Comparative analysis of two genomes of Chlamydia pecorum isolates from an Alpine chamois and a water buffalo

To date, whole genome sequencing has been performed mainly for isolates of Chlamydia trachomatis, C. pneumoniae, C. psittaci and C. abortus, but only a few isolates of C. pecorum have been entirely sequenced and this makes it difficult to understand its diversity and population structure. In this study the genome of two C. pecorum strains isolated from the lung of an Alpine chamois affected with pneumonia (isolate PV7855) and the brain of a water buffalo affected with meningoencephalomyelitis (isolate PV6959), were completely sequenced with MiSeq system (Illumina) and analyzed in their most polymorphic regions. The genome length and GC content of the two isolates were found to be consistent with other C. pecorum isolates and the gene content of polymorphic membrane proteins and plasticity zone was found to be very similar. Some differences were observed in the phospholipase genes for both isolates and in the number of genes in the plasticity zone, such as the presence of some hypothetical proteins in PV6959, not present in any other genomes analyzed in this study. Interestingly, PV6959 possesses an extra pmp and has an incomplete tryptophan biosynthesis operon. Plasmids were detected in both isolates. Genome sequencing of the two C. pecorum strains did not reveal differences in length and GC content despite the origin from different animal species with different clinical disease. In the plasticity zone, the differences in the genes pattern might be related to the onset of specific symptoms or infection of specific hosts. The absence of a tryptophan biosynthesis pathway in PV6959 may suggest a strict relationship between C. pecorum and its host.

Initially, the main method for typing chlamydiae was by using monoclonal and polyclonal antibodies [6], but with the development of molecular biology, new genotyping methods have been applied, including PCR-RFLP and sequencing of individual genes, e.g. ompA and incA [7]. In the last few years, the advent of whole genome sequencing (WGS) has allowed the analyses of the entire genome and specific regions and genes associated with virulence, tissue tropism and host range. In chlamydial species, WGS has enabled the comparison of gene content and synteny, as well as the identification of singlenucleotide polymorphisms (SNPs), detailed analysis of specific regions in the chlamydial genome where nucleotide variations and differences occur more frequently. Such regions are thought to be involved in differences in virulence, pathogenesis and niche specificity, and include the plasticity zone (PZ), the loci encoding the polymorphic membrane protein (Pmp) gene family, the tryptophan biosynthesis operon (trp), which is responsible for the tryptophan production and biotin biosynthesis gene region, which leads to biotin production [4]. Another feature of difference between chlamydial species and strains lies in the presence or absence of a plasmid, which has been linked to virulence and pathogenesis [8][9][10].
Currently 14 genomes of C. pecorum isolated from ruminant and swine livestock species and koalas are deposited in public databases, six of which are complete ( Table 1). The aim of this study was to compare the full sequenced and analyzed genomes of two strains of C. pecorum isolated from a wild and a farmed ruminant, exhibiting different disease manifestations. PV7855 was isolated in 1996 from the lung of an Alpine chamois (Rupicapra r. rupicapra) found in the Adamello Brenta Park in the Alpine region of northern Italy and affected with pneumonia [11] belonging to a population that shared the grazing area with farmed ruminants (sheep and cattle). PV6959 was isolated from the brain of a farmed water buffalo (Bubalus bubalis) affected with meningoencephalomyelitis in a farm in Southern Italy [12] where several buffalo calves were affected with sudden depression, recumbency, paralysis of the limbs and amaurosis.

Genome features
Sequencing of the genome of PV7855 resulted in a single circular chromosome of 1,107,260 bp with a read depth of 458x, while assembly of the genome of PV6959 resulted in two gaps and two chromosomal contigs which were closed by PCR and sequencing giving a total length of 1,108,369 bp and a read depth of 327x. The circular representation and the general features of these genomes compared with those previously sequenced from sheep, cows and koalas affected with different disease pathotypes are shown in Fig. 1 and Table 1.
