P1 alone is not sufficient for mycoplasma attachment to host cells. A number of spontaneously arising mutants have been isolated by screening individual colonies for hemadsorption, a convenient indicator for adherence to respiratory epithelium [16,17]. All but one of these mutants possess P1, and for most of those, P1 fails to localize to the attachment organelle but is widely scattered elsewhere on the mycoplasma surface [18–20]. This observation led to the conclusion that proper trafficking of P1 requires cytadherence-accessory proteins, and more recent analysis has shed additional light on the defect in P1 localization . In wild-type mycoplasmas the P1 precursor was detectable in pulse-chase labeling experiments, and processing to yield mature P1 was essentially complete within 1 h. However, the processing of pre-P1 occurred considerably more slowly in the cytadherence mutants, where the P1 precursor was detectable at substantial levels for several hours post-labeling. It is not clear whether the failure of P1 to localize correctly is due entirely to this defect in processing, or perhaps vice versa. Nevertheless, if one assumes that the final folding of P1 takes place only once it reaches the proper subcellular compartment, perhaps the long leader peptide is necessary to stabilize pre-P1 in an intermediate conformation. The inability to form a fully developed terminal organelle might therefore delay the localization of P1 to the proper subcellular compartment, thereby restricting the final steps in processing and folding.
The number of proteins associated with cytadherence is surprisingly large (Table 1), raising the possibility that most are not directly involved with receptor binding but might be structural elements of the attachment organelle . These cytadherence-associated proteins are largely components of the cytoskeleton, based on their partitioning in the Triton X-100-insoluble fraction [2,22]. Furthermore, loss of these cytadherence-associated proteins is accompanied by dramatic changes in cell morphology affecting the presence and appearance of the terminal organelle [2,12,19].
One such protein is HMW1, which garnered much initial attention largely out of convenience in that it is both highly immunogenic and easily discernible by SDS–PAGE. Analysis by immunoelectron microscopy established a bipolar localization for HMW1 to both the leading and trailing filaments of the mycoplasma cell . More recent examination by immunofluorescence microscopy confirmed a polar location for HMW1 but failed to detect HMW1 along the trailing filaments, perhaps reflecting a difference in the sensitivity of the two techniques . Loss of HMW1 results in a striking change in morphology, with the attachment organelle no longer distinguishable by scanning electron microscopy . Recombinant HMW1 restores a more wild-type morphology, but a slight C-terminal truncation renders this recombinant HMW1 non-functional. Significantly, the C-terminus of HMW1 seems to be required both for full function and for proteolytic targeting in non-cytadhering mutants that lack HMW2 (Table 1 and discussed below [19,21]). HMW1 contains an unusual central domain that is dominated by repeating acidic and proline-rich motifs . This structural feature is shared with cytadherence-associated proteins HMW3 and P65 (Table 1) as well as protein P200 [22,25], and while its significance in the function(s) of these proteins is not clear, all four proteins are components of the detergent-insoluble, cytoskeletal fraction [22,25]. HMW1 localizes to the mycoplasma cell surface, but how it crosses the cell membrane is not known, as HMW1 has neither a leader peptide sequence nor a significantly hydrophobic domain.
Protein P30 is likewise essential to cytadherence, and P30-specific antibodies block attachment to host cells , suggesting a direct role in receptor binding. P30 is a transmembrane protein oriented with its C-terminus to the cell exterior and having what is probably a small intracellular domain . The cell-surface domain of P1 is dominated by proline-rich repeats that are essential for normal P30 function [17,27]. Particularly striking is the cascade of effects associated with loss or truncation of P30, including loss of cytadherence and gliding motility, a dramatic, branched morphology, and a diffuse nucleoid [12,20]. Similar changes in morphology and appearance of the nucleoid were described recently in other M. pneumoniae cytadherence mutants , consistent with a developmental defect . Nevertheless, normal localization of P1 [12,20], HMW1, HMW3, B and C  was observed in the mutant having a truncated P30, the possible significance of which is addressed below.
