SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The contribution of the human microbiota to health and disease is poorly understood. Propionibacterium acnes is a prominent member of the skin microbiota, but is also associated with acne vulgaris. This bacterium has gained recent attention as a potential opportunistic pathogen at non-skin infection sites due to its association with chronic pathologies and its isolation from diseased prostates. We performed comparative global-transcriptional analyses for P. acnes infection of keratinocytes and prostate cells. P. acnes induced an acute, transient transcriptional inflammatory response in keratinocytes, whereas this response was delayed and sustained in prostate cells. We found that P. acnes invaded prostate epithelial cells, but not keratinocytes, and was detectable intracellularly 7 days post infection. Further characterization of the host cell response to infection revealed that vimentin was a key determinant for P. acnes invasion in prostate cells. siRNA-mediated knock-down of vimentin in prostate cellsattenuated bacterial invasion and the inflammatory response to infection. We conclude that host cell tropism, which may depend on the host protein vimentin, is relevant for P. acnes invasion and in part determines its sustained inflammatory capacity and persistence of infection.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The Gram-positive bacterium Propionibacterium acnes is a ubiquitous member of the skin microbiota and is found in sebaceous follicles located on the face and back of the majority of the human population. In addition to its prevalence on human skin, P. acnes has also been detected at other body sites such as the stomach, large intestine, oral cavity, conjunctiva and prostate (Hentges, 1993; Cohen et al., 2005; Hori et al., 2008; Delgado et al., 2011; Perry and Lambert, 2011).

Propionibacterium acnes is generally regarded as a commensal of the skin. Certain properties suggest a mutualistic role of the bacterium (Cogen et al., 2008); however, the bacterium is known for its association with acne vulgaris, a highly prevalent skin condition with an inflammatory component that affects up to 80% of adolescents (Kurokawa et al., 2009). Despite this association, the exact role and significance of P. acnes in acne vulgaris remains undetermined. Cell culture experiments using skin-derived keratinocytes and sebocytes showed that P. acnes can trigger an inflammatory response, including the production of a range of pro-inflammatory chemokines and cytokines (Graham et al., 2004; Nagy et al., 2006; Lee et al., 2010). Pattern recognition receptors of the toll-like receptor (TLR) protein family have been identified as P. acnes responsive receptors, and the expression of TLR2 and TLR4 is elevated in P. acnes infected keratinocytes (Jugeau et al., 2005). Furthermore, P. acnes infection triggers the production of reactive oxygen species in keratinocytes, which is associated with interleukin (IL)-8 production and host cell apoptosis (Grange et al., 2009a).

Besides its role in inflammatory acne initiation and/or progression, emerging evidence suggests that P. acnes can also act as an opportunistic pathogen in other diseases. It has been reported to be causatively involved in cases of endocarditis, intravascular and central nervous system infections, endophthalmitis, and has been particularly associated with infections following surgical intervention and the implantation of prosthetic devices (Perry and Lambert, 2011). Recently, independent studies reported a link between P. acnes and prostate pathologies. P. acnes was found to be associated with histological inflammation in the prostate, and we, along with others, detected it in cancerous prostates (Cohen et al., 2005; Alexeyev et al., 2007; Fassi Fehri et al., 2011). Investigations with prostate epithelial cells showed that P. acnes induced the secretion of cytokines and chemokines such as IL-6, IL-8 and GM-CSF (Drott et al., 2010; Fassi Fehri et al., 2011). Moreover, infection with P. acnes modulated host cell adhesion and proliferation properties, which resulted in the initiation of cellular transformation (Fassi Fehri et al., 2011). These studies indicate that the pathological outcomes of P. acnes infection might be highly dependent on the induction of tissue-specific host responses.

The consequences of P. acnes’ omnipresence for the human host are poorly understood; circumstantial evidence suggests that the host response to P. acnes depends on host predisposition (e.g. immune status), P. acnes strain-specific properties, and the anatomical site of infection (Dessinioti and Katsambas, 2010; Lomholt and Kilian, 2010; Szabó and Kemény, 2011). Here, we have compared the global transcriptional profiles of skin and prostate epithelial cells during infection with P. acnes strain P6, a clinical prostate isolate. Microarray analyses revealed that the host cell responses to infection fundamentally differed in keratinocyte and prostate cell lines, particularly with regard to the temporal regulation of inflammatory response genes. Subsequent analyses showed host cell-specific modifications of the canonical MAP kinase and NF-κB pathways by P. acnes. Immunofluorescence and electron microscopy demonstrated host cell-specific invasion and intracellular persistence of P. acnes. Bacterial invasion was partially dependent on the prostate expressed host cell protein vimentin. This study reveals, for the first time, that P. acnes host cell tropism has biological significance for the pathogenic properties of this ubiquitous bacterium.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Transcriptional profiles of P. acnes infected HaCaT and RWPE1 cells