The GC content of PV7855 is 41.1%, while for PV6959 is 41.2%. Moreover, PV7855 presents an additional tRNA gene (tRNA-Cys (aca)) compared to the other genomes considered in this study. The annotation identified 990 predicted CDSs (of which 261 are hypothetical proteins) in PV7855 and 987 predicted CDSs (of which 263 are hypothetical proteins) for PV6959 (Table 1).
Both isolates sequenced in this study harbour a plasmid (Additional file 1: Fig. S1): pCpecPV7855 and pCpecPV6959. The annotation of the two sequences revealed a similar structure to that of chlamydial plasmids with 8 CDSs predicted in total and a length of 7674 bp with a coverage of 1050x for pCpecPV7855 and a length of 7676 bp with a coverage of 171x for pCpecPV6959 and the GC content is 31.7% for both of Fig. 1 Circular representation of the comparison between chromosome of C. pecorum PV6959 and PV7855. Circles from the outside in represent the positions of protein-coding genes (blue), tRNA genes (green) and rRNA genes (pink) on the positive (circle 1), and negative (circle 2) strands respectively. It is shown the position of BLAST hits detected through blastn comparisons of PV6959 and PV7855 against W73 (circle 3), PV3056/3 (circle 4), P787 (circle 5), E58 (circle 6) with default settings; and GC skew (Circle 8). The image was generated with CGView Server them. The plasmids consist of the following genes: three hypothetical proteins, a virulence protein pGP3-D, a DNA helicase, two tyrosine recombinase (XerC) and a sporulation inhibition protein (Soj). The two copies of XerC present in pCpecPV7855 are very different, with an amino acid identity of 38,43%.
The circular representation of these plasmids compared with those belonging to genomes completely sequenced and included in this study and the plasmid of WA/B31/ Ileal, are shown in Fig. 2.

Phylogenetic and SNPs analysis
To determine the relationship between the two sequenced C. pecorum genomes and the other representative Chlamydiaceae genomes selected for comparative analysis, a phylogenetic tree was constructed based on orthologous genes (Fig. 3). A strain of C. abortus, C. pneumoniae, C. psittaci and C. trachomatis were used as outgroup. The phylogenetic reconstruction obtained from the 753 single copy orthologs genes present in all organism (evolutionary model: LG + I + G4) is highly supported (all the nodes are supported by 100 bootstraps, with the exception of two nodes supported by 89 bootstraps) and shows a tight clustering of the C. pecorum strains. Phylogenetic analysis revealed that PV6959 is clearly separate from the other strains, including PV7855 with 8664 non synonymous SNPs. In contrast PV7855 is very close to E58 with only 125 non synonymous SNPs. The three C. pecorum strains isolated from Australian koala clustered separately from the majority of ruminant strains.
Our NeighbourNet analysis ( Fig. 3 and Additional file 1: Fig. S2) using the same genomes as our phylogenetic tree confirmed that PV7855 is most closely related to strain E58, while PV6959 is well separate.
Even though there is a wide similarity across the genomes ( Table 2 and Fig. 3) of C. pecorum, there are a significant number of SNPs, which contribute to variation among the genomes and, interestingly, Fig. 2 Circular representation of the comparison between plasmid pCpecPV6959 and pCpecPV7855. Circles from the outside inside represent the plasmids used for the comparison: pCpecWA/B31/Ileal (red), pCpecW73 (light blue), pCpecMC/MarsBar (pink), pCpecDBDeUG (yellow), pCpecPV6959 (blue) and pCpecPV7855 (aqua green). The image was generated with CGView Server PV7855 has a very low number of SNPs compared to E58 (125). The analysis of SNPs revealed also a great difference between PV7855 and PV6959 (8664 non synonymous SNPs). When PV7855 is compared with the C. pecorum genomes considered in this study, it shows a lower number of total SNPs and non-synonymous SNPs (179 with W73, 188 with P787, 198 with DBDeUG, 191 with MC/MarsBar), as opposed to PV6959, which has more differences (746 with PV3056/3, 820 with E58, 830 with W73, 834 with P787, 818 with DBDeUG and 823 with MC/MarsBar).