A striking cascade of effects is likewise associated with a frameshift in the hmw2 gene in cytadherence mutant I-2, where loss of HMW2 is accompanied by reduced levels of HMW1, HMW3, P65, and P30 [16,28,29]. HMW1 and HMW3 are actually synthesized at wild-type levels in this mutant but are subject to accelerated proteolysis in the absence of HMW2 , and recent studies indicate that the same is true for P65 (manuscript in preparation). More recent analysis of HMW1 by pulse-chase studies  clearly demonstrates that newly synthesized HMW1 begins in the cytoplasmic (Triton X-100-soluble) fraction, rapidly shifts to the cytoskeleton fraction, and eventually localizes to the surface of wild-type mycoplasma cells. In the absence of HMW2 the newly synthesized HMW1 is not processed efficiently to the cell surface, leading to accelerated turnover in the cytoplasmic pool.
Synthesis of HMW2 at normal levels from a recombinant wild-type hmw2 allele introduced by transposon delivery fully restores cytadherence and wild-type levels of HMW1, HMW3, and P65 . However, low-level production of HMW2 in some transformants only partially restored a normal phenotype, with the levels of P65 and HMW1 correlating directly with the amount of recombinant HMW2 produced. Thus, proper stoichiometry of these cytadherence-associated proteins appears to be particularly important.
HMW2 is predicted to have a periodicity that is highly characteristic of a coiled-coil conformation, which is typical of filamentous domains of known cytoskeletal proteins . However, the apparent disruption of the coiled-coil motifs in several places suggests a conformation that is more like a flexible chain than a rigid rod. Five leucine zipper motifs are dispersed in the middle of HMW2, providing additional potential for dimerization interactions. The subcellular location of HMW2 in M. pneumoniae has been an enigma for some time. HMW2 is a major cell component and therefore available in substantial quantities (by mycoplasma standards) but is poorly immunogenic. Antibodies produced against fusion proteins containing N- and C-terminal sequences of HMW2 react with denatured HMW2 in Western immunoblots  but not in various subcellular fractions or in thin sections. As an alternative approach to localize HMW2 in M. pneumoniae, we engineered a translational sandwich fusion of HMW2 and green fluorescent protein (GFP). The fusion protein was stable and functional in an hmw2 mutant and was localized by fluorescence microscopy to the attachment organelle (manuscript in preparation). Immunoelectron microscopy studies are under way using antibodies to the GFP domain to localize this fusion protein more precisely in the mycoplasma cell. Based upon that information it should be possible to conjecture how HMW2 may function to effect efficient delivery of HMW1 to the mycoplasma surface. In the meantime, these findings clearly demonstrate that GFP can be a powerful tool even in cells as small as mycoplasmas.
M. pneumoniae gene MP012 (; formerly orf6 or E07_orf1218 in the P1 operon ) encodes a 130-kDa polypeptide that is subsequently processed to yield proteins B and C (also known as P90 and P40, respectively; reviewed in ). Translation of the 130-kDa polypeptide appears to be coupled to the translation of P1, the gene for which immediately precedes orf6. Loss of B and C results in the inability to cytadhere, failure to localize P1 to the terminal organelle, and an atypical appearance to the tip structure . Seto et al., however, report that P1, as well as HMW1, HMW3, and P30, all localize properly in an independently isolated mutant lacking proteins B and C . The reason for the discrepancy regarding P1 localization is not clear but may reflect an undefined difference between the two mutants. Nevertheless, chemical cross-linking analysis has established that B, C, and P1 exist in very close proximity on the mycoplasma surface (reviewed in ), a finding now confirmed and extended to include HMW1 and P65 . Finally, the loss of B and C appears to have no effect on the stability of other cytadherence-associated proteins, and the stability of B and C is unaffected by the loss of P30, HMW1, or HMW2 (Table 1; unpublished data).