We identified 24 distinct proteins that are phosphorylated in the mycoplasmas, 22 in M. genitalium and 18 in M. pneumoniae, by combining the results of 2D gel electrophoresis analysis of soluble proteins using both the Pro-Q Diamond stain and 33P labeling.
Fluorescence-based analysis of phosphoproteins of mycoplasmas
The primary method used to assay phosphorylation in both M. pneumoniae and M. genitalium was the Pro-Q diamond stain. We used 2D PAGE to separate soluble proteins derived from each of these species at both exponential growth phase and stationary phase, staining the gels using Pro-Q Diamond to detect the phosphorylated fraction then SYPRO Ruby to detect the total protein complement. The computer-generated overlays of the images for M. pneumoniae proteins obtained at stationary phase are shown in Figure 1, where the Pro-Q detected phosphoproteins are shown as blue/white spots and the SYPRO Ruby stained spots are shown in red. The ratio of the Pro-Q Diamond dye signal to the SYPRO Ruby dye signal (here abbreviated as D/S ratio) was used as an approximate measure of the level of phosphorylation with respect to the amount of each protein in a sample . In some cases the strong SYPRO Ruby signal made it difficult to discern the overlaid Pro-Q signal (e.g. spot N26), so panels B and C of Figure 1 show the Pro-Q signal for gels with smaller PI ranges (6–11 and 4–7, respectively) without a SYPRO Ruby overlay.
Figure 2 shows the results for M. genitalium in three separate panels, the top being the total protein stain by SYPRO Ruby, the middle being for the Pro-Q Diamond stained proteins, and the bottom for the 33P labeled proteins.
Combining the results from 2-D gel electrophoresis with IPG gradients of 3–10, 4–7 and 6–11, approximately 230 protein spots (for M. genitalium) and 310 protein spots (for M. pneumoniae) were observed with the SYPRO Ruby stain. There were about 30 protein spots in each strain reproducibly detected by the Pro-Q Diamond stain, with many of these apparently due to multiple isoforms of the single phosphoproteins. A few protein spots stained intensely with Pro-Q Diamond but not with SYPRO Ruby, which may have resulted from proteins with multiple phosphorylation sites, resulting in high-sensitivity detection by Pro-Q Diamond.
Thirty-one of the labeled spots from the M. genitalium gels (Figure 2), and 20 spots from M. pneumoniae (Figure 1), were selected for MALDI-TOF-TOF mass spectrometry (MS) identification. Six spots from the M. pneumoniae gel, that stained as phosphoproteins by Pro-Q Diamond, were directly identified by correspondence to a previously published 2D proteome map, so mass spectrometry was not performed on these spots [6,14]. To confirm the correspondence mapping, we also determined that the identity of the protein was the same in each case as for the corresponding M. genitalium protein we identified by our own mass spectrometry efforts.
A total of 24 distinct proteins were identified between both strains, with an example peptide mass fingerprint-based identification shown in Table 1. There were 22 distinct phosphoproteins identified in M. genitalium (Table 2), with eight proteins presenting multiple spots due to different isoforms present. In M. pneumoniae, 12 independent proteins were identified by MS and in addition to the six identified based on correspondence to identified spots in the previously published gel (Table 3). All but two of the analyzed proteins had one or more isoforms with significant (p M. pneumoniae spots N22 and N26 not producing protein identification scores above the significance threshold, we located phosphorylated homologs with significant scores in the M. genitalium gels. Of the 51 spots analyzed, only spot N16 did not produce any Mascot identification results. Based on its position in a row of spots with the same molecular weight (N17, N18, N19, and N20), it may be an isoform of the hypothetical protein (H10_orf220L) that was identified in these other spots.
