Host Cell Responses to Persistent Mycoplasmas - Different Stages in Infection of HeLa Cells with Mycoplasma hominis

Mycoplasma hominis is a facultative human pathogen primarily associated with bacterial vaginosis and pelvic inflammatory disease, but it is also able to spread to other sites, leading to arthritis or, in neonates, meningitis. With a minimal set of 537 annotated genes, M. hominis is the second smallest self-replicating mycoplasma and thus an ideal model organism for studying the effects of an infectious agent on its host more closely. M. hominis adherence, colonisation and invasion of HeLa cells were characterised in a time-course study using scanning electron microscopy, confocal microscopy and microarray-based analysis of the HeLa cell transcriptome. At 4 h post infection, cytoadherence of M. hominis to the HeLa cell surface was accompanied by differential regulation of 723 host genes (>2 fold change in expression). Genes associated with immune responses and signal transduction pathways were mainly affected and components involved in cell-cycle regulation, growth and death were highly upregulated. At 48 h post infection, when mycoplasma invasion started, 1588 host genes were differentially expressed and expression of genes for lysosome-specific proteins associated with bacterial lysis was detected. In a chronically infected HeLa cell line (2 weeks), the proportion of intracellular mycoplasmas reached a maximum of 10% and M. hominis-filled protrusions of the host cell membrane were seen by confocal microscopy, suggesting exocytotic dissemination. Of the 1972 regulated host genes, components of the ECM-receptor interaction pathway and phagosome-related integrins were markedly increased. The immune response was quite different to that at the beginning of infection, with a prominent induction of IL1B gene expression, affecting pathways of MAPK signalling, and genes connected with cytokine-cytokine interactions and apoptosis. These data show for the first time the complex, time-dependent reaction of the host directed at mycoplasmal clearance and the counter measures of this pestering pathogen.


Introduction
Mycoplasma hominis is the second smallest, self-replicating mycoplasma species that colonizes humans. This facultativepathogenic cell wall-less bacterium is found as a commensal in the urogenital tract of sexually active people, but is also associated with bacterial vaginosis, pelvic inflammatory disease, arthritis and even neonatal meningitis [1]. The patho-physiological mechanisms that enable this commensal to become pathogenic are mostly unresolved. In bacterial vaginosis shifts to a higher pH in vaginal flora are often accompanied by higher M. hominis titers. However, whether higher colonisation rates are the consequence or the reason for such changes in the milieu is still unknown.
For the last twenty years we have been interested in the characterisation of pathogenic factors of M. hominis. As attachment to host epithelial cells is thought to be the crucial step in infection, we began our studies with the identification of cytoadhesive membrane proteins, such as the P80 secretin and the lipoproteins P50/Vaa, P60 and OppA [2,3,4], and identified the multifunctional OppA protein as a pathogenic factor of M. hominis. Besides its function as the substrate-binding domain of an oligopeptide importer [5], OppA additionally induces ATP release and induces damage in host cells while its unique ecto-ATPase activity is a prerequisite for OppA-mediated cytoadhesion [6,7]. The capacity of M. hominis to invade cells was firstly described in 1991 by Taylor-Robinson and coworkers, who used HeLa cells as host in an in vitro infection model [8]. Fifteen years later invasion into spermatozoa, leading to abnormal sperm morphology [9], was demonstrated [10]. With the detection of intracellular localisation and replication in another venereal pathogen, Trichomonas vaginalis, a symbiotic association between T. vaginalis (as Trojan horse) and M. hominis was elucidated [11]. This association was suggested to be a benefit for both, influencing the metronidazole susceptibility of the protozoan [12] and defending the invading mycoplasma from immune responses.
Detailed descriptions of the patho-physiological effects of a M. hominis infection on the host at different stages of infection (adhesion -invasion -survival) are still missing. Sequencing of the whole genome of the M. hominis type strain PG21 in 2009 led to the annotation of only 537 protein-encoding genes, of which 220 were predicted to be M. hominis-specific [13]. This reduced protein set suggests that M. hominis is an excellent model organism for studying host-pathogen interactions in detail. To study the cellular effects of a urogenital tract infection by M. hominis more closely, we established an in vitro infection model using the human cervix carcinoma cell line HeLa as host cell and the M. hominis isolate FBG as pathogen.

