Unveiling genome plasticity and a novel phage in Mycoplasma felis: Genomic investigations of four feline isolates

Abstract Mycoplasma felis has been isolated from diseased cats and horses, but to date only a single fully assembled genome of this species, of an isolate from a horse, has been characterized. This study aimed to characterize and compare the completely assembled genomes of four clinical isolates of M. felis from three domestic cats, assembled with the aid of short- and long-read sequencing methods. The completed genomes encoded a median of 759 ORFs (range 743–777) and had a median average nucleotide identity of 98.2 % with the genome of the available equid origin reference strain. Comparative genomic analysis revealed the occurrence of multiple horizontal gene transfer events and significant genome reassortment. This had resulted in the acquisition or loss of numerous genes within the Australian felid isolate genomes, encoding putative proteins involved in DNA transfer, metabolism, DNA replication, host cell interaction and restriction modification systems. Additionally, a novel mycoplasma phage was detected in one Australian felid M. felis isolate by genomic analysis and visualized using cryo-transmission electron microscopy. This study has highlighted the complex genomic dynamics in different host environments. Furthermore, the sequences obtained in this work will enable the development of new diagnostic tools, and identification of future infection control and treatment options for the respiratory disease complex in cats.


Introduction:
Mycoplasmas are the smallest free-living bacteria and have some of the smallest bacterial genomes (0. .They are thought to approach the minimal gene sets essential for independent life, making them ideal model organisms for studies on the fundamental requirements of a living cell [1]. They evolved from Gram positive bacterial ancestors 600 million years ago [2] and have undergone considerable genome reduction during their evolution [3,4].They lack a cell wall, are pleomorphic and can infect and cause acute and chronic disease in a diverse range of animal and plant species [5].Some mycoplasmas have been found to exhibit significant genetic plasticity and to evolve rapidly, mainly driven by intra-species spontaneous mutation and recombination or horizontal gene transfer (HGT) between and within species, facilitating swift adaptation to environmental changes, enhancing survival and/or virulence [6][7][8][9].Studies examining the genomic stability of Mycoplasma synoviae have revealed that mutations occur frequently during infection in birds [10], and that rapid thermoadaptive evolution can occur in vitro and in vivo [11], presumably to regain fitness and pathogenicity [12,13].HGT appears to occur more frequently between species that share an ecological niche, irrespective of their phylogenetic divergence.Examples of significant HGT include the transfer of the genes encoding the phase variable VlhA cell surface lipoproteins between the poultry pathogens Mycoplasma gallisepticum and M. synoviae [14], and the transfer of several cell surface lipoproteins involved in host colonisation between the ruminant pathogens Mycoplasma capricolum subsp.capricolum and Mycoplasma mycoides subsp.Mycoides, and Mycoplasma agalactiae and Mycoplasma bovis [15].
At least four species of mycoplasmas have been isolated from cats (Mycoplasma felis, Mycoplasma gateae, Mycoplasma arginini, and Mycoplasma feliminutum), in most cases as commensals, but they have been associated with conjunctivitis and, increasingly, feline respiratory disease [16][17][18][19].Feline respiratory disease complex (FRDC) is the leading cause of euthanasia of cats in animal shelters [20] and is commonly associated with several viral (feline herpesvirus 1 and feline calicivirus) and bacterial pathogens, including M. felis [21,22].M. felis infections are common in most cat populations and the ability of this organism to persist in an animal enhances opportunities for recurrent opportunistic infections.Vaccines against M. felis are not currently available and treatment is limited to the prolonged treatment with tetracyclines or fluoroquinolones [23,24].Currently full genomes of three isolates of this species have been published: strain Myco-2 (GenBank accession number: AP022325.1),an isolate from a case of respiratory disease in a horse in Japan [25], which has been completely assembled, and multiple assemblies of isolates 16-040612-4 (CP104192, CP104193, CP104194, CP11027) and 16-057065-4 (CP103988, CP103993, CP110269), which were each isolated from feline bronchoalveolar lavages in Canada [26].Additionally, there is an incomplete genome of ATCC strain 23391, assembled into scaffolds (GenBank Biosample: SAMN02841218).
The aim of this study was to sequence and assemble four high quality, complete M. felis genomes, using hybrid assembly methods.These genomes would represent the first complete assemblies of feline isolates of this species.The objective was to a significantly expand on our limited understanding of the genomic structure and plasticity of M. felis, which may lead to enhanced development of diagnostic tests and vaccines, as has been seen in other mycoplasma species.

