Chronic Prostatic Infection and Inflammation by Propionibacterium acnes in a Rat Prostate Infection Model

Chronic inflammation in the prostate, seen as infiltration of inflammatory cells into the prostate gland in histological samples, affects approximately half the male population without indication of prostate disease, and is almost ubiquitous in patients diagnosed with benign prostate hyperplasia and cancer. Several studies have demonstrated the Gram-positive bacterium Propionibacterium acnes to be frequently present in prostate tissue from men suffering from prostate disease. P. acnes has been shown to be associated with histological inflammation in human prostatectomy specimens, and also to induce strong inflammatory response in prostate-derived tissue culture models. The present paper describes a rat model for assessment of the pathogenic potential of P. acnes in prostate. Prostate glands of Sprague Dawley rats (n = 98) were exposed via an abdominal incision and live P. acnes or, in control rats, saline were injected into the ventral and dorso-lateral lobes. Rats were sacrificed 5 days, 3 weeks, 3 months and 6 months post infection, and prostate tissue was analyzed for bacterial content and histological inflammation. Rat sera were assessed for levels of CRP and anti-P. acnes IgG. Live P. acnes could be recovered from the dorso-lateral lobes up to 3 months post infection, while the ventral lobes were cleared from bacteria at that time. In samples up to 3 months post infection, the dorso-lateral lobes exhibited intense focal inflammation. CRP and IgG levels were elevated throughout the span of the experiment, and reached maximum levels 3 weeks and 3 months post infection, respectively. We show that P. acnes have the potential to cause chronic infection in previously healthy prostate, and that the infection has potential to cause chronic histological inflammation in the infected tissue. The high prevalence of P. acnes in human prostate tissue calls for resolution of pathogenic details. The present rat model suggests that complications such as chronic inflammation may be induced by P. acnes infection.


Introduction
Chronic inflammation in the prostate, seen as infiltration of inflammatory cells into the prostate gland in histological samples, affects approximately half the male population without indication of prostate disease [1], and is almost ubiquitous in patients diagnosed with benign prostate hyperplasia and cancer [2,3]. Accumulating evidence suggests that prostatic inflammation contributes significantly to the etiology of prostate cancer [4,5] as well as benign prostatic hyperplasia (BPH) [1,6]. Bacterial colonization and infection of the prostate have been implicated as contributing to the initiation and maintenance of chronic inflammation [7,8,9]. Asymptomatic or subclinical bacterial infections in the prostate appear to be relatively common, yet largely under-diagnosed [10,11]. Several studies have demonstrated high prevalence rates of the Gram-positive bacterium Propionibacterium acnes (P. acnes )in prostate tissue from men diagnosed with prostate disease [12,13,14]. Serum titres of P. acnes antibodies correlate positively with PSA in cancer-negative patients [15], thus in-dicating P. acnes involvement in prostatic inflammation. Furthermore, P. acnes has been shown to be associated with histological inflammation in human prostatectomy specimens [13] and to induce a strong inflammatory response in prostate derived tissue culture models [14,16]. However, well-characterized models of acute and chronic prostate infection are yet to be developed. The present paper describes a rat model of prostatic P. acnes infection for the assessment of acute and chronic infection/inflammation in wild-type animals.

Propionibacterium acnes Cultivation
Two batches of Propionibacterium acnes bacteria were cultivated from frozen stock in BHI +5% horse serum at 37uC under microaerophilic conditions; type 1A (CCUG 41530) and a mixture of four human prostate isolates, two of type 1 and two of type 2 [17], respectively. Exponentially growing bacteria were collected after two passages in fresh medium, washed with sterile saline by centrifugation and resuspended into saline at a density of 1?10 7 CFU/ml.

Animals and Animal Treatment
Adult male Sprague Dawley rats (age 3-4 months, weight: 400-500 g) (n = 98) (B&K, Stockholm Sweden) were anesthetized with pentobarbital (50 mg/kg) and an incision was made in the lower abdomen to expose the prostate. Propionibacterium acnes (5 ml) (5?10 7 CFU) (type 1A in animals to be infected for 5 days and 3 weeks, respectively, and prostate isolate-P. acnes mixture in animals to be infected for 3 weeks, 3 months and 6 months, or saline (5 ml) was injected with a Hamilton syringe into the left ventral prostate (VP L ) and into the dorso-lateral prostate (DLP) lobes. After 5 days, 3 weeks, 3 months and 6 months, blood samples were collected by cardiac puncture and, subsequently, animals were sacrificed. Left ventral (VP L ), right ventral (VP R ), and dorso-lateral (DLP) prostate lobes were excised and treated for bacterial counts or fixed in formalin for subsequent histological analysis.

Ethics
The rats were maintained at the animal facility at Umeå University and all experiments involving animals were approved by the local Animal Review Board (Umeå, Sweden) (approval Ids: 2008/293, date:081029, A81-06, date:060818, A82-06, date:060818). All surgery was performed under sodium pentobarbital anesthesia, and all efforts were made to minimize suffering.

