The putative HKs and RRs in Psy B728a, Pto DC3000 and Pph 1448A were identified by searching the complete genome sequences for proteins containing HK and RR domains using Pfam HMM profiles. Four CheA-like HKs in each P. syringae genome were identified in BLASTP searches using as template the CheA HK of E. coli 31 (Table 1). In addition, BLASTP searches of the HKs and RRs found in each P. syringae pathovar against the genomes of the other two pathovars allowed the identification of additional HKs and RRs. The genomes of P. syringae pathovars were found to contain large numbers of genes encoding TCS proteins: 68 HKs and 93 RRs in Psy B728a, 69 HKs and 95 RRs in Pto DC3000, and 70 HKs and 92 RRs in Pph 1448A (Table 1; see Additional File 1 and 2). The number of genes encoding hybrid HKs (REC-HKs) was 20 in Psy B728a, 22 in Pto DC3000 and 24 in Pph 1448A (Tables 1 and 4). The HMM search method used in this work retrieved hybrid HKs as well as RRs (Table 1). No TCS proteins were identified on any of the plasmids of Pto DC3000 and Pph 1448A. In recent studies, similar numbers of TCS proteins for Pto DC3000 have been reported: 69 HKs 2829 and 71 RRs, 21 of which were hybrid HKs 12; or 64 HKs in a study not including CheA-like HKs, 20 of which where hybrid HKs 30. Although the number of ECF sigma factors in all three P. syringae genomes (10 ECF sigma factors) is only about half that found in other Pseudomonas species 2432, the number of TCS proteins is close to that found in other Pseudomonas genomes 33.

HK and RR genes were scattered over the entire chromosomes of the three P. syringae pathovars. Conservation of the genetic organization between HK and RR genes was analysed in the genomes of Psy B728a, Pto DC3000 and Pph 1448A allowing the identification of gene clusters containing HKs and RRs that constitute putative TCSs (Table 2). Like in other bacterial species, many P. syringae HKs and RRs were encoded by clusters of adjacent genes: 37 putative clusters of complete TCS genes in Psy B728a, 34 in Pto DC3000, and 33 in Pph 1448A (Table 2). For the remaining HK or RR genes, their partner genes could not be predicted from genetic organization and, therefore, they were considered as orphan HKs or RRs. The orphan HKs were 11 in each P. syringae genome, and the number of genes encoding orphan RRs was very high: 36 in Psy B728a, 38 in Pto DC3000 and 35 in Pph 1448A (Table 3). Finally, the comparative genomic analysis allowed the identification of a core of complete TCS protein orthologues among the three P. syringae pathovars, that is composed by 30 putative TCS clusters (HK and RR) (Table 2), 11 orphan HKs, 33 orphan RRs (Table 3), and 16 hybrid HKs (Table 4).

HKs have been classified on the basis of phylogenetic analyses and the sequence relationships of the residues surrounding the H-box 781734. Furthermore, several new domains with putative biological functions have been described in HKs, and domain architecture has proven particularly informative for analysing multi-domain proteins involved in signal transduction 2111235. The phylogenetic analysis and examination of the region around the H box of P. syringae HKs showed that three of the five major HK types found in E. coli 8 were present in P. syringae: Type I (IA, IB, IC), III, and CheA-like HKs (Table 1; see Additional File 1). In contrast, Type II and IV HKs were totally absent from P. syringae. However, the LytS-like HK FimS/AlgZ and HKs containing GAF domains did not cluster within any of the defined HK types of E. coli 8, and formed two separate HK groups: LytS-like HKs and GAF-HKs. GAF sensor domains are commonly found cytoplasmic signalling domains in the N-terminal region of HKs 234, and appear to act as binding sites for small ligands, such as cyclic nucleotides (cAMP and cGMP) and small molecules, which modulate the catalytic activity of the target protein 3637. In addition, analysis of domain architecture of P. syringae HKs showed a conserved core structure for each HK type in P. syringae (Figure 1). The conserved core of Type III HKs and LytS-like HKs only had a HK-like ATPase (HATPase_c) catalytic domain and a His_kinase domain, respectively. The conserved core of CheA-like HKs contained a C-terminal CheA regulatory domain but lacked the HisKA domain. The conserved core of Type I HKs and GAF-HKs had a central region with HisKA and HATPase_c domains fused to additional domains on the N-terminal end: a HAMP domain in Type IA, a PAS domain in Type IC, and GAF plus phytochrome (PHY) binding domains in GAF-HKs (Figure 1).

Orphan HKs fell into two HK types: Type I (IA, IB and IC), and GAF-HKs (Table 3); and hybrid HKs of P. syringae belong either to the Type I (IB and IC) or CheA-like HKs (Table 4). PSYR3504 (BphP1) and PSYR2385 (BphP2) HKs have been previously described as bacteriophytochromes (BphPs) that belong to the HWE_HK family 438. Similar to other BphPs, the bphP1 (PSYR3504) gene of P. syringae pathovars is located in an operon downstream from a bphO gene, encoding a putative heme oxigenase.

RRs show a great variety of output domains and domain combinations. Recently, bacterial and archaeal RRs have been classified into families based in their domain architectures 12. RRs typically consist of an N-terminal REC domain fused to a C-terminal HTH DNA-binding output domain (OmpR, NarL, NtrC, LytR, AraC, Spo0A, Fis, YcbB, RpoE, and MerR) that activates or represses transcription of specific target genes 212. In addition, prokaryotic genomes encode a variety of RRs with unusual domain organization: RRs with enzymatic output domains (GGDEF, EAL, HD-GYP, CheB, CheC, PP2C, and HisKA), RRs with RNA-binding output domains (ANTAR and CsrA), RRs with protein- or ligand-binding output domains (CheW, PAS, GAF, TPR, CAP_ED, and Hpt), RRs with the REC domain as a stand-alone module, and RRs with domains of unknown function 12. The RRs identified from the genomes of P. syringae pathovars were assigned to these different RR families 12 according to the domain architecture and phylogenetic analysis (Table 1; see Additional File 2).

Bacterial RRs without a REC domain are extremely rare, but a number of enhancer-binding proteins (EBPs) lack the REC domain and normally function as RRs 39. EBPs are involved in the activation of the bacterial transcription by interaction with the sigma-54 RNA polymerase holoenzyme 40. In P. syringae, the HrpR and HrpS proteins show a high sequence similarity to the NtrC family of transcriptional RRs and have been previously identified as unusual EBPs lacking the N-terminal REC domain; however, similar to other EBPs, they retain the domain that interacts with the sigma-54 RNA polymerase holoenzyme plus the C-terminal DNA-binding domain 39404142. In addition, the NarL-like RR CorP of Pto DC3000 that is involved in the regulation of coronatine biosynthesis 4344 also lacks the REC domain. Thus, HrpR, HrpS and CorP proteins were not identified during the search of RRs in P. syringae genomes with the HMM profile that targets the RR REC domain, nevertheless these proteins were considered orphan RRs (Table 3).

A close analysis of the distribution of genes encoding TCS proteins revealed that there are important differences in TCS proteins among the three pathovars of P. syringae that may contribute to their diverse host ranges and association with particular host plants. A number of the identified TCS genes were unique to each P. syringae pathovar without counterparts in the other two pathovars. The corRSP regulatory region (PSPTO4704-4706) of coronatine biosynthesis and the copRS TCS (PSYR1497/1498) regulating copper resistance were only present in Pto DC3000 and Psy B728a, respectively. Other TCS genes unique to each P. syringae pathovar were: PSYR2114, PSYR2939, PSYR2940 and PSYR3084 in Psy B728a; PSPTO0785/0786, PSPTO2329, PSPTO4079, PSPTO4080 and PSPTO5573/5574 in Pto DC3000; PSPPH0770, PSPPH0778, PSPPH0944 and PSPPH1261 in Pph 1448A. The unique hybrid HKs PSPPH0770 and PSPPH0944 were flanked by transposases. However, the unique RRs PSPTO2329 and PSPTO5574 were disrupted by transposases 2527, and it is unlikely that these genes encode functional products. Finally, 11 TCS proteins were only shared between two of these P. syringae pathovars.

Variations among P. syringae pathovars were also produced by the insertion of mobile genetic elements or point mutations in TCS genes resulting in disrupted reading frames. PSPTO2326 and PSPPH2083 encoded truncated hybrid HKs by comparison with the length of their orthologue PSYR2113 (Table 4) that is located next to the unique RR PSYR2114. PSPTO2326 and PSPPH2083 were located adjacent to a transposase and to a site-specific recombinase, respectively. Probably these elements caused the disrupted hybrid HKs and the lack of PSYR2114 orthologues in Pto DC3000 and Pph 1448A. Similarly, PSPTO2983 (baeS2) and PSPPH2510 encoded truncated HKs compared to the length of their P. syringae orthologues, and PSPPH2980 was interrupted by an ISPsy18 transposase. PSPTO2983, PSPPH2510 and PSPPH2980 HKs were unpaired without a RR gene in its vicinity, whereas their P. syringae orthologues are located on TCS gene clusters with adjacent RRs (Table 2).

Although the PSPPH1362 gene was disrupted by an authentic frameshift, Psy B728a (PSYR1292) and Pto DC3000 (PSPTO1482) orthologues encoded intact hybrid HKs with similarity to BvgS of Bordetella species that controls the regulation of many virulence factors 45. In each pathovar, these hybrid HK genes were adjacent to orphan RR genes transcribed in the same direction (PSYR1293, PSPTO1482 and PSPPH1363), and their encoded proteins exhibited significant homology to the PvrR RR of P. aeruginosa PA14 which controls antibiotic susceptibility and biofilm formation 46, and to the virulence related protein VieA of Vibrio cholerae 47.

The identification of HKs and RRs is based on the computational domain analysis of protein sequences. The approach used to identify putative HKs and RRs from the complete genome sequences of Psy B728a, Pto DC3000 and Pph 1448A was similar to that described previously 33 with slight modification. Briefly, five different HMM profiles (accession numbers PF00512, PF07568, PF07730, PF07536 and PF06580) were found in Pfam database that target different families of HKs (HisKA, HisKA_2, HisKA_3, HWE_HK and His_kinase). The HWE_HK domain is defined by the absence of a recognizable F box, and the presence of a highly conserved H residue and a WxE motif within the N and G1 boxes of the C-terminal transmitter domain, respectively 4. These five different HMM profiles were used to recognize the different HKs in the P. syringae genomes, and hits with a E-value below a selected cut-off (10^-6) were extracted. A profile HMM downloaded from Pfam protein families database 62, which targets the RR REC domain (accession number PF00072), was used to recognize the RRs in each P. syringae genome. Hits with an E-value below a selected cut-off (10^-12) were extracted. Additionally, the CheA HK of Escherichia coli 31 was used as template in BLASTP searches to identify CheA-like HKs in the P. syringae genomes and hits with an E-value below a selected cut-off (10^-10) were extracted. Hybrid HKs (REC-HKs) were determined by the presence of complete HK transmitter and REC domains in a single protein. Detection of orthologues of the identified HKs and RRs between the genomes of Psy B728a, Pto DC3000 and Pph 1448A was determined by BLASTP 63 based on the reciprocal best hits of each P. syringae genome against each other genome, completed by the phylogenetic analyses. Finally, functional domains of the HKs and RRs were identified by search the Conserved Domain Databases (CDD) with Reverse Specific Position BLAST 64.

Multiple sequence alignments and phylogenetic trees of HKs and RRs were constructed using the ClustalW program 65, and aligned sequences were imported into the MEGA 3.1 program 66 where phylogenetic trees were inferred. Default parameters were used. Phylogenetic trees were subdivided into groups of orthologues, and co-clustering with members of specific TCS proteins allowed a definitive assignation to a given HK type or RR family.