Mycoplasma isolation
Swabs were collected for bacteriological culture from clinical cases of respiratory disease (3 cases) and an unresolved post-surgical infection (1 case) (Table 1).Swabs were used to inoculate sheep blood agar (SBA) plates, which were incubated aerobically at 37°C.Following observation and presumptive identification of small mycoplasma-like colonies, single colonies were transferred to Mycoplasma Broth (MB) containing 10% swine serum (Sigma-Australia) and 0.01% nicotinamide adenine dinucleotide (NAD) (Sigma-Australia), based on the formulation of Frey's medium, with minor modifications [27].Acidification of the MB medium was used to confirm the growth of Mycoplasma spp.and the cultures were stored at -80°C for future investigation.

DNA extraction, sequencing, and genome assembly
Mycoplasma isolates were cultured axenically in MB for 18 h at 37°C prior to DNA extraction.Bacteria were pelleted and the medium removed, and then resuspended in the appropriate extraction kit lysis buffer.The DNA for short read sequencing was extracted using the DNeasy Blood and Tissue Kit (QIAGEN), following the manufacturer's protocol, and 100 ng of the extracted DNA was used to prepare sequencing libraries using Illumina's Nextera Flex DNA Library Prep Kit.Sequencing was performed on the NovaSeq platform at Charles River Laboratories, Victoria, Australia, to generate 150 bp paired-end reads.Two rounds of Oxford Nanopore Technologies (ONT) long read sequencing were performed.The DNA was extracted using the Promega Wizard DNA Purification Kit (Round 1) or the Promega Wizard High Molecular Weight DNA Purification Kit (Round 2).DNA libraries were generated using the ONT Rapid Barcoding (SQK-RBK004) Kit.Sequencing was performed on a MinION Mk1b device fitted with a FLO-MIN106D Flow Cell (R9.4.1 chemistry).Base calling, de-multiplexing, and adapter removal were performed using the Guppy (version 6.2.1+6588110a6) high accuracy base caller with a minimum quality score of 9 (default settings).The Illumina reads were trimmed and adaptors removed with fastp v0.20.1 [28] using the default settings.The ONT reads from the two rounds were merged for each isolate and processed using Filtlong v0.2.0 [29] to remove all reads shorter than 1000 bp, retaining 90% of input data (after removal of short and low-quality reads).

Genome annotation and sequencing analysis
The genomes were annotated using the Prokaryotic Genome Annotation Pipeline v2022- [12][13].build6494[34] for primary analysis.A Tenericutes specific protein database, containing both Swiss-Prot (4464 sequences) and TrEMBL (343,598 sequences) annotations, was downloaded from UniProt and converted to a Diamond format protein database [35].This database was used as input to Pydeel (https://github.com/alegione/pydeel) to determine the coding ratio of our annotated genomes in comparison with M. felis genomes on GenBank.This tool identifies the coding ratio of annotated genes compared to their best hit within the provided database to identify the proportion of annotated genes that may have erroneous insertions or deletions as a by-product of the use of long read sequencing.From here, annotated genomes were processed through Panaroo v1.3.0 [36], along with previously sequenced M. felis genomes (Myco-2 and ATCC23391), to identify the core genome.The core genome results were visualised using Pagoo [37].The annotated genomes from this study were investigated for pathogenicity islands and integrative conjugative elements using Island Viewer 4 [38] and ICE finder [39], respectively.Antimicrobial resistance genes and virulence factors were detected by generating a Diamond database of the CARD [40] and full virulence factor databases [41], and using Diamond BLASTp to identify best hits with a cut-off bit score of 50.The UniProt webserver was used to identify homologs of specific open reading frames (ORFs) in other genomes (https://www.uniprot.org).