Comparative analysis
Whole genome comparisons (Additional File 1: Fig. S3) show that the two C. pecorum genomes sequenced in this study are highly conserved and syntenic and have similar gene content to each other and to the other genomes included in the analysis.
Some regions reveal increased variability, including the PZ and the pmps. Some other regions are more conserved, such as the bioBFDA system, which is complete and highly homologous in both isolates and the trp operon of PV7855 which is complete and homologous to that found in the other genomes. Instead, in PV6959 the tryptophan synthase alpha chain (trpA), tryptophan synthase beta chain (trpB) and tryptophan repressor (trpR) genes are present, while the trpE/G genes are missing (Fig. 4), as observed for other C. pecorum genomes studied to date [13], but we have also found that PV6959 lacks the prsA, kynU, and the trpFCD genes.

Polymorphic membrane proteins
A total of 15 pmps, which constitute one of the hotspots for SNPs accumulation, were observed in PV7855 and are characterized by one gene belonging to subtype B, one to subtype A, two to subtype E, one to subtype H, nine to subtype G and one to subtype D. The genomic organization of pmps of the genomes sequenced in this study is shown in Fig. 5, while the gene designations and annotations are listed in Additional File 2 Table S1. The comparison between pmps of PV7855 with the ones belonging to the genomes sequenced from ruminants, shows the degree of amino acid similarity between 87 and 100%. In contrast, pmp16 and pmp15 of PV7855 compared to the correspondent of P787 and W73 have a lower degree of amino acid similarity, of 79% in both cases. The comparison of this region with the loci of pmps of DBDeUG and MC/MarsBar revealed a rather high amino acid similarity between 98 and 91%, except for pmp6 (Cpecorum_PV6959_00975; Chlamydia_peco-rum_2-7855_00564) (82%).
In PV6959, however, we detected the presence of 16 pmps, indeed an extra pmp belonging to the G family was identified, pmpG5 (Cpecorum_PV6959_00982). The region comprising the gene was amplified and sequenced with Sanger method and the consensus sequence was analyzed with BLAST showing 76% of nucleotide identity with P787, E58, W73, PV3056/3 and DBDeUG. The presence of two frameshifted pmps was observed in the same genome (corresponding to pmpG7 to pmpG8) while no frameshift was observed in the pmps of PV7855. The annotation identified two hypothetical proteins in both genomes, corresponding to pmpG4 and pmpG8 in PV7855 and pmpG4 and pmpG9 in PV6959. The pmps of PV6959 compared to those of the other strains are more variable in the amino acid sequence, especially the two pmps belonging to E family, pmp16 (Cpecorum_ PV6959_00005) and pmp15 (Cpecorum_PV6959_00004), and the two pmps belonging to G family (pmp9 and pmp10). Indeed, when pmp16 and pmp15 are compared with the amino acid sequences of the correspondent of Comparative analysis of trp system. Trp system comparison between C. pecorum PV7855, P787 and PV6959 showing comparison of nucleotide matches between complete 6-frame translations (computed using Megablast blastn) using ACT. Grey bars represent the forward and reverse strands of DNA with CDSs marked as arrows. The scale is marked in base pairs. The red bars represent homology matches, the white ones represent the non-homology matches the genomes considered, the degree of identity is between 83% with E58 and 72% with P787 and W73. When we considered the pmps of DBDeUG and MC/MarsBar, we noticed there is a rather high similarity with PV6959. The lower identity is observed always in pmp16 (84%) and in pmp15 (82%).
We noticed that pmp6 appears as the pmp having the most differences in the amino acid sequence when it is compared with that of the other genomes analyzed. Indeed, in PV7855 its homology ranges between 78.22% (with PV6959) and 99.71% (with E58), while in PV6959 it ranges between 77.04% (with PV3056/3) and 87.97% (with P787).

