Identification and sequence analyses of the gliding machinery proteins from Mycoplasma mobile

Mycoplasma mobile, a fish pathogen, exhibits its own specialized gliding motility on host cells based on ATP hydrolysis. The special protein machinery enabling this motility is composed of surface and internal protein complexes. Four proteins, MMOBs 1630, 1660, 1670, and 4860 constitute the internal complex, including paralogs of F-type ATPase/synthase α and β subunits. In the present study, the cellular localisation for the candidate gliding machinery proteins, MMOBs 1620, 1640, 1650, and 5430 was investigated by using a total internal reflection fluorescence microscopy system after tagging these proteins with the enhanced yellow fluorescent protein (EYFP). The M. mobile strain expressing a fusion protein MMOB1620-EYFP exhibited reduced cell-binding activity and a strain expressing MMOB1640 fused with EYFP exhibited increased gliding speed, showing the involvement of these proteins in the gliding mechanism. Based on the genomic sequences, we analysed the sequence conservativity in the proteins of the internal and the surface complexes from four gliding mycoplasma species. The proteins in the internal complex were more conserved compared to the surface complex, suggesting that the surface complex undergoes modifications depending on the host. The analyses suggested that the internal gliding complex was highly conserved probably due to its role in the motility mechanism.


Results
Subcellular localisation of the internal structure proteins. In order to examine the involvement of candidate proteins in the gliding machinery, we tagged MMOBs 1620, 1640, 1650, and 5430 with a yellow fluorescent protein. Previously, we had optimised and developed an EYFP tagging system for M. mobile 9,17,18 . The system included an eyfp gene which was codon-optimised for Mollicutes (opteyfp). To achieve higher fluorescence intensity, opteyfp was placed under the control of a strong promoter of the gene encoding for the elongation factor Tu of M. mobile (Mtuf) resulting in Mtuf-opteyfp fragment. In the present study, the Mtuf-opteyfp fragment was inserted into pMTnGm, a Tn4001 transposon-derived vector system that harbours transposase (tnp) gene outside of the transposable element by using the Infusion subcloning method. The resulting fluorescence may be dependent on whether the tag was fused at N or C-terminal of the protein. Considering this possibility, we created N-terminally and C-terminally tagged fusion proteins as different clones (Fig. 3A).
The C-terminal fusions of MMOBs 1620, 1640, and 5430 and the N-terminal fusion of MMOB1650 showed weak signals which are not suitable for analyses (Fig. S1). Then, the N-terminal fusions of MMOBs 1620, 1640, 5430 and the C-terminal fusion of MMOB 1650 were selected for further analyses. Next, we examined the subcellular localisation of candidate proteins, MMOBs 1620, 1640, 1650, and 5430 by comparing their positions relative to Gli349, which is known to localise at the gliding machinery ( Fig. 3B) 12,13 . Gli349 was labelled by Cy3 using a monoclonal antibody, mAb7 12,13 .
For localisation controls, we examined the localisation of Type 1 and Type 3 F-type ATPase/synthases, the α-subunit paralogs. MMOB2110 (Type 1) and MMOB2970 (Type 3) were tagged at their N-terminus by EYFP, as previously done for MMOB1660 (Type 2) 9 . Next, we confirmed the expression of the fusion proteins by immunoblotting using an anti-EYFP antibody. We detected EYFP-MMOB1620, EYFP-MMOB1640, MMOB1650-EYFP, EYFP-MMOB5430, EYFP-MMOB2110, EYFP-MMOB1660, and EYFP-MMOB2970 (Fig. 4A) at their expected molecular size. MMOB1650-EYFP exhibited two bands, one of them was the expected size, while the other band was of smaller molecular size (Fig. 4A), suggesting that MMOB1650-EYFP was either processed or digested in the cell. The amounts of EYFP and fusion proteins were consistent with their fluorescence intensities, suggesting that the fluorescence intensities of directly reflected the expression levels of the fusion proteins (Fig. 4B).
