In order to identify the genes concerned with wall teichoic acids synthesis in S. aureus, strains Mu50, MW2, N315, MRSA252, MSSA476 and COL, amino acid sequences of all tar genes in B. subtilis W23 strain and tag genes in B. subtilis 168 strain from GenBank were BLASTed against the Refseq ORFs of Mu50, MW2, N315, MRSA252, MSSA476 and COL. Significant hits were identified (data not shown), and further subjected to phylogenetic analysis. The analysis led to the identification of the corresponding tar or tag orthologs in the S. aureus strains we examined. "Ortholog" here is technically defined as those with the best phylogenetic similarity. By deleting branches of less homologous hits, the trees in Figure 1 were generated, indicating the orthologous tar or tag genes in those S. aureus strains. tarB, tarF, tarL, tarK from B. subtilis W23 and tagB, tagF from B. subtilis 168 share common BLAST hits in these S. aureus strains. We grouped these genes as well as their BLAST hits together to perform the phylogenetic analysis and built an integral tree (Figure 1e) to show their respective orthologous relationship.

Interestingly, these S. aureus orthologs, identified by phylogenetic analysis, shown in Figure 1, are actually the best BLAST hits (not shown). (We subsequently performed a "reverse BLAST hit" analysis on the identified S. aureus ORFs.) We took the S. aureus ORFs (the best BLAST hits) and BLASTed them back against B. subtilis strain 168 ORFs. Their corresponding tag genes were also identified as the best BLAST hits (not shown). The reverse BLAST was not performed for W23 strain since its genome sequence is not currently available. The reverse best BLAST hit analysis was in agreement with phylogenetic clustering in identifying the orthologous genes involved in wall teichoic acid synthesis in examined S. aureus strains (Figure 1). The orthologous relationship is also summarized in Table 1. We further performed dot-blot analysis, which also support the identified corresponding orthologous relationships (Figure 2a, 2b).

The protein alignment of S. aureus tagF/tarF orthologs with B. subtilis TagF and TarF (Figure 3b) shows that tagF/tarF orthologs in these S. aureus strains are fully aligned with the W23 TarF in length, and also like W23 TarF, are about only half the size of 168 TagF. And only the C-terminal part of the larger TagF is significantly homologous to TarF2.

In B. subtilis, despite sharing 60% identity, the size difference between TarF and TagF implies a functional difference. TagF and TarF both use CDP-glycerol as a substrate but do not carry out identical functions. In strain W23, TarF is likely to be responsible for the addition of the second glycerol-phosphate which completes the linkage unit process. While in 168, TagF polymerizes the complete glycerol-phosphate chain onto the first residue 6, which requires a bigger protein. Thus the size difference correlates with their functional difference between TagF and TarF. The size difference between TagF and TarF may also be used to differentiate tar or tag pathway in teichoic acid synthesis 5.

In the analyzed S. aureus strains, the size of tagF/tarF orthologs suggests the existence of TarF like function rather than TagF like function, and additional enzymes are required to complete teichoic acid synthesis2.

In B. subtilis strain W23, tarI, tarJ. tarK and tarL are specific to tar pathway, which are responsible for the synthesis of RboP and the addition of poly(RboP) to the linkage unit 2 and are absent in strain 168's tag pathway. The identification of tarI, tarJ and tarL orthologs (Figure 1 and Table 1) and the tarF kind of function (instead of that of tagF) strongly suggest that the tar rather than the tag pathway is employed for cell wall teichoic acid synthesis in analyzed S. aureus strains. In the rest of the paper, these S. aureus wall teichoic acid synthesis genes are all referred as tar genes (Tar prefix will be used to refer to their protein products).

There are two copies for each tar J, tarI and tarL gene in those S. aureus strains (Figure 1, Table 1 and Figure 2). Interestingly, the tarK ortholog is missing from the analyzed S. aureus strains. In W23, tarK and tarL are identified to catalyze a similar function, but tarL could take a bigger substrate enzymatically (poly(Rbop)) 6. Thus the absence of tarK ortholog in S. aureus could either mean it is functionally replaced by one of the two tarL or be compensated for by the extra copy of tarL.

According to multiple sequence alignment analysis and BLAST analysis of B. subtilis TarB/TagB (protein) with the corresponding S. aureus orthologs, the TarB ORFs in Mu50 and N315 were found to be notably shorter than B. subtilis TarB/TagB. They were also shown to lack half of the amino acid residues from the N-terminal of B.subtilis TarB/TagB (Figure 3a). One would expect that the translation start site would be further upstream in both strains. The ORF prediction was thus rerun on this genomic region (see methods), and the correct tarB/tagB in Mu50 and N315 were identified (see Additional file 1). The correct tarB in N315 is from 687949 to 689052, producing an ORF of 367 amino acids; the correct tarB In Mu50 is located between 712200 and 713303, also encoding an ORF of 367 amino acids. Both two new TarB ORFs were subsequently confirmed by a TBLASTN analysis against these two genomes with B. subtilis TarB as query. The new TarB ORFs in Mu50 and N315 strains and their alignment with the original incorrectly predicted ones are shown in Figure 3a.