Phage isolation and visualisation
A phage-like contig was identified in the annotation results of one of the assemblies (MF329).In order to confirm whether the detection of a bacteriophage genomic sequence corresponded to the presence of an integrated prophage or to replicative virions, bioinformatic methods were first utilised.The probable terminal ends were investigated using short and long read mapping to the contig with BWA-MEM and minimap2 v2.26-r1175 [42], respectively, and the contig's starting point adjusted based on the depth of coverage.The tool PhageTerm [43] was also used to investigate the potential phage start and end points.To investigate whether this contig was a prophage, the long reads were mapped, using minimap2, to the updated phage contig.Any reads that had overhangs at the 5' and 3' ends of the contig were extracted and subsequently mapped to the assembled MF329 genome to determine whether there was a common point of overlap.To detect the presence of bacteriophage virions, transmission electron microscopy was performed on a culture of MF329.
Isolate MF329 was cultured in 50 mL of Mycoplasma Broth, as described above, for 20 hours.The culture was centrifuged for 15 mins at 10,000 x g at 4°C and the medium fraction separated from the bacterial cell pellet.The bacterial cells were gently lysed by resuspension in 150 µL of sterile water, followed by centrifugation at 10,000 x g at 4°C for 5 mins to remove cellular debris.This was repeated twice, with the supernatant fraction captured each time for ultracentrifugation.The culture medium and the cell lysate supernatant fractions were separately ultra-centrifugated at 100,000 x g for 90 min at 4°C, and the pellets resuspended in 0.2 M HEPES buffer, pH 7.7.Negativeky stained samples were examined at the Ian Holmes Imaging Center (Melbourne, Australia) using a FEI Talos L120C cryoTEM (Talos L120C) microscope.

Characterisation and comparisons of genomes
All four isolates were successfully assembled into circularised, complete genomes.In addition to these four genomes, a novel M. felis phage genome was identified that utilised the mycoplasma translation code.Metrics for long and short read sequencing output are provided in Table 2.The assembly statistics for the four isolates are provided in Table 3.Briefly, the circularised isolates were significantly larger than the reference genome (Myco-2; 841,695 bp), with a median length of 940,935 bp (range: 905,741 -948,716) and a median of 759 coding domain sequences (CDSs) (range: 743 -777), compared to the reference genome which has 740 CDSs.The NCBI Prokaryotic Genome Annotation Pipeline incorporates a CheckM [44] completeness analysis and an average nucleotide identity taxonomic classification step to identify any potential issues with an assembled genome.For all four genomes the CheckM completeness mirrored that of the reference Myco-2 genome (99.21% completeness), and all assemblies were classified as Mycoplasma felis with high confidence (Table 3).The coding ratios of annotated assemblies were investigated to determine the frequency of potential indel errors.The two published M. felis genomes (Myco-2 and ATCC23391) had mean coding ratios of 1.0 (standard deviation ±0.02) and 0.96 (±0.32), respectively.Our completed genomes, MF47, MF219, MF329, and MF632, had mean coding ratios of 1.0 (±0.28), 0.99 (±0.30), 1.0 (±0.27), and 1.0 (±0.26), respectively.Seven recently sequenced M. felis isolates that utilised only Oxford Nanopore sequencing had a substantial number of indel errors in their annotations based on coding ratio analysis (mean coding ratios between 0.44 and 0.47) and were not used in further comparisons for this reason (Figure 1).
The genomes of two isolates, MF047 and MF632, had an additional tRNA compared to the other isolates, for tRNA-Trp and an unknown tRNA respectively.The genomes of the newly sequenced isolates had 1-3 secretory systems (in all cases these were annotated as Type IV secretory systems), whilst that of Myco-2 has no annotated secretory systems.The newly annotated genomes also had an increased number of transposases, typically annotated as IS1634 or IS256, or in a few cases as IS3.
In comparison, the Myco-2 genome only had IS3 and IS1634 family transposase annotations.
A progressive MAUVE alignment of the four genome assemblies in comparison to that of Myco-2 (Figure 2) identified two significant genome inversions in the reference sequence compared to the isolates assembled here, as well as more localised inversions within each of the isolates.These inversions, and the genomic regions present in individual isolates, were typically flanked by transposases.Comparison of the average nucleotide identities (ANIs) of the four new assemblies, and with previously published M. felis genomes (Table 4) found that the best match for each of the four Australian isolates was another isolate from this study, with the exception of MF329 (MF047 and MF219, 99.8%; MF632 and MF329, 98.3%, MF329 and ATCC23391 98.4%), noting that MF047 and MF219 were obtained from the same animal two months apart.
The core genome of M. felis, as determined by Panaroo (genes present in >99% of genomes) contained 595 genes, with 301 shell genes (genes present in 15% -95% of genomes).Collapsing paralogues increased the number of core genes by one and reduced the number of shell genes to 219.Principal component analysis (PCA) highlighted clustering of the two published genomes, as well as clustering of isolates that were obtained from the same individual (Figure 3).All genes, except those annotated as hypothetical proteins, identified in the newly sequenced isolates but not present in the reference sequence are listed in Table 5, with the homologs found in other genomes.The variations between the genomes were mostly observed in sequences encoding putative proteins involved in DNA transfer, metabolism, DNA replication, mycoplasma-host cell interactions, and restriction modification systems (Figure 4).