Microscopic View of M. hominis Attachment to and Invasion in HeLa Cells
Initially, M. hominis adherence to and colonisation of HeLa cells were characterised over time, from 4 h to 2 weeks post infection, using scanning electron microscopy and confocal laser microscopy. As shown in Figure 1A, M. hominis cells attached to the glassadherent HeLa cells preferentially on the convex side of the cell body (4 h) and then dispersed over the surface of the host cell. Colonisation led to a pronounced shortening of filopodia and contraction of the cell, which resulted in disruption of the cell monolayer (24 h). In a chronically infected cell line (i.e. 2 weeks post infection, perm) adherence of the infected HeLa cells to glass was less strong and the proportion of rounded host cells increased (Fig. 1A perm). In addition, empty HeLa shells with a hole in the membrane appeared. Cultivation of a M. hominis-infected HeLa cell line for more than 3 months did not lead to clearance of the pathogen.
Immunofluorescence studies using confocal microscopy confirmed these observations. As depicted in Figure 1B, actin microfilaments of the host, stained with Alexa Fluor 488 phalloidin (green), became condensed over the course of infection. The primarily surface-localised M. hominis cells (4 h post infection) were increasingly found intracellularly after 24 hours and were found predominantly in the cytoplasm of the chronically infected HeLa cells (perm). As shown in Figure 1B, M. hominis cells mainly adhered to the HeLa cell surface (depicted in magenta) and in only a few cases could mycoplasmal invasion be observed at this early stage of infection (4 h) (shown in red and marked by an arrowhead), at which three-fourth of all HeLa cells were colonised by M. hominis. To differentiate between intra-and extracellular mycoplasmal cells, staining was performed with a mycoplasmaspecific antibody (BG11) without permeabilization and fixation of the cells, thus staining the surface-localised mycoplasmas. After fixation with paraformaldehyde and permeabilization of the host cells with Triton-X-100, staining of extra-and intracellular mycoplasma cells was performed. Mycoplasmas localised on the surface were coloured magenta and mycoplasmas residing inside the host cell are shown in red. Loosely bound HeLa cell shells were removed by this procedure. As seen in an overlay (yellow) of actin-(green) and mycoplasma-(red) staining, mycoplasma entry into the host cell seems to occur at and affect the cytoskeleton (Fig. 1B, 24 h). The ratio of intracellular and extracellular mycoplasmas, using the gentamicin assay as described by Winner et al. (2000) [14], indicated that 0.5% of the mycoplasmas appeared to be intracellular 24-48 h after infection. Thus, based on the 30-fold excess of M. hominis to HeLa cells (see Figure S1) and a colonisation rate of the HeLa cells of 95%, one-sixth of the HeLa cells should carry intracellular mycoplasmas at 48 h post infection.
As seen in confocal microscopy and calculated by qPCR, nearly all HeLa cells of the chronically infected HeLa cell line were colonised by a 50-fold excess of mycoplasma cells, 10% of which reside intracellularly as estimated by gentamycin assay.

Differentially Expressed HeLa Cell genes over a Time Course of M. hominis Infection
To elucidate the host-reactions at the different stages of M. hominis infection, four time points post infection were chosen for microarray gene expression analyses: 0 h to monitor the baseline of transcription, 4 h to examine host reactions to mycoplasma attachment, 48 h to capture in addition the initiation of invasion, and 2 weeks post infection to examine a chronically infected host cell. Starting with a 100-fold multiplicity of infection the cell counts of M. hominis and HeLa cells recovered from each time point of infection kept each within the same range (10 5 -10 6 M. hominis genome equivalents/PCR and 10 3 -10 4 HeLa genome equivalents/PCR, see Figure S1). In calculating the ratio of M. hominis to HeLa cells a remarkable decrease of mycoplasma cells compared to HeLa cells was detectable from 4 h to 48 h post infection, which might be the result of the host's effort to eliminate the pathogen. Interestingly, within the chronically infected HeLa cell the ratio of mycoplasma to HeLa cells increased again.
Total RNA was prepared from these infection assays in parallel with uninfected HeLa cells and the host cell transcriptomes were determined by comparative microarray analyses. As depicted in Figure 2, a more than twofold change in the expression of 4283 HeLa cell genes was observed over the course of Mycoplasma hominis infection compared to the uninfected cell line (all genes, regulated above or below baseline, are listed in Table S1). At 4 h post infection, which corresponds to M. hominis attachment and initial colonisation, 67.6% (489/723) of the affected genes were upregulated, whereas at 48 h post infection, when mycoplasma reduction and invasion started, the ratio of up-(859/1588) to down-(729/1588) regulated genes was nearly equal. Almost four times as many genes were exclusively downregulated 48 h after infection and in the chronically infected HeLa cells compared to 4 h post infection, while upregulation of infection-time dependent genes continuously increased from 326 (4 h) to 618 (48 h), to 1101 genes (perm) (see Figure 2, B and C).

