliminary screening for a potential NolA binding site upstream to nodD ₂ found no corresponding conserved cis-element in this region, contrary to current literature.

NodD protein sequences and structure conservation

In addition to cis-regulatory regions in the bacterial genome, trans-acting factors can also affect transcriptional regulation. The high degree of protein conservation observed for NodD₁ and NodD₂ among Bradyrhizobium genomes highlighted the evolutionary importance of the mechanism by these regulators operate. Indeed, the NodD amino acid sequence showed extensive conservation that ranged from > 90% sequence similarity among NodD proteins from organisms of the same genus to < 50% for NodD sequences from distantly related organisms (Göttfert et al., 1992). Although the global sequence similarity for NodD can vary considerably among different organisms, a general feature is the high-to-low conservation in the amino acid sequence within the protein from the amino-terminus to the carboxy-terminus of the polypeptide. This structural pattern occurs in many rhizobia species and is consistent with the function attributed to each domain of the protein (Burn et al., 1987, 1989; Horvath et al., 1987).

The amino-terminal region, the most conserved part of the protein, anchors the HTH DNA-binding motif responsible for recognition of the nod box cis-elements in bacterial DNA. Despite this higher conservation in sequence similarity, subtle amino acid substitutions nevertheless occur among B. elkanii genomes and may ultimately modify the affinity of this motif for the DNA, thereby altering the pattern of binding and consequent activity of these regulators. On the other hand, the NodD carboxy-terminus region is believed to be primarily involved in protein oligomerization, an important aspect since it is generally accepted that NodD₁ is active in a tetrameric form that arises from the cognate homodimers (Peck et al., 2013). Multiple weak interactions occur among non-specific amino acids in this region of NodD proteins and may contribute to oligomerization (Ezezika et al., 2007; Peck et al., 2013). As observed in other rhizobia, the NodD carboxy-terminus region was clearly less conserved in the B. elkanii genomes. Even though this lower sequence similarity did not markedly affect protein structure, it may play a role in oligomerization.

Based on the information available for the B. diazoefficiens USDA 110 reference genome, the similarities between B. diazoefficiens and B. elkanii, and the results obtained from this comparative genomic analysis, an extension in the model of nod gene regulation proposed for B. diazoefficiens can be applied to B. elkanii with some new added features (Figure 5). Besides the activity of all known elements in the core mechanism of nod operon regulation by the already discussed nod regulon, the presence of a partial nod box sequence upstream to nodD ₂ suggests a putative NodD₁ binding site in this region, with a possible regulatory effect on nodD ₂ as a transcriptional activation. Such a mechanism may possibly function under higher concentrations of NodD₁ as a consequence of the autoregulation of nodD ₁ expression by its respective product. Under these conditions, a surplus of NodD₁ may be “trapped” by this partial nod box sequence; NodD₁ may show lower affinity for the regulator protein relative to the conserved consensus nod box sequence and ultimately activate transcription of the repressor protein NodD₂ to balance expression of the nod operon.

Figure 5. A model illustrating the modulation of nod gene expression proposed for B. elkanii. Expression of the nod operon containing the structural nod genes is regulated by regulatory nod genes (nod regulon) in the presence of the respective flavonoid inducer. Transcriptional activation of the nod operon is mediated by NodD₁ and NodVW, resulting in the biosynthesis of lipochitooligosaccharides. Negative regulation of the nod operon is due to the action of NodD₂ and indirectly by NolA. Based on Loh and Stacey (2003).

The prediction of highly conserved structures between NodD₁ and NodD₂ and their structural alignment suggested the possible formation of NodD₁-NodD₂ heterodimers since their protein structures are very similar, but no specific amino acid residues responsible for oligomerization have yet been identified. If NodD₁-NodD₂ heterodimers are formed in Bradyrhizobium, they may well affect the NodD₁-driven regulation in this symbiont, in addition to interfering with nod operon gene expression. Although the data are suggestive of these events, there is still no conclusive evidence for this hypothesis and experimental confirmation of such mechanisms is required to prove the validity of this model.

Our knowledge of the genetic control of nodulation and our understanding of the mechanisms involved in the early events of a new symbiotic relationship have undoubtedly expanded considerably in the last two decades. Although a general framework for nodulation and symbiosis has not yet been developed for all diazotrophic microorganisms, given their genetic and phenotypic diversity, comparative studies may nevertheless help to clarify these processes and provide some useful preliminary evidence for future investigations. As microbial genome sequencing projects are quickly delivering an ever-increasing amount of data, comparative analyses of cis-elements and their cognate trans-acting factors could provide new insights into the activation and/or repression of nod gene expression.

This work has identified several similarities between B. elkanii genomes and the closely related B. diazoefficiens. Based on these similarities, it was possible to identify and correlate functional mechanisms and key elements that play an essential role in regulating nod gene expression. In addition, new genomic features that had not been clearly explored before, some of which were unique to certain B. elkanii genomes, have raised new questions for future research.

Supplementary material

The following online material is available for this article: