INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Adherence of the human pathogen Mycoplasma pneumoniae to host respiratory epithelial cells constitutes a critical step in the pathway leading to tracheobronchitis and atypical (walking) pneumonia. A polar extension of the mycoplasma cell membrane, the terminal organelle, is the major site of attachment and contains proteins responsible both directly and indirectly for attachment. Although the M. pneumoniae adhesin P1 mediates receptor binding (1, 5, 13), Triton X-100 (TX)-insoluble (hereafter referred to as triton shell or cytoskeletal) cytadherence-accessory proteins are required for both the proper formation of the attachment organelle and the localization of P1 to the attachment organelle (reviewed in reference 17). Loss of some of these proteins results in the failure to accumulate P1 at the attachment organelle and confers irregular cell shapes (9, 21, 34, 35), though the means by which this is manifested is unknown. Therefore, characterization of the biochemical properties and the order of assembly of the cytadherence-accessory proteins is necessary for a fuller understanding of both the regulation of adherence and the structure and formation of the attachment organelle.

The cytadherence-accessory protein HMW1 is a 112-kDa phosphoprotein (3, 4) that is concentrated along mycoplasma cell filaments, including the attachment organelle, as revealed by immunoelectron microscopic analyses (34). HMW1 has a modular structure (Fig. 1), including a large central acidic and proline-rich (APR) domain (4), which likely contributes to the irregular migration of HMW1 by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (24). Although the function of this domain is unknown, similar domains are present in other M. pneumoniae proteins that partition in the triton shell (24, 27, 28). Analysis of several M. pneumoniae noncytadhering mutants has begun to elucidate HMW1 function. Studies of the M6 mutant, in which HMW1 is absent and the cytadherence-accessory protein P30 is truncated (22), indicate that HMW1 is required for proper attachment organelle structure and function (9). In another mutant, designated I-2 (18), proteolytic turnover of HMW1 and several other proteins is accelerated (25), correlating with the absence of the cytadherence-accessory protein HMW2 (6, 18, 19). The accelerated turnover of HMW1 requires its C-terminal domain, as the absence of that region renders recombinant HMW1 stable in mutant I-2 (9). The timing and nature of the interactions involving HMW1, HMW2, P1, and other cytadherence-accessory proteins are poorly characterized, and the basis for their accelerated turnover in some mutants is unknown. Furthermore, analysis of the deduced amino acid sequences of these proteins reveals very little about how they might interact with each other or with other proteins of the triton shell.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   The deduced amino acid sequence of HMW1 suggests that there are three domains (4). Domain I (the N-terminal 170 amino acids of HMW1; white box) is predicted to consist of mostly -strand structure. Domain II (residues 171 to 522; black box [this boundary has been revised from that in reference 4 upon closer examination]) is the acidic, proline-rich (APR) domain common to several M. pneumoniae cytoskeletal proteins (see text). Domain III (the remainder of HMW1; medium gray box) includes a C-terminal domain postulated to be involved in targeting for proteolytic degradation (25). The dark gray box within domain I represents the location of the EAGR box (residues 106 to 136; see Discussion and Fig. 8). The light gray boxes within domain III represent the locations of predicted coiled-coil domains (residues 778 to 818 and 842 to 881; see Discussion).

The purpose of the present study was to correlate the stability of HMW1 with its subcellular location and fractionation characteristics. We determined that, despite the absence of known secretion or transmembrane signals (4), HMW1 is a peripheral membrane protein associated with the outer surfaces of wild-type M. pneumoniae cells. In both wild-type and mutant I-2 cells, the TX-soluble pool of newly synthesized HMW1 decreased over time at comparable rates. However, most of this HMW1 in wild-type cells was incorporated into the cytoskeleton, whereas in I-2 cells it was lost. These data suggest that interconversion of HMW1 between TX-soluble and cytoskeletal forms occurs in both wild-type and mutant I-2 cells but that stabilization of HMW1 in the cytoskeletal fraction, where it is protected from degradation, is dependent upon proteins absent or reduced in the I-2 mutant.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and culture conditions. Wild-type M. pneumoniae strain M129 (broth passage 17 since original culture) (23) and mutant I-2 (18) were cultured in Hayflick broth (10) at 37°C to mid-log phase (phenol red indicator was orange) and harvested as previously described (8).

Antibodies. Preparation and characterization of the anti-HMW1 serum (34), anti-P1 serum (18), and anti-HMW2 serum (19) were described previously. Anti-elongation factor G (EF-G) serum was a kind gift of R. Herrmann (Universität Heidelberg, Heidelberg, Germany). A tetrameric multiple antigenic peptide (36) with the amino acid sequence KAIVNGMMTQDQKSNNGTEL, corresponding to predicted amino acids 131 to 150 of M. pneumoniae membrane-associated protease FtsH, was synthesized for preparation of rabbit anti-FtsH serum. Freund's complete adjuvant was used for initial immunization, and incomplete adjuvant was used for booster immunizations. After two booster immunizations, the antiserum tested positive by immunoblotting for a 75-kDa band in M. pneumoniae lysates (not shown), in good agreement with the predicted mass of M. pneumoniae FtsH (77.7 kDa).

Membrane fractionation. Cells were fractionated by a procedure adapted from Razin (30, 31). Wild-type M. pneumoniae stock was inoculated into five 162-cm² tissue culture flasks (Corning Costar, Cambridge, Mass.), each containing 50 ml of Hayflick broth, and grown to mid-log phase. The medium was decanted, and the cell monolayers were rinsed in 10 ml of PBS. The cells were scraped into a total of 10 ml of PBS and centrifuged for 15 min at 17,400 × g at 4°C; pellets were then suspended in PBS and recentrifuged. The pellets were suspended in 10 ml of 2 M glycerol using a syringe with a 22-gauge needle and centrifuged for 15 min at 17,400 × g. The pellets were suspended in 2 ml of 2 M glycerol by using a syringe, and the cell suspension was injected into 60 ml of water at 37°C, causing cell rupture by osmotic lysis. Twenty-two milliliters was set aside as lysate. Forty milliliters of the ruptured cell suspension was centrifuged for 1 h at 100,000 × g, and the supernatant was pooled and saved as cytosol. The pellets were suspended in 200 to 500 µl of  buffer (7.5 mM NaCl, 0.5 mM -mercaptoethanol, 2.5 mM Tris-HCl [pH 7.4]) (diluted 1:20) by using a syringe with a 25-gauge needle. This pool of pelleted material was saved as membrane.

RIP. Wild-type and mutant I-2 M. pneumoniae cells were grown in 75-cm² flasks and pulse-labeled with [³⁵S]methionine (>1,000 Ci/mmol; 1 Ci = 37 GBq; Amersham Pharmacia Biotech, Piscataway, N.J.) in Hanks' balanced salt solution (HBSS) (Sigma, St. Louis, Mo.) for 30 min as described previously (25). After the cells were washed once with HBSS containing 1 mM methionine and three times with PBS, the pellets from each flask were suspended in 2 ml of PBS. Samples were split, with part being immediately subjected to immunoprecipitation (whole-cell radioimmunoprecipitation [RIP]) (4) and part being osmotically lysed as described above before immunoprecipitation (lysate RIP) (25). Antibodies used were anti-HMW1 (C-terminal domain) and anti-FtsH (predicted surface loop). For whole-cell RIP, after the suspended cells were incubated with antiserum overnight at 4°C while rocking, the samples were spun for 30 min at 100,000 × g to remove free antibodies. The resulting pellets were resuspended in TDSET (1% TX, 0.2% sodium deoxycholate, 0.1% SDS, 10 mM tetrasodium EDTA, 10 mM Tris-Cl [pH 7.8]) and incubated with protein G-Sepharose beads, hydrated according to the manufacturer's protocol (Sigma), while rocking, 2 h at 4°C. For the lysate RIP, osmotic lysates containing membrane and cytosol were incubated with antiserum overnight at 4°C while rocking. Subsequently, the lysates were brought to 1 × TDSET by the addition of 0.2 volume of 4× TDSET, and protein G-Sepharose beads were added. The mixture was incubated for 2 h at 4°C. For both RIPs, after the beads were washed three times with TDSET, samples were heated in SDS-PAGE sample buffer and subjected to SDS-PAGE. The dried gels were subjected to autoradiography.

Indirect immunofluorescence microscopy. M. pneumoniae cells were grown on glass coverslips; coverslips used for mutant I-2 were pretreated with 0.1% poly-L-lysine. After growth, coverslips were washed three times with PBS for 10 min each time and then blocked for 1 h at room temperature with 4% bovine serum albumin in PBS. After three washes in PBS, coverslips were incubated for 1 h at 37°C with anti-P1 (1:100), anti-HMW1 (1:1000), anti-EF-G (1:1000), or preimmune serum diluted in PBS containing 1% bovine serum albumin with gentle shaking. After three washes in PBS, coverslips were incubated for 1 h at 37°C with fluorescein isothiocyanate-conjugated goat anti-mouse (1:100) or anti-rabbit (1:50) immunoglobulin G (Sigma). Coverslips were rinsed three times in PBS, air dried, and embedded in a drop of PBS-buffered 90% glycerol. Coverslips were visualized by epifluorescence microscopy.

TX fractionation and pulse-chase analysis. Cultures of wild-type M. pneumoniae cells were metabolically labeled 15 min with [³⁵S]methionine as described previously (25). Samples were split three ways in fresh Hayflick medium containing 1 mM unlabeled methionine. One sample (0 h) was immediately washed and subjected to TX solubilization. After centrifugation, the insoluble pellet was suspended in buffer to the same volume as the soluble material, such that loading equal volumes onto gels would reflect the physiological ratio between soluble and insoluble material. The other samples were incubated for 1 or 4 h before similar treatment. Samples were heated in SDS-PAGE sample buffer, subjected to SDS-PAGE, and evaluated by autoradiography. M. pneumoniae mutant I-2, which does not adhere to plastic, was grown as described above and transferred to HBSS with [³⁵S]methionine for 30 min, after which cells were centrifuged, washed, recentrifuged, and resuspended in Hayflick broth with unlabeled methionine before proceeding as described above for wild-type cells.

Protein preparation, SDS-PAGE, autoradiography, and immunoblotting. Protein in cytosol samples following osmotic lysis (see above) was precipitated in 9% trichloroacetic acid for 3 min on ice and microcentrifuged for 3 min; the resulting pellets were neutralized by suspension in 1 M Tris. Samples were heated in SDS-PAGE sample buffer for 7 to 15 min at 68°C prior to SDS-PAGE (20). Gels were stained in Coomassie blue, or protein from gels was transferred electrophoretically to nitrocellulose sheets (MSI, Westborough, Mass.) (37) and stained in Ponceau S to visualize protein. Membranes were blocked and blotted with antisera as previously described (19). For autoradiography, gels were fixed for at least 45 min in 8.3% acetic acid-20% methanol and dried before exposure to film.

Sequence analysis. Identification of domains within HMW1 and other proteins were evaluated using GCG software (Wisconsin Package version 10.1; Genetics Computer Group, Madison, Wis.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

HMW1 is a peripheral membrane protein. Wild-type M. pneumoniae cells were separated into membrane and cytosol fractions by osmotic lysis after glycerol loading (30, 31). The pellet collected after high-speed centrifugation of the lysate was subjected to sucrose density gradient centrifugation, yielding a white band with a density of 1.17 g/ml, consistent with previously published values (30). The material constituting this band was demonstrated by immunoblot analysis to contain the transmembrane protein P1 (1, 5, 13) (Fig. 2A) and both cytadherence-accessory proteins HMW1 (18) (Fig. 2C) and HMW2 (18) (Fig. 2D), but not the cytosolic protein EF-G, in significant quantities (Fig. 2B).

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Immunoblot analysis of wild-type M. pneumoniae subcellular fractions. Anti-P1 serum (1:1,000) (A), anti-EF-G serum (1:1,000) (B), anti-HMW1 serum (1:10,000) (C), and anti-HMW2 serum (1:2,000) (D) were used. Lane 1, lysate; lane 2, postlysis supernatant (cytosol); lane 3, major membrane band collected from the sucrose gradient; lane 4, pooled material above the major gradient band; lane 5, pooled material below the major gradient band. Each lane contains 11 µg of protein. The migration positions (in kilodaltons) of molecular mass markers are shown to the left of the gels.

View larger version (20K): [in this window] [in a new window]   FIG. 3.   Immunoblot analysis of susceptibility of M. pneumoniae HMW1 and FtsH to alkali solubilization. Lanes S, supernatant (alkali-soluble peripheral membrane fraction) lanes P, pellet (alkali-insoluble integral membrane fraction). Anti-HMW1 serum (1:10,000) and anti-FtsH serum (1:500) were used.