In order to compare the transcriptional responses of skin and prostate epithelia to P. acnes infection we infected the keratinocyte-derived cell line, HaCaT (Boukamp et al., 1988) and the prostate-derived cell line, RWPE1 (Bello et al., 1997). First, we performed a comparative analysis of the genome-wide transcriptional profiles of these two cell lines in response to infection with P. acnes strain P6. The transcriptional responses at 24 h post infection (h p.i.) (short-term) and 7 days post infection (d p.i.) (long-term) were determined by microarray (Fig. 1). It was ascertained that cell viability (HaCaT and RWPE1 cell lines) was not adversely affected by P. acnes infection (Fig. S1). Our analysis focused on genes that were not differentially expressed at a basal level (i.e. in the absence of infection) between the two cell lines. A total of 1212 and 1308 genes were differentially expressed between the two cell lines at 24 h p.i and 7 d p.i., respectively, including 307 genes that were differentially expressed at both time points. The data showed an upregulation of inflammation-associated genes such as IL-8, IL-6, LTB and IL-1β at 24 h p.i. in both HaCaT and RWPE1 cells (Fig. S2). However, P. acnes upregulated inflammatory targets more strongly in HaCaT cells compared with RWPE1 at 24 h p.i.; for example, the upregulation of IL-8 was 22-fold stronger in HaCaT cells at 24 h p.i. (Fig. 1A). This was also true for several other chemokines, including CCL1, CCL3, CCL8 and CXCL1. At 7 d p.i. the situation was reversed and P. acnes infection upregulated inflammation-associated genes in RWPE1 cells more strongly compared with HaCaT cells. For example, IL-6 and TLR2 were 9.5- and 6-fold, respectively, more strongly upregulated in RWPE1 cells. Functional network analysis was performed with de-regulated genes in the two cell lines at 24 h p.i and 7 d p.i. (Fig. S3). This again illustrated the upregulation of concerted functions of the inflammatory response in HaCaT cells at 24 h p.i and in RWPE at 7 d p.i. All significantly de-regulated genes were categorized according to their assigned biological functions/diseases (Fig. 1B). Besides the terms ‘inflammatory responses’ and ‘inflammatory disease’ the most enriched functions/diseases were ‘dermatological diseases and conditions’, ‘cancer’ and ‘cellular growth and proliferation’, all of which assigned to de-regulated genes in RWPE1 cells at 7 d p.i.

figure

Figure 1. Gene expression differences in P. acnes-infected HaCaT and RWPE1 cells.

A. Direct comparison of microarray analyses of transcript levels between P. acnes-infected HaCaT and RWPE1 cells at 24 h p.i and 7 d p.i. revealed cell type-specific upregulation of inflammation-associated genes in a time-dependent manner. Green and red indicate stronger expression in HaCaT and RWPE1 cells respectively. Numbers are fold changes. Only genes that were not differentially expressed at a basal level (i.e. in the absence of infection) between the two cell types were considered in this analysis. Grey indicates non-significant (n.s.) expression differences.

B. Functional analysis of host cell transcriptome responses to P. acnes infection. Biological functions and diseases were assigned to de-regulated genes upon infection of HaCaT and RWPE1 cells. P-values indicate significance of enrichment of biological functions/diseases in the respective transcriptomes. At 24 h p.i. the transcriptional response in HaCaT cells dominates over that of RWPE1 cells in most pathways. However, at 7 d p.i. the transcriptional response in RWPE1 cells is largely dominant. The microarray data are based on eight and four experiments for the 24 h and 7 days time points respectively. In order to compensate specific effects of the dyes and to ensure statistically relevant data analysis, a colour-swap dye reversal was performed.

Download figure to PowerPoint

Thus, HaCaT cells strongly upregulated inflammatory response genes during acute infection with P. acnes; however, this response did not persist considerably beyond 24 h p.i. In contrast, the transcriptional response of RWPE1 cells to acute infection was limited, but became more pronounced up to 7 d p.i.

P. acnes-dependent activation of the MAPK pathway differs in RWPE1 and HaCaT cells

The observed differences in the acute (24 h p.i.) transcriptional response to P. acnes infection between the two cell lines prompted us to investigate the initial cellular signalling events during infection. MAP kinase (MAPK) cascades play crucial roles in the response to external stimuli, and mediate a first line reaction towards pathogens (Pearson et al., 2001). There are three major MAPK signalling pathways that employ the ERK, p38 and JNK kinases; the activation of each of these kinases, and hence pathway specificity, can be determined by protein phosphorylation at specific sites. P. acnes infection of HaCaT cells led to phosphorylation of ERK1/2, p38 and JNK (Fig. 2A and B) in a time-dependent manner as determined by immunoblot. Increased phosphorylation was observed at 30 min p.i. (p38) and 4 h p.i. (ERK1/2 and JNK). Phosphorylation of these three MAP kinases decreased at 8 h p.i., but was strongest at 24 h p.i., indicating an oscillatory activation pattern of MAPK cascades in HaCaT cells in response to P. acnes infection. In contrast, basal expression levels of phospho-ERK1/2 and phospho-JNK were higher in RWPE1 cells in the absence of infection and only phospho-p38 increased in RWPE1 cells 24 h p.i. (Fig. 2C and D). Moreover, ERK was slightly deactivated from 8 h p.i., as judged by decreasing phosphorylation levels of ERK1/2. Similar to observations in HaCaT cells, p38 activation oscillated: the first peak of p38 phosphorylation occurred approximately 2 h p.i., and strongest phosphorylation was observed at 24 h p.i.

figure

Figure 2. Host cell-dependent activation of MAPK pathways by P. acnes.

A and B. In P. acnes infected HaCaT cells, three MAPK pathways are activated in a time-dependent manner, determined by phosphorylation of ERK1/2, JNK and p38.

C and D. In P. acnes infected RWPE1 cells, only p38 is activated at 24 h p.i.

Quantification of Western blots was based on the band intensity normalized to β-actin (B, D). Representative blots of three independent experiments are shown. NI, non-infected.