Further, we estimated the fluorescence intensity of a cell with a typical flask shape in phase-contrast along the cell axis and determined the distribution range (Fig. S2) 17 . To map the fluorescent foci on the cell axis, the Three types of gene clusters related to F-type ATPase/synthase in M. mobile are shown in the red box. Type 2 codes for the internal structure of the gliding machinery. Gene clusters for three species homologous to the internal structure of M. mobile gliding machinery are shown in the lower positions. All of these species have five (six in case of M. pulmonis, because the gene homologous to MMOB1650 is split into two ORFs) conserved hypothetical proteins coded from the upstream of the genes of αand βsubunit paralogs. The phosphate acetyltransferase (eutD) and acetate kinase (ack) genes are predicted to be transcribed as an operon with the αand βsubunit paralogs in M. mobile and M. pulmonis 15,43 , while they are distantly positioned in the genomes of M. agassizii, and M. testudineum.
phase-contrast image was used to determine the front-end position of a cell assuming that the internal structure was uniform in shape and density among individual cells. The fluorescence intensities were measured for 10 images of each construct and the average intensity was calculated (Fig. 3B). The signal of MMOB1660 overlapped with that of Gli349, as reported previously 9 . The signals of MMOBs 1620, 1640, and 1650 mostly overlapped with that of Gli349. The signal of MMOB5430 also overlapped with Gli349, but extended beyond Gli349 in the direction of the cell body. The signals of the α-subunit paralogs, MMOBs 2110 and 2970, were distributed through the cell, distinct from those of MMOB1660 and other component candidates.

Effects of fusion proteins on cell binding and gliding speed.
The fluorescent protein tag on the fusion protein might cause steric hindrance in the target proteins affecting their function 9,17,18 . To examine the effects of fluorescence tagging for the candidate proteins, the gliding motility, the binding activity and the gliding speed of transformant cells were examined and compared to the wild-type (WT) strain. All strains were cultured to www.nature.com/scientificreports www.nature.com/scientificreports/ an optical density of 0.08 at 600 nm. Then, cells were collected, washed, and suspended in phosphate-buffered saline supplemented with glucose (PBS/G). The cell suspensions were inserted into tunnel slides, and videos were recorded on a phase-contrast microscope (Fig. 5A). An area of glass 64.0 μm wide and 85.3 μm long was selected and the cells bound to it were counted. The cells from 10 independent fields were counted for each strain. The average gliding speed was calculated for 30 s and 120 independent cells including fusions of MMOBs 1620, 1640, 1650, and 5430. The cells with MMOB1620-EYFP exhibited a 25% reduced binding activity and the cells with EYPF-MMOB1640 exhibited 33% increased gliding speed (Fig. 5B). These results are consistent with MMOBs 1620 and 1640 being a part of the gliding machinery, with changes in their functions caused by the fluorescent protein tagging, affecting some steps in the gliding mechanism.

Genome and sequence analyses of the new gliding species. The class Mollicutes represented by
Mycoplasma are subdivided into four subgroups. Some species in the Pneumoniae and Hominis subgroups have gliding capability 1 . These two subgroups are phylogenetically distant from each other and do not share any genes involved in gliding 5,6,19-21 . The orthologs of M. mobile gliding proteins have been reported in the genome of M. pulmonis, but not in other species 10 . To obtain information related to protein functions based on the amino acid sequences, we conducted a bioinformatic search for the candidate protein orthologs in the genomes of other mycoplasmas. First, we searched for the orthologs of Type 2 F-type ATPase/synthase α subunit of M. mobile using BLASTP (Accelerated protein-protein BLAST) and PSI-BLAST (Position-Specific Iterated Blast) algorithms within Mycoplasma (taxid:2085) non-redundant protein sequence database. We could not find any other Type 2 F-type ATPase/synthase α-subunit orthologs. Then, we searched for the draft genome sequences in the Joint Genome Institute genome portal and Type 2 F-type ATPase/synthase gene clusters were identified along with Type 1 and Type 3 clusters in Mycoplasma agassizii and Mycoplasma testudineum (pathogens of the desert tortoise), which belong to the Hominis subgroup ( Fig. 2A).
Further, we compared the Type 2 gene clusters of M. mobile, M. pulmonis, M. agassizii, and M. testudineum. All of these species except M. pulmonis have five conserved hypothetical proteins upstream from the αand βsubunit paralogs. The ortholog of MMOB1650 is split into two genes in M. pulmonis, MYPU_2330 and MYPU_2340. Two genes, phosphate acetyltransferase (eutD) and an acetate kinase (ack) are predicted to be transcribed as an operon with the αand βsubunit paralogs in M. mobile and M. pulmonis, but these genes were distantly positioned in the genomes of M. agassizii and M. testudineum.
