Identification of core TCS genes in Wolbachia pipientis

The widespread use of TCS by eubacteria raises the question of how widely these genes have been retained in endosymbiotic Wolbachia bacteria. Prior studies indicate that the Wolbachia relatives A. phagocytophilum and E. chaffeensis carry the TCS pairs: cckA/ctrA, pleC/pleD and ntrY/ntrX (Kumagai et al. 2006; Lai et al. 2009). This annotation is based on deduced amino acid sequences, which exhibit 55–67% similarity to the TCS homologs in C. crescentus (Table S1). In accordance with this, we used predicted amino acid sequences from the closer phylogenetic relative, E. chaffeensis, to identify TCS homologs in Wolbachia (Brouqui and Matsumoto 2007). We searched the genomes of 12 completely or near completely sequenced Wolbachia strains, which are classified in supergroups A–D, and represent symbiosis with a range of insect and nematode hosts (Table 1). This revealed that, in addition to a few previously examined strains, all Wolbachia lack detectable homologs for ntrY and ntrX, whereas corresponding homologs for the four other TCS genes were ubiquitously detected (Table 2). One of the exceptions was wOo, in which pleC is annotated as a pseudogene. In three other cases, a single TCS gene is predicted to be split into multiple open reading frames (ORFs). This is seen for pleD of wOo as well as for cckA of wAlbB and wMel.

A split ORF in any Wolbachia TCS gene could dramatically affect signaling processes in a system lacking functionally redundant genes. To confirm the basis for the split ORF predictions, we re-examined the deposited sequences of wOo pleD, wAlbB cckA, and wMel cckA genes. For wOo pleD, nucleotide sequence alignments and visual inspection revealed multiple nucleotide substitutions leading to four stop codons between wOo_06950 and wOo_06960. A frameshift was also detected that positions these ORFs in different reading frames. Because these data cannot be substantiated by any single sequencing error in relation to other Wolbachia pleD genes, these findings are consistent with a split ORF prediction in the wOo pleD locus.

Investigating the basis for the prediction in wAlbB cckA locus revealed five in-frame stop codons, partitioning the gene into two annotated ORFs, WALBB_620009 and WALBB_620010. Because all of these changes could be attributed to a single nucleotide deletion, it was unclear whether this change was genuine or reflected an artifact in the deposited sequence. Our re-sequencing of this cckA region, using wAlbB DNA isolated from both A. albopictus tissue culture cells and intact mosquitoes, revealed an exact match with the deposited sequence. Thus, data obtained from our two independent samples confirm the split ORF prediction for wAlbB cckA.

Analysis of the genomic region for wMel cckA also indicated that the split ORF prediction was potentially attributable to a single nucleotide addition in the deposited sequence, creating a stop codon that partitioned wMel CckA into the ORFs WD1215 and WD1216. To verify whether this split ORF prediction is accurate, we sequenced cckA of wMel carried by D. melanogaster (Serbus and Sullivan 2007) and in a transinfected D. simulans strain (Poinsot et al. 1998). As controls, cckA was also sequenced from wMelPop and wRi, attained from lab strains of D. melanogaster and D. simulans, respectively. We found that the ∼2.5-kB fragment sequenced from wMelPop and wRi cckA exactly matched the Genbank record. This was also the case for nearly all of the wMel cckA sequence from both Drosophila hosts, including the wMel-associated SNP found at position 2402 (Chrostek et al. 2013). However, both of the re-sequenced wMel cckA samples lacked the frame-shifting cytosine at position 1149 of the deposited wMel cckA sequence (Figure S2). This indicates that wMel cckA is more likely encoded by a single ORF, analogous to cckA in other Wolbachia strains. Further analysis of wMel CckA, presented below, is done in accordance with this finding.