HMW1 shifts from the TX-soluble fraction to the cytoskeleton. In the M. pneumoniae cytadherence mutant I-2 (18), a frameshift in the hmw2 gene terminates translation of HMW2 prematurely (6). The absence of HMW2 results in decreased steady-state levels of other cytadherence-accessory proteins including HMW1 by what is thought to be proteolytic degradation (25). In preliminary studies with this mutant, we observed a decrease in the amount of TX-soluble metabolically labeled HMW1 with no corresponding increase in the insoluble fraction over time (data not shown). These findings suggested that HMW1 is normally converted from a soluble, unstable form to an insoluble, stable form and that this conversion was largely impaired in mutant I-2. In order to clarify the process by which HMW1 is normally rendered stable in M. pneumoniae, lysates of wild-type and mutant I-2 mycoplasmas pulse-radiolabeled with [³⁵S]methionine were separated into TX-soluble and cytoskeletal components after chase periods of up to 4 h. To offset potential gel-loading variation, thereby optimizing accuracy in quantitating labeled HMW1, we densitometrically compared the levels of HMW1 and a control protein band. This protein, which migrated at ~150 kDa on SDS-polyacrylamide gels, was deemed acceptable for this purpose because of both its reproducible levels in both TX-soluble and -insoluble fractions and its apparent stability over a 4-h period (not shown). For a given sample, the levels of HMW1 are expressed as the ratio of the level of HMW1 to the level of this reference protein, which is unidentified.

View larger version (54K): [in this window] [in a new window]   FIG. 4.   Accumulation of HMW1 in the triton shell in wild-type M. pneumoniae but not in mutant I-2. Each time point is the average of three experiments. Zero-hour totals were independently normalized to 100% for the wild type and mutant 1-2; all data in the graph are in relation to these zero-hour values. For the wild type, the standard deviation (indicated by error bars) for the total amount of HMW1 compared with 0 h is 12.0% for 1 h and 6.2% for 4 h. For mutant 1-2, these values are 17.1% for 1 h and 10.2% for 4 h. Gray bars show the percentages of total HMW1 at 0 h that is present in the triton shell (TX insoluble); white bars show the percentages of total HMW1 at 0 h that is TX soluble. Gray bars plus white bars are the total HMW1 (insoluble plus soluble) compared to the 0-h time point.

HMW1 fails to accumulate stably in the triton shell in the absence of HMW2. In mutant I-2, levels of soluble HMW1 decreased at approximately the same rate as in wild-type M. pneumoniae at 4 h (Fig. 4). However, in contrast to wild-type cells, in mutant I-2 this decrease was not accompanied by a concomitant increase in the cytoskeletal fraction. Instead, a decrease of 30% of this initial amount of HMW1 over 4 h was observed (Fig. 4), indicating that in this mutant, the loss of soluble HMW1 is not due primarily to stable translocation to the triton shell. The 45% decrease in total cellular HMW1 over this period (Fig. 4) suggests that a substantial pool of HMW1 is lost, in agreement with previous observations (25), although the rate of loss was slightly lower here than previously observed. HMW1 was not detected in media in which wild-type or mutant I-2 M. pneumoniae cells were grown (not shown), leading us to conclude that proteolytic degradation of soluble HMW1, and not expulsion into the medium, was responsible for reduced levels of HMW1 in mutant I-2. As in wild-type M. pneumoniae, slightly more than half of the total newly synthesized HMW1 in mutant I-2 was soluble immediately following pulse-labeling (Fig. 4), indicating that the initial association of HMW1 with the triton shell was unaffected in the mutant. The portion of total cellular HMW1 that was TX soluble decreased over a 4-h period to 42%, comparable to that in wild-type mycoplasmas (Fig. 4).

HMW1 is a cell surface protein whose surface exposure is reduced in the absence of HMW2. Immunoelectron microscopy of whole wild-type M. pneumoniae cells previously demonstrated that HMW1 was present on the outer surface at both the attachment organelle and the trailing end of the cell (34). To confirm the surface exposure of HMW1, we performed both immunofluorescence and immunoprecipitation analyses. Antibodies against both P1 and HMW1, but neither preimmune serum nor antibodies to EF-G, labeled wild-type M. pneumoniae microcolonies strongly in immunofluorescence experiments (Fig. 5A to D). Mutant I-2 microcolonies were labeled by anti-HMW1 serum (Fig. 5F) but at a considerably lower intensity than were wild-type M. pneumoniae microcolonies. Similar results were observed with antisera against fusion proteins containing segments of domains I and II (Fig. 1; also data not shown). The anti-HMW1 serum used was reactive in immunoblots against fusion proteins containing segments of each domain (Fig. 1; also data not shown).

View larger version (104K): [in this window] [in a new window]   FIG. 5.   Indirect immunofluorescence of whole wild-type (A to D) and mutant I-2 (E to H) M. pneumoniae colonies. Anti-P1 serum (A and E), anti-HMW1 serum (B and F), anti-EF-G serum (C and G), and rabbit preimmune serum (D and H) were used.

View larger version (30K): [in this window] [in a new window]   FIG. 6.   Immunoprecipitation of M. pneumoniae HMW1 and FtsH from lysates and whole cells. Whole-cell RIP and lysate RIP were performed. For each RIP, immunoprecipitation with anti-FtsH serum (left lane) and with anti-HMW1 serum (right lane) was done. The migration positions (in kilodaltons) of molecular mass markers are shown to the left of the gel.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Adhesion of M. pneumoniae to host cells is a complex process, especially considering the limited number of proteins encoded by this organism. HMW1 and other cytadherence-accessory proteins, while not directly involved in the attachment process, impact the function of the adhesin P1 by affecting its localization to the attachment organelle (1, 9). Understanding the nature of the interactions among these proteins in the assembly and function of the attachment organelle is expected to provide insight into the basic biology and virulence of M. pneumoniae.

Preparation of M. pneumoniae membranes had been performed previously (30, 31), but to our knowledge the fractions had never been examined for markers by immunoblot analysis. The pellet fraction following osmotic lysis lacked EF-G, a cytosol marker protein. Density gradient centrifugation of this fraction resulted in a single visible band with a density of 1.17 g/ml. Significantly, however, the material in this band did not account for the total amount of proteins P1, HMW1, and HMW2 compared with the original postlysis pellet. Accordingly, these membrane-associated proteins were found in gradient fractions both more and less dense than the visible band. Thus, after osmotic lysis, the membrane fraction appears to be present in multiple forms, the predominant form accounting for the major visible band. A reasonable possibility is that whereas the material present in this band contains structures similar to membrane ghosts, consisting of whole or nearly whole cell membranes, the gradient fractions of different densities contain membrane fragments in the form of vesicles or membrane subdomains. Importantly, the attachment organelle, whose protein components constitute the focus of these experiments, is a distinct membranous subdomain. If isolated attachment organelles were found to be enriched in any of these gradient fractions, it might be possible for future workers to exploit this fractionation property in order to obtain pure structures. Nonetheless, for the present experiments the appearance of P1, HMW1, and HMW2 in these other fractions precludes performing this gradient step, and the absence of EF-G from the fraction loaded onto the gradient obviates it.

Although HMW1 is essential for cytadherence (9), a membrane-mediated function, its deduced amino acid sequence provides no indication as to how HMW1 might be either exported or associated with the cell membrane. HMW1 lacks predicted transmembrane domains and secretion signals and is predicted to bear considerable negative charge at physiological pH (4). All domains of HMW1 were surface accessible by either immunofluorescence or immunoprecipitation (Fig. 5 and 6 and data not shown), suggesting that all or nearly all regions of HMW1 are present on the cell surface. These predictions, along with the characterization of HMW1 as a peripheral membrane protein, suggest that HMW1 interacts with the membrane indirectly, probably through other proteins. Since HMW1 is present in both the cytoskeletal and membrane fractions in all cytadherence mutants studied (data not shown), proteins presently known to function in cytadherence are excluded as essential for cytoskeletal or membrane anchoring of HMW1. This is consistent with the predicted early role of HMW1 in attachment organelle assembly (33).

In wild-type M. pneumoniae we observe that newly synthesized HMW1 exhibits net movement from the TX-soluble fraction to the TX-insoluble fraction over a long period of time (Fig. 4). Failure of HMW1 to do so in mutant I-2, together with reduced steady-state levels of HMW1 in this mutant, suggests that association of HMW1 with the cytoskeleton correlates with its stability. We envision two stages of incorporation of HMW1 into the triton shell, reflecting two modes of association with this material (Fig. 7). The first step following synthesis of HMW1 is a transient association with the triton shell, which is reflected in the HMW2-independent appearance of nearly half the newly synthesized HMW1 in the cytoskeletal fraction. Thus, HMW1 is decreased in mutant I-2 over time in both the TX-soluble fraction, where proteolysis likely takes place, and the TX-insoluble fraction, from which HMW1 is presumably returning to the soluble fraction for degradation (discussed below). The second stage is a stable, likely irreversible incorporation of HMW1 into the cytoskeleton, evidenced by both the slow continued accumulation of HMW1 in the insoluble fraction over time and the previous observations that most HMW1 is TX-insoluble at steady state (34). This second step appears to be HMW2 dependent, since a sharp difference is indicated at this stage between wild-type and mutant I-2 M. pneumoniae.

View larger version (13K): [in this window] [in a new window]   FIG. 7.   Model for HMW1 trafficking in wild-type and mutant I-2 M. pneumoniae. In wild-type M. pneumoniae, newly synthesized HMW1 equilibrates between a TX-soluble pool susceptible to degradation and a TX-insoluble pool which is likely membrane associated and awaiting export to the cell surface, where it remains insoluble. Little HMW1 is degraded (dotted arrow) in wild-type cells. In the absence of HMW2 (mutant 1-2), HMW1 is inefficiently exported, as evidenced by the decreased amount detected by immunofluorescence (Fig. 5) or whole-cell RIP (data not shown), so preexport HMW1 accumulates in the soluble pool, where it is degraded, accounting for the low steady-state levels of HMW1 in the mutant.

Thus, we conclude that there are three pools of HMW1: a soluble one, a transiently insoluble one, and a stably insoluble one. We envision that the soluble pool, in which newly synthesized HMW1 appears, contains the small amount of steady-state HMW1 that is not membrane associated (Fig. 2C) and is susceptible to proteolytic turnover if it accumulates. The transiently insoluble pool is membrane associated and awaiting export to the cell surface but is not committed to export and may return to the soluble pool; this pool is in equilibrium with the soluble pool of HMW1. Finally, the pool of HMW1 that has become stabilized as TX insoluble represents HMW1 that has been exported to the cell surface and might also include HMW1 that is in transit to the surface. In an alternative model, HMW1 might be expelled into the medium in mutant I-2 rather than degraded. Were this the case, we might expect to observe in mutant I-2 a transient increase in TX-insoluble HMW1 during the chase period as it passes through the cytoskeleton to the medium; however, no such increase is observed, and in fact the opposite is true (Fig. 4). In consideration of both this and the failure to detect HMW1 in spent medium (data not shown), we favor the proteolysis model.

Although the favored model proposes a means of proteolytic targeting of HMW1 in mutant I-2, the reason for proteolysis remains unexplained. In other bacteria, some proteins that fail to become incorporated into a complex of which they normally are a constituent are susceptible to degradation by housekeeping protease activity: for example, FtsH degrades SecY that is in stoichiometric excess over its binding partners in Escherichia coli (16). It is likewise reasonable to hypothesize that the enhanced degradation of HMW1 in mutant I-2 is also the result of a housekeeping function, reflecting the failure of HMW1 to become incorporated efficiently into a cytadherence-regulating complex. While it is also reasonable to envision a directed turnover of HMW1 that can be induced under some unknown conditions in wild-type M. pneumoniae, both protein production and degradation are energetically costly, making posttranslational regulation of this large protein inefficient.