Plasticity zone
A genomic island, which is the most variable region in gene content and sequence, is the PZ, located near the terminus of replication. In C. pecorum strains, the PZ is generally around 42 kb in size and included between the acetyl-CoA-carboxylase genes (accB and accC) and the inosine-5′-monophosphate dehydrogenase (guaAB/add) genes. The two C. pecorum genomes sequenced in this study have PZs distinct from each other: the PZ of PV7855 is most similar in gene order to that of W73 with 18 genes, while in PV6959 we identified 24 genes, in particular a second set of pld genes upstream of the tox genes (Fig. 6). The comparison of our genomes with the others, all from ruminant hosts, resulted in a syntenic gene order and a high genetic relatedness. In contrast, while the gene order was syntenic with the PZs in koala derived C. pecorum genomes, the genetic relatedness was lower.
We identified 5 hypothetical proteins in PV7855 which have a degree of amino acid identity between 95 and 100% with the correspondent proteins in the genomes considered. The correspondent of Chlamydia_peco-rum_2-7855_00297 in PV3056/3 and PV6959 represents the only exception with, respectively, the 62% and the 59% of amino acid similarity.
In PV6959, we identified 6 hypothetical proteins, one of which is located between the two set of pld genes. The degree of amino acid identity with the correspondent proteins in the genomes considered ranged between 99 and 97%.
In PV7855 we identified 4 copies of the pld genes and a gene labeled as cardiolipin synthase (cls), which is also part of the phospholipase family. In contrast, in PV6959 we identified three pld genes (pld1, pld2 and pld3) plus two genes noted as cardiolipins (cls1 and cls2) and the degree of amino acid homology are listed in Table 3. Searches against GenBank using BLAST confirmed their annotation as cardiolipins.
In both isolates sequenced in this study, we identified two copies of the cytotoxic genes. In PV7855, the  Table S1 Fig. 6 Visual representation of the genomic island of the plasticity zone in C. pecorum. Comparative analysis of the genes in the plasticity zone of C. pecorum PV7855 and PV6959. Comparison of the nucleotide matches (computed using blastn) between the genes guaB (pink) and accB (orange) in C. pecorum strains. The brown genes indicate hypothetical proteins. The orientation of coding sequences in the forward and reverse frames are indicated by the direction of the block arrows. The level of BLAST identity between the sequences is indicated by the degree of grey shading in the vertical bars. The figure was generated using EasyFig [14]

Phylogenetic and comparative analysis of plasmids
The phylogenetic analysis based on alignment of the C. pecorum plasmids listed in Table 4, revealed that pCpecPV6959 is most closely related to the plasmid from WA/B31/Ileal (Fig. 7), which was isolated from cattle, while the plasmid of PV7855 is well separate from all the others, in line with the core gene phylogeny. Moreover, the tree shows two separate lineages of the plasmids of C. pecorum isolates from Australian koala.

Discussion
C. pecorum is recognized as one of the most widely distributed chlamydial species, with a wide host range that includes livestock (sheep, goats, cattle and swine) and wildlife [15]. Despite this, very few genome sequences of C. pecorum have been deposited and thus made available, which is a limiting factor for investigating genetic diversity in this chlamydial species.
In this study, we sequenced and examined the genomes of two C. pecorum isolates, PV7855 from an Alpine chamois, which is the only C. pecorum isolate fully sequenced from a wild ruminant to date and PV6959 from a water buffalo [11,12]. The genomes of our isolates were compared to those of E58, PV3056/3, P787 and W73, all recovered from farmed ruminants [13,16], which have been selected for analysis because neither C. pecorum sequences from wild ruminants nor from water buffaloes are available in sequence repositories. Furthermore, PV6959 was isolated from a water buffalo reared in a farm where the observed clinical signs, the seropositivity to Chlamydia, the histological findings, the detection of C. pecorum by immunofluorescence assay in the brain tissue and eventually the isolation of the organism in cell culture, confirmed the diagnosis of chlamydial meningoencephalomyelitis [12]. The two isolates were compared also versus the two genomes obtained from Australian koala (DBDeUG and MC/MarsBar) with regard to the two main variable regions, PZ and pmps.
The genome length and GC content of PV7855 and PV6959 isolates are similar to those of the other C. pecorum isolates deposited in the genome repositories. The C. pecorum genome, like other chlamydial species, has a conserved order and gene content [13] and an explanation for this could be that Chlamydia tried to reduce the exposure to frequent lateral gene transfer events, which can cause phenotypic modifications [17]. The phylogenetic analysis based on the orthologous genes shows a tight cluster of the C. pecorum strains, but interestingly, PV6959 appears to be well separate from the other C. pecorum strains considered in this analysis, while PV7855 is very close to E58. This confirm the data from sequence analysis and the number of SNPs, which demonstrate that PV7855 and E58 are very similar in genetic structure. PV7855 and PV6959 are both isolates from ruminants and, as already demonstrated [4], are well separate from C. pecorum isolates from Australian koala, maybe due to their different geographic origin and host preference. NeighborNet analysis gave further confirmation that PV7855 and E58 are evolutionarily close, in fact, the genome analysis revealed a high similarity (range from 97 to 100%) in the amino acid sequences of pmps, the PZ, the trp system and bioBFDA system, furthermore, they present a very low number of SNPs. To better clarify the phylogenetic distance between isolates from koala and wild ruminants more NGS analysis on different isolates are needed.
Firstly, we focused our analysis on the pmp loci, which are known to be one of the most polymorphic loci in the entire chlamydial genome. The members of the pmps, which are part of Type V Secretion System, are unique to chlamydial species [18] and possess a conserved domain structure that includes the C-terminal autotransporter beta-barrel domain, a central m-domain unique to this family of proteins and an N-terminal passenger domain that is involved in adhesion [19].
Pmps are involved in the adhesion to the host's cell as autotransporter surface-exposed proteins, but are also involved in the immunopathogenesis of Chlamydia infections as potent antigenic proteins [20]. The number The number of pmps identified in PV7855 was 15, identical to the number identified in the isolates of C. pecorum deposited in the public databases. In contrast 16 pmps were identified in PV6959, one of which, Cpe-corum_PV6959_00982 corresponding to pmpG5, is not present in any other whole genomes currently available. Even though PV6959 and E58 are characterized by the same disease pathotype and are evolutionarily very close (see NeighborNet analysis Fig. 3B), this Pmp is not present in E58. We can hypothesize that this protein is not involved in this specific pathogenesis, but more analyses from ruminants affected with the same disease are needed to better understand a possible correlation. The pmpG5 detected in PV6959 showed a low degree (76%) of nucleotide identity with P787, E58, W73, PV3056/3 and DBDeUG when it is run on BLAST and the Sangersequence of the region demonstrated that the gene is present and it is not an assembly or sequencing error.
PmpG6 is the most polymorphic Pmp in PV7855 and PV6959 with respect to the Pmps of the other C. pecorum isolates. This data supports other published studies showing that the Pmps belonging to the G family are those most subject to diversity in the sequence [8] and that the pmps are the most rapidly evolving genes in C. pecorum [4]. To date few studies have been carried out to investigate the function and the expression of pmps genes in C. pecorum [21].