The 22 phosphorylated proteins detected in M. genitalium represent about five percent of the 480 M. genitalium ORFs [2,16], whereas the 18 identified in M. pneumoniae represent about three percent of the 688 predicted M. pneumoniae ORFs. However, since not all expressed proteins were observed in the gels, this is likely an underestimate. Also, since the M. pneumoniae strain we used is attenuated and lacks P90 and P40, the phosphorylation patterns might be altered from wild-type. A more realistic number might be obtained from the ratio of phosphoproteins observed to total proteins observed on the gels, which is 9% (22/230) for M. genitalium and 6% (18/310) for M. pneumoniae. For comparison, E. coli has over 130 phosphoproteins, which is about three percent of its annotated ORFs , and Corynebacterium glutamicum also has three percent of its cytoplasmic proteins phosphorylated . Although the mechanisms of and targets for phosphorylation are different between these species, the ratio of proteins phosphorylated is quite consistent, adding evidence that regardless of bacterial species, protein phosphorylation plays an important signaling role in the cell.
Twenty-two of the phosphoproteins we identified had not been previously reported in the mycoplasmas. There were eight phosphoproteins that occurred in only one of the strains (Table 4), six from M. genitalium and two from M. pneumoniae. Since M. pneumoniae contains a strict superset of orthologs to M. genitalium , a case such as MPN295 (an M. pneumoniae specific product) only detected in the former (spots N17-N20) is not surprising . For the other phosphoproteins only detected in one strain, it is possible that the ortholog in the other species is also phosphorylated, but that we did not detect it.
We also examined the MS analyzed peptides matched by Mascot and FindMod, and in all but six of the identified proteins, at least one peptide mass was found that matches a predicted phosphopeptide for the protein (see additional file 1: Putative_Phosphopeptides.pdf for a list of all predictions). Most of these corresponded with low-intensity peaks, as is typically the case for phosphopeptides due to poor ionization efficiency in the presence of phosphoryl groups .
Comparison of Pro-Q fluorescence detection with in vivo radiolabeling
Pro-Q Diamond is a relatively new method for rapid and global protein phosphorylation analysis in gels , which may overcome some of the challenges with 33P labeling or immunodetection with antibodies, such as cross-reactivity in immunodetection and the difficulty of obtaining the right growth conditions for radiolabeling . Because of limited reports of its use in the literature, we used 33P to perform M. genitalium in vivo radiolabeling of phosphoproteins for comparison with and verification of the Pro-Q Diamond stain results. The generation time of M. genitalium is generally 12–16 h, so for the radiolabeling experiments we incubated 24 h for full uptake of 33P by cells. Figure 2B shows the Pro-Q Diamond stained control sample grown under identical conditions, and Figure 2C shows the radioactive phosphoproteins detected by scanning the dried gel after exposure to storage phosphor screen (Amersham) for 3 days. We found approximately 30 protein spots by each of the two methods, and most of the spots are well matched between the two detection methods. However, spots G4, G21, G22, G29, G30 and G31 were seen only on the radiolabelled gels, and spot G3 was stained by Pro-Q Diamond but not observed by radiolabelling. Possible reasons for the differences in detection include: (i) insufficient accessibility to the site of phosphorylation in some proteins for the Pro-Q Diamond dye, for example due to a steric hindrance effect; (ii) the presence of phospho-enzyme intermediates that can only be detected by metabolic radiolabeling, for example when phosphoglucomutase forms phosphoseryl enzyme intermediates during the catalytic cycle  they are detected by radiolabeling but not immunostaining ; (iii) in vivo isotope labeling may alter cell cycle progression and morphology, potentially influencing the modification state of proteins [22,23]. Despite these few differences, the > 80 % agreement between the methods indicates that Pro-Q Diamond dye is accurate for detection of phosphoproteins, and has the advantage of speed and simplicity. Most importantly, the radiolabelling adds supporting evidence for phosphorylation of most of our reported phosphoproteins.
Biological categorization of identified phosphoproteins
We functionally categorized the identified phosphoproteins using Regula et al.  and COGs [24,25] as guides, as shown in Table 4. Phosphoproteins we identified are involved in energy metabolism, carbohydrate transport, carbohydrate metabolism, translation, transcription, heat shock response/chaperoning, cytadherence, and other unknown functions. Interestingly, fourteen of our identified phosphoproteins had previously been associated with the cytoskeleton-like structure in M. pneumoniae . These proteins include pyruvate dehydrogenase E1 alpha and beta chains, elongation factor Tu, heat shock protein GroEL, DnaK, cytadherence accessory proteins, trigger factor, osmotic inducible protein-C-like family and hypothetical protein H10_orf 220L.