Identification of TCS-related genes in Wolbachia pipientis

The presence of TCS genes in Wolbachia raises other questions about how well the overall TCS regulatory network is conserved. In the Caulobacter system, a complex network of kinases and phosphotransfer proteins affects the signaling ability of CckA and PleC (Ausmees and Jacobs-Wagner 2003; Biondi et al. 2006). These include DivL, an HK-related tyrosine kinase that promotes CckA signaling; ChpT, an intermediary phosphotransfer protein; CpdR and DivK, response regulators that can also interact with CckA; and DivJ, an HK whose activity directly opposes that of PleC. No homologs for chpT, cpdR, divK, or divJ have been reported for Anaplasma or Ehrlichia, and our analyses did not identify homologs in Wolbachia (Brilli et al. 2010). However, coding sequence homologous to Caulobacter divL was widely shared between the Anaplasmataceae and Wolbachia (Table 2, Table S1). This sequence, encoding an approximately 400-amino-acid-long N-terminal fragment of DivL, will be referred to as dvlL (for DivL-like) in this analysis. A. phagocytophilum, E. chaffeensis, and 11 of 12 Wolbachia strains analyzed all contained dvlL. The status of dvlL was inconclusive in the wUni Wolbachia strain due to lack of sequence coverage in that region of the genome (Table 2). The importance of DivL in well-characterized bacterial systems and the conservation of dvlL in Wolbachia open the possibility that DvlL interacts with other Wolbachia TCS components.

α-Proteobacteria are known to carry other factors that modulate CtrA activity as well (Christen et al. 2006; McGrath et al. 2006; Gorbatyuk and Marczynski 2005). These include CcrM, a methyltransferase that modifies the ctrA promoter region; GcrA, a transcriptional activator of ctrA; and SciP, a transcriptional repressor of CtrA-regulated genes. Neither Anaplasma nor Ehrlichia has been reported to carry homologs for ccrM, gcrA, or sciP (Brilli et al. 2010; Tan et al. 2010; Fioravanti et al. 2013; Stephens et al. 1996). However, the majority of sequenced mosquito and fruit fly Wolbachia strains contained anywhere from one to three copies of the ccrM gene (Table S2). Because these strains also carried 2 CcrM methylation sites within 400 base pairs of the ctrA start site (unpublished observation), the presence of ccrM has possible implications for Wolbachia TCS and cell cycle regulation.

Many α-proteobacteria have been shown to use additional regulatory proteins to drive shutdown of CtrA and PleD outputs through degradation (Christen et al. 2006; McGrath et al. 2006; Gorbatyuk and Marczynski 2005). These include ClpX/P, a protease that degrades CtrA, clearing the origin of replication (ori) for DnaA to bind and initiate DNA replication; RcdA and PopA, which facilitate CtrA interaction with ClpX/P; and EAL-domain phosphodiesterase proteins, which hydrolyze the c-di-GMP second messenger produced by PleD (McGrath et al. 2006; Ryan et al. 2004; Jenal and Fuchs 1998; Simm et al. 2004; Christen et al. 2005). Consistent with prior analyses of other Anaplasmataceae, rcdA, popA, and any EAL domain–encoding genes could not be identified in sequenced Wolbachia strains (Taylor et al. 2009; Ozaki et al. 2014; Cheng et al. 2006). However, homologs were identified for clpX and clpP, as well as for dnaA in 12 of 12 sequenced Wolbachia strains (Table S1, Table S2). These results taken together indicate that Wolbachia have retained a subset of factors that regulate TCS activity.

Genome-wide positioning of TCS-related genes in Wolbachia pipientis

The positioning of genes throughout the bacterial genome has a strong impact on relative expression throughout the cell cycle (Condon et al. 1992). Given the evidence that Wolbachia share core TCS-related genes with Anaplasma and Ehrlichia, we asked whether the overall positioning of these genes is also conserved in Wolbachia. To address this, we created syntenic alignments using the genomes of completely assembled Wolbachia strains. These were aligned with respect to the ori locus and oriented according to the proximal hemE gene (Ioannidis et al. 2007). The relative positions of conserved TCS-related genes were then plotted on this map, with the ori for all genomes shown at position 0′ and the terminus at the relative position of 6′ (Figure 1, Table S3).