The second region analyzed in our study is the PZ, where most of the studies focus on to evaluate the presence and/or absence of a range of established chlamydial virulence factors [22]. In the PZ of the C. pecorum, the number of pld genes can vary from 4 to 5, which is a matter not unique to C. pecorum [13], indeed also C. trachomatis and C. muridarum have a variable number of pld genes [23]. The PV7855 and PV6959 isolates each possess five copies of the phospholipase genes, of which one and two, respectively, are noted as cardiolipin. The gene annotated as cardiolipin synthase in PV7855 corresponds to a pld gene in the other strains. Cardiolipin is present in three copies in Escherichia coli [24] and has been identified in C. trachomatis, where it appears to be expressed at 16 h after infection [25]. The role of mac and pld genes is still unclear but their respective functions appear to be linked [26] and they could be involved in the evasion of the host immune system and in facilitating entry and exit from the host cell. The mac genes, which encode membrane attack complex/perforin proteins, are not present in all Chlamydia species. Indeed, mac genes are present in C. pecorum, C. felis, C. ibidis, C. muridarum, C. pneumoniae, C. psittaci, C. suis and C. trachomatis [27]. In the two C. pecorum isolated in this study, the differences detected in Pld amino acid sequence compared with strains already present in repertoires, may suggest a role of these proteins in the difference of pathogenicity and virulence, but experimental studies are necessary to evaluate this hypothesis. Interestingly, PV6959 is the only strain, among those considered in this study, to have a hypothetical protein between the two tox genes instead of the two copy of pld genes. In contrast, among the mac/perforin domain, of the strains analyzed, there is a very high degree of amino acid identity, comprised between 95 and 99%.
The toxB cytotoxicity gene is present in a different number of copies depending on the chlamydial species, for example, there is only one copy in C. psittaci, C. felis and C. caviae and three copies in C. muridarum. In C. pecorum, toxB cytotoxicity gene is present in two copies, but its function is not yet fully known, it may contribute to the ability of the organism to switch from persistent infection to acute disease [13] and it could have different biological functions or a link to host specificity. Another peculiarity of PV6959 genome is the absence of the two pld genes between the two toxB genes. It represents the only case, between the genomes considered in this study, so the sequencing of more C. pecorum could be relevant also to understand the meaning of this aspect.
Genetic diversity also occurs in biotin biosynthesis, which leads the biotin production from pimeloyl-CoA, which shows significant variability among the different chlamydial species and is absent in some of them (C. caviae, C. trachomatis and C. muridarum) [22]. Biotin is involved in many central cell metabolism pathways and this could indicate its role in the host specificity [22]. In PV7855 and PV6959, it is present and highly homologous to the other systems belonging to the C. pecorum isolates.
Chlamydia species can be characterized for their ability to synthetize tryptophan and this ability depends on trp system, which is not complete in all chlamydial species. The trp system consists of prs, kynU, trpD, trpF, trpC, trpA, trpB and trpR and its aim, theoretically, is allowing the production of tryptophan from an anthranilate substrate [24]. In C. pecorum PV7855 sequenced in this study, trp system is almost intact, consisting of trpD, trpF, trpC, trpA, trpB and trpR, prs and kynU. This complement of genes would permit the production of tryptophan from the substrate anthranilate. However, this system does not allow the first step in the production of tryptophan, the conversion from chorismate to anthranilate, which would be catalyzed by trpE/G. It is hypothesized that the acquisition of anthranilate could be achieved through the capture of kynurenine from the host cell with a transport amino acid, tyrP [13]. In the isolate PV7855, trp system is complete and highly homologous to the one of the genomes considered in the comparison. In PV6959, however, the system is incomplete and in particular lacks the genes prs, kynU, trpF, trpC and trpD. The absence of these genes in a C. pecorum genome is detected in the present study for the first time, since C. felis and C. caviae represent two of the species, along with C. pecorum, where the trp system is complete. The system is absent in the PZ of C. muridarum and C. psittaci suggesting that the ability to synthesize trp de novo is not mandatory for the transmission or survival of these species [22]. PV6959 is likely unable to metabolize and produce tryptophan by itself, unless it manages to obtain indole in some other way.
