Molecular Structure of Isolated MvspI, a Variable Surface Protein of the Fish Pathogen Mycoplasma mobile

M. mobile strain 163K (ATCC 43663) was grown at 25°C in Aluotto medium, consisting of 2.1% heart infusion broth, 0.56% yeast extract, 10% horse serum, 0.0025% thallium acetate, and 0.005% ampicillin, to an optical density of around 0.1 at 600 nm (3, 25).

We modified the Gli349 purification procedure to fit MvspI isolation (1, 29). All procedures were done on ice except the gel filtration, which was performed at room temperature (RT). Cells from 1 liter of culture were collected by centrifugation at 14,000 × g for 10 min and washed twice with phosphate-buffered saline (PBS) consisting of 75 mM Na-phosphate (pH 7.3) and 68 mM NaCl. The cells were suspended to an optical density of 20 at 600 nm in 10 mM Tris-HCl (pH 8.0)–0.1 mM phenylmethylsulfonyl fluoride (PMSF) and then were mixed with Triton X-100 to 0.5% (vol/vol). After gentle shaking for 1 h, the suspension was centrifuged at 450,000 × g for 30 min (step i). The supernatant was fractionated by stepwise salting out with ammonium sulfate of 35% and 40% saturations. The insoluble fractions of 40% saturation were recovered by centrifugation at 22,000 × g for 15 min (step ii). The recovered fraction was dissolved and dialyzed overnight by 10 mM 2-(N-morpholino)ethanesulfonic acid (MES) (pH 5.9). The insoluble fraction caused by this pH shift was removed by centrifugation at 22,000 × g for 15 min (step iii). The soluble fraction was applied to a HiLoad 16/60 Superdex 200 prep grade column set on an ÄKTA prime or ÄKTA purifier (GE Healthcare, Milwaukee, WI) and eluted with a buffer consisting of 0.2 M NaCl, 0.1% Triton X-100, and 10 mM Tris-HCl, pH 8.0, with a flow rate of 1 ml/min at RT. The sample elution was monitored by absorbance at 280 nm and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (step iv). The homogeneity of protein fractions was estimated by densitometry of Coomassie brilliant blue (CBB)-stained SDS-PAGE gels with the use of a GT-9800F scanner (Epson, Nagano, Japan) and analysis software (ImageJ, version 1.44p; NIH). MvspI fractions were pooled and concentrated using a Biomax-10 instrument (Millipore, Bedford, MA) to 1 to 3 mg/ml, followed by removal of Triton X-100 by Carbiosorb adsorbent (Calbiochem, Darmstadt, Germany).

To analyze the associating properties, MvspI at a concentration of 1 mg/ml was applied to a gel filtration column (TSK G5000PWXL gel set; Tosoh, Tokyo, Japan), equipped with a series 1200 high-performance liquid chromatograph (Agilent Technologies, Palo Alto, CA) and was eluted with 0.2 M NaCl, 0.1% Triton X-100, and 10 mM Tris-HCl, pH 8.0, at a flow rate of 0.5 ml/min at 20°C, with the absorbance monitored at 280 nm (29). Thyroglobulin (669 kDa), ferritin (440 kDa), aldolase (158 kDa), and albumin (68 kDa) (with each mass shown in parentheses) were used as the standards for Stoke's radii of 8.5, 6.1, 4.8, and 3.6 nm, respectively (9). We examined the effect of Triton X-100 on the result of gel filtration and found that Triton X-100 does not influence the behavior of isolated MvspI. The analytical ultracentrifugation was performed by an XL-I analytical ultracentrifuge equipped with an An-60Ti rotor (Beckman-Coulter Inc., Fullerton, CA). MvspI at 0.75 μM (0.17 mg/ml) in the PBS was centrifuged at 42,000 rpm at 20°C for 360 min and scanned for the absorbance at 220 nm with 210-s intervals. The sediment coefficient [C(s)] distribution was analyzed by SEDFIT, version 12.2 (33), and the dimensions of the molecule were calculated by sednterp, version 1.09, assuming that the hydration of the molecule is 0.3 g/g.

MvspI at 20 to 200 μg/ml in 33% (vol/vol) glycerol and 0.3 M ammonium acetate was sprayed to a freshly cleaved mica surface. The following procedure was performed as described previously (1, 4, 29). For the analysis of MvspI decorated by an anti-MvspI monoclonal antibody (MAb14), the antibody was purified from ascites fluid by using a HiTrap Protein G HP column (GE Healthcare, Milwaukee, WI) (13). MvspI and MAb14 were mixed to 0.2 and 0.6 mg/ml, respectively, incubated for 1 h at 4°C, and sprayed onto a mica surface. Each particle image was picked up by EMAN, version 1.6 (http://ncmi.bcm.tmc.edu/∼stevel/EMAN/doc/) (16). The lengths, angles, and profiles of molecular images were analyzed by ImageJ, and image averaging was performed using Adobe Photoshop, version 7.0.1 (Adobe, San Jose, CA).