Next, we searched for the genes coding for the surface proteins of gliding machinery in the genomes of M. agassizii, and M. testudineum (Fig. 2B). In the genome of M. mobile, the genes for the surface structure are present in the order of gli123-gli349-gli521-gli42. This gene alignment is mostly conserved in M. pulmonis. We analysed M. agassizii and M. testudineum genomes by BLAST search and found only the gli349-gli521 genes, but not the gli123 and gli42 genes were not detected in the genome of M. agassizii and M. testudineum.
Sequence analyses of the surface structure proteins. By using the multiple sequence alignment algorithm T-Coffee, we compared the amino acid sequences of Gli349 and Gli521 orthologs from M. mobile, M. pulmonis, M. agassizii, and M. testudineum genomes (Fig. 6) to examine the relationship of the orthologs and to predict their function.
Gli349 protein (3183 amino acids) of M. mobile functions as a "leg" during gliding and is essential for binding and gliding 12,13,22,23 . The Gli349 protein is composed of two short rigid arms, a long flexible part, and a C-terminal oval 'foot' which are linked tandemly (Fig. 6A) [24][25][26] . A foldable hinge links the two short arms. An additional feature of Gli349 is that it has 18 repeats of about 100 amino acids 26 .
The sequence alignment of the four Gli349 orthologs showed 72 similarity in the main score of T-Coffee. The N-terminal region, Rigid arm 1 (RA1) comprising the repeat sequences A through E (118-727 aa) showed 62 T-Coffee score. Rigid arm 2 (RA 2) comprising the repeat sequences G through I (830-1161 aa) showed a 74. The long flexible part (LFP) comprising the repeat sequences J through V (1248-2720 aa) showed 80 T-Coffee score. The binding positions of the previously described two inhibitory antibodies mAb3 and mAb7, and two mutation points inhibiting binding and gliding were found on Gli349 in m13 and m23 strains 23,27 . These constituted the highly conservative regions corresponding to J-K-L (1248-1546 aa) repeats in the LFP. The N-terminus of the foot has a leucine-rich repeat (LRR), as observed in Toll-like receptors and contains a serine residue (at position 2770) which is essential for binding 5,22 . The regions corresponding to J-K-L showed a 95 T-Coffee score (Fig. 6B) 23,27 .
Gli521 protein (4727 amino acids) of M. mobile, is also essential for binding and gliding, functions as a crank in force transmission through interaction with Gli349 (Fig. 6A) 11,28,29 . Limited proteolysis showed that Gli521 consists of three parts, domain I (oval, 44-715 aa), domain II (rod, 889-3035 aa) and domain III (hook, 3244-4560 aa) which are connected by flexible hinges 28 . The full-length protein sequence alignment comparing Gli521, WP_084232748, WP_094254639, and MYPU_2120-2140 showed 65 T-Coffee score, while domain I, domain II, and domain III showed 72, 77, and 77, respectively. The gap regions between the domains showed low similarities (~ 41 T-Coffee score). The flexible region at the C-terminal end of domain III corresponding to 4573-4662 amino acids showed high similarity (70 T-Coffee score), indicating that this flexibility was essential for the protein function. Previously, our group has identified mutations and epitopes in the monoclonal antibody against Gli521 which influenced binding activity and gliding speed. Interestingly, these sequences were never mapped onto the highly conserved regions 23,27 . We then focused on analysing the conservativity of individual mutation points to understand its effect on the function of the protein. Proline at 476 th and serine at 859 th positions in Gli521 are totally diverse but proline at 1461 st position is totally conserved among the four species. Serine at 2770 th position in Gli349 is totally conserved, consistent with the fact that m23 mutant which has this serine mutated to leucine is incapable of binding 23,27 . Using genome sequences of two new species, we evaluated the conservativity of each position. This will give us a clue to design new experiments to elucidate this unique gliding mechanism.