Download figure to PowerPoint

P. acnes activates the NF-κB pathway in RWPE1, but not in HaCaT or HEKa cells

The NF-κB signalling pathway is another important mediator of the inflammatory response and is often activated during innate immune recognition of bacterial pathogens. NF-κB activation requires rapid phosphorylation, ubiquitination and subsequent degradation of IκBα, the inhibitor protein of NF-κB (Natoli and Chiocca, 2008). To monitor NF-κB signalling in infected host cells, IκBα degradation was determined by immunoblot analysis. No notable IκBα degradation was observed in P. acnes infected HaCaT or RWPE1 cells during the first 24 h of infection (Fig. 3A and B). However, degradation of IκBα could be detected in RWPE1, but not in HaCaT cells, at later infection time points (48 h to 7 d p.i.) (Fig. 3C and D), which was indicative of cell type-dependent activation of NF-κB. Since HaCaT cells are reported to have an aberrant NF-κB signalling pathway (Lewis et al., 2006), we repeated the experiments with primary human epidermal keratinocytes (HEKa cells). As in HaCaT, no IκBα degradation could be observed in HEKa cells upon P. acnes infection (data not shown).

figure

Figure 3. P. acnes triggers NF-κB activation in RWPE1 but not in HaCaT cells.

A–D. Western blot analysis reveals no degradation of IκBα in HaCaT of RWPE1 cells during acute infection (≤ 24 h). During long-term infection (48 h to 7 days) significant degradation could be observed in RWPE1, but not in HaCaT cells. β-Actin was used as loading control.

E and F. Immunofluorescence analysis of p65 translocation in infected RWPE1 and HaCaT cells at 48 h p.i. Nuclear translocation of p65 could only be detected in infected RWPE1 cells. p65, green; P. acnes, red; cellular nuclei, blue. NI, non-infected; inf, infected. Scale bar, 20 μm. Representative blots and images of three independent experiments are shown.

Download figure to PowerPoint

IκBα forms a dimer with the transcriptional regulator p65; once IκBα is degraded upon stimuli, the dimer disaggregates and p65 translocates into the nucleus to control DNA transcription of NF-κB target genes (Hoffmann and Baltimore, 2006). Thus, p65 nuclear translocation is a sign of NF-κB activation. By immunofluorescence analysis we observed p65 nuclear translocation in P. acnes-infected RWPE1 cells at 48 h p.i., but not in HaCaT cells (Fig. 3E and 3F). These data indicated that NF-κB was exclusively activated in RWPE1 cells following long-term infection with P. acnes.

Intracellular localization and persistence of P. acnes in RWPE1 cells, but not in HaCaT or HEKa cells

Based on the results obtained, we suspected that the infection process was fundamentally different in the two cell lines, which could reflect cell type-specific invasion and/or adhesion properties of P. acnes. P. acnes has previously been shown to invade RWPE1 cells (Fassi Fehri et al., 2011), but investigations were not extended to keratinocytes. Thus, we compared the invasiveness of P. acnes in both cell lines, using a streptomycin/penicillin protection assay. Invasion was 6- and 10-fold higher in RWPE1 compared with HEKa and HaCaT cells, respectively, as determined from cfu counts at 24h p.i. (Fig. 4A). To confirm the invasion phenotypes, we performed extra-/intracellular double staining of P. acnes. Immunofluorescence microscopy (after stringent washing to remove unattached bacteria) revealed higher numbers of intracellular, as well as extracellularly attached, P. acnes bacteria in RWPE1 cells at 24 h p.i. compared with HaCaT cells (Fig. 4B–E). This indicated cell type-specific adhesion and invasion properties of P. acnes. Furthermore, P. acnes bacteria could be detected intracellularly up to 3 weeks p.i. in RWPE1 cells (Fig. 5A). Heat-inactivated bacteria were not detected 3 weeks p.i., which demonstrated that bacterial viability was required for intracellular persistence. Notably, P. acnes could not be detected in HaCaT cells 3 weeks p.i. (data not shown). Intracellular P. acnes could also be detected by electron microscopy at 24 h p.i. in RWPE1 cells (Fig. 5B–E). Thus, P. acnes intracellular invasion and persistence is cell type-specific; this host cell tropism could explain the observed differences in sustained NF-κB activation and transcriptional de-regulation.

figure

Figure 4. Host cell type-specific attachment and invasion of P. acnes.

A. A streptomycin/penicillin protection assay revealed a 5- to 10-fold difference of viable intracellular P. acnes in RWPE1 and HaCaT/HEKa cells at 24 h p.i. **P ≤ 0.01.

B–E. Confocal immunofluorescence microscopy revealed cell type-specific differences in attachment and invasion of P. acnes.

B and D. No extracellular and very few intracellular P. acnes (red) could be detected in infected HaCaT cells after stringent washing.

C and E. More extra- (green) as well as intracellular (red) bacteria are found in infected RWPE1 cells. Scale bar, 20 μm.

D and E. Higher magnification of P. acnes infected HaCaT and RWPE1 cells. Scale bar, 8 μm. Actin stained in blue (blue).

Download figure to PowerPoint

figure

Figure 5. Persistence of intracellular P. acnes in RWPE1 cells.

A. Viable, intracellular P. acnes (green), but not heat-inactivated bacteria, were detected in RWPE1 cells after infection for 21 days and stringent washing. Nuclei, blue; actin, red. Scale bar, 20 μm.

B–E. Electron microscopy images of infected RWPE1 show P. acnes in intracellular vacuoles (double arrowhead) as single bacterium or in small clusters. P. acnes localizes close to the nuclear (n) envelope (single arrowhead). Scale bars, 2 μm and 8 μm. Representative images from three independent experiments are shown.

Download figure to PowerPoint

Functional analysis of P. acnes host cell tropism

To investigate the mechanistic basis for the observed host cell tropism of P. acnes we identified endogenous differences in the transcriptional profiles of RWPE1 and HaCaT cells. Employing a stringent fold change cut-off, we identified a number of genes that were highly differentially expressed between the two cell lines. We focused on genes that were upregulated at least 25-fold in RWPE1 cells, i.e. RPS4Y1, VIM, IGFBP2, LOC150763, IL1B and FN1 (Table S1). Since previous studies had demonstrated a role for vimentin (VIM) in host–pathogen recognition and/or interactions (Kumar and Valdivia, 2008; Miller and Hertel, 2009; Das et al., 2011; Ghosh et al., 2011), we focused further functional analyses on this protein.

Vimentin facilitates P. acnes host cell invasion

Vimentin is a type III intermediate filament protein and has various functional roles; for example, it is involved in regulating cell shape and integrity, adhesion, migration and signalling events (Satelli and Li, 2011). We initially confirmed the microarray data with the observation that vimentin was expressed in RWPE1 cells, but was not detectable in HaCaT cells by immunoblot (Fig. 6A). To investigate its function in the context of P. acnes infection we ectopically overexpressed vimentin in HaCaT cells (Fig. 6B). In parallel, vimentin knock-down (siVIM) RWPE1 cells were generated using siRNA (Fig. 6C). Employing the streptomycin/penicillin protection assay, we then compared numbers of intracellular P. acnes in vimentin-overexpressing HaCaT cells and siVIM RWPE1 cells with those in control cells (HaCaT carrying a GFP-containing vector and RWPE1 transfected with AllStars siRNA respectively) at 24 h p.i. Overexpression of vimentin in HaCaT cells led to a 1.7-fold increase in intracellular bacteria (Fig. 6D). Conversely, silencing of vimentin expression in RWPE1 cells led to a 2.3-fold decrease in intracellular P. acnes (Fig. 6E). Both experiments showed that manipulation of vimentin expression altered the number of intracellular bacteria at 24 h p.i. The data revealed that P. acnes invasion depends on the expression of vimentin. To further address the role of vimentin in the uptake of P. acnes, we used an anti-VIM antibody to block membrane-exposed vimentin. Invasion of P. acnes into anti-VIM-treated cells was approximately twofold decreased at 24 h p.i. compared with anti-IgG-treated cells (Fig. S4A), indicating that the invasion process of P. acnes is partially vimentin-dependent. Since vimentin is an intermediate filament protein with supposed structural roles, for example in supporting the cellular membrane and the cytoskeleton integrity (Eriksson et al., 2009), we investigated if disturbance of the cytoskeleton network would influence P. acnes invasion. Indeed, P. acnes invasion was significantly reduced at 24 h p.i. in RWPE1 cells treated with chemical inhibitors of actin and microtubule polymerization, e.g. up to 10-fold in the case of cytochalasin B treatment (Fig. S4B). This implicated vimentin and the cytoskeleton network in the adhesion and cellular invasion of P. acnes.

figure

Figure 6. Role of vimentin in host cell invasion of P. acnes.

A. Western blot analysis revealed that vimentin (VIM) is expressed in RWPE1 but not in HaCaT cells.

B. HaCaT cells transformed with a VIM-containing expression vector express vimentin. Cells transformed with a GFP-containing plasmid were used as a control.

C. Knock-down of VIM in RWPE1 by RNAi successfully reduced vimentin protein levels. AllStars siRNA- transfected cells were used as a control.

D. Overexpression of VIM in HaCaT cells increased P. acnes invasion.

E. Knock-down of VIM in RWPE1 cells reduced P. acnes invasion.

β-Actin was used as loading control. Representative blots and results of at least three independent experiments are shown. **P ≤ 0.01.

Download figure to PowerPoint

Vimentin-dependent bacterial host cell entry and its role in the inflammatory response of RWPE1 cells to P. acnes infection

Given the finding that vimentin is involved in P. acnes invasion, further investigations into the effects of vimentin expression manipulation were performed to elucidate whether this would influence the inflammatory response of RWPE1 cells to P. acnes. Therefore, we compared gene expression profiles of P. acnes-infected RWPE1 cells transfected with either vimentin-specific siRNA or AllStars control siRNA at 24 h p.i. We focused on genes that were not de-regulated in non-infected siVIM RWPE1 (compared with control cells). A total of 1329 genes were differentially expressed between infected siVIM RWPE1 and infected AllStars transfected control cells, indicating a strong influence of vimentin on the host cell response to P. acnes. Inflammation-associated genes were a prominent group of downregulated genes (Fig. S5A and B); for example, expression of IL-8 was reduced more than 40-fold in infected siVIM RWPE1 cells compared with infected control cells. These data show that silencing of vimentin expression in RWPE1 cells – thus partially preventing P. acnes internalization – reduced the inflammatory response. Taken together, we have shown that vimentin plays an important role in establishing intracellular P. acnes infection, by facilitating or supporting bacterial invasion in a host cell-dependent manner. The data further indicate that the capability of the microorganism to invade intracellularly has direct consequences on host cell-derived inflammation.

Vimentin is expressed in prostate tissue

Vimentin facilitated the invasion of P. acnes into prostate cells in vitro. It has been previously reported that vimentin is focally expressed in prostate gland epithelium (Wernert et al., 1987; Heatley et al., 1995). In addition, vimentin was found to be abundantly expressed in poorly differentiated prostate cancer samples (Wei et al., 2008; Zhao et al., 2008); the latter studies suggested that vimentin affects prostate cancer cell motility and invasiveness. We immunostained clinical skin and prostate tissue samples to analyse tissue-specific vimentin expression patterns. Vimentin was absent from skin epithelium (Fig. 7A), but was focally expressed in prostate epithelium (Fig. 7B and C). Interestingly, a distinct subset of cells that lined glandular structures were vimentin-positive (Fig. 7C).

figure

Figure 7. Vimentin expression in skin and prostate tissues.

A. Skin tissue sample showing some scattered VIM-positive stromal cells; epidermal cells (outer layer) are VIM-negative.

B. Prostate tissue sample with VIM-positive cells, representing blood vessel-lining endothelial cells and stromal cells.

C. Benign prostatic hyperplasia tissue sample with strong VIM expression in cells in the glandular epithelium.

Green, VIM; blue, nuclei.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

We have shown that a P. acnes clinical prostate isolate exhibits profound host cell tropism. Alongside the role of P. acnes strain determinants in predisposition to, and pathogenesis of, P. acnes infection (Holland et al., 2010; Lomholt and Kilian, 2010; Brzuszkiewicz et al., 2011; McDowell et al., 2011), host factors are also important for the pathological outcome of infection (Szabó and Kemény, 2011). Moreover, the likelihood exists that the pathogenic potential of P. acnes depends on the anatomical site of infection. Experimental comparison of host cell-specific responses to P. acnes infection had not previously been explored; thus, we compared infections of keratinocyte (HaCaT) and prostate (RWPE1) derived cell lines using a P. acnes clinical isolate. We observed clear differences in both the magnitude of the transcriptional responses and their temporal regulation between the two cell lines. In essence, P. acnes-induced inflammation in keratinocytes is acute but transient, whereas the bacterium triggers delayed but sustained inflammation in prostate cells, which is associated with prolonged NF-κB activation. Other inflammatory stimuli such as lipopolysaccharide (LPS) do not elicit such differential effects; LPS triggers an acute NF-κB-dependent response in both HaCaT and RWPE1 cells (Seo et al., 2001; Grange et al., 2009b; Kim et al., 2011), indicating that the here described differential inflammatory responses are P. acnes-specific.

The capacity of P. acnes to induce a sustained de-regulation of the transcriptional response in RWPE1-infected cells was associated with invasion. To date, the invasion capacity of P. acnes has not been well studied. We previously showed that P. acnes isolates could be visualized intracellularly at 24 h p.i. in prostate epithelial cells (Fassi Fehri et al., 2011). In a study attempting to find an aetiological link between sarcoidosis and P. acnes infection, invasion of HEK293T cells by P. acnes was correlated with bacterial serotype and genotype, although no association with disease could be made (Furukawa et al., 2009). In another sarcoidosis study, Tanabe and co-workers observed the invasion of A549 cells by clinical P. acnes isolates (Tanabe et al., 2006). Here, we present evidence that host cell entry marks the beginning of a persistent stage of infection, characterized by continuous NF-κB activation that implicates sustained upregulation of inflammatory markers. The data add to our previous findings that long-term infection of RWPE1 cells by P. acnes altered host cell fate; in particular, increasing cell proliferation and reducing E-cadherin expression and anchorage-independent growth (Fassi Fehri et al., 2011), which are markers of the epithelial-to-mesenchymal transition (EMT) (Kalluri and Weinberg, 2009).

Here, vimentin was expressed in RWPE1 but not in HaCaT cells. Moreover, we confirmed that expression of vimentin was absent from human keratinocyte tissue, but present in the prostate. Our data suggest that vimentin expression is partially responsible for the observed differences in P. acnes invasion between the two cell lines. Vimentin has been described as a cytoskeletal component responsible for maintaining cell integrity in mesenchymal cells (Eriksson et al., 2009). It is expressed in a wide range of cell types including pancreatic precursor cells, neuronal precursor cells, fibroblasts, endothelial cells, renal tubular cells, macrophages, neutrophils, leucocytes and renal stromal cells (Satelli and Li, 2011). In addition, vimentin has been recognized as an EMT marker (Thiery, 2002) and is overexpressed in various epithelial cancers, including those of the prostate and gastrointestinal tract (Wei et al., 2008; Zhao et al., 2008). Our observation that vimentin was not produced in keratinocytes, which are of ectodermal origin, supports the general consensus (Katagata et al., 1999). RWPE1 cells are secretory epithelial cells and are derived from the endoderm; although lineage dependence has not been unequivocally determined, an endothelial origin has been excluded (Bello et al., 1997). RWPE1 cells are frequently used as a prostate epithelial cell model since they organize into acini in 3D matrigel and secrete PSA into the lumen when exposed to androgen (Bello-Deocampo et al., 2001). A recent study revealed that mesenchymal and epithelial cadherins were coexpressed in RWPE1 cells (Härmä et al., 2010); the authors speculated that these cells (or subpopulations within RWPE1) might have undergone a (partial) EMT. Interestingly, all cell lines, which were reported to be invaded by P. acnes, including HEK293T and A549 cells, are vimentin-positive (Lahat et al., 2010; Bhattacharyya and Hope, 2011), supporting our findings that vimentin is required for efficient bacterial invasion.

Investigations into the role of vimentin in microbial infections have shown that infectious agents can modulate vimentin to establish an intracellular niche and for optimal positioning of bacteria-containing vacuoles. Such a mechanism has been studied for African swine fever virus infection, which initiated vimentin rearrangement to form a cage surrounding viral factories (Stefanovic et al., 2005). Similar observations have been made with Chlamydia trachomatis, which remodelled and recruited cytoskeletal components of the host cell, including vimentin, to form a dynamic scaffold that provided structural stability to the inclusion (Kumar and Valdivia, 2008) and Salmonella typhimurium, which induced the formation of aggresome-like structures and dramatically remodelled vimentin and cytokeratin networks in epithelial cells and macrophages (Guignot and Servin, 2008). Interestingly, that study indicated that vimentin remodelling was required to maintain Salmonella microcolonies in the juxtanuclear area, a prerequisite for the initiation of Salmonella replication. We also observed that P. acnes bacteria preferentially localized close to the nucleus, but the significance of this localization for the intracellular fate of P. acnes requires further exploration. Studies have also shown that pathogens can exploit vimentin for the purpose of invasion; it has been shown that vimentin is found on the extracellular surface of various cells, and some bacteria interact specifically with surface-exposed vimentin. For example, meningitic Escherichia coli K1 strains bound to surface-exposed vimentin as primary receptor to gain entry into human brain microvascular endothelial cells (Chi et al., 2010). Our data, in particular the effect of antibody-mediated neutralization of surface-bound vimentin on P. acnes’ invasion, also point towards a role for vimentin as a cell surface-associated mediator of bacterial invasion. The mechanistic basis of this interaction is an interesting question for future investigation.

Despite its ubiquitous presence on the human skin, our understanding of the role of P. acnes in health and disease, at both skin and non-skin sites, remains fragmentary. Although an aetiological role for P. acnes in benign prostatic hyperplasia, prostate cancer or other inflammatory conditions is not established, mounting evidence suggests that this bacterium has strong inflammatory capacity, which could initiate or contribute to disease. The pronounced pro-inflammatory potential of P. acnes is also the key argument for its driving role in inflammatory acne. Our data contribute to the ongoing debate concerning the putative role of P. acnes in prostate cancer. We have demonstrated that P. acnes can induce a sustained inflammatory response in prostate cells permissive to bacterial invasion in a host cell-specific manner. Given that vimentin is expressed in the human prostate, our data support the hypothesis that P. acnes acts as an opportunistic invader of prostate cells, and could contribute to prostate cancer aetiology owing to its inflammatory profile. In summary, we propose that host cell tropism is an important facet of P. acnes-associated pathologies. It is very likely that host cell tropism is affected by bacterial heterogeneity, i.e. phylogenetically distinct P. acnes isolates might differ substantially in eliciting host cell-specific responses. The exact interdependence of bacteriaI and host cell entities and its relevance for health and disease has to be investigated in the future.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Bacteria, cell cultures and reagents

Propionibacterium acnes strain P6 (type I-2, ST33, according to the MLST scheme of Lomholt and Kilian, 2010), an isolate from a cancerous prostate (Fassi Fehri et al., 2011), was cultured on Brucella agar plates for 3 days at 37°C under anaerobic conditions. For heat inactivation P. acnes was incubated at 60°C for 30 min.

The human keratinocyte cell line HaCaT (CLS) was cultured in DMEM medium (Gibco) supplemented with 10% heat-inactivated FCS. HEKa (Invitrogen) cells are primary human epidermal keratinocytes isolated from adult skin; they were cultured in EpiLife® Medium supplemented with human keratinocyte growth supplement (Gibco). The human prostate epithelial cell line RWPE1 (ATCC CRL-11609) was cultured in keratinocyte medium supplemented with 50 μg ml−1 bovine pituitary extract and 5 ng ml−1 epidermal growth factor (Gibco).

Antibodies and inhibitors

Polyclonal anti-P. acnes antibody (Fassi Fehri et al., 2011) was diluted 1:1000 for immunofluorescence analysis. Polyclonal rabbit antibodies against phospho-JNK, phospho-p38, IκBα (Cell Signaling), against vimentin (H84 and V9) (Santa Cruz), and a polyclonal mouse antibody against phospho-ERK1/2 (Sigma-Aldrich) were diluted 1:1000 for Western blotting. The monoclonal anti-β-actin antibody (Sigma-Aldrich) was used at a 1:5000 dilution. Polyclonal rabbit antibody against p65 (Santa Cruz) and monoclonal mouse antibody against vimentin (Zymed) were diluted 1:50 and 1:400, respectively, for immunofluorescence. For neutralization of vimentin, host cells were treated with anti-vimentin H84 antibody (Santa Cruz) at a 1:10 dilution immediately prior to P. acnes infection. The following chemical inhibitors (dissolved in DMSO) were used: cytochalasin B (Serva) and cytochalasin D (Fluka) at 1 μg ml−1; colcemide (Roche) at 10 ng ml−1 and nocodazole (Sigma) at 100 ng/μl. All inhibitors were added to RWPE1 cells immediately prior to bacterial challenge.

Infection of epithelial cells

HaCaT, HEKa and RWPE1 cells were seeded into 12- or 24-well plates. P. acnes, cultured for 3 days on Brucella plates, was harvested, washed in the respective cell culture medium and diluted. Viability of bacteria was assessed by colony-forming unit (cfu) counts; it was ascertained by cfu counts that P. acnes was viable in all used cell culture media over the indicated infection time periods. Cells were infected at a multiplicity of infection (moi) of 50 in a humidified 5% CO2 atmosphere at 37°C. Infections were stopped 30 min, 1 h, 2 h, 4 h, 8 h, 24 h or 48 h after initial infection. For long-term infections (1–3 weeks), media were changed every third day and cells were split once a week.

Host cell viability

A WST-1 assay (Rapid Cell Proliferation Kit, Merck) was used to determine host cell viability in the absence and presence of P. acnes. The assay measured the increased activity of cellular mitochondrial dehydrogenases that can cleave the tetrazolium dye WST-1 to formazan. Formazan formation was quantified by measuring the change in absorbance at 450 nm in a microplate reader (procedure as described by the manufacturer).