The Most Regulated genes at Each Time Point after Infection
In order to obtain a more detailed understanding of the Mycoplasma induced gene expression changes within the host cells we next focused at genes with a more than 10 fold regulation at each time point after infection.
At 4 h post infection 41 genes had a .10 fold difference in expression compared to uninfected cells. Of these 41 genes, 66% (27/41) belonged to one of the three cellular systems: heat shock proteins, the immune system, or cell-cycle regulation. Heat shock protein Hsp70, which plays a role in MAPK signalling and endocytosis pathways, was downregulated in nearly all its subdomain encoding genes (HSPA1A, HSPA1B, HSPA1L and HSPAS2), whereas genes of the immune response, such as the interleukins (IL6, IL11, IL20), their receptors (e.g. IL7R) and several chemokines (CXCL1 and CXCL2), as well as genes involved in cell cycle regulation, cell growth or death (GADD45A, AREG, EREG and metallothioneins), were highly upregulated (Table 1). It should be mentioned that most of the encoded proteins play important roles in more than one system or pathway of the host cell.
Eleven of the 12 genes with the greatest differences in expression that did not belong to one of the above mentioned systems were up-regulated, including ANGPTL4. Expression of this gene is known to be upregulated in infection, inflammation or malignancy [15]. The encoded angiopoietin-like 4 protein affects cell-matrix communication [16], interacts with extracellular matrix (ECM) proteins, inhibits proliferation, migration, and tubule formation of endothelial cells [17], and protects endothelial cells from apoptosis [18].
At 48 h post infection 33 genes were differentially expressed more than 10 fold compared to uninfected HeLa cells. In contrast to the host's reaction at 4 h post infection, the genes differentially regulated at 48 h were associated with a diversity of functions and more than a third (13/33) of these were downregulated. Depending on the multiplicity of infection, a rearrangement of the host's actin cytoskeleton occurred, resulting in co-localisation with invading mycoplasmas (Fig. 3). The fact that angiopoietin is known to limit the formation of actin stress fibres correlates well with our findings that at 48 h post infection transcription of ANGPTL4 further increased and transcription of gamma actin (ACTG1 and ACTG2) declined. Repression of the heat shock response was less prominent, whereas expression of lysosomespecific markers changed, with upregulation of the lysosomeassociated membrane protein LAMP3, more so than LAMP1 and LAMP2, which are delivered to phagosomes during the maturation process. Moreover, expression of the lysosomal cathepsin D (CTSD), which is involved in resolution of inflammation by initiating neutrophil apoptosis [19], increased. Overall, these changes in gene expression correlate well with the reduction of M. hominis cells per host cell and demonstrate the host cells' effort to eliminate the invading pathogen.
The strongest reaction, in terms of gene expression, of the host to M. hominis infection was observed in chronically infected cells (2 weeks). Within this chronically infected HeLa cell line, differential transcription was seen in 56 genes (. 10 fold), with 49 upregulated and 7 downregulated. Of these genes 39% (22/56) were functionally related to the immune system, although the genes involved were quite different to those involved at the beginning of the infection. In these chronically infected cells prominent transcription of IL1B, IL24 and IL3RA, all of which influence pathways of MAPK signalling pathways, cytokine-cytokine interactions and apoptosis, was seen. Similarly induced were genes encoding the S100 calcium binding proteins A8 and A9, which are associated with inflammation, antimicrobial activity and apoptosis (reviewed by Goyette & Geczy [20]), and those of the serum amyloid proteins SAA1 and SAA2, which are thought to affect inflammatory responses, to activate TLR2-and TLR4-dependent signalling and to have an intrinsic bactericidal action [21]. Also highly expressed were membrane-embedded proteins, such as the bradykinin receptors B1 and B2, which are involved in signal transduction, and also lipocalin 2 (LCN2), which is associated with pro-as well as anti-apoptotic processes [22].
In intact tissue the extracellular matrix (ECM), which provides structural support for host cells, is covered by epithelial or endothelial cells and is therefore not accessible for bacterial binding and colonisation. However, tissue damage leads to exposure of the ECM, which may facilitate bacterial colonisation [23]. Thus, it is not surprising that in the chronically infected HeLa cells, transcription of genes involved in ECM-receptor interactions was affected. Expression of SV2B, which has been shown to be involved in the regulation of synaptic vesicle exocytosis [24], was more reduced at 48 h post infection than in the early stages of infection (Table 1), whereas expression of the phagosome-related integrin (ITGB3), which functions as an ECM receptor, and the serpins B3 and B4, which bind heparin, leading to inhibition of the lysosomal cysteine protease cathepsin L, was markedly increased.
To verify the microarray results, transcript levels of specific genes were confirmed by RT-qPCR using RNA purified from the same infection assays used for the microarrays and, additionally, RNA from two separate infection assays conducted months later. As shown in Figure 4, some differences became obvious when using RNA from independent infection assays. In particular some of the most highly regulated genes (e.g. IL6) did not exhibit the same expression profiles. Expression levels at 4 h and 48 h and in the chronically infected HeLa cells (perm) measured by microarray corresponded with those quantified by RT-qPCR when using the same RNA preparation. The differences are more likely to result from slightly different infection conditions, than of inaccuracies in measuring expression levels by microarray, and may be an indicator of variation in the capacity of the host cell to modulate the transcriptional response to the infectious pathogen used.