Genomic islands, virulence genes and antimicrobial resistance genes
Genomic islands were detected in three of the four newly sequenced genomes (Table 7).The average length of these islands was 22,145 bp (± 5818 bp) and each contained an average of 16 (±5) annotated genes.In comparison, the Myco-2 strain of M. felis had two predicted genomic islands of 3052 bp and 10,027 bp.Further inspection of these predicted genomic islands identified matching regions present in isolates in alternative locations.For example, MF329 had a predicted genomic island of 24 kb (nucleotide position [nt] 311,409 -335,810) containing 19 genes, including a type IV secretion system and an insertion sequence (IS1634) at its midpoint.Extraction and remapping of the sequence of this genomic island showed that it was repeated three times within a wider 122 kb region (nt 213,995 -335,810) of the MF329 genome.This same genomic region occurred, without the IS element present, in the genomes of MF047 (twice), MF219 (once), and MF632 (once).In the case of MF632, the region was also a predicted to be a genomic island, but in this instance was flanked with restriction endonuclease genes and either predicted transposases (3') or an IS256 element (5').
Diamond BLAST searches for antimicrobial resistance genes and virulence genes identified a few high 237 identity hits ( Table 6 and Supplementary Tables 2 and 3).Only one hit with the virulence factor database had an amino acid sequence identity of greater than 90%, and this was to the gene for the elongation factor Tu in Mycoplasma synoviae (90.1% average identity), which has been found to be an adherence factor and is present in all genomes.Only five genes had hits with >50% amino acid sequence identity.

Phage/prophage detection and identification
A 33.6 kb contig (vB Mfe PM329; Mycoplasma felis strain PM329) was assembled and identified as a likely phage or prophage in the sequenced M. felis MF329 isolate.It had an average depth of coverage of 40,847.7 reads, with 82.6% of the MF329 Illumina reads mapping to the phage genome.
Interrogation of the nanopore read dataset to determine its genome integration status detected only 63/18134 (0.35%) reads that mapped to both the phage and the bacterial genome.Of these, most were found to map across the bacterial genome at mostly single-read depth, with the exception of five reads that clustered mid-genome in the region of nt 456,500.If this was indicative of a true prophage insertion site it would disrupt ORF MF329_000403, which encodes a putative 16S rRNA pseudouridine synthase gene.The phage genome had a 27.1% G+C content and encoded 36 predicted ORFs using the mycoplasma translation code.It shared 94.8% global pairwise nucleotide sequence identity with two nanopore contigs in the publicly available unassembled M. felis contig dataset (NCBI Biosample: SAMN30346842), and there was 96.8% and 97.4% pairwise amino acid sequence identity between each of their encoded putative recombinase and portal proteins, respectively.Interrogation of the read datasets of the other three isolates sequenced in this study with the phage genome identified only 4/9515 (0.04%; MF047), 19/19128 (0.1%; MF632) and 22/65383 (0.03%; MF219) mapped nanopore reads, within the range of barcode error.No Illumina reads mapped to the phage genome, indicating that the isolates did not carry the phage.Phylogenetic analyses using VipTree [45] and vContact2 [46] was not able to assign vB Mfe PM329 to any known phage genus or family, but its gene arrangement was similar to those of the MAgV1 phage from M. agalactiae and MAgV1-like prophage sequences detected in mycoplasma species in the Hominis phylogenetic group.In particular, both the protein and gene arrangement similarity were greatest to those of prophages found in Mycoplasma mustelae and Mycoplasma molare, including the central position of the recombinase gene.BLASTX analysis of predicted ORFs detected pairwise protein sequence identities of 48% to 69% between core conserved ORFs encoded within the M. mustelae and M. molare prophage sequences [47] (Supplementary Table 1).
Confirmation of productive virus formation was achieved using cryoTEM on medium and cell lysate supernatant fractions, which revealed the presence of non-enveloped contractile-tailed polyhedral viral particles approximately 100 nm in length, with a 50 nm diameter head and tails of varying lengths.Higher numbers of virions were observed within the cell lysate supernatant fraction than in the culture medium fraction, with clusters of virions observed seemingly embedded within bacterial membrane debris (Figure 5).These long tailed polyhedral viral structures were consistent with those seen in siphoviruses and myoviruses within the recently ratified Caudoviricetes class of bacteriophage.The contractile tail and polyhedral head structure is consistent across all visualised mycoplasma phages, with the main differences being the head diameter and tail lengths.M. pulmonis Phage P1 has an isometric head with a diameter of 28 nm and a short tail, Br1 has head diameter of 77 nm and a long tail of 104 nm [48], and the M. hyorhinis Hr1 phage is a short-tailed (14 nM) phage with a head diameter of 34 nm [49].