Hematoxylin/Eosin-and Immunofluorescence Stainings
Sample tissue was fixed in formalin, dehydrated and embedded in paraffin. Four micron thick sections were deparaffinized and rehydrated. The tissue sections were stained with hematoxylin & eosin according to standard procedures. Tissue was examined for histological inflammation (see below) with an Olympus AX-70 microscope and documented with an ALTRA 20 CCD camera. For IF, deparaffinized sections were antigen-retrieved by boiling in citrate buffer (10 mM, pH 6.0) at 2 atm for 1 h. Following blocking with 1% BSA in PBS, slides were incubated with P. acnes antiserum diluted 1:1000 in blocking solution for 1 h. Slides were washed in PBS and incubated for 1 h with goat anti-rabbit monoclonal antibodies labeled with alexa 488 (Invitrogen) diluted 1:1000 in blocking buffer. Following washing and dehydration, the slides were mounted and examined with epifluorescence (Zeiss Axioskop) or confocal fluorescence microscope (Leica). The overlay pictures were created with the Adobe Photoshop software.

Histological Characterization of Inflammation
Slides stained with hematoxylin & eosin were assessed microscopically. Inflammatory patterns were qualitatively categorized as focal or diffuse, and intensity of inflammation was scored based on the amount of infiltrating inflammatory cells. Patterns were categorized as diffuse when covering half the tissue or more, with no sharp borders between inflamed and non-inflamed regions. Patterns were categorized as focal when set off from surrounding normal tissue by clearly defined borders. Intensity of inflammation in each prostate specimen was categorized as minimal (#5 leukocytes/5000 mm 2 ), moderate (5-50 leukocytes/ 5000 mm 2 ), or severe (. 50 leukocytes/5000 mm 2 ), based on total leukocyte counts in five randomly selected 1000 mm 2 areas. A panel of infected prostate glands illustrates the histological correlates of these criteria (Fig. 1). The volume of inflamed foci, as a percentage of total prostate volume, was determined microscopically by the method described in [18].

Bacterial Counts and P. acnes Biotype Identification
Whole lobes were aseptically homogenized in PBS and 1/10 of the slurry obtained was subjected to 10-fold serial dilutions and plated on anaerobic blood agar plates. Plates were incubated for 1 week in 37uC under anaerobic atmosphere, whereafter P. acnes colonies were counted on the 1-2 plates with maximum resolution for each sample. Strains from biotype 1 & 2 differs in recA gene at base 71 (type 1 = G, type 2 = A), base 183 (type 1 = A, type 2 = G), base 214 (type 1 = C, type 2 = T) and base 424 (type 1 = A, type 2 = G). These SNPs were used to type the recovered isolates. Single colonies were transferred from the anaerobic blood agar plates into 100 ml PBS, and nucleic acid was prepared in a NorDiag Arrow preparation robot, using the Viral N/A extraction kit according to the manufacturer's instructions (NorDiag AB, Hä gersten, Sweden). 1 ml extracted DNA was template for a PCR reaction with primers parecAForw: AGCTCGGTGGGGTTCTCTCATC (0,3 mM) and pare-cARew: GCTTCCTCATACCACTGGTCATC (0,3 mM) [19], ABI SYBR Green PCR MasterMix (ABI, Warrington, UK), and water to a final volume of 25 ml. The PCR was performed on a 7900 HT Fast Real-Time PCR System, (Applied Biosystems) with a program consisting of: 95uC, 10 minutes (1 cycle), 95uC 60 s +50uC 30 s +72uC 90 s (35 cycles), 72uC 10 minutes (1 cycle). Sequencing of the PCR product was performed with parecAForw as primer (Eurofins MWG Operons sequencing service, Eurofins MWG GmbH, Ebersberg, Germany). All sequences could unambiguously be assigned type 1 or type 2. QPCR assessment of P. acnes genome numbers were performed on the remains of the prostate lobe slurry, or, for the animals infected for 6 months that were not directly processed for CFU counts, whole lobes that were stored at 270uC until used. Tissue was disintegrated with a Fast Prep-24 sample preparation system with tubes containing Lysing Matrix M (MP Biomedicals, Solon,Ohio, USA), and nucleic acid was prepared in a NorDiag Arrow preparation robot, using the Viral N/A extraction kit according to the manufacturer's instructions (NorDiag AB, Hä gersten, Sweden). 5 ml (of total 50 ml) extracted DNA was template for a quantitative PCR reaction as described earlier [20]. A standard curve was created from VP lobes from non-infected rats spiked with a defined number of P. acnes bacterial cells (CCUG 41530) prepared from serial dilutions of a liquid broth culture (1.75 10 9 CFU/ml). The standard curve, tissue processing and qPCR was performed as described above, and the Ct values vs CFU was plotted in a graph described by: CFU = 7E+13e 20,914Ct (R 2 = 0,9956). The calculations were performed with MS Excel.

Preparation of Rabbit Anti-Propionibacterium acnes Polyclonal Antiserum
A washed suspension of Propionibacterium acnes in PBS corresponding to approximately 1 10 9 cells/ml were treated with formaldehyde at a final concentration of 0.01 M. After incubation at 37uC on a slow shaker for 2 h followed by overnight shaking at room temperature, the bacteria were washed three times in PBS and then resuspended in PBS to an optical density of 1 and stored at 4uC. The inactivated bacterial culture was plated on anaerobic blood agar plates and incubated under anaerobic conditions for 10 days. No colonies were detected, indicating that the bacteria were completely inactivated. The inactivated bacteria were used as antigen to raise a polyclonal rabbit antiserum (Agrisera, Umeå, Sweden). The rabbits were maintained at the animal facility at Agrisera and all experiments involving animals were approