Random transposon insertion in the Mycoplasma hominis minimal genome

Mycoplasma hominis is an opportunistic human pathogen associated with genital and neonatal infections. Until this study, the lack of a reliable transformation method for the genetic manipulation of M. hominis hindered the investigation of the pathogenicity and the peculiar arginine-based metabolism of this bacterium. A genomic analysis of 20 different M. hominis strains revealed a number of putative restriction-modification systems in this species. Despite the presence of these systems, a reproducible polyethylene glycol (PEG)-mediated transformation protocol was successfully developed in this study for three different strains: two clinical isolates and the M132 reference strain. Transformants were generated by transposon mutagenesis with an efficiency of approximately 10−9 transformants/cell/µg plasmid and were shown to carry single or multiple mini-transposons randomly inserted within their genomes. One M132-mutant was observed to carry a single-copy transposon inserted within the gene encoding P75, a protein potentially involved in adhesion. However, no difference in adhesion was observed in cell-assays between this mutant and the M132 parent strain. Whole genome sequencing of mutants carrying multiple copies of the transposon further revealed the occurrence of genomic rearrangements. Overall, this is the first time that genetically modified strains of M. hominis have been obtained by random mutagenesis using a mini-transposon conferring resistance to tetracycline.


Restriction-modification system analyses.
Many restriction-modification systems (R-M systems) have been described in the genomes of mollicutes species ( Supplementary Fig. S1), with several having been functionally characterized 8,[16][17][18] . These systems provide immunity against foreign DNA and may therefore constitute an important barrier to both natural and artificial bacterial transformations 14,19 . Thus, to facilitate the development of a transformation procedure for M. hominis, we sought a strain containing the fewest of these defense mechanisms.
Twenty sequenced M. hominis genomes (13 from clinical strains described in this study, six already published 3,[20][21][22] , and the M132 reference strain) were analyzed and compared using Molligen and Rebase databases 23,24 . Many putative R-M systems were identified (Table 1), with each strain studied containing between two and six complete R-M systems. Interestingly, two putative R-M systems were shared by all strains, the type I EcoR124II-like system and a type III system. The type II Sau96I-like system is present in 19 out 20 strains, but 7 seem to harbor an incomplete system. Other predicted systems were rather diverse in terms of their type and specificity (three different type I and seven different type II systems), and/or copy numbers (four copies of an incomplete type II DpnII-like system were predicted in the strain 3364) ( Table 1). R-M systems were considered to be incomplete if one or more subunits were missing.
This analysis suggested a great diversity of R-M systems in the M. hominis lineage. Since no strain had a very low number or no R-M systems, we decided to pursue further investigations with the strains available in the laboratory: two reference strains (PG21 and M132) and the 13  www.nature.com/scientificreports www.nature.com/scientificreports/ between cells and PEG were investigated (1, 2, 5, 10, 30, and 60 min). The most transformants were obtained when cells were incubated for 30 min with the methylated plasmid in presence of 40% PEG 8000 (Table 2). This table focuses only on the conditions that yielded transformants. The transformation efficiency average from at least three independent experiments is shown as well as the minimal and maximal transformation efficiency obtained in each condition. Several other parameters were investigated, including the culture volume (3 to 10 mL), the number of washes (1 to 3) and the quantity of plasmid (1 to 20 µg) (Fig. 1), but their impact on transformation efficiency was unclear.
The results of these experiments led to the development of a reproducible PEG-mediated transformation protocol (see Material and Methods) that yielded M132 transformants at an efficiency of about 10 −10 transformants/ cell/µg of plasmid, with the highest score reaching 2.3 × 10 −9 transformants/cell/µg of plasmid ( Table 2).
The protocol was tested using 13 other clinical strains available in the laboratory that were susceptible to tetracycline, with only transformants obtained for strains 4016 (three mutants) and 5012 (one mutant).
Genotypic analyses of M. hominis transformants. Colonies generally appeared on selective plates after 3 to 14 days of incubation, with most colonies exhibiting a typical "fried-egg" morphology that is characteristic of the wild-type M. hominis strain.
All of the transformants tested were positive by PCR for the tet(M) gene, suggesting that the transposon in the pMT85-Tet plasmid was present into the M. hominis cells. Moreover, the transformants were confirmed as M. hominis via sequencing of the 16S rDNA PCR products.
Further analyses were performed to precisely identify the site of transposon insertion. Single-primer PCRs using the gDNA of 24 transformants was followed by Sanger sequencing of the PCR products, with results obtained for 16 of the 24 mutants. For these 16 mutants, the sequencing data showed that only one copy of the transposon was present in the bacterial genome and that its insertion seemed to occur at random position in the genome (Table 3). Transposons were found inserted into intergenic regions, a hypothetical protein, lipoprotein-encoding genes, or at the 3′ end of essential genes (Table 3). For the remaining eight isolates, insertion sites for two could not be determined because of technical problems, while six had sequencing profiles showing several superimposed peaks, suggesting that several copies of the transposon were present in the genome (see below).
Phenotypic analyses of M. hominis mutant 28-2. The M. hominis mutant 28-2 was of particular interest, as the tet(M) gene was integrated in the middle of a gene encoding a precursor of the P75 lipoprotein that is potentially involved in the pathogenicity of M. hominis (Fig. 2). The expression of the gene encoding P75 was investigated by semi-quantitative RT-PCR. A small transcript corresponding to the beginning of the gene (before the tet(M) insertion) could be detected, whereas no transcript was produced around the insertion site of the transposon, suggesting that only a short transcript was produced for the gene (Supplementary Fig. S3).
The ability of this mutant to adhere to HeLa cells was tested together with the M. hominis M132 wild-type strain. No difference in adhesion was observed between the M132 wild-type strain and the 28-2 mutant, suggesting the mutant fully retained its ability to adhere to eukaryotic cells. For both samples, the quantity of adhered M. hominis increased linearly as a function of the inoculum (Fig. 3). As expected, the correlation disappeared at high concentrations of M. hominis cells.

Detection of large DNA rearrangements by WGS.
To confirm that the genomes of some mutants carried multiple copies of the transposon, the genomes of two mutants (28-1 and 29-1) were completely sequenced. The WGS data revealed the presence of variable copy numbers of the transposon throughout the genomes M. hominis of the two assayed clones. Indeed, five copies of the transposon were identified in the genome of transformant 28-1, while three copies were identified in the genome of transformant 29-1. Among the five transposon copies present in transformant 28-1, four were part of tandem repeats, three of which included copies of the entire pMT85-Tet plasmid (Fig. 4), while the remaining copy of the transposon was located 300 kb away. Surprisingly,  www.nature.com/scientificreports www.nature.com/scientificreports/ this 300-kb region flanked by transposons was inverted compared to the wild-type strain. This inversion resulted in the disruption of the tyrosine recombinase-encoding xerC gene into two sections. For clone 29-1, a perfect duplication of approximately 9 kb flanked the double insertion of tet(M) at position 256,474 of the genome (Fig. 4). The third copy was observed to be inserted within a conserved hypothetical protein-encoding gene at position 208,266 of the genome. Altogether, these results suggested that large DNA rearrangement events in the M. hominis genome occurred adjacent to insertions of multiple copies of the transposon.

Analyses of the antibiotic resistance of clones carrying several copies of the tet(M) gene.
Minimal inhibitory concentrations (MICs) to tetracycline were determined to assess whether the copy number of the tet(M) gene influenced the level of tetracycline resistance. Compared to M132, whose MIC for tetracycline was 0.125 µg/mL, the 28-2 mutant, which carries only one copy of tet(M) gene, had an MIC of 8 µg/mL. This value was similar to that of the 29-1 mutant (16 µg/mL), which had three copies of the transposon. Interestingly, the clone 28-1, harboring five insertions of the tet(M) gene, had an MIC for tetracycline that was two to four times higher than that of the other mutants (32 µg/mL).

Discussion
In this study, we developed a reproducible PEG-based transformation protocol for M. hominis. We showed that the procedure can be used for random mutagenesis using a plasmid carrying a mini-transposon. In the future, we believe that the use of this procedure could lead to a protocol for the directed mutagenesis of M. hominis, which has been accomplished for other mycoplasma species 26 .
To achieve this goal, we first sought to identify an M. hominis strain containing a low number of R-M systems. These systems are composed of a restriction endonuclease and a DNA methyltransferase, the latter of which    www.nature.com/scientificreports www.nature.com/scientificreports/ prevents DNA cleavage by the cognate endonuclease (except for the type IV R-M system), and are well known to be a barrier to DNA transformation 14,27 . The genomes of 20 M. hominis strains were scrutinized for the presence of such systems, with the in silico analysis predicting the presence of numerous R-M systems encoded within the assayed strains (Table 1 and Supplementary Table S1). All of the studied genomes harbored a minimum of four systems total (complete plus incomplete), with up to nine systems identified for some clinical isolates (331 and 3364). Although no information is available regarding their activity, the high prevalence of R-M systems in the assayed M. hominis genomes combined with their high diversity (types I, II, III and potentially IV) may explain the difficulties encountered by the research community in developing strategies to genetically modify M. hominis.  www.nature.com/scientificreports www.nature.com/scientificreports/ Regarding the transformation protocol, several parameters appeared to be crucial to be successful (Fig. 1). First, we observed that the success of transformation depended on the selected strains. The PG21 strain was initially selected for our experiments as the most widely studied M. hominis representative 7,12 . Unfortunately, because no satisfying results were obtained, our efforts were redirected toward other strains. Among the 15 other strains tested, only three (M132, 4016 and 5012) were successfully transformed, although the reason for the observed success using these strains could not be ascertained. Indeed, the genomes of those three strains contained 5, 4 (7 including incomplete copies) and 3 (8 including incomplete copies) different complete R-M systems respectively; that is to say, a similar number of R-M systems as most of the other strains assayed (Table 1 and  Supplementary Table S1). Moreover, all three strains contained different types of R-M systems (types I, II, III), sometimes in duplicate or triplicate (clinical strains 4016 and 5012). Although some R-M systems are perhaps not active, this is certainly not all of the systems detected, mainly because, bacteria that are subjected to degenerative evolution, such as mycoplasmas, have a tendency to lose genetic material that is not essential for their survival. Thus, may all these observations suggest that barriers other than R-M systems, not yet identified, may exist and prevent artificial DNA transfer in M. hominis?
It should be pointed out that a potential type IV R-M system (members of which are only composed of a restriction enzyme that cuts methylated DNA) was predicted in the M. hominis M132 genome. This system shares similarities with the E. coli McrBC system, which cleaves DNA containing methylcytosine on one or both strands 28 . In our assays, successful transformation uniquely occurred when the plasmid DNA was treated with the CpG methyltransferase SssI, suggesting that the McrBC-like system is present but either not functional under our experimental conditions or not active in the provided substrate. In any case, all assays performed with unmethylated plasmid or plasmid methylated with the GpC methyltransferase CviPI did not yield any transformants.
Another important parameter to consider was the growth phase of the bacteria used in transformation assays. Indeed, transformants were only obtained when an equal volume of the early, mid and late log phases were combined ( Supplementary Fig. S2). Although this result was reported for U. parvum 13 , early or mid-log phase bacterial cultures are generally preferred in many other well-established transformation protocols [29][30][31][32] . Together with the bacterial growth phase, we noticed that some wash buffers were more suited to the transformation of M. hominis species than others. The T-Buffer used in the M. arthritidis transformation 16 was successful, whereas other buffers commonly used for mycoplasmas (PBS or 10 mM Tris-HCl and 0.5 M sucrose, pH 6.5) were not.
The last crucial parameters to optimize for the M. hominis transformation procedure were the molecular weight and the concentration of the fusogenic chemical agent (PEG) used for transformation. The time of contact between the bacterial cells and the plasmid mixture was also of high importance. The optimal procedure developed required the use of PEG with an average molecular weight of 8,000 g/mol (PEG 8000) at high concentrations (40%, 50%, or 60% in T-Buffer), where the time of contact with bacterial cells/plasmid mixture lasted for 10 min or more. This need for the latter adjustment was unexpected, as most of the previously published PEG-mediated transformation protocols recommended that the incubation time of contact not be allowed to proceed for more than 2 min due to the presumed toxicity of the PEG toward the cells. In our experiments, we did not observe a higher mortality of M. hominis cells when their membranes were permeabilized by PEG for 10 to 30 min compared to 2 min (data not shown). Finally, for the transformation experiment to succeed, we noticed that more than 10 µg of plasmidic DNA are required since lower quantity did not yield any transformants.
Two antibiotic resistance-encoding genes were tested for selection during the transformation experiments. In contrast to the tetracycline determinant, the gentamicin resistance gene only allowed for the recovery of false-positive clones corresponding to spontaneous resistance mutants (data not shown). The development of a replicative plasmid containing the origin of replication of M. hominis was attempted, although convincing results were not obtained (data not shown).
By combining all of these parameters, we generated a reliable transformation protocol for M. hominis. Relatively low efficiencies were obtained for each experiment (up to 2.3.10 −9 transformants/cell/µg of plasmid (corresponding to 1 to 3 transformants per experiment), indicating that the protocol should be further optimized. However, the results obtained using this protocol are reproducible and resulted in the generation of genetically modified M. hominis cells by transposon mutagenesis. The protocol may be further refined by counteracting of R-M systems. For example, the transformed DNA could be protected by in vitro methylation before its entry into the cells. In the current protocol, plasmids are methylated with the commercial methyltransferase SssI, which only provides protection against some restriction endonucleases (those recognizing CpG sites), but not all. DNA protection with other types of cytosine methyltransferases, but also with adenine methyltransferases should be considered, both type of R-M systems being found in M. hominis genomes (Tables 1 and S1). The best way to tackle the problem would be to prepare M. hominis cellular crude extracts (or M. hominis recombinant methyltransferases) in order to be able to methylate the exogenous DNA with M. hominis endogenous methyltransferases just before transformation 8 . However, it could be laborious and time-consuming to obtain crude extracts with fully functional methyltransferases or determine the laboratory conditions in which recombinant methyltransferases are active. The incoming DNA can be degrading by R-M systems once in the cell, but can also be degraded by surface nucleases before its entry into the cell 33 . Most nucleases require a divalent cation as a cofactor to be fully active (usually Mg 2+ or Ca 2+ ). Washing the cells with the chelators EDTA or EGTA may help neutralizing their activity and increasing the number of transformants. Finally, reaching better transformation efficiencies may also rely on the construction of a plasmid that is more adapted to M. hominis. Different promotors for driving the tet(M) expression gene could be tested, and a codon optimized version of the tet(M) gene could also be designed.
Mutants of interest were isolated over the course of our transformation experiments, such as mutant 28-2, which harbors the tet(M)-carrying transposon within the P75 lipoprotein gene that is potentially involved in the cytoadherence of M. hominis 34,35 . After verifying that the gene was knocked-out by PCR and RT-PCR, the adhesion ability of the mutant strain was tested using HeLa cells (Fig. 3). The results showed no difference in adhesion between the M132 wild-type strain and the derivative mutant 28 www.nature.com/scientificreports www.nature.com/scientificreports/ respect to this result: (i) the truncated protein may still be expressed at the surface with a conformation allowing adhesion to HeLa cells, antibodies against the P75 protein could be helpful to confirm this hypothesis by Western Blot; (ii) the adhesion test to HeLa cells was not sensitive enough, a possibility for which a negative control using a nonadherent strain (which is not currently available) would allow us to confirm or rule out this hypothesis; and, (iii) the absence of one protein is not sufficient to observe an altered adhesion, as other surface proteins have been shown to be important in M. hominis adhesion 36,37 . Additionally, adhesion experiments using other cell lines could certainly be done, but those tests would require some optimizations.
The use of next generation sequencing for deciphering the genome sequence of two transformants (mutants 28-1 and 29-1), confirmed the presence of multiple copies of the transposon throughout their genomes. A large 300-kb genomic inversion was observed in mutant 28-1, which caused the inactivation of a gene annotated as putative tyrosine recombinase XerC. This type of recombinases is known for its role as a mediator in site-specific recombination events and inversion events in bacteria [38][39][40][41][42][43] . Interestingly, the transposon was shown to be integrated into this gene in three mutants (two with a single insertion and one with a multiple insertion) out of the 24 obtained during this study. We concluded that xerC was certainly a "hotspot" of integration of this transposon in M. hominis M132 species, similarly to the four genes MG339 (recA), MG414 (Hypothetical protein), MG415 (hypothetical protein), and MG428 (putative regulatory protein) in M. genitalium. which constituted about 31% of the total transposon insertions during gene essentiality study 44 . We could also hypothesize that mutations in this locus facilitate fitness and growth. Multiple insertions of transposons as well as whole plasmid integration have been previously observed in mycoplasmas using the transposons Tn4001 and Tn916 13,[45][46][47][48][49][50] . In the pMT85-Tet plasmid, the transposase-encoding gene was moved from the transposon to limit multiple insertions and full plasmid integration 25,45 . To the best of our knowledge, we showed for the first time the presence of multiple transposon integrations, whole plasmid recombination and highlighted large DNA rearrangements in the genomes of M. hominis cells that were certainly induced by the presence of the plasmid. It would be interesting to analyze more mutants with suspected multiple insertions (i) to look for the presence of small or large genomic rearrangements, (ii) study their nature (insertions, duplications, inversions) and finally (iii) to understand the underlying mechanisms.
Another interesting outcome of this work was that the tet(M) copy number may influence the level resistance of a strain to an antibiotic. In our experiments, clone 28-1, which had the highest copy number of the tet(M) gene, has a tetracycline MIC that is two to four times higher than that observed of the other mutants (32 µg/mL).
Altogether, these data demonstrate that we are now capable of obtaining some genetically modified M. hominis mutants by random mutagenesis. Even though the transformation efficiency obtained using this protocol is currently low, these results are encouraging and lay the foundation for the function studies of genes in this bacterium. These results may also pave the way toward the development of genome transplantation to permit targeted mutagenesis and direct inactivation of genes of interest in M. hominis.  51 . Transformants were cultured in the same medium containing 2 µg/mL of tetracycline. Bacterial titers were evaluated both by determining color changing units (CCU) and colony-forming units (CFUs) as previously described 51 . MICs were determined according to the Clinical & Laboratory Standards Institute guidelines 52 . plasmid and in vitro methylation. The vector used for M. hominis transformation was the plasmid pMT85-Tet, which was derived from the Tn4001-based mini transposon plasmid pMT85 25 . The pMT85-Tet plasmid carried the tetracycline resistance gene tet(M), the expression of which was driven by the spiralin promotor (PS) instead of the gentamicin resistance gene 53,54 . Before transformation, the plasmid was methylated by the CpG methyltransferase from Spiroplasma sp. strain MQ1 (M. SssI, New England Biolabs, Ipswich, Ma, United States) according to manufacturer's recommendations.

M. hominis transformation protocol.
After determining the optimal conditions, the transformation protocol was performed as follows: a 10 8 CFU/mL pre-culture was diluted from 10 −1 to 10 −10 and incubated for 24 h at 37 °C to obtain cultures at three different growth phases (e.g., early, mid-log and late log phase). The pool of these three different growth stages (which corresponded to the 10 −3 , 10 −4 and 10 −5 dilutions in three milliliters of culture) was centrifuged at 10,000 g, 4 °C for 20 min. The pellet was washed twice with cold T-Buffer (Tris 10 mM, pH 6.5) and centrifuged at 10,000 g, 4 °C for 10 min. After centrifugation, the cells were resuspended in 400 µL of cold 0.1 M CaCl 2 and incubated at 4 °C for 30 min. Cold CaCl 2 -incubated cells (100 µL) were gently mixed with 10 µg of yeast tRNA (Life technologies, Carlsbad, CA, United States) and 10 µg of plasmid methylated with the methyltransferase SssI (New England Biolabs, Ipswich, Ma, United States). This mixture was aliquoted onto the surface of 1.5 mL of 40% PEG 8000 (Sigma-Aldrich, Saint-Louis, MO, United States) for 30 min. The contact was stopped by the addition of 7.5 mL of Hayflick arginine liquid medium and the cells were incubated for 3 hours at 37 °C. The transformation reaction was centrifuged at 8,000 g, room temperature for 10 min. The pellet was suspended in 1 mL of Hayflick arginine liquid medium and 200 µL was plated onto selective solid medium supplemented with 2 µg/mL of tetracycline (Sigma-Aldrich, Saint-Louis, MO, United States) and incubated at 37 °C with 5% CO 2 . Transformed M. hominis colonies appeared 3 to 7 days after transformation.
Colonies obtained on selective plates were picked and transferred into 1 mL of Hayflick arginine plus tetracycline (2 µg/mL) medium and incubated at 37 °C for 24 to 96 hours. Cultures were stocked at −80 °C. PCRs were performed with DNA extracts obtained using a NucleoSpin ® Tissue kit (Macherey-Nagel, Düren, Germany).