Discussion
In this study, the genomes of four new M. felis isolates were characterised and compared; three of the isolates were derived from clinical cases of respiratory disease and the remaining one from an unresolved post-surgical infection.Comparisons between these genomes and with several published datasets derived from other isolates of M. felis detected genomic variations and inversions between the Australian felid isolates and the only assembled M. felis reference isolate, from a horse.These variations and inversions were flanked by transposases, which are commonly involved in HGT and genome reassortment in other mycoplasmas.There were differences in the number of copies of putative transposases from several families, including the IS1634, IS256 and IS3 (Figure 4).Insertion sequences (ISs) are transposable elements that are often present in multiple copies in a genome and involved in chromosomal rearrangements by deletion, insertion, or inversion.The substantial similarity between various isoforms of ISs enables them to facilitate homologous recombination [50].
IS families encoded in mycoplasma genomes play an important role in genomic plasticity with several reported instances of IS exchange between strains or species [15,[51][52][53][54].This is in contrast to early research that suggested that some ISs, such as IS1634, which was found in abundance in the sequenced genomes of our isolates, were unique to specific mycoplasma species [55].The differences between our Australian M. felis genomes, which were obtained from three different cats, based on gene presence/absence and the number of coding sequences, underscores the genetic diversity present amongst the sequenced felid M. felis isolates (Figure 3).Multiple gene insertions, relative to the equid M. felis reference genome, were detected within the cluster of Australian field isolates (Table 5).These included genes involved in DNA transfer, replication, metabolism, and restriction modification (RM) systems, as well as a number of proteins of unknown function (Figure 4).