This analysis indicated that a subset of TCS-related genes was similarly positioned with respect to the ori in A. phagocytophilum, E. chaffeensis, and Wolbachia. This includes ctrA, positioned approximately 2′–3′ distant from the ori, dvlL, closely associated with ctrA in Wolbachia; and pleD, positioned approximately 3′–5′ from the ori. Positioning trends for cckA, pleC, and clpX/P were also visible between A. phagocytophilum and E. chaffeensis, as well as between Wolbachia strains, but not between the three genera collectively. Copies of the ccrM gene, absent from A. phagocytophilium and E. chaffeensis, were generally positioned 4′–5′ distant from the ori in fly and mosquito Wolbachia strains. Wolbachia cckA and pleC were positioned closer to the ori, whereas clpX/P was positioned more distantly than in A. phagocytophilium and E. chaffeensis. In addition, the clustering of dvlL and clpX/P genes seen in A. phagocytophilium and E. chaffeensis was not shared by the Wolbachia genomes, which consistently showed dvlL proximal to the ctrA locus (Figure 1, Table S3). This differential positioning raises the possibility that Wolbachia TCS gene dosage may differ appreciably from A. phagocytophilum and E. chaffeensis during the cell cycle (Couturier and Rocha 2006).

Immediate context of the core Wolbachia TCS genes

To further evaluate the genomic context immediately flanking the TCS genes of A. phagocytophilum, E. chaffeensis, and Wolbachia, we aligned these regions and analyzed them with several operon prediction programs (Table S4) (Price et al. 2005a,b; Dam et al. 2007; Mao et al. 2009). This revealed some variation in the context of all shared TCS loci. For A. phagocytophilum and E. chaffeensis, the cckA gene was closely flanked by the genes o-methyltransferase and cutA, which encode a cation tolerance protein (Figure 2A). However, cckA in all Wolbachia strains was neighbored at its 5′ end by the hemF gene, which supports heme biosynthesis (Heinemann et al. 2008). Furthermore, all sequenced Wolbachia genomes, except the phylogenetically distant strains wBm and wOo, showed cckA as being flanked at its 3′ end by parA and parB, which encode chromosomal partitioning proteins (Figure 2A) (Foster et al. 2005).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Synteny and operon predictions for the genetic regions surrounding the (A) cckA-, (B) ctrA-, (C) pleC-, and (D) pleD- genes for A. phagocytophilum (APH), E. chaffeensis (ECH), and the Wolbachia strains indicated. Each line represents a genetic region from the organism/strain indicated. *Region surrounding ctrA in wOo is adjacent to the region surrounding the predicted pseudogene for pleC. Color-filled arrows are predicted ORFs in their respective orientations; white arrows are predicted pseudogenes. Similarly colored arrows are ORFs predicted to share a common operon based on data from Table S4; open arrows indicate an ORF that extends beyond the region shown. Gene names are referenced along with locus tag information in Table S4.

A similar type of contextual variation was evident for Wolbachia ctrA. In A. phagocytophilum and E. chaffeensis, ctrA was flanked upstream by a gene encoding a helix-turn-helix (hth) DNA binding protein and downstream by xnse, which encodes a 3′–5′ exonuclease family protein (Figure 2B). However, in nearly all sequenced Wolbachia strains, ctrA appeared to share an upstream region with an operon that contains dvlL, as well as the genes glycerol-3-phosphate dehydrogenase, phosphotidylglycerophosphatase A, and an acetyltransferase (Table S4). This genomic arrangement was similar in wBm and wOo, although the neighboring operon may be fragmented or incomplete. The 3′ end of Wolbachia ctrA was flanked by a variety of genetic regions that differed according to supergroup (Figure 2B). Thus, the genomic context of cckA and ctrA is generally conserved between Wolbachia strains, although not between Wolbachia and other Anaplasmataceae.

In contrast, the immediate context of pleC and pleD appeared relatively more conserved. Analysis of the A. phagocytophilum and E. chaffeensis pleC region suggested that pleC shares a promoter with the nitrogen metabolism gene argD (Velasco et al. 2002), with its 3′ end flanked by either hypothetical genes or the mutL membrane protein gene (Figure 2C). Interestingly, in all sequenced Wolbachia genomes except wOo, which lacks detectable homologs for both genes, pleC ORFs were predicted to share a promoter with argD, analogous to Anaplasma and Ehrlichia. However, Wolbachia pleC was also flanked by peroxiredoxin and the recombination gene recF at its 3′ end, indicating that the pleC genomic region is not entirely conserved (Figure 2C).

Examination of the pleD region suggested a similar extent of conservation between species. In A. phagocytophilum and E. chaffeensis, pleD was neighbored at the 5′ end by glutamate dehydrogenase B and a short hypothetical protein ORF denoted as hp (Figure 2D). This gdhB-hp-pleD cluster was predicted to form an operon in Ehrlichia (Table S4). Interestingly, a gdhB-hp-pleD–containing operon was also consistently predicted in Wolbachia, with the addition of a chaperonin gene, clpB, included at the 5′ end of the operon (Figure 2D). Thus, considerable homology is evident in the genomic context of pleC and pleD among Wolbachia strains, some of which is shared with other Anaplasmataceae.

Comparison of domain structure between TCS homologs

If Wolbachia TCS proteins are functional, then the predicted products should carry the domains and key residues important for activity. To resolve this issue, we compared the predicted functional domains of the Caulobacter TCS proteins against A. phagocytophilum, E. chaffeensis, and Wolbachia. A. phagocytophilum and E. chaffeensis CckA exhibited features typical of a hybrid-HK (Figure 3A, Figure S1) (Dutta et al. 1999). The N-terminal sensor region of CckA contained a transmembrane domain, followed by a region of predicted secondary structure indicating classic PAS-fold architecture (see Materials and Methods; Table S5, Table S6). Two of these “PAS-like” domains were found in A. phagocytophilum and one was found in E. chaffeensis. Neighboring this N-terminal “sensor” portion, a dimerization/histidine phosphotransfer (DHp) domain was predicted. The DHp contained the conserved His residue, as well as two closely flanking residues that impart both kinase and phosphatase capabilities to the DHp (Figure 4A) (Willett and Kirby 2012). Following the DHp domain was an internal ATP-catalysis domain (CA) with a conserved asparagine (Asn), and a C-terminal REC domain with a conserved Asp (Figure 3A, Table S5) (West and Stock 2001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Domain architecture of deduced amino acid sequence for the TCS pairs (A) CckA/CtrA and (B) PleC/PleD. For comparison, C. cresentus (C.cr.) domains, as well as those from A. phagocytophilum (APH) and E. chaffensis (ECH), are shown. wMel is represented by predicted architecture for the independently sequenced strains from this study. CA, catalytic-ATPase domain; CC, coiled-coil; DHp, dimerization and histidine-phosphotransfer domain; GGDEF, di-guanylate cyclase domain; HTH, helix-turn-helix DNA-binding domain; PAS, P(er) A(rnt) S(im)-like sensor domain fold; REC, response-receiver domain; TM, trans-membrane region; D, aspartate; H, histidine; Mg²⁺, magnesium; N, asparagine; P, phosphate; Y, tyrosine

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Analysis of CckA/CtrA and PleC/PleD cognate pairing-residue alignments. (A) CckA HK domain, (B) CtrA REC domain, (C) PleC DHp, and (D) the PleD N-terminal REC domain. Identical/similar residues are similarly colored. Amino acid numbers shown above are for the C. cresentus sequence. Asterisk indicates conserved phosphorylation sites, K indicates residue necessary for kinase function, P indicates residue necessary for phosphatase function (Willett and Kirby 2012). Boxed residues in alignments indicate covarying residues critical in specifying cognate HK/RR interaction (Capra et al., 2012a; Podgornaia and Laub 2013).

Analogous to other Anaplasmataceae, most Wolbachia CckAs were predicted to carry internal DHp and CA domains, a C-terminal REC domain, and all the key functional residues associated with those domains (Figure 3A, Table S5) (Cheng et al. 2006; Kumagai et al. 2006). One exception to this was wAlbB, truncated partway into the C-terminal REC due to a split ORF and lacking the conserved Asp residue, confirmed by our sequencing results. All Wolbachia CckAs were predicted to have two N-terminal transmembrane domains, except wNo. Predicted secondary structures also indicated that all Wolbachia CckAs carried at least one PAS-like domain (Figure 3A, Table S5, Table S6). The conservation of these structural features suggests a functional role for CckA has been conserved in Wolbachia. Furthermore, examination of DvlL domain structure indicated the three previously identified PAS domains, as well as complete conservation of DvlL between all Wolbachia strains (Figure S3, Table S5) (Childers et al. 2014). This raises the possibility that CckA regulation, as seen in the well-defined free-living α-proteobacterial model Caulobacter, may be at least partly conserved in Wolbachia as well.

The response regulator CtrA was strikingly conserved in its domain structure between C. crescentus, A. phagocytophilum, E. chaffeensis, and Wolbachia. In all cases, CtrA was predicted to carry an N-terminal REC domain with a conserved Asp residue (Figure 3A, Figure S1, Table S5). The C-terminal helix-turn-helix (HTH) domain was also confirmed, and all Wolbachia strains carried the conserved α3-helical residues required for DNA binding (Figure 3A, Figure 5A) (Martinez-Hackert and Stock 1997; Quon et al. 1996; Lang and Beatty 2000; Bird and Mackrell 2011). This conservation suggests that the phospho-acceptor and DNA-binding properties of Wolbachia CtrA are analogous to CtrA in other α-proteobacteria. Analysis of seven Wolbachia genomes also identified 34 to 55 ORFs with upstream consensus CtrA binding sites, further supporting a role for Wolbachia CtrA in vivo (Table S7).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Analysis of CtrA and PleD output domain alignments. (A) CtrA HTH domain. (B) PleD GGDEF domain. Identical/similar residues are similarly colored. Amino acid numbers shown are for the sequence from C. cresentus. The HTH-DNA-binding motif in (A) is boxed with a dashed line, and the DNA-sequence-recognition α3 helix is boxed with a solid line. In the PleD GGDEF alignment in (B), the active site residues are boxed with a solid line and the positions of I-site and metal-binding (MB) residues are marked.

Predicted structural domains were also examined in PleC and PleD. A. phagocytophilum and E. chaffeensis PleC domain structure was similar to C. crescentus PleC, with predicted N-terminal transmembrane domains, an internal DHp domain, and a C-terminal CA domain, all carrying key functional residues, although no PAS or PAS-like domains were detected (Figure 3B, Figure 4C, Figure S1, Table S5, Table S6). The Wolbachia PleCs were similarly organized in nearly all strains, carrying a pair of transmembrane domains, an internal DHp domain, a C-terminal CA domain, and all key residues. PleC of wPip JHB was distinctive in the loss of a transmembrane domain, and wOo was, as noted, predicted to lack PleC altogether (Figure 3B). This suggests that most Wolbachia PleC proteins function similarly to PleC in other Anaplasmataceae (Kumagai et al. 2006; Lai et al. 2009).

The predicted domain structure of the PleD RR also appears widely conserved. As detected in C. crescentus, A. phagocytophilum, and E. chaffeensis, nearly all Wolbachia PleD proteins were predicted to carry two N-terminal REC domains with conserved Asp residues (Figure 3B, Table S5). One exception was wAlbB PleD, which carried an Asp-to-Tyrosine substitution in the internal REC domain. The other exception was wOo PleD, in which the REC domains were separated by a split ORF, and the dissociated REC carried an Asp to Asn substitution. The GGDEF domain at the PleD C-terminus was also shared between Wolbachia and other Anaplasmataceae (Figure 3B). Twelve out of 14 key catalytic residues in the GGDEF were identical between all species and strains examined (Figure 5B) (Chan et al. 2004). Complete conservation was observed in all key residues of the GGDEF I-site, which is known to inhibit catalytic function in response to c-di-GMP binding (Christen et al. 2005; Christen et al. 2006). These results suggest that the majority of Wolbachia PleDs have similar functional and regulatory capacity as PleD of related bacteria.

Analysis of cognate specificity residues in Wolbachia TCS proteins

The conservation of key functional domains in Wolbachia TCS proteins raises the question of whether they interact as exclusive functional pairs or are capable of cross-talk. Prior work comparing HK/RR pairs from 200 bacterial genomes has indicated a subset of residues that specify interaction within a cognate pair (Skerker et al. 2008; Capra et al. 2010). Nine residues in the HK DHp domain form a spatially constrained interface with seven residues in the REC domain of the cognate RR. Pairs of residues within this interface have been shown to co-vary between species. In vitro studies also show that mutating two to three residues in the HK DHp domain or three to four residues in the RR REC domain changes the specificity of HK/RR interaction (Skerker et al. 2008; Bell et al. 2010; Capra et al. 2010, 2012a). To assess the likelihood of exclusive CckA/CtrA and PleC/PleD interactions in Wolbachia, we examined the cognate specificity residues in these proteins through amino acid alignments with other Anaplasmataceae homologs informed with data from co-crystalized HK/RR pairs of major model systems (Casino et al. 2009; Capra et al. 2012a; Capra et al. 2010).

Analysis of CckA DHp cognate specificity residues revealed that seven out of nine key amino acids were identical between other Anaplasmataceae and Wolbachia (Figure 4A). Both of the nonhomologous amino acids in Wolbachia CckA were at positions known to co-vary in other species (Figure 6A) (Bell et al. 2010; Capra et al. 2010, 2012b). Furthermore, the amino acid identities of these key residues were identical in all Wolbachia strains (Figure 4A). By contrast, the cognate specificity residues of the CtrA REC domain displayed little homology between Anaplasmataceae and Wolbachia, with only two out of seven amino acid identities shared between the genera (Figure 4B). The majority of these nonconserved residues in Wolbachia CtrA were not explainable by covariation (Figure 6A). However, the identity of cognate specificity residues in CtrA was shared between all Wolbachia strains (Figure 4B). This indicates that, overall, CckA and CtrA residues that specify cognate pairing are highly conserved within Wolbachia. However, it is unclear whether they have retained an exclusive pairing affinity (Cheng et al. 2006; Kumagai et al. 2006).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Comparison of co-evolving residues in cognate pairs from Wolbachia. Amino acid residues that specify cognate pairing for the (A) CckA/CtrA pair and the (B) PleC/PleD pair are listed. Change in Wolbachia sequences from E. chaffeensis identities are indicated by a neighboring triangle (Δ). *The majority of residues at that position are unchanged. Lines connecting HK and RR positions in the alignment indicate potential covariation for Wolbachia pairs corresponding with an “adjusted mutual information score” of higher than 3.5 using high-value pairing of canonical histidine kinases and response regulators (Capra et al. 2012a).

We also investigated potential for specificity of Wolbachia PleC/PleD interaction. Compared against E. chaffeensis PleC, most Wolbachia PleC proteins were homologous at six out of nine cognate specificity residues in the DHp domain (Figure 4C). Supergroup B Wolbachia strains were distinct, showing homology at five out of nine residues (Table 1, Figure 4C). These divergent Wolbachia PleC residues corresponded to sites of predicted covariation (Figure 6B) (Capra et al. 2012a). By contrast, the PleD N-terminal REC domain was less conserved, with only two to four out of seven cognate specificity residues shared between E. chaffeensis and Wolbachia (Figure 4D). The nonhomologous residues varied along phylogenetic lines, with wOo PleD of supergroup C showing the greatest divergence. Interestingly, the four positions with strongest potential for covariation did coincide with Wolbachia PleD polymorphisms (Figure 6B). These data indicate that PleC/PleD cognate specificity residues are less conserved between Wolbachia than those seen for CckA/CtrA. However, as divergence of Wolbachia PleC and PleD sequences could largely be explained by covariation, it remains possible that PleC/PleD function as a cognate pair.