For epitope mapping of MAb14, purified MvspI was digested with trypsin or V8 protease at a 1/25 (wt/wt) ratio to MvspI at 37°C for 5 h. For the domain analysis of MvspI, purified MvspI was digested with proteinase K (ratio of 1/100, 1/50, 1/10, and 1/5 [wt/wt] to MvspI), chymotrypsin (1/100, 1/50, 1/25, and 1/10 [wt/wt] to MvspI), or trypsin (1/25 [wt/wt] to MvspI) at 37°C for 30 min. The digests were analyzed by SDS-PAGE, Western blotting (WB), and peptide mass fingerprinting (PMF), as reported previously (27–29, 43). Briefly, in PMF, protein bands in SDS-PAGE are digested by a residue-specific protease, for example, trypsin. The resulting peptide mixture is extracted and subjected to mass spectrometry. The combination of mass measurements of many peptides and the genome database allows us to identify the corresponding open reading frame (ORF) for each protein band. In the case that the peptide band of SDS-PAGE is part of the protein, we can identify the corresponding position in the whole sequence from the covering regions of peptides detected by the mass spectrometry.

The MvspI protein was isolated using the four-step procedure as presented in Fig. 1. In step i, MvspI was extracted with 0.5% Triton X-100 treatment, and most of it was recovered as a soluble fraction, as shown in lanes 1 to 3 in Fig. 1. In step ii, the MvspI fraction was applied to stepwise ammonium sulfate fractionation. The MvspI protein was recovered between 35% and 40% saturations as shown in lane 4 in Fig. 1. In step iii, the ammonium sulfate precipitate was dissolved and dialyzed against the gel filtration buffer. No obvious precipitation was found in this dialysis. In step iv, the MvspI protein was purified through gel filtration column chromatography in the presence of 0.1% Triton X-100 and used in the following experiments. The MvspI sample did not show any other major bands in the gel image of CBB-stained SDS-PAGE, as shown in lane 6, with 92.0% homogeneity, based on densitometry. The major protein band was confirmed to be MvspI by WB and PMF. Finally, about 100 μg of MvspI was obtained from 1 liter of mycoplasma culture.

The isolated MvspI protein was applied to a gel filtration assay (Fig. 2A). The elution pattern showed a single peak with an estimated Stoke's radius of 8.6 nm. Next, the protein was analyzed by analytical ultracentrifugation (Fig. 2B). The 94.6% protein sedimented with a sedimentation constant of 5.2 S_20.w (sedimentation coefficient standardized to 20°C in water) with a Stoke's radius of 9.7 nm, which is consistent with the result from the gel filtration. Based on the results of centrifugation, the mass of MvspI was calculated as 200 kDa, corresponding to the 218 kDa which was predicted from the mature amino acid sequence. Therefore, we concluded that MvspI behaves as a monomer in water. The molecular dimensions were estimated as a length of 66 nm and a diameter of 2.7 nm, assuming that the molecule is a cylinder rather than an ellipsoid.

The purified MvspI protein was rotary shadowed and observed by EM (Fig. 3A). Gooseneck particle images were found basically in all fields of a grid, and the frequency of their appearance changed in correspondence with the MvspI concentrations in the specimen. These facts suggest that the gooseneck particles are MvspI molecules. Some of the images featured a bulge at one end and a bend around the center. We picked 845 isolated images and examined their features. The frequencies of images with and without a bulge were 15.4% and 84.6%, respectively, and the frequencies of those with and without a bend were 45.5% and 54.5%, respectively. The images were classified into four types, bulge and bend (bulge-bend), no bend (bulge-none), no bulge (none-bend), and no bend or bulge (none-none), with frequencies of 5.0%, 10.4%, 40.5%, and 44.1%, respectively (Fig. 3). Next, we measured the total length, bend position, and bulge size as shown in Fig. 3B. The averages of total length were 66.4 ± 6.2, 68.8 ± 7.8, 76.4 ± 8.7, and 77.0 ± 9.5 nm for bulge-bend, bulge-none, none-bend, and none-none images, respectively. The total lengths shown in the images of the bulge-bend and bulge-none types were shorter than those of none-bend and none-none types by 10.0 and 8.2 nm, respectively, suggesting that these differences are caused by the alignment of the bulge part relative to the rod, as shown schematically in Fig. 3B (see also Fig. 7A). These molecular lengths are consistent with the estimation of a cylinder 66 nm long and 2.7 nm thick, based on the analytical centrifugation (Fig. 2B).