Gene gains compared to the equid M. felis reference strain
Of particular interest are the gene acquisitions involved in DNA transfer, such as the multiple copies of genes encoding putative type IV secretory system (T4SS) conjugative DNA transfer family proteins within the Australian felid genomes, which were not detected in the publicly available equid and felid M. felis genomes.T4SS conjugation systems translocate DNA through a process dependent on cell to cell contact and have been identified as one of the core genes within mycoplasma Integrative and Conjugative Elements (ICEs), which play a significant role in genomic plasticity in the species in which they have been found [56,57].ICEs have been shown to be self-transmissible in M. agalactiae, providing conjugative properties from ICE-positive cells to ICE-negative cells, while concurrent chromosomal transfers (CTs) occur in the opposite direction [58].Protein sequence analysis of the putative T4SS identified in the Australian felid M. felis isolates detected 43.2% amino acid sequence identity with ORF5 of the ICE in Mycoplasma cynos.Whilst M. cynos predominantly infects the respiratory tract of dogs, it has also been detected in cats in association with conjunctivitis and upper respiratory tract disease [59].Taken together, it is plausible that a potential HGT event may have resulted in transfer of the T4SS conjugative element of an ICE from a M. cynos-like species to M. felis in a shared feline host.This T4SS was part of a defined region of genes that, along with 15 hypothetical proteins and a DUF87 domain-containing protein, was repeated three times in the genome of isolate MF329.DUF87 domain-containing proteins are associated with conjugative elements in other organisms, and include TraJ in the virulence plasmid of Salmonella enterica serovar Typhimurium [60], adding further weight to the likelihood that this region is an ICE, even though the bioinformatics tools used here did not detect it.
Several genes involved in metabolism were newly detected in the felid M. felis isolates sequenced in this study.A type 2 glycerol-3-phosphate oxidase (GlpO) was identified within the Australian M. felis genomes, while in the equid M. felis isolate, only its FAD-dependent oxidoreductase domain was detected, accompanied by two transposases, located upstream.GlpO converts glycerol-3-phosphate to dihydroxyacetone phosphate (DHAP), releasing hydrogen peroxide as a by-product, which has been shown to play a role in cytotoxicity in vitro in M. gallisepticum and Mycoplasma pneumoniae [61,62], and in the virulence of M. mycoides subsp.mycoides small colony type and mosquitoassociated Spiroplasma species [63,64].The significant reduction or complete loss of H2O2 production in highly passaged M. bovis, M. agalactiae and M. gallisepticum strains, and in attenuated vaccine strains of M. gallisepticum, compared to wild type strains suggests that glycerol metabolism can rapidly become redundant in some mycoplasmas [61,65,66].It is possible that the sequences coding for GlpO were lost in the equid M. felis isolate, possibly because of differing levels of available glycerol in horses compared to cats.Likewise, variations were seen in the counts of genes coding for nutrient transport systems, including ATP-binding cassette (ABC) transporters and the PEPdependent phosphotransferase transport system (PTS), as well as numerous other metabolic enzymes.A number of these nutrient transport systems, and many cell surface enzymes, are multifunctional in mycoplasmas, playing critical roles in pathogenesis and survival in the host [13, [67][68][69][70][71][72]]. The differences observed in the PTS and ABC transporters of the Australian felid M. felis isolates, in contrast to the equid M. felis isolate, may also be indicative of rapid metabolic gene gains and losses in response to the varying nutrient availabilities within felid and equid ecological niches.
In the Australian felid M. felis genomes, five new coding sequences related to replication were identified that were absent in the equid origin M. felis genome, including genes encoding the putative anti-termination protein NusB, a Hsp70 family protein, a C5 methylase, the 16S rRNA methyltransferase RsmG, and an HNH endonuclease.The presence of these additional proteins suggests potential differences in bacterial growth within the feline and equine hosts, potentially due to distinct replication strategies.Future experiments comparing growth patterns of these isolates in felid or equid derived cells are indicated.The potential origins of these additional sequences have been detected in mycoplasmas isolated from mink, pigs, and dogs, which share a common ecological environment with cats, or pigeons, which are a common prey of cats.The HGT events resulting in the acquisition of these genes by the Australian felid M. felis isolates may have been facilitated by the close ecological proximity or may have occurred within the feline gastrointestinal tract (GI) following the ingestion of infected prey.HGT is frequently observed between diverse microorganisms within the mammalian gastrointestinal tract, possibly due to the high population density in this niche, which presumably facilitates genetic exchange [73].Furthermore, the conjugation mechanism favoured for HGT within the GI tract protects DNA from nucleases and heavy metals, and thus is more likely to result in a successful genetic exchange [74].
Two additional restriction endonucleases (Eco47II and Sau3AI) were annotated in the Australian felid M. felis genomes, and they lost a DpnII family type II restriction endonuclease, compared to the equid isolate.Restriction modification (RM) systems play a pivotal role in regulating bacterial HGT by protecting the genome against incoming mobile genetic elements, or by facilitating recombination with the introduced DNA [75].A strong correlation has been reported between the presence of active Type II RM systems and the presence of mobile genetic elements and ICEs within M. agalactiae genomes, indicating the facilitative role of Type II RM systems in HGT in mycoplasmas [76].The diversification of RM target recognition sites can regulate HGT by promoting genetic exchanges among those organisms with RM systems that recognise the same target motifs, and by reducing genetic transfer between lineages that possess distinct RM systems [77].Taken together, it is plausible that following a host switch into horses, M. felis may have diversified its RM systems to enhance the level of genetic exchange from phylogenetically distant bacteria.