Finally, an increasing number of recent studies have linked the chlamydial plasmids to pathogenesis [8][9][10]. Plasmids are also recognized as carriers of virulence factors and are almost ubiquitous in chlamydial species [9], thus, we also included plasmids in our analysis. Plasmids in Chlamydia are small, usually about 7.5 kbp in length, highly conserved, non-integrative and non-conjugative and they have been observed in many chlamydial species, including C. pecorum [8,10]. They consist of non-coding RNA and eight open reading frames (ORF 1-8) [28] and their main role, in C. trachomatis, is the contribution in glycogen accumulation [29].
In this study, the plasmids found confirmed the presence of eight CDSs and a GC content in line as the other C. pecorum plasmids.
A comparison among the plasmids of PV6959, W73 and PV7855 shows slight differences. The W73 plasmid consists of eight genes, some noted as virulence factors that correspond to those noted as hypothetical proteins in plasmids of the isolates studied. In Chlamydia plasmid, CDSs 4 (the most polymorphic locus), 5 and 6 (the most conserved loci) are associated with virulence [30] and these data are confirmed in pCpecPV7855, which is composed of eight genes of which three are hypothetical proteins.
One limitation of our study was that, unfortunately, among the genomes selected for the chromosomal phylogeny only 3 strains (W73, DBDeUG and MC/MarsBar) have both a complete chromosome and plasmid. This meant our plasmid phylogeny is not directly comparable with the chromosomal phylogeny and highlights the lack of complete genomes for C. pecorum in public databases.

Conclusions
The WGS and comparative analyses of PV7855 and PV6959 revealed that the two genomes are very similar in length and GC content and are highly preserved in the PZ and the pmps, the most polymorphic regions, in spite of the differences in the clinical manifestations of infections and the difference of host. Notwithstanding, in PV6959 an extra pmp was identified, confirmed by sequencing and genome analyses, but further studies will be necessary to clarify its role in host range. Some differences in the sequence and number of genes were also notice in the PZ, particularly in pld genes and hypothetical proteins. Since these genes are involved in virulence and evasion of immune response-system, their presence/absence might be related to the onset of specific symptoms or infection of specific host(s) and further studies are necessary to confirm or discard some of these hypotheses and to better understand their role.
Concerning the absence of a tryptophan biosynthesis pathway in PV6959, the host cell might provide tryptophan supplementation to Chlamydia. Tryptophan is an essential amino acid required for development of all Chlamydia species and their dependency on the host suggests a strict relationship between C. pecorum and its host. Even more interestingly, tryptophan metabolism has been implicated in chlamydial persistence and tissue tropism.
We consider it would be useful to investigate whether the genomic differences that we observed in our isolates are present also in others from ruminants with the same pathology. In addition, a genomic comparison with isolates from non-ruminant species would allow an investigation on whether the differences noted are linked to the pathology rather than to the host species.

Strain features, propagation on cell cultures and EB purification
Strain PV7855 was isolated in 1996 from the lung of an Alpine chamois found in the Adamello Brenta Park in the Alpine region of northern Italy and affected with pneumonia [11]. The isolate was extracted from a tissue suspension with a mortar and sterile quartz powder and then centrifuged at 108 x g for 10 minutes. The supernatant was inoculated on LLC-MK2 cells monolayer which were cultured in EMEM (Eagle's Minimum Essential Medium) with the supplement of L-glutamine 1% (v/v), gentamicin 10 mg/L, vancomycin 10 mg/L, glucose 720 mg/L (Sig-maAldrich), and inactivated Fetal Calf Serum (FCS) at 10% (Gibco ® , Life Technologies) and incubated at 37 °C and 5% CO 2 . L-glutamine, gentamicin and vancomycin were supplied by AppliChem. Infection was verified by immunofluorescence with a monoclonal antibody (Merifluor ® Chlamydia) conjugated with fluorescein specific for the lipopolysaccharide antigen of Chlamydia. The cell suspension was then recovered and the elementary bodies (EBs) were purified [31].