Several proteins we identified were previously shown as phosphorylated in other prokaryotic cells, including phosphopyruvate hydratase (enolase), ATP synthase, and RNA polymerase [12,26]. Ribosomal protein S2 was identified as a phosphoprotein in both strains (spots N21 and G30), though it co-separated with a phosphorylated isoform of L-lactate dehydrogenase in M. genitalium, causing concern that the phosphorylation could be on the latter protein. However, in M. pneumoniae, these two spots were resolved separately, and the S2 spot remained stained by Pro-Q Diamond. Though S2 has not previously been reported as phosphorylated, there is some precedent for ribosomal protein phosphorylation, with S1 being phosphorylated in E. coli [27,28]. Also, the NetPhos WWW server[29,30] predicted 10 strong phosphorylation sites on S2, one of which matched a detected MALDI peak corresponding to the putative phosphopeptide "ENLMLSR", where there was a high score for predicted phosphorylation on the serine residue (Ser-153).
Estimated phosphorylation levels in different growth states
As cells enter the stationary growth phase, nutrients become depleted and inter-cellular competition is increased. In E. coli and Thiobacillus ferrooxidans, phosphorylation of the heat shock proteins DnaK and GroEL was shown to be modulated by environmental stresses including temperature, nutrient starvation and pH change . To determine whether there was a similar effect upon phosphorylation in the mycoplasmas, we compared protein abundances for samples obtained during stationary growth phase to those from cells harvested at exponential growth phase by comparing the Pro-Q Diamond (D) to SYPRO Ruby (S) ratio between growth conditions (Table 4, 3rd column: Stat/Exp growth variation). While there was an expected reduction in overall protein synthesis for stationary phase growth , the phosphoproteins DnaK, GroEL (Figure 3, see spot N8, N9), Elongation factor Tu and the pyruvate dehydrogenase complex were up-regulated in both species. In addition, putative structural protein MPN474, heat shock protein DnaK, GroEL, P65 (Figure 3, spots N1, N2, N8, N9, and N4 from M. pneumoniae), elongation factor Tu and pyruvate dehydrogenase all appeared to have increased phosphorylation levels during stationary phase. The phosphorylation level of HMW3 in both M. genitalium and M. pneumoniae (Figure 3, spot N3 from M. pneumoniae) was slightly reduced during stationary growth, as were the restriction modification enzyme EcoD and the DNA-directed RNA polymerase delta subunit. The increase in phosphorylation during stationary phase for mycoplasmas involves some of the same proteins as the response in E. coli (DnaK and GroEL), pointing at a role for these proteins in stress sensing and response.
Protein phosphorylation in mycoplasma cytadherence and cytoskeleton-like proteins
Both M. pneumoniae and M. genitalium have a complex terminal organelle structure that enables them to adhere to a host cell for colonization and nutrient acquisition [33-35]. The terminal organelle is a membrane-bound extension of the cell, characterized by an electron-dense core comprised of rod-like structures oriented laterally within the tip , and thin fibrous structures extending into the cell body. The proteins P1, HMW1, HMW2, HMW3, P90, P40, P30 and P65 were found to be localized to the complex terminal organelle by a variety of techniques such as immunofluorescence microscopy [33,37-39], with P1 playing a major receptor-binding role in M. pneumoniae [9,40,41], and the others also shown necessary for cytadherence [33,36-38]. Of these cytadherence proteins, HMW1 and HMW2 were found to be phosphorylated, and there was weaker evidence that P1 may also be phosphorylated . We found two more cytadherence proteins phosphorylated in both species studied, HMW3 (spots N3, G11, and G12), and P65 (spots G3, N4). Adding these to the previous identifications, at least four out of the eight cytadherence-critical proteins appear to be phosphorylated, indicating an important role for phosphorylation in the process of host-cell attachment.