Microarray and transcriptome analysis

Microarray experiments were performed as dual-colour hybridizations. To compensate for dye-specific effects, a dye-reversal colour-swap was applied (Churchill, 2002). Samples were isolated with TRIzol (Invitrogen) and total RNA prepared according to manufacturer's instructions using glycogen as carrier. Total RNA yield and purity was assessed using an Agilent 2100 bioanalyser (Agilent Technologies) and a NanoDrop 1000 spectrophotometer (Kisker). RNA labelling was performed with the Quick Amp Labelling Kit (Agilent Technologies). In brief, mRNA was reverse transcribed and amplified using an oligo-dT-T7-promotor primer and resulting cRNA was labelled with either Cyanine 3-CTP or Cyanine 5-CTP. After precipitation, purification and quantification, 1.25 μg of each labelled cRNA was fragmented and subsequently hybridized to whole human genome 44k microarrays (AMADID-014850), according to the supplier's protocol (Agilent Technologies). Hybridized microarrays were washed using the SSC washing protocol (Agilent Technologies). Scanning of microarrays was performed with 5 μm resolution using a DNA microarray laser scanner (Agilent Technologies). Raw microarray image data were analysed with the Image Analysis/Feature Extraction software G2567AA (Version A.9.5.1, Agilent). The extracted MAGE-ML files were further analysed with the Rosetta Resolver Biosoftware, Build 7.2 (Rosetta Biosoftware). Ratio profiles comprising single hybridizations were combined in an error-weighted fashion to create ratio experiments. A twofold change expression cut-off for ratio experiments was applied together with anti-correlation of ratio profiles rendering the microarray analysis highly significant (P-value < 0.01). The data presented in this publication have been deposited in NCBIs Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series Accession No. GSE33731. Ingenuity Pathway Analysis (IPA; http://www.ingenuity.com) software was used for functional analyses of transcriptome data. The tool identified the biological functions and/or diseases that were most significant to the expression data. Genes which were de-regulated at least twofold and with P-values < 0.01 were included in the analysis and associated with biological functions and/or diseases in the Ingenuity Knowledge Base. Right-tailed Fisher's exact test was used to calculate a P-value determining the probability that each biological function and/or disease assigned to data sets was due to chance alone.

Western immunoblotting

Cells were lysed in Laemmli buffer. Resulting protein extracts were loaded on SDS-PAGE gels; separated proteins were transferred to PVDF membranes. Membranes were probed with primary antibodies and HRP-conjugated secondary antibodies (Amersham), and detected with ECL reagent (ICN). The monoclonal anti-β-actin antibody (Sigma) was used as a loading control. All blots shown are representative of at least three independent experiments.

Immunofluorescence

RWPE1 and HaCaT cells were grown on 12 mm coverslips. After P. acnes infection, cells were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. Cells were then stained by permeabilization with 0.2% Triton X-100 for 15 min and blocking with 0.2% BSA (Biomol) in PBS for 10 min, all at room temperature. Primary antibodies (anti-P. acnes and anti-p65) were incubated for 1 h, followed by a detection step using Cy2- or Cy3-conjugated anti-mouse/anti-rabbit IgG secondary antibodies (1:150, 1 h; Jackson Immunoresearch). Actin was stained with Phallodin (Invitrogen). Nuclei were stained with Draq5 (Cell Signaling). Coverslips were mounted in Mowiol and analysed by confocal laser scanning microscopy using a Leica TCS SP and epifluorescence microscopy. Images were taken using appropriate excitation and emission filters for the fluorescent dyes used. Overlay images of the single channels were obtained using Adobe Photoshop. Extra-/intracellular staining was achieved as follows: after fixation of infected cells with 4% paraformaldehyde and blocking with 0.2% BSA for 10 min, extracellular bacteria were stained with primary anti-P. acnes antibody for 1 h and subsequently detected with secondary Cy2-conjugated anti-mouse IgG antibody (1:150, 1 h; Jackson Immunoresearch). Cells were then permeabilized with 0.2% Triton X-100 for 15 min and intracellular bacteria were stained with primary anti-P. acnes antibody for 1 h, and detected with secondary Cy3-conjugated anti-mouse IgG antibody (1:150, 1 h; Jackson Immunoresearch). Coverslips were mounted in Mowiol and analysed.

For immunofluorescence-staining of human skin and prostate tissues, formalin-fixed, paraffin-embedded tissue samples were deparaffinized and hydrated through a graded ethanol series before undergoing heat-induced antigen retrieval for 45 min in target retrieval solution (Dako). Tissues were then incubated with mouse anti-vimentin antibody at a dilution of 1:400. Slides were then incubated with Alexa-488-conjugated secondary antibody and counterstained with DAPI. All tissue samples were obtained under a Johns Hopkins Institutional Review Board (IRB) approved protocol.

Transmission electron microscopy

Propionibacterium acnes infected RWPE1 cells were fixed with 2.5% glutaraldehyde, post-fixed with 1% osmium tetroxide, contrasted with uranylacetate and tannic acid, dehydrated and embedded in Ultra-Low Viscosity Embedding Media (Polysciences). After polymerization, specimens were cut at 60 nm and contrasted with lead citrate. Specimens were analysed in a Leo 906E transmission electron microscope (Zeiss SMT) using a digital camera (Morada, SIS).