M. hominis Infection Affected HeLa Cell Pathways
As the quantification of the expression levels of distinct genes does not necessarily provide a complete picture of what is happening in the host cell at different stages of infection, and many regulated factors are known to function at several sites in the cell, sometimes fulfilling different functions in different locations, we looked at which functional systems, networks or pathways the more regulated genes clustered. We mapped the differentially expressed transcripts derived from our microarray analyses onto KEGG pathway maps [25], which represent current knowledge of molecular interactions and reaction network systems for Metabolism (1), Genetic Information Processing (2), Environmental Information Processing (3), Cellular Processes (4) and Organismal Systems (5). We also calculated the ratio of regulated genes per total number of genes in each pathway (Table 2).
At all stages of infection, Genetic Information Processing (such as transcription, translation, folding, sorting and degradation, and replication and repair) was the least affected system of the host (Fig. 5). This correlates well with previous observations of reduced growth in M. hominis-infected HeLa cells [6]. At an early stage of infection (4 h), the regulated genes were found predominantly in pathways belonging to Environmental Information Processing (7% of which were regulated), such as signal transduction, and signalling molecules and interaction. In the second most regulated system, Cellular Processes,   5.7% of the genes were differentially expressed, and in Organismal Systems 4.2% of the genes were regulated (Fig. 5, 4 h). Metabolic pathways were mainly affected at later stages of infection (9% of the genes being differentially regulated at 48 h post infection and 8.6% in the chronically infected cells). With the exception of Genetic Information Processing, more regulated genes were found in pathways in the chronically infected HeLa cell, demonstrating the host's ongoing struggle against the pathogen (Fig. 5, perm).

Identification of Regulated Pathways in Environmental Information Processing
As might be expected as part of the host reaction to attachment of mycoplasmas 4 h post infection, most of the differentially expressed genes (constituting more than 10% of the genes in the pathway) occurred in signal transduction. In the Jak-STAT signalling pathway 12.4% of the genes were differentially expressed at the beginning of infection, including greatly increased expression (.10 fold) of IL6, IL7R, IL11 and IL20. The number of regulated genes At 48 h post infection signal transduction was spread over several signalling pathways. More than 10% of the genes in each of the following signalling pathways were differentially expressed: the Notch signalling pathway (hsa04330), the TGF-beta signalling pathway (hsa4350), the VEGF signalling pathway (hsa04370) and the phosphatidylinositol signalling system (hsa04070). Regulation of genes' expression involved in ECM-receptor interactions was highest in the chronically infected HeLa cells, in which 13.1% of the pathway genes were differentially expressed including SV2B, which was highly downregulated and the integrins which were highly upregulated in number (n = 6) and concentration (ITGB3 = 13 fold).
A high proportion of highly upregulated genes was also found in the cytokine-cytokine receptor interaction pathway (hsa04060), which belongs to the signalling molecules and interaction group of the Environmental Information Processing system, with 27 of the 258 pathway genes (10.5%) upregulated 4 hours post infection, eight of them, encoding IL1B, IL6, IL11, IL7R, IL20, CCL20; CXCL1 and CXCL2, by more than tenfold. Although the proportion of regulated transcripts remained constant in the chronically infected cell line, the pattern of highly expressed cytokines (.10 fold) changed; IL11, IL7R, IL20, CCl20 and CXCL2 decreased, and CCL5, CCL26, IL3RA and IL24, which increased in expression.

Identification of Regulated Pathways in Cellular Processes and Organismal Systems
The total proportion of regulated transcripts in Cellular Processes was 5.7% (95/1672) at 4 h post infection with 19.1% (13/68) of those genes being concentrated in the p53 signalling pathway (hsa04115), demonstrating the importance of this pathway in early host responses to the attached pathogen. In the apoptotic pathway (hsa04210) 10.3% of the genes (9/87) were regulated, with BIRC3 and IL1B more than 10-fold upregulated.
Differentially expressed genes in the Organismal System network clustered in two pathways of the immune system group, those for NOD-like receptor signalling and the complement cascade. In the NOD-like receptor signalling pathway (hsa04621) 10 genes (of a total of 57 pathway genes) were differentially expressed at 4 h post infection and five of them were upregulated more than 10-fold, encoding IL6, IL1B, BIRC3, CXCL1 and CXCL2. The NOD-like receptor signalling pathway seemed to represent an important route for host cell identification of M. hominis as a pathogen and for the generation of an innate immune response. This was supported by the findings that in the chronically infected HeLa cells the proportion of regulated genes in this cluster increased to 21.1% (12/57), with little change in the highly induced genes encoding IL1B, IL6, CCL5 and CXCL1.
The complement and coagulation cascades (hsa04610) were mainly affected in the chronically infected HeLa cells, with 19.1% of the pathway genes differentially expressed compared to the uninfected host cells, with five genes highly upregulated. These encode the complement factors C4BPA, a component of the classical pathway, CFB and CFH, components of the alternative pathway, C3, a component of both complement activation pathways, and BDKRB1, which, in healthy human tissues, is rarely expressed at significant levels [26].

Discussion
To study the host's reactions to infection with M. hominis in an in vitro model, we performed scanning electron microscopy, confocal microscopy and transcriptome analyses of HeLa cells infected with M. hominis strain FBG for 4 h, 48 h and two weeks. We were able to study the host reaction to infection with a bacterium harbouring only 537 annotated proteins (the second smallest known mycoplasmal genome). Infection-stage dependent changes in the host   cell were elucidated by mapping the regulated genes to the KEGG Pathways and developing a M. hominis-specific infection model (Fig. 6). This goal was complicated by the fact that at the different time points post infection, mixtures of HeLa cells were analysed that were either uninfected, or attached by, or attached and invaded by M. hominis. Furthermore the up-and downstream neighbours of a highly regulated signalling pathway component were often not regulated, which may be due to the different halflives of transcripts or, alternatively, unknown regulatory mechanisms. Differential expression of genes was confirmed by RT-qPCR and correlated well when using the same RNA, but differed in some cases when using RNA isolated from different infection assays at different times, possibly demonstrating the variability of individual host-pathogen interactions.
As depicted in Fig. 6, mycoplasmal infection of HeLa cells starts with cytoadhesion which has previously been shown to be mediated by cytoadhesins such as P50/Vaa, P60/P80 and OppA [2]. A prominent induction of pro-and anti-apoptotic cytokines occurs and HSP70 is highly repressed. Besides their action as molecular chaperones, heat shock proteins are known to modulate the innate and adaptive immune responses after exposure to microbial pathogens. Therefore, downregulation of HSP70 has been interpreted as a mechanism of immune evasion and it has been suggested that this promotes chronic infection [27]. This is in accordance with observations that the course of a M. hominis infection is rarely acute, and that the M. hominis induced cytokine profile in human dendritic cells corresponds to that of a pathogen associated with chronic infections [28]. Targosz and coworkers demonstrated in 2006 that Helicobacter pylori also inhibits expression of HSP70 in human epithelial cells [29]. They hypothesised that downregulation of HSP70 might be a primary cause of Helicobacter-associated mucosal cell damage.
The data of Axsen and coworkers confirmed a reduced HSP expression in H. pylori-infected cell lines and in animal models, which was interpreted as a mechanism of immune evasion [30]. In 2009 Kang et al. showed that downregulation of HSP70 protects  rheumatoid arthritis fibroblast-like synoviocytes from NO-induced apoptosis [31]. Interestingly, M. hominis is known to be associated with arthritis [32] with symptoms similar to rheumatoid arthritis [33] known to be the result of an autoantibody response.
Cytoadhesins of M. hominis have been shown to bind to sulphated structures on human cells and ECM molecules [3,34]. This is in good accordance with the upregulation of laminin, thrombospondin and collagen as ligands, and integrins as receptors in the host cell membrane for internalisation (Table  S1). M. hominis neither possesses a type III secretion system like Salmonella or Shigella (which enables uptake according to the trigger model) nor induces e-cadherin nor prominent MET expression, as shown for Listeria invasion (which follows the zipper model) [35]. Nevertheless, confocal microscopy analyses in this study clearly demonstrated that M. hominis invades the HeLa cell and that, depending on the multiplicity of infection, this is associated with rearrangement of the actin cytoskeleton (see Fig. 3.A).
Invasive pathogens have evolved efficient strategies to promote their uptake in non-phagocytic cells such as HeLa cells. To do so they mimic or bind endogenous ligands of host cell receptors, thus activating signalling cascades for uptake. Phagocytosis promoting factors and receptors were found to be upregulated at an early stage of M. hominis infection and in chronically infected HeLa cells (Fig. 6). Interestingly, the S100 proteins A8 and A9, the serpins B3 and B4 and bradykinin receptors B1 and B2 were highly upregulated only within the chronically infected HeLa cells. In endothelial cells the human S100 A8/A9 heterocomplex has been shown to induce expression of adhesion molecules and a weak increase in pro-inflammatory cytokines [36], in accordance with our findings that integrins and a distinct set of cytokines, including IL-1 beta (IL1B), were upregulated. The complexity of the network is evident from the observation that IL-1 beta upregulates the expression of the bradykinin receptors B1 and B2 [37]. In infection with Listeria, immunity to the bacterium has been shown to be via activation of the BDKRB2 receptor [38] and in infections with Staphylococcus aureus or Pseudomonas aeruginosa, secreted components of the pathogens have been demonstrated to upregulate human BDKRB1 in epithelial cells via NFkB-pathways [39,40]. Thus, upregulation of the bradykinin receptors B1 and B2 seems to be an important host defense mechanism against invading microbes including M. hominis.