We did not detect HMW1, HMW2, P30, and P40/P90, corroborating a previous report that these proteins were not found when IEF was used for separation in the first dimension. In previous studies, HMW1  and P40/P90  were found using NEPHGE (Non-Equilibrium pH Gradient Electrophoresis) instead of IEF for the first dimension. And HMW2 (Mr ≈ 216,000), which has the tendency to form multimeric complexes, has only been identified using a 1-D polyacrylamide gel . A faint signal was detected in the SYPRO Ruby stain for P1 indicating its presence in the gel, but it was not detected by Pro-Q Diamond or radiolabeling, giving no evidence for its phosphorylation, unlike the previous report. The recently reported phosphoprotein Hpr, was also not detected. It was previously reported that this protein was phosphorylated in the presence of glycerol , so our culture medium may not have promoted its phosphorylation.
These cytadherence-related proteins, except P30, were also detected in a study of the Triton X-100 insoluble protein fraction from M. pneumoniae . As in eukaryotic cells , it is thought that the Triton X-100 insoluble proteins comprise cytoskeleton-like structural elements, of which the complex terminal organelle is a part. We found a number of the other reported Triton X-100 insoluble mycoplasma proteins to be phosphorylated, including pyruvate dehydrogenase E1, elongation factor Tu, heat shock proteins DnaK and GroEL. Of the 24 phosphorylated proteins identified in our study, fourteen (58%) were reported as Triton X-100 insoluble.
Further supporting these associations with cytadherence and a cytoskeleton-like structure, it was previously demonstrated that elongation factor Tu and pyruvate dehydrogenase E1 beta are expressed on the M. pneumoniae cell surface and involved in binding to eukaryotic cell surfaces . In E. coli, EF-Tu functions as a translation/transcription protein, but it has also been observed to polymerize to form filaments and bundles with actin-like properties , indicating that it is likely to play more than one role in bacteria. DnaK, GroEl, enolase and elongation factor EF-Tu have been reported as phosphorylated in Corynebacterium glutamicum, Thiobacillus ferrooxidans and E. coli, respectively [12,31,46].
We also found a high molecular weight (118 kDa) putative structural protein to be phosphorylated in M. pneumoniae, MPN474 (spots N1, N2 in Fig.1), along with a homologous protein (88 kDa), MG328 (spot G1 in Fig. 2), in M. genitalium. This protein was found in the Triton X-100 insoluble fraction, and has been annotated as "coiled coil putative structural protein involved in cytoskeleton" like HMW2, which is typical of filamentous domains of known cytoskeletal proteins . According to a protein motif scan by "MyHit" , these proteins contain several leucine zipper motifs as is the case in HMW2, providing potential for dimerization interactions. Given all the evidence, it appears that this protein may be important in the formation of a cytoskeleton-like structure.
The combination of previous results and our own indicate that a significant fraction of the mycoplasma cytadherence and cytoskeleton-like proteins are phosphorylated. This may be analogous to eukaryotic cells where phosphorylation and dephosphorylation of cytoskeletal elements has been associated with changes in cell morphology .
In the studies by Seto and Miyata  as well as Kenri et al. , the locations of the proteins correlated with cytadherence in M. pneumoniae were determined. The electron-dense core in the attachment organelle of M. pneumoniae is thought to be the same as a rod-like structure comprised of cytadherence-related proteins (including HMW1, HMW2, and HMW3) that was observed in mycoplasma cells treated with detergent . The wheel-like structure resembles a flagellar motor, and may be connected to the filamentous network extending into the cytoplasm of M. pneumoniae cells. While the composition of the filamentous network  is not definitively known, several of the proteins identified as phosphorylated in this study show evidence of being involved in the formation of filamentous bundles, including EF-Tu, DnaK, and pyruvate dehydrogenase[41,45].
We combined these lines of evidence into Figure 4, representing the prevailing hypotheses and data regarding the cystoskeleton-like structural elements in the mycoplasmas, with the phosphorylation data overlaid. While the figure is speculative due to the fact that the cellular location of the elements represented have not all been definitively pinpointed, there is a strong overall trend apparent in the association of phosphorylated proteins with the cytoskeleton-like structure. We hypothesize that this correlation is because phosphorylation plays an important role in the formation and/or regulation of the structure, and that it may represent a signaling network in the cell. Whether or not the hypothesis proves true, this evidence warrants further study regarding the functional role of phosphorylation in host attachment, stress response, and cell morphology.