Gene losses compared to the equid M. felis reference strain
A sequence encoding a putative ComEC/Rec2 family competence protein was absent in the Australian felid M. felis isolates.This protein is one of the components of DNA translocation via transformation, enabling DNA transport by forming a channel across the cytoplasmic membrane [78].
While the ComEC component has been reported to be essential for DNA transport in Gram positive bacteria, deletion of homologous genes did not affect DNA uptake in bacteria without a cell wall [79].
Furthermore, the competence domain seems to be missing in many mycoplasmas, possibly due to the reductive evolution of these bacteria [80].Characterisation of two mycoplasma genomes recovered from gut microbiota of a deep-sea isopod detected the comEC gene adjacent to genes encoding enzymes producing dTMP from thymidine or dUMP, suggesting that the role of the ComEC protein might be importing extracellular DNA as a nutrient source [81].The presence of ComEC in the equid M. felis isolate, and its absence in the Australian felid M. felis isolates suggests that during evolutionary adaptation DNA translocation via a membrane channel might have become redundant and perhaps was replaced with T4SS conjugation-mediated DNA transfer, as discussed above.
The Australian felid M. felis genomes exhibited a total absence of four coding sequences associated with host cell interactions, those encoding a putative variable surface lipoprotein, a fibronectin type III domain-containing protein, a FIVAR domain-containing protein, and a GA module-containing protein, in contrast to the equid M. felis genome.The FIVAR (Found In Various Architectures Region) and GA domains are essential constituents of extracellular matrix-binding proteins (Embp), facilitating biofilm formation and attachment to fibronectin in some bacteria such as Staphylococcus epidermidis and S. aureus [82,83].The presence of more than 30% amino acid sequence identity between the FIVAR and GA domains in the equid M. felis reference isolate Myco-2 and these domains in Embp of S. aureus raises the possibility that the equid M. felis isolate might have potential for biofilm formation and/or fibronectin attachment.Conversely, the absence of these proteins, and of the surface lipoprotein and fibronectin type III domain-containing protein, suggests that these capacities may have become obsolete in the Australian felid M. felis isolates, possibly indicating different modes of host cell interaction in the two hosts.

Rapid evolution within two months in one host:
Two of the Australian M. felis isolates sequenced in this study, MF047 and MF219, were obtained from separate bronchoalveolar lavages (BAL) of a feline patient with clinical signs consistent with pneumonia.Isolate MF047 was obtained from a BAL collected during the initial consultation, whereafter the cat was treated with 25 mg of doxycycline twice a day orally for approximately two months.Because the response to treatment was unsatisfactory, a second BAL was collected and isolate MF219 was obtained.While no genome reassortment was detected between the MF047 and MF219 genomes, minor differences were observed in the counts of some genes encoding proteins involved in RM systems, DNA transfer, and methylation (Figure 4).These variations may suggest rapid host adaptation of M. felis after only two months.The growing utilisation of whole genome sequencing in mycoplasma diagnostics is revealing links between genetic mutations and antimicrobial resistance.Genomic comparisons of MF047 and MF219 revealed the presence of five single nucleotide polymorphisms (SNPs) within the gene encoding the ribosomal RNA small subunit methyltransferase A (RsmA/KsgA), along with one SNP in the gene encoding the 16S rRNA and another SNP in the gene encoding the 23S rRNA.In other studies, mutations within the 16S RNA methyltransferase family have been reported to confer aminoglycoside resistance in M. bovis and M. gallisepticum [84,85], while mutations in the 16S rRNA or 23S rRNA genes have been shown to be associated with resistance to tetracyclines in M. pneumoniae, M. hominis, M. genitalium, and M. bovis [84,[86][87][88].MF047 and MF219 both contained five protein coding regions with low confidence matches to tetracycline resistance genes, with the highest bit score of the five associated with a predicted elongation factor G protein with amino acid sequence identity/similarity to tet(T) from Steptococcus pyogenes of 25.8%/47.7%.It is important to emphasise that the application of novel genome sequencing techniques for mycoplasma diagnostics, particularly in the context of antimicrobial resistance, is still in its early stages, and the availability of data for comprehensive analysis of all mycoplasma species (and particularly animal mycoplasmas) is currently limited.Future research determining the phenotypic susceptibility of isolates MF047 and MF219 to tetracyclines is necessary to determine whether there is a correlation between the genomic variations observed and tetracycline resistance.