Discussion
The protein components in the internal structure of gliding machinery were analysed. The MMOB5430 signal was slightly different from other component proteins, as shown in Fig. 3B. This difference in the signal pattern might be due to the small size of MMOB5430. In general, compared to the larger proteins smaller proteins tend to occupy smaller spaces in the cell, EYFP tagging to such proteins might result in steric clashes with other components leading to the formation of aggregates from the excess protein molecules and ultimately resulting in patchy signals as seen for MMOB5430 9,17,18 . However, cellular localisation based on the fluorescent signals showed the involvement of MMOB5430 in the gliding machinery.
The EYFP fusions caused a reduction in cell binding for MMOB1620 and acceleration in gliding speed for MMOB1640. These results can be explained if we consider the working model explaining the gliding mechanism, where a cycle consists of four steps of interaction including catch, pull, drag and release 5,6,8,27,31 with the binding target, i.e. sialylated oligosaccharides (SOs) [32][33][34] . The EYFP fusion to MMOB1620 might reduce the catch step or activate the release step, and the EYFP fusion to MMOB1640 may activate the frequency of the pull step.
We have shown that 10 proteins localised to the gliding machinery of M. mobile. Proteins MMOBs 1620, 1630, 1640, 1650, 1660, 1670, 5430, and 4860 constitute the internal structure, and Gli123 (MMOB1020), Gli349 (MMOB1030), and Gli521 (MMOB1040) constitute the surface structure. Another component protein of the gliding machinery, Gli42 (MMOB1050) is translated together with the surface proteins, but the localisation has not been clarified. Although we do not have any direct evidence for MMOB1610, we think that it is probably involved in the gliding machinery, because proteins transcribed as an operon frequently form a complex in bacteria. It is likely that the internal structure of gliding machinery evolved as Type 2 F-type ATPase/synthase gene cluster from the genuine cluster 5,8,15,22 . As structural details of all the internal complex are not available, it is difficult to establish the similarities with F-type ATPase/synthase but the rough image obtained by electron microscopy suggested a pair of particles featured by rotational symmetry 8,14 which can be related to the feature of F-type ATPase/synthase 15,35 .  Table S1 for details.
Further, we performed sequence analyses of gliding proteins based on an amino acid sequence from the four species. We acquired the genomic sequence of two new Mycoplasma species, which glide probably by a mechanism common with M. mobile. We were able to investigate conservativity in each region of the proteins for the first time by comparing the amino acid sequences of the proteins required for gliding from four Mycoplasma species (Figs. 2, 6, 7). The summary of the analysis is as follows: (i) The proteins of internal structure are highly conserved compared to the proteins of the surface structure. The surface structure should change depending on the environments, as observed in the difference in the conservativity levels between the surface and core structures of bacterial flagellins 36 . (ii) Gli123 and Gli42 were not essential for the mechanism of gliding, while Gli349 and Gli521 were essential. (iii) The N-terminal region in the foot domain of the Gli349 protein was well conserved, suggesting that it should be essential for gliding 5,22 . (iv) The phylogenetic relationship of each protein is in common with the phylogenetic relationship of 16S rDNA, suggesting that the gliding mechanism evolved with the Mycoplasma genomes.
In this study, we identified new proteins involved in the internal and surface machineries for an M. mobile gliding mechanism. This information will be a clue to clarify the mechanism of this specialized gliding mechanism.

Materials and Methods
Strains and culture conditions. M. mobile 163 K (ATCC 43663) strain was cultured in Aluotto medium at 25 °C 37,38 . For the selection of M. mobile transformants, gentamicin sulfate was used at a final concentration of 50 μg/ml 9,37-39 . Escherichia coli strain DH5α and Stellar cells were used for all DNA manipulation procedures 9 .