In Listeria infection, serpins and clathrin have been shown to be mediators of bacterial endocytosis rather than phagocytosis [41]. As clathrin was moderately down-regulated in M. hominis-infected HeLa cells, activation of serpins B3 and B4 seems more likely to be associated with the phagocytotic process. Both serpins have been shown to inhibit the cysteine protease cathepsin L, which is normally highly expressed in lysosomes [42] and, indeed, in M. hominis-infected HeLa cells, cathepsin L2 is downregulated 48 h post infection. Viable Salmonella Typhimurium are found in special phagosomes of macrophages that diverge from the ''normal'' degradative endocytic pathway in lacking cathepsin L2 and neither interact with incoming endocytic components such as transferrin and dextran nor fuse with functional lysosomes [43], thus preventing their elimination.
Once within the host cell, bacteria have evolved different mechanisms for survival. Legionella pneumophila, Brucella abortus and Porphyromonas gingivalis invade endothelial cells and replicate in vacuoles that resemble autophagosomes. Here the late endosomal markers and lysosomal cathepsin L are absent [44]. The late endosomal marker LAMP3 (lysosome-associated marker protein 3) was markedly upregulated in HeLa cells chronically infected with M. hominis. As viable, proliferating Pseudomonas aeruginosa are only found in perinuclear vacuoles that do not contain LAMP3 [45], upregulation of LAMP3 may either be indicative of M. hominis residence in a lytic surrounding or a prerequisite for host lysosomal clearance of the mycoplasma. Interestingly, it could be demonstrated in confocal microscopy that the invaded M. hominis cells colocalise with LAMP3 in the chronically infected HeLa cell line ( Fig. 3. C). Thus, a survival in this vacuole becomes most likely and is supported by the findings that an additional acidification of the lysosome by V-ATPases, which is known to affect a pathogen's clearance, probably does not take place. Transcripts of the lysosomal V-ATPases are nearly all downregulated in the chronically infected HeLa cells (see Fig. 6). In addition, the few upregulated V-ATPase genes reach transcript levels of only up to 1.3 fold higher than in the uninfected HeLa cells. It would be of great interest to examine whether resistance in or escape from the late phago(lyso)some into the cytoplasm is successful and which mycoplasmal effectors are involved. Listeria monocytogenes escapes from the phagosome into the cytosol of the invaded cell, where it replicates and, by the use of an actin-based motility mechanism, spreads from cell to cell [46]. The strategy used by the internalized M. hominis cell to survive within the HeLa cell can be better understood by examining the published data about its symbiotic relationship with Trichomonas vaginalis and our data on the regulated genes of the phagosomal and lysosomal pathways. Vancini and Benchimol published a study in 2008 describing the entry and intracellular localisation of M. hominis in Trichomonas vaginalis [47]. Furthermore, using transmission electron microscopy, they were able to visualize endocytotic uptake involving coated pits. Depending on the Trichomonas strain used, the internalized mycoplasmas were digested, or were able to survive or even to escape from the vacuole into the cytoplasm. One could argue that this endosymbiotic situation is not compatible with a hostpathogen interaction in which an ongoing battle for survival occurs. However, as Brunell et al. reported in 2002, S. Typhimurium replicates in vacuoles of non-phagocytic epithelial cells and phagocytic macrophages [48] even though only the latter affects bacterial clearance. Budding of the trichomonal plasma membrane bearing Mycoplasma hominis [47] looks similar to the budding observed in this study by confocal laser microscopy (Fig. 3  B) and scanning electron microscopy (inset in 3 B) in chronically infected HeLa cells, suggesting that M. hominis could exit HeLa cells in the same way as it leaves Trichomonas.
The present study provides an initial overview of the reactions of the M. hominis-infected HeLa cell at different time points of infection. The findings suggest a complex, infection-stage dependent intervention of the host directed at mycoplasmal clearance and counteracted by the nature of this pestering pathogen. In the future, studies will be required to confirm whether our model of M. hominis infection in HeLa cells is fully representative. The host reactions to intracellular mycoplasmas will need to be characterized separately from the membrane adhesive mycoplasmas (e.g. in eliminating the surface colonising mycoplasmas via gentamycin treatment), the components of M. hominis-containing vacuoles will need to be characterized in more detail, and the proposed mycoplasmal escape into the cytoplasm of HeLa cells and the presence of M. hominis within these budding-like structures will need to be proven. By characterizing the mycoplasmal effectors that are involved, we will deepen our understanding of the survival strategies of this facultative human pathogen.

Cell Culture Conditions
The human cervical carcinoma cell line HeLa S3 (ATCC CCL2.2) was cultivated in DMEM medium supplemented with 10% (v/v) horse serum and 0.01% (w/v) penicillin/streptomycin in a humidified 10% CO 2 -enriched atmosphere at 37uC. Cells were harvested by trypsinisation, suspended in 10 ml buffered saline (PBS) and pelleted by subsequent low speed centrifugation (1,1006g, 10 min). The proportion of damaged cells was ascertained using trypan blue staining. The HeLa cell line chronically infected with M. hominis was cultivated in the same culture medium. M. hominis strain FBG was grown in argininemedium as described previously [2].

Infection Assay
Two days pre-infection 30 ml of a mid-logarithmic-phase broth culture of M. hominis was suspended in 300 ml DMEM-medium supplemented with 10% (v/v) horse serum, 0.01% (w/v) penicillin/streptomycin and 75 ml arginine-medium and cultivated for 48 h at 37uC. After sedimentation of the mycoplasma cells (4,0006g, 30 min) the sediment was washed with PBS, suspended in 300 ml DMEM/10% (v/v) horse serum/0.01% (w/v) penicillin/ streptomycin; and 10 ml M. hominis cell suspension was added to 1610 7 HeLa cells that had been seeded 20 h pre-infection in 75 cm 2 cell culture flasks. For scanning and confocal laser microscopy, HeLa cells were seeded in 24 well tissue culture plates on glass cover slips at 1-2610 5 cells/well and after 5-48 h of cultivation the cells were infected with 1-2610 7 M. hominis cells. The multiplicity of infection (MOI) was controlled by measurement of the number of colour changing units (CCU) of M. hominis, as previously described [49]. In general, the procedure described above resulted in an MOI of 100. Recovery rates of M. hominis and HeLa cells at each time point of infection were estimated by TaqMan based quantification of HeLa (hgap) and M. hominis (hitA) genome equivalents [50], enabling the calculation of cell counts and the respective cell ratios.

Nucleic Acid Preparations and Gene Expression Analyses
Nucleic acids were prepared from each 75 cm 2 cell culture flask after twice washing the adherent HeLa cells with 10 ml PBS and subsequent lysis in 650 ml RLT Buffer (RNeasy Kit; Qiagen GmbH, Hilden, Germany). Genomic DNA of each sample was isolated using 50 ml lysate/RTL Buffer adjusted to 200ml with G2-Buffer of the EZ1 DNA Tissue Kit followed by purification on an EZ1 Biorobot after proteinase K digestion according to the instructions of the manufacturer (Qiagen). 600ml lysate/RTL Buffer was used for total RNA preparation according to the instructions of the manufacturer. All RNA samples that were analysed by RT-PCR were pretreated with DNaseI to digest contaminating traces of DNA, as described previously [5]. RNA integrity was checked using an Agilent 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). All samples in this study had high quality RNA integrity numbers (RIN 9.7-10). RNA was quantified photometrically using a Nanodrop ND1000 (Thermo Fisher Scientific GmbH, Dreieich, Germany).
Synthesis of cDNA and subsequent fluorescent labelling of cRNA was performed on four replicates of each experimental condition according to the manufacturers protocol (One-Color Microarray-Based Gene Expression Analysis / Low Input Quick Amp Labeling; Agilent Technologies). Briefly, 100 ng of total RNA was converted to cDNA, followed by in vitro transcription and incorporation of Cy3-CTP into the nascent cRNA. After fragmentation, labelled cRNA was hybridized to Agilent SurePrint G3 Human GE 8660k Microarrays (Agilent Technologies, Boeblingen, Germany) for 17 h at 65 uC and scanned as described in the manufacturers protocol.
Signal intensities on 20 bit tiff images were calculated using Feature Extraction (FE, Vers. 10.7.1.1; Agilent Technologies). Data analyses were conducted with GeneSpring GX (Vers. 11.5; Agilent Technologies). Probe signal intensities were quantile normalized across all samples to reduce inter-array variability [51]. Input data pre-processing was concluded by baseline transformation to the median of all samples.
After grouping of replicates according to their respective experimental condition a given transcript had to be expressed above background (called ''detected'' by FE) in at least three of four replicates in any one of two or both conditions to be further analysed in pairwise comparisons of conditions (t4 infected vs. uninfected, t48 infected vs. uninfected, perm infected vs. t0 infected). Differential gene expression was assessed statistically using unpaired T-tests. Resulting P values were corrected for multiple testing [52]. Figure 2 shows the distribution of differentially regulated HeLa cell genes (. 2 fold), while more detailed analyses are shown in Table S1.