Antibiotic protection assay

To quantify bacterial entry into host cells an antibiotic protection assay was performed. Both HaCaT and RWPE1 cells were seeded to 24-well plates prior to infection. At 24 h p.i., 30 μg ml−1 streptomycin/penicilin (Invitrogen) was added for 2 h to kill extracellular bacteria. Cells were washed and 1% saponin (Sigma) was added to permeabilize cells, followed by plating of appropriate dilutions of lysate on Brucella agar. Assays were performed in duplicate wells. Each experiment was repeated at least three times. Data were tested for significance using unpaired t-test (GraphPad Prism).

siRNA and plasmid transfection

For RNA interference experiments RWPE1 cells were seeded into 12-well plates (1 × 105 cells well−1) 1 day prior to transfection. Cells were transfected with 10 nM siRNAs using HiPerFect transfection reagent (Qiagen), according to the manufacturer's instructions. Silencing efficiency on protein level was validated after 48 h by Western blot. The following siRNAs were used: siVIM 5′-CAGGTTATCAACGAAACTTCT-3′ and AllStars (Qiagen No. 1027281).

For vimentin overexpression keratinocytes were seeded into 12-well plates (5 × 104 cells well−1) 1 day prior to transfection. HaCaT cells were transfected with 1 μg of vimentin cDNA (Origene No. SC111054) or control GFP plasmids using 4 μl of FuGENE (Roche) according to the manufacturer's instructions. Overexpression of vimentin and GFP was determined after 48 h by Western blot.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

The authors thank Meike Sörensen for excellent technical assistance and Kate Holden-Dye for critically reading the manuscript.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
cmi1833-sup-0001-si.tif1153K

Fig. S1. P. acnes infection did not reduce host cell viability. Cells were infected with P. acnes P6 at an moi of 50. After 48 h (A) and 7 days (B), cell viability was assessed by a WST-1 assay.

cmi1833-sup-0002-si.tif555K

Fig. S2. Selection of genes strongly de-regulated in HaCaT and RWPE1 cells upon P. acnes infection. Comparison of gene expression profiles of P. acnes infected and non-infected cells at 24 h p.i. (in contrast to Fig. 1, where genes are listed whose expression differs between infected HaCaT and infected RWPE1 cells). In HaCaT 1269 genes are significantly de-regulated at 24 h p.i., 58% of these are upregulated upon infection. In RWPE1 only 688 genes are significantly de-regulated at 24 h p.i. (73% are upregulated). Two hundred and sixty-four genes are commonly de-regulated in both cell lines upon infection compared with non-infected cells. The data illustrates that P. acnes induces a much stronger response in HaCaT than in RWPE1 cells at 24 h p.i., in terms of the number of de-regulated genes as well as the fold change differences. Upregulated genes are shown in red, while downregulated ones are depicted in green. Non-significant expression differences are marked in black. All genes depicted in Fig. 1 are included in this figure.

cmi1833-sup-0003-si.tif4232K

Fig. S3. Network analysis of de-regulated genes in P. acnes-infected HaCaT and RWPE1 cells. Interactions between at least threefold upregulated genes were depicted with STRING (EMBL), which visualizes known and predicted protein–protein interactions based on reports within the literature, database entries and/or experimental evidence. The blue line thickness indicates the strength of interaction. Extensive networks largely based on inflammation-associated genes can be found in P. acnes-infected HaCaT cells at 24 h p.i. (A; key nodes are: IL-8, JUN, ICAM1, CXCL1, CCL4, NFKBIA, ATF3, NOS2) and in RWPE1 cells at 7 d p.i. (D; key nodes are: IL-8, JUN, FOS, PTGS2, ICAM1, MMP9, CSF2, CSF3, STAT1, TLR2). In contrast, only few genes de-regulated in RWPE1 at 24 h p.i. (B) and in HaCaT cells at 7 d p.i. (C) were functionally connected.

cmi1833-sup-0004-si.tif993K

Fig. S4. Anti-VIM antibody and cytoskeleton-interfering inhibitors partially block P. acnes invasion into RWPE1 cells. (A) P. acnes less efficiently invaded anti-VIM antibody-treated-RWPE1 cells. Non-treated and anti-IgG antibody-treated cells were used as controls. (B) Various chemical inhibitors interfering with actin polymerization (CB, cytochalasin B; CD, cytochalasin D) and microtubule polymerization (Col, colcemide; Noco, nocodazole) block P. acnes invasion. Representative results are at least three independent experiments are shown.*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.005.

cmi1833-sup-0005-si.tif2218K

Fig. S5. Vimentin mediates the inflammatory response to P. acnes in RWPE1. Gene expression profiles of P. acnes-infected siVIM RWPE1 cells and infected AllStars knock-down control cells were compared at 24 h p.i. The inflammatory response to P. acnes is strongly reduced in siVIM RWPE1 cells. (A) Interactions between genes downregulated at least fourfold were depicted with STRING (EMBL), which visualizes known and predicted protein–protein interactions based on reports within the literature, database entries and/or experimental evidence. The thickness of the blue line connecting genes correlates with the strength of interaction. (B) Expression fold changes upon VIM knock-down in infected (INF) RWPE1 cells for some prominent inflammation-associated genes are given. The data were adjusted to fold change differences between non-infected (NI) siVIM RWPE1 cells and NI AllStars knock-down control cells.

cmi1833-sup-0006-si.tif879K

Table S1. Strongly differentially expressed genes between HaCaT and RWPE1 cells. Comparative microarray analysis on non-infected HaCaT and RWPE1 cells revealed six genes that are highly expressed in RWPE1 but not in HaCaT. Only genes with expression fold changes of ≥ 25 were considered here.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.