The average width of the bulge was 25.0 nm, as shown in Fig. 3B. The bend positions measured from the closer end were 28.8 nm and 31.5 nm, on average, for the bulge-bend and none-bend types, respectively. These observations suggest that the molecules have a bend at the corresponding positions in both the bulge-bend and none-bend types.

To determine the orientation of the amino acid sequence on the molecular shape, we examined the binding positions of a previously isolated anti-MvspI monoclonal antibody, MAb14, on the amino acid sequence and EM images (13). Purified MvspI was partially digested by trypsin and V8 protease, applied to SDS-PAGE gels, and analyzed by CBB staining and WB (Fig. 4A). Protein bands ii and iii (Fig. 4A), with apparent masses of 31 kDa and 43 kDa, were reactive to MAb14, while bands i and iv of 43 kDa and 30 kDa were not. PMF showed that the reactive bands, ii and iii, cover the regions of amino acids 1679 to 1985 and 1635 to 1985, respectively, while the nonreactive bands, i and iv, cover the region of amino acids 62 to 429 and both amino acids 62 to 303 and 1724 to 1985, respectively, showing that band iv was a mixture of two peptides (Fig. 4A, right). The peptides detected in the PMF covered 19.1%, 38.2%, 28.9%, and 28.0% lengths of each region for bands i to iv, respectively. The overlapping region among the reactive bands, except the regions of nonreactive protein bands, was the region of amino acids 1679 to 1723, located at the 0.85 position to the whole sequence length from the N terminus. Therefore, we concluded that the epitope of MAb14 is at this position.

Next, MAb14 was mixed with MvspI at twice the molar ratio, rotary shadowed, and observed by EM (Fig. 4B). MvspI molecules were found to be decorated by a globular particle of about 17.5 ± 2.5 nm (n = 35) in diameter, presumably MAb14, at one end of the molecular image, with a frequency of one per five particle images. The molecules bound by the antibody did not appear to have a bulge. This observation can be explained by either of the following possibilities: the binding of the antibody inhibits the bulge formation or the antibody hides the bulge, with its binding to the bulge end. Considering that the epitope of MAb14 lies at 0.85 from the N terminus of the total amino acid sequence, we concluded that the bulge end should be the C terminus.

To characterize the domain structures, the purified MvspI was partially digested with various amounts of proteinase K, chymotrypsin, and trypsin and then analyzed by WB using MAb14 (Fig. 5). Most of the major products were reactive to MAb14 (marked as bands v to ix). The masses of these bands were 144, 149, 141, 144, and 135 kDa, respectively, for bands v to ix. These protein bands were subjected to PMF and shown to cover at least the regions of amino acids 614 to 1985, 422 to 1985, 614 to 1985, 614 to 1985, 614 to 1985, and 698 to 1985, respectively, with length coverages of 23.3%, 14.7%, 20.9%, 24.7%, 25.7%, and 51.7% (Fig. 5B). Major peptide bands smaller than 135 kDa were not produced during the digestion of the N-terminal region. This shows that the N-terminal region of amino acids 28 to 613, including repeats a to g (Fig. 5B), was digested into small pieces, suggesting that this region is sensitive to proteases, probably because of its flexibility.

We analyzed the amino acid sequence of Mvsps to infer the topology of the molecule on the cell membrane using SMART (15) and DOLOP (17) (Fig. 6). MvspI was predicted to be cleaved at the C-terminal side of the 27th alanine with the modification of the following cysteine by diacylglycerol for lipid anchoring. MvspA, -H, -J, -K, -L, and -P were also predicted to be processed and modified in similar ways. Other Mvsps except MvspG were predicted to have a transmembrane segment at the N terminus. Mvsp proteins are also known to be featured with sequence repeats of about 90 amino acids, which are common in most Mvsp proteins (10). To list all repeats in Mvsp proteins, we analyzed the sequences of MvspA to MvspP, using the HMMER package (http://hmmer.janelia.org/) based on a hidden Markov model (8) in the same way that we did for Gli349 sequence analysis (18), and identified 52 repeats of about 90 amino acids, ranging from 72 to 105 amino acids whose E-values were less than 0.001 (Table 1 and Fig. 6). The sequence of MvspI had 16 repeats, which were clustered mostly on the N-terminal side, as shown in Table 1 and Fig. 5 and 6. The N-terminal ends of protease-resistant regions exist in repeats d, f, and g (Fig. 5B).

We produced an image of an MvspI molecule (Fig. 7A) by integrating the EM images and the sequence analysis. The molecule is a 77-nm rod with a bend at the position two-fifths from the N terminus and the occasional bulge at the C terminus. A plausible lipid anchor was suggested at the N terminus. The 16 repeat sequences, a to p, of about 90 amino acids exist mainly at the N-terminal side, and the region of repeats a to g is more flexible than the other parts.