Pathway Analyses
Primary pathway analysis was performed online using the Kyoto Encyclopedia of Genes and Genomes (KEGG) website (http://www.genome.jp/kegg/pathway.html; vers. 7/21/2011). Entrez GeneIDs (including aliases) of differentially expressed genes were searched against human reference pathways within the KEGG basic pathway mapping tool. HTML output was extracted and integrated into a local database for detailed enrichment analyses of regulated genes.
RNA (1 mg) was converted to random-primed cDNA in a total volume of 40 ml according to the instructions of the manufacturer (Invitrogen, Life Technologies, Darmstadt, Germany), followed by threefold dilution in 10 mM Tris/HCl, pH 7.5. The qPCR assays were then carried out in a total volume of 25 ml consisting of 16MesaGreen MasterMix, 5 mM MgCl 2 , Amperase, 300 nM of each primer and 2.5 ml of the cDNA solution, which was derived from 20 ng RNA. Thermal cycling conditions were as follows: 1 cycle at 50uC for 10 min, 1 cycle at 95uC for 10 min followed by 45 cycles of 95uC for 15 s and 60uC for 1 min for amplification, and 1 cycle at 95uC for 15 s, 1 cycle at 60uC for 1 min. The product was than heated from 65uC to 95uC with an increment of 0.5uC/15 s and the plate read for melt curve analysis to check the identity of the amplicon. Each sample was analysed in duplicate. Cycling, fluorescent data collection and analysis were carried out in an iCycler from BioRad Laboratories (Munich, Germany) according to the manufacturer's instructions.

Western Blot Analysis
Proteins of cell lysates derived from infection assays were separated on 9.5% polyacrylamide gels, transferred to nitrocellulose (Schleicher and Schuell, Dassel, Germany) using a semi-dry blotting procedure and immunostained as formerly published [3].

Confocal Laser Microscopy
Cells adhering to glass cover slips from the infection assays were triply washed with PBS (taking care to prevent detachment of the cells) and blocked with 2% (w/v) BSA/PBS for 20 min. To differentiate between intra-and extracellular mycoplasma cells the cells were firstly stained with primary antibodies for 45-60 min at 37uC that targeted extracellular antigens of M. hominis, washed with PBS, followed by Alexa Fluor-633 conjugated to rabbit antimouse IgG (H+L) secondary antibody staining (1:1000 dilution in 2% (w/v) BSA/PBS; Invitrogen) for 45 min in the dark. Cells were then fixed with 4% (w/v) paraformaldehyde/PBS for 20 min in the dark at room temperature, permeabilized with 0.2% (v/v) Triton-X-100/PBS for 15 min at 37uC, followed by a second antibody, which targeted intra-or extracellular antigens of M. hominis, with similar incubation steps to those used for the first staining, but detected by a Cy3-conjugated secondary antibody (1:1000 dilution in 2% (w/v) BSA/PBS; Dianova, Hamburg, Germany) for 45 min in the dark. For co-localisation studies, cells were fixed and permeabilized as described above and subsequently stained with rabbit mAb MEM-259 (anti-LAMP3; 1:200) detected by an anti-rabbit Cy2-conjugated secondary antibody (1:200) followed by a second staining with murine mAb BG11 (anti-OppA of M. hominis) detected by an anti-mouse Cy3-conjugated secondary antibody (1:1000; Dianova). Staining of the cytoskeleton was effected by 30 min incubation with phalloidin conjugated to Alexa Fluor 488 (1:40 dilution; Invitrogen), staining of membranes with Alexa Fluor 594 Wheat Germ Agglutinin (1:1000 dilution; Life Technologies) and staining of cell nuclei with 49,6-diamidino-2-phenylindole (DAPI). Each incubation step was followed by triple washing with PBS. Coverslips were once rinsed with water, positioned with the cell side down on glass slides in Fluoromount-G (Southern Biotech, Birmingham, Alabama, USA) and stored overnight at 4uC in the dark for fixation.
Images were collected using an LSM 510 Meta or an LSM 780 confocal microscope (Zeiss, Jena, Germany). To avoid cross-talk in the detection of the different fluorophores, the multitracking scanning mode was used. Image analysis and processing was performed with ZEN (Zeiss).

Scanning Electron Microscopy
Scanning electron microscopy (SEM) was performed on sputtered gold coated M. hominis-infected HeLa cell samples to evaluate the effects of M. hominis infection on the cell shape of HeLa cells using a Hitachi S 3000 N SEM and associated software (Hitachi High Technologies America, Pleasanton, CA, USA).