The same procedure was used for the strain PV6959, isolated from the brain of a farmed water buffalo affected with meningoencephalomyelitis in a farm in Southern Italy [12]. The supernatant was inoculated onto a McCoy cells monolayer and the infection was confirmed by real-time PCR. The DNA was amplified using primers and probe specific for the ompA gene of C. pecorum at a concentration of 0.6 μM and 0.25 μM, respectively and GoTaq ® Probe qPCR Master Mix (Promega).
The EBs of both isolates were purified according to the protocol of Fukushi et al. [31] and resuspended in 200 μl of transport medium for Chlamydia SPG (Sucrose Phosphate Glutamate).
DNA was extracted from EB using a NucleoSpin ® Tissue kit (Macherey-Nagel) according to manufacturer's instructions.

Whole genome sequencing and genome assembly
The DNA extracted from the isolates, was used to prepare the genomic libraries using the Nextera XT kit (Illumina), quantified with KAPA SYBR FAST Universal qPCR Kit (Roche) and sequenced on a Miseq system (Illumina) in a 2x250bp run generating paired-end reads.
The quality of the reads obtained was assessed using FastQC [32], while the presence of any contaminants was checked through the Kraken software [33] and an inhouse pipeline based on the blast of the reads with subsequent analysis using the MEGAN software [34].
The reads were separate from those of the host, following the "blobology" bioinformatics pipeline [35] and then assembled using SPAdes v. 3.10 [36]. Possible assembled contig joinings were evaluated using Bandage [37], tested by PCR and then sequenced for confirmation. The primers used were designed on non-repeated regions of the ends of the contig and their specificity was verified by BLAST.
The PCR products were purified with the Nucle-oSpin ® Gel and PCR Clean-up kit (Macherey-Nagel) and sequenced by the Sanger method. The electropherograms obtained were analyzed using the BioEdit software v.7.2.5 and the sequences obtained were compared with the previously assemblies and manually inserted into the junction points of the contigs. The assembly of the reads and the visualization of the assembly graph also allowed the identification of a plasmid for each of the isolates.
Artemis Comparison Tool (ACT) software v.17.0.1 [44] was used to compare each genome with others present in NCBI and selected on the basis of assembly level.
The amino acid sequences of the genes of the PZ, the pmps cluster, the bioBFDA system and trp system, were compared by ClustalW to evaluate the degree of homology with the corresponding sequences of the other genomes.
The SNPs detected in the comparison of genomes, including the number of synonymous and non-synonymous substitutions, were identified with the Purple pipeline [45].
The presence of an additional pmp was checked bioinformatically by mapping the reads to the coding region of the pmp and the flanking regions (1000 nt upstream and 1000 nt downstream) using Bowtie2. The mapped reads were sorted and indexed using Samtools and sequence coverage observed graphically using BamView (Artemis release: 18.1.0) [46]. In addition, a walking PCR was performed to confirm the presence of the gene, the primers (Table 5) and the thermic cycle were designed ad hoc. The PCR products were excised from agarose gel and sequenced with Sanger method. The obtained sequences were concatenated with BioEdit Software v. 7.2.5 and the consensus sequence analyzed with BLAST tool (NCBI, http:// blast. ncbi. nlm. nih. gov/).
The evolutionary history was inferred by using the Maximum Likelihood method and Tamura 3-parameter model. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances  6959_1  GTG CAT TGG GAA AGC AAT TTGA  CAC TAC CTG TCG TGG AGT TGT  723 bp   6959_2  GGG AAC AAT AGC GAT GGT GC  GTT TCT AGG GTT GTT CCT GCATT  677 bp   6959_3  GAC GGA GTG ACT CTA CAA GCA  AGA TGG CGG AAT CCC TGT T  668 bp   6959_4  GGA AAT CAT AGC AGC TTG GGA  AGG GTT AAT TTC GTG TAG GTGG  576 bp   6959_5  CTC CCT ATC GCT GCA AGT  TGA GAC ATT TCG GGT CGT G  422 bp