In B. subtilis, the wall teichoic acid synthesis genes are organized into a divergon, tarABIJKL-tarDF in W23 and tagAB-tagDEF in 168 (Figure 4, Additional file 4). However, in S. aureus, the tar genes seem are rather organized by genomic distance as tarIJL-tarF-tarIJL-tarA-tarB-tarD in cis orientation (Figure 4, see Additional file 4). This genomic organization is conserved in all six analyzed S. aureus strains.

As shown above, BLASTP and phylogenetic analysis identify two copies of tarI, tarJ and tarL in each analyzed S. aureus strain, which are clustered into two tarIJL regions with the same gene order. Alignment and dot blot analysis of these two tarIJL regions in each S. aureus strain confirm this gene duplication (Figure 5, N315 is shown. The others with similar results are not shown). The relevant genomic sequences of tarIJL regions including the intergenic regions among I, J, K genes and the upstream 275 bp were used as an input for Dotmatcher program to perform dot-blot analysis of the two tarIJL regions (see Materials and Methods). We also ran the NCBI BLAST2seq program to align the two regions (not shown). These analyses further confirmed the homology between those two tarIJL regions, which strongly suggests the whole tarIJL region is duplicated. The high homology indicates the duplication should not be an evolutionary (or biologically) distal event. Why and how this gene duplication occurred is still a question that remains to be answered.

The dot blot analysis also demonstrated that a small section of the C-terminal of tarJ is not very conserved in the two copies of tarJ. The enzymatic implication of this is not yet clear. Similarly, the N-terminal of tarL (almost half the size of tarL) is neither homologous between the two tarL genes. It implies that one of the tarL is very likely to be the missing tarK.

Six bacteria strains are included in this analysis: Staphylococcus aureus Mu50, MW2, N315, MRSA252, MSSA476, COL and Bacillus subtilis 168. Their respective Refseq proteomes are downloaded from NCBI microbial genomes website 15. The complete geonome sequence of Bacillus subtilis W23 is not available yet. The Tar proteins in W23 (A, B, D, F, I, J, K, L) are also downloaded from NCBI microbial genomes website 15.

The eight Tar proteins and five Tag proteins (A, B, D, E, F) were used as queries to BLASTP against the proteomes (ORFs) of S. aureus Mu50, MW2, N315, MRSA252, MSSA476 and COL. For each querying protein, using E value cut off of 1e-10, we collected all the significant hits.

To define the identified possible Tar/Tag homoglous proteins in six S. aureus strains, we also took the above identified S. aureus proteins (ORFs) to BLAST back against proteome of B. subtilis168. Mutual best BLAST hit is another evidence of the orthologous relationship.

The above selected BLAST hits from the combined dataset plus the relevant Tar protein are taken to do ClustalW alignment. The multiple sequence alignment result is further taken as input to do phylogenetic analysis.

Mega3 package is used for phylogenetic analysis and tree building. First the ClustalW produced .aln files were transformed into Mega readable .meg files, followed by performing neighbour-joining phylogenetic analysis taking .meg files as inputs. The default Mega parameters were used when making NJ trees and performing bootstrap test (bootstrap replication time at 500).

Since Tar/TagB, Tar/TagF, TarL, and TarK share certain significant BLAST hits, we put them and their BLAST hits together to do ClustalW alignment and Mega analysis.

MUMmer package V3.0 was downloaded from TIGR ftp sites 16, which was used to align between the B. subtilis 168 genome and each genome of the six S. aureus strains. Promer program was selected for its relatively high sensitivity. Promer generates amino acid alignments between two DNA input files which contain multiple sequences in FASTA format. Mummerplot was used to produce the dot-blot of the MUMmer alignments (Figure 2).

The TarB orthologs identified from S. aureus Mu50 and N315 Refseq proteomes are remarkably shorter than TarB in B. subtilis W23, MRSA252, MSSA476 and COL, missing the N-terminal part and only about half the size of W23 TarB. We applied NCBI's ORF finder program 14 to analyze the TarB regions of N315 and Mu50. Two extended ORFs were identified which were similar to W23 TarB in length. Alignment analysis confirmed they are TarB orthologs in these two S. aureus strains (Figure 3).

The ORF in N315 for TarB locates in the genome from 687949 to 689052, and ORF in Mu50 for TarB from 712200 to 713303.

The genomic localization information of Tar and Tag genes in B. subtilis 168, W23 and S. aureus strains Mu50, MW2, N315, MRSA252, MSSA476, COL were retrieved from NCBI microbial genomes website 15. Genomic organization maps of the five bacteria were then made (Figure 4, Additional file 4).

BLASTP and phylogenetic analysis identified two copies of TarI, J and L in S. aureus Mu50, N315, MW2, MRSA252, MSSA476 and COL. From genomic analysis, TarI, TarJ and TarL in the analyzed S. aureus strains, are clustered into two tarIJL regions. To confirm that it is caused by gene duplication, we further performed alignment and dot-blot analysis of those two tarIJL regions in each of the S. aureus strains. 14

We first cut out the relevant genomic sequences of tarIJL regions, including the intergenic regions among I, J, K genes and upstream 275 bp. Then we used dotmatcher program from EMBOSS package17 to perform dot-blot analysis of the two tarIJL regions (Figure 5). We also ran NCBI bl2seq program to align the two regions (not shown).