The amino acid sequence can be mapped onto the molecular image, based on their features and the orientation of amino acid sequence, as shown in Fig. 7. The flexible repeats, a to g, are assigned to the 30-nm rod at the other side of bulge, and the solid repeats, h to p, are assigned to the 38-nm rod at the bulge side. The segments of 246 and 195 amino acid residues on each side of repeat p are assigned to the bulge. This assumption is consistent with the fact that the bulge appears to be composed of two small globules (Fig. 3A).

We carried out a PSI-BLAST search (2) to find a known three-dimensional (3D) structure whose sequence is homologous to Mvsps and to obtain more concrete images of repeats. We repeated the PSI-BLAST search six times until we found no new hits with the default parameters and thus obtained the position-specific site matrix for Mvsps. Unfortunately, with the use of the matrix, there was no hit against the sequences in the Protein Data Bank, where known 3D structures of proteins and nucleic acids have been deposited.

The region including repeats a to g was sensitive to protease, while the other repeats were not (Fig. 5B). To find out the possible causes of this sensitivity in the amino acid sequence, we carried out multiple sequence alignments of the repeats. Although the repeat regions did not show any features specific to the protease-sensitive regions, the lengths of the amino acid sequences connecting the repeats were different between the sensitive (repeats a to g) and resistant (repeats h to o) regions. That is, in the sensitive region, the lengths were 4 amino acids for a to b, 1 for b to c, 4 for c to d, 4 for d to e, 8 for e to f, and 4 for f to g, while in the resistant region, all the lengths were 1 or 2 amino acids. The longer connecting sequences that are unlikely to form a specific structure will cause flexibility, resulting in structures sensitive to proteases.

In a previous EM study of M. mobile cells, we found filamentous structures on the cell surface (28, 32, 35), consistent with the molecular image of MvspI obtained in the present study. However, we could not identify the MvspI molecule on the cell surface because at least MvspN, -O, and -K exist at the overlapping regions on the cell (10, 13).

The flask-shaped cell of M. mobile can be divided into three parts: the head, neck, and body from the pole of the membrane protrusion (13, 36). MvspI localizes at the head and body, and at least three proteins involved in gliding motility exist at the neck (34, 36, 37). In previous studies, the molecular shapes of Gli349 and Gli521, responsible for binding and force transmission, respectively, have been clarified (1, 29). The present study showed that the MvspI protein molecule shares some features with these molecules although they do not share the amino acid sequences. The molecules have a rod-shaped morphology about 100 nm long, an oval structure at one terminus, and a transmembrane segment at the other terminus. Low isoelectric points are also common features of 12 Mvsps and gliding proteins, i.e., pI values of 5.26, 5.05, and 5.38 for MvspI, Gli349, and Gli521, respectively. Moreover, sequence repeats of about 100 amino acids exist also in Gli349 (Fig. 7B) (1, 18, 38). Although these proteins play different roles in cellular activities, some aspects of protein functions are common; i.e., they stick out from the cell membrane and have some interactions at their distal ends with other molecules, antibodies for MvspI, sialyllactose for Gli349, and Gli349 for Gli521 (13, 14, 26, 29, 38). The common features in the structures and functions of these proteins may suggest that they evolved from a common ancestor. Alternatively, these features are essential for the functions and localization of the surface-exposed proteins of M. mobile and became common among unrelated proteins as a result of convergent evolution.

The structures of proteins responsible for surface variation represented by phase and antigenic variations may be critical to understanding the strategies of parasitic bacteria because surface variation is one of the determinants of their survival. However, these molecular shapes have not been examined experimentally. The present study may be the first example of such studies. Vlps, the antigenic variation proteins of M. hyorhinis, have been illustrated based on the features in amino acid sequences, where a filamentous molecule is anchored to the membrane at its N terminus while the other parts float outside (40). These predicted features of Vlps are common with those of MvspI although the total amino acid numbers after processing, ranging from 75 to 344, are much smaller than 1975, the number of amino acids in MvspI. We showed that the MvspI molecule is flexible, i.e., that it is a flexible rod with an occasional central bend and the occasional C-terminal bulge. Generally, the proteins for surface variations are known to be modified frequently through the modifications of the genome sequence (7). The flexibilities of the rod and the switch between the forms of the MvspI molecule with and without the bulge may produce additional structural variations on the molecules without genome modifications. Recently, MvspI has been shown to be involved in a novel mechanism of surface variation, designated mycoplasmal antigen modulation, where MvspI decreases in a reversible way, responding to binding of MAb14 (41). The molecular structure of MvspI clarified in the present study would be essential information to elucidate this novel mechanism of surface variation.