Likelihood-based significance testing of tree topologies was performed by the pairwise one-sided Kishino–Hasegawa (1sKH) test that has been shown to be superior to the original two-sided Kishino–Hasegawa (2sKH) test (Kishino & Hasegawa, 1989) if evaluated tree topology sets are permutatively incomplete (Goldman et al., 2000) as is the case in the present study. A
set of 297 candidate topologies for significance testing (Table S3) was generated manually in Newick format according to the rationale outlined in Fig. S1. The 1sKH test was performed as implemented in the Tree-Puzzle 5.2 software package applying a 5% significance threshold. Based on the previous phylogenetic GPCR Compound Library cell line analysis of 211 families of single-copy orthologous genes (SCOG) identified in the order Legionellales (Leclerque, 2008a), six genes, namely dnaG, gidA, ksgA, rpoB, rpsA, and sucB (Table S1), were chosen as potential MLST markers as the respective buy Selumetinib SCOG families (i) were found to be sufficiently informative in both phylogenetic analysis and significance testing at the suprageneric level, (ii) at this level clearly fulfilled the dN/dS < 1 criterion, (iii) did not give rise to any detectable sign of lateral gene transfer (LGT) when explored across a set of
alpha- and gammaproteobacterial as well as chlamydial genomes, and (iv) the respective gene loci are widely dispersed across the R. grylli genome. More exactly, all potential protein-encoding MLST markers were located in a single gene copy on the major contig 637 that comprises > 99% of the R. grylli genome sequence (1 581 239 bp). The marker genes are oriented in a way forming three linked marker pairs (ksgA-gidA, rpsA-sucB, dnaG-rpoB), an arrangement that increases the probability to detect
LGT in future studies (Table S2). Moreover, the R. grylli genome contains two identical rRNA operons located at a distance of 370 000 bp from each other on contig 637. Using the primer pairs listed in Table S1, PCR products of expected size (Table S2) were obtained from Rickettsiella pathotypes ‘R. melolonthae’ and ‘R. tipulae’. The triplicate raw sequences generated a reliable consensus for internal partial sequences of genes dnaG, gidA, ksgA, rpoB, rpsA, sucB, and ftsY as well as a 3′-terminal partial sequence of the 23S rRNA-encoding gene rrl. The 16S Methamphetamine rRNA-encoding sequences from both Rickettsiella strains had been determined previously (Leclerque & Kleespies, 2008a, c). Expectedly, amino acid sequences deduced from the protein-encoding marker sequences as well as the rrl nucleotide sequences from both strains unambiguously identified the respective orthologous R. grylli sequence as most closely related GenBank database entry. For each marker, the three Rickettsiella sequences were aligned to two orthologs each from Coxiella and Legionella genomes together with three further gamma- and four alphaproteobacterial as well as three chlamydial orthologs under particular consideration of arthropod-associated bacterial genera.