Plasmid construction and transformation. For pMTnPE construction, tuf-eyfp was amplified from pMobopt using primers: 5′-CACACGAATTCTTAAAAAAGCTTGAACATAAGA-3′ and 5′-CACACGTCGACGAGCTCGGCCTATATGGCCAGATCTTTTATATAATTCATCCATACCTAAT-3′. The amplified tuf-eyfp fragment was inserted into pMTnGm using EcoRI and SalI sites. M. mobile genomic DNA was prepared using the Genomic-tip System (Qiagen, Hilden, Germany). For the construction of the N-terminal fusions of each targeted gene (MMOBs1620, 1640, 1650, and 5430), the DNA fragments were amplified from the genomic DNA and inserted into the BglII-SacI sites (the 3′ end of eyfp gene) of pMTnPE. For the construction of the C-terminal fusions of each targeted gene, the amplified DNA fragments were inserted into the 5′ end of the eyfp gene by using In-Fusion EcoDry PCR Cloning Kit (TaKaRa Bio, Shiga, Japan). Electroporation, colony PCR, and transposon insertion site detection were performed as previously described 9,17,18,40 . Procedure for making the frozen stock of the transformants. A single colony was picked up and mashed in 10 μl of Aluotto liquid medium. The mashed colony-agar mixture was inoculated into 1 ml of Aluotto medium containing 15 μg/ml of gentamicin for 3-5 days in 24-well plates. Then, 1 ml of cultivated transformants were inoculated into 10 ml of fresh Aluotto medium containing 15 μg/ml of gentamicin and were cultivated for 3-4 days. The culture was divided into 500 μl aliquots and was centrifuged at 10,000 × g. The supernatant was removed completely, and the cell pellet was resuspended in 100 μl of PBS consisting of 75 mM sodium phosphate (pH 7.3), 68 mM NaCl, and 10 mM glucose. Further, the cells were centrifuged at 10,000 × g and resuspended in 500 μl of fresh Aluotto medium containing no antibiotics. The suspension was frozen immediately in liquid nitrogen and kept in a deep freezer.
Fluorescence microscopy and protein analysis. Chemical fixation of the cells and immunofluorescence microscopy were performed as described previously 9,41,42 . TIRF imaging of the fixed samples was performed using an Olympus IX83 (Olympus, Tokyo, Japan) inverted optical microscope in TIRF mode. A 100 × TIRF objective (Plan-APO 100 ×/1.45 Oil, TIRFM, Olympus) was used to collect fluorescence onto an electron-multiplying charge-coupled device (EMCCD) camera (iXon X 3 , Andor, Oxford Instruments, UK) and excitation was provided by 514 and 532 nm lasers for EYFP and Alexa Fluor 555, respectively 16 . SDS-PAGE and western blotting analysis were performed as described previously 9 .
Analyses of binding and gliding. Cultured cells were collected by centrifugation at 12,000 × g for 4 min at room temperature (RT) and then were washed and suspended in PBS/G. The cell suspension was inserted into a tunnel chamber with 5-mm width, 22-mm length, and 86-µm thickness. The tunnel chamber was constructed with a coverslip and a glass slide, assembled with double-sided tape and precoated with a growth medium containing 10% horse serum for 60 min. Mycoplasma gliding was observed by phase-contrast microscopy using a BX50 microscope (Olympus) and recorded with a Wat-120N CCD camera (Watec, Yamagata, Japan). All video data were analysed by using ImageJ software, version 1.52a (http://rsb.info.nih.gov/ij/), as previously described 27,31,33,37 . Sequence and phylogenetic analyses. Genome analyses and comparisons were carried out with MolliGen, a database for Mollicutes genomes (http://molligen.org). Sequence similarities were searched using PSI-BLAST (http:// blast.ncbi.nlm.nih.gov/Blast.cgi) against the NCBI non-redundant protein sequence databank. The sequences of homologs were aligned by T-Coffee (http://www.igs.cnrs-mrs.fr/Tcoffee/tcoffee_cgi/index.cgi), and their phylogenetic trees were constructed by ClustalW (http://www.genome.jp/tools/clustalw/). The whole-genome sequences of M. agassizii and M. testudineum were provided by The Genome Portal of the Department of Energy Joint Genome Institute (http://genome.jgi.doe.gov). Protein sequences of M. pulmonis and M. mobile were named by their locus tags. PSIPRED v.3.0 (http://bioinf.cs.ucl.ac.uk/psipred/) was used to predict secondary structure. The TMpred (http:// www.ch.embnet.org/software/TMPRED_form.html) and TMHMM (http://www.cbs.dtu.dk/services/TMHMM) methods were used in combination for the prediction of transmembrane segments. Species and protein phylogenies were determined by the neighbour-joining (NJ) and the maximum likelihood (ML). The reliability of each tree topology was checked by 1000 bootstrap replications using the MEGA7 package (http://www.megasoftware.net/). Operon constructions were predicted by DOOR 2 (Database of prOkaryotic OpeRons, Version 2.0) 43 .