Phage in mycoplasmas
In other bacteria, phages have been demonstrated to provide critical virulence factors to their hosts, and play a key role in the pathogenesis of diseases associated with a diverse range of bacteria (e.g.cholera, diphtheria, botulism, and haemolytic-uremic syndrome).Approximately 20 mycoplasma phages/prophages have been reported across diverse ruminant, avian, canine, pig, human and rodent mycoplasmas [47,89].Of these, most are unclassified, only three have had their virion structure visualised (P1, Hr1 and Br1), and only one has been classified by the International Committee for the Taxonomy of Viruses (ICTV), the temperate M. pulmonis phage P1 (12 kb genome), as the sole member of the floating genus Delislevirus [90].The well-described temperate MAV1 phage found in M. arthritidis (16 kb genome) has not yet (at the time of publication) been classified within the new system.In other mycoplasmas, phages have demonstrated dynamic mobility within the bacterial chromosome, with loss of integrated prophages seen with clonal passage (MFV1 in M. fermentans), and variable expression of unique phage-encoded membrane surface proteins (MFV1 and MAV1 in M. fermentans and M. arthritidis, respectively) [91], and may have a role in bacterial adaptation and survival.Detection of a similar contig within M. felis 16-057065-4 (NCBI Biosample: SAMN30346842) from Canada indicates that this phage is probably carried by M. felis strains globally, although in that study it was not identified as a phage/prophage, but rather as M. felis bacterial genomic DNA.No known selective advantage or disadvantage of the presence of this phage within

Figure 1 .
Figure 1.Coding ratio frequency counts across the Mycoplasma felis genomes from this study and those currently available on GenBank.Coding ratios were determined by dividing the amino acid sequence length of annotated proteins by the best hit match in a custom UniProt Mycoplasma database.The coding ratio axis is log transformed for visualisation.Title names for NCBI origin data include the accession number and strain name (e.g.NZ_CP104194, and16-040612-4).For Oxford Nanopore data from NCBI, the depth of coverage is added from NCBI (e.g.58 x).

Figure 2 .
Figure 2. A progressive Mauve alignment of the Mycoplasma felis reference genome (Myco-2) and the genomes sequenced in this project.Locally co-linear blocks (LCBs) are separated by colour within genomes, and are the same colour in each genome, with links to identify placement and inversion differences.Solid blocks overlaying the genomes represent the presence of transposases, type IV secretion systems, and genomic islands, whilst solid blocks of blue that break up the colour gradient within the genomes are representative of LCBs not present in linked genomes

Figure 3 .
Figure 3. Analysis of the Mycoplasma felis core genome and gene presence/absence using Panaroo and plotted using Pagoo.A) the frequency of clustered genes within the six genomes used in the analysis; B) principal component analysis of gene presence/absence using Bray Curtis dissimilarity identifies clustering of the previously published genome sequences; C) clusters of genes present (blue) or absent (yellow) in the six genomes included in analysis.

Figure 5 .
Figure 5. A) Genome of the bacteriophage identified in Mycoplasma felis strain MF329 compared to MAgV1-like prophage sequences found in M. mustelae, M. molare and M. agalactiae.Genome annotations depicting conserved phage/prophage gene sequences are indicated in black.H = helicase; Pol = DNA polymerase; D = DNA primase; X = Xer recombinase; C = prohead protein; P = portal protein; T = terminase; HP = hypothetical protein.B) and C) cryotransmission electron photomicrographs of bacteriophages detected in a cell lysate of M. felis isolate MF329, showing that the phage structure consists of a contractile tail and polyhedral head.

Table 1 .
Mycoplasma felis isolates sequenced in this study.

Table 2 .
Summary of DNA sequencing results after trimming and quality filtering

Table 4 .
Average nucleotide identities (ANI) of Australian M. felis isolate genomes, the genome of the equid M. felis reference strain (Myco-2) and felid M. felis contig

Table 5 .
Coding sequences detected in the genomes of Australian M. felis isolates but absent in the equid M. felis reference strain (Myco2), and their possible species of origin based on amino acid sequence identity.

Table 6 .
Quantity and range of amino acid sequence identities and similarities of putative virulence and antimicrobial resistance genes (AMR) based on significant BLASTP hits (bit scores >50) compared to the virulence factor database and CARD database, respectively.

min -max) Median similarity (min -max) Total Median identity (min -max) Median similarity (min -max)
Presence and absence of genes within the genomes of Australian felid M. felis isolates compared to the genome of the Myco-2 equid M. felis reference strain, 538 clustered by putative protein functions.Only genes that differ between isolates or differ from the reference are listed.Counts indicate the number of gene copies detected