Whereas stabilization of HMW1 on the cell surface is anticipated to occur through contacts with some elements of the cytoskeleton, both the processes of delivery of HMW1 to the surface and the assembly of the attachment organelle remain intriguing. Lacking recognizable secretion signals, HMW1 must employ a novel means of export. Our model of HMW1 maturation (Fig. 7), which implicates HMW2 in the stable incorporation of HMW1 into the cytoskeleton, suggests that HMW2 may serve as a cytoskeletal organizer that is necessary for efficient coordination of HMW1 export, analogous, for example, to agrin, which associates with components of the vertebrate myocyte cytoskeleton and clusters certain membrane proteins at the neuromuscular junction (reviewed in reference 12). However, the data in the present study do not preclude other roles for HMW2 in M. pneumoniae but suggest nonetheless that the presence of HMW2 at the attachment organelle is a prerequisite for full HMW1 function and by analogy for full function of HMW3 and P65 as well. While this implies that HMW2 must arrive at the nascent attachment organelle prior to these other proteins, further work must be carried out to address this question specifically. Seto et al. (33) did not address HMW2 in their study of order of assembly of attachment organelle components, though they do propose that HMW1 acts early in this process based on localization of cytadherence proteins in M. pneumoniae cytadherence mutants. The observation of a small amount of anti-HMW1-derived fluorescence on the surfaces of mutant I-2 colonies (Fig. 5F) indicates that while HMW2 is not essential for HMW1 export, HMW2 clearly promotes it. Along with parallel studies of P65, P30, and HMW3, further experiments will elucidate the mechanism by which HMW2 promotes HMW1 export, the mechanism of HMW1 proteolysis in mutant I-2, and the significance of these phenomena with regard to M. pneumoniae morphology, P1 clustering, and cytadherence (9, 32).

In the absence of homology with proteins of known function, it has been challenging to designate with any certainty regions of the HMW1 molecule that mediate interactions with other proteins of the triton shell. Nonetheless, HMW1 possesses features identifiable as candidates for such domains. First, along with the cytadherence-accessory proteins HMW2, P65, and P30, domain III of HMW1 contains predicted coiled-coil regions that could be involved in either homo- or heteromerization (Fig. 1). Second, the cytoskeletal proteins HMW1, HMW3, P65, and P200 all contain APR domains, which are enriched in proline and acidic amino acid residues (4, 24, 27, 28). Neither P200 nor HMW3 has thus far been demonstrated to be exposed on the cell surface, in contrast to P65 (27) and HMW1 (Fig. 5 and 6), which are at least partly surface exposed. Therefore, it is plausible that this APR domain, designated domain II of HMW1 (Fig. 1), confers interactions with cytoskeletal components. Significantly, proline-rich domains often mediate protein-protein interactions in eukaryotic cells (reviewed in reference 14). Finally, within domain I of HMW1 is a motif of 31 amino acids (Fig. 1) enriched in aromatic and glycine residues, which we therefore designate the EAGR box. We have identified EAGR boxes in only three M. pneumoniae proteins and in their orthologs in Mycoplasma genitalium, but in proteins of no other organisms. These are HMW1, P200 (in which there are multiple EAGR boxes), and a third protein, designated MP119 in M. pneumoniae and MG200 in M. genitalium (Fig. 8) (2, 7, 11). The last is a novel protein consisting of an N-terminal J domain, which is characteristic of cochaperones (reviewed in reference 15), an EAGR box, and an APR domain, followed by weaker homology to DnaJ toward the C terminus; these features suggest potential involvement in the proper folding of cytoskeletal proteins. The presence of EAGR boxes exclusively in proteins with APR domains suggests that the functions of these two types of sequences are linked. We plan to focus on each of these domains in the future in order to elucidate the function of HMW1.

View larger version (80K): [in this window] [in a new window]   FIG. 8.   CLUSTAL W 1.8 alignment of EAGR boxes (named EAGR for enriched in aromatic and glycine residues) from M. pneumoniae HMW1, P200, and MP119 and their M. genitalium orthologs. Except for MP119 (M. pneumoniae) and MG200 (M. genitalium), M. pneumoniae and M. genitalium sequences are indicated with the letter P or G after the name of the protein, respectively. M. pneumoniae P200 EAGR boxes are numbered from the N terminus to the C terminus; M. genitalium P200 ortholog EAGR boxes are numbered according to their positional counterparts in M. pneumoniae, but this protein lacks an EAGR box corresponding to the third box in M. pneumoniae P200. Consensus sequence symbols: @, aromatic residues; 0, nonpolar residues; !, polar residues; dots, any residue. Numbers surrounding EAGR box sequences indicate the amino acid position within the protein. The M. genitalium homologs of HMW1 and P200 are MG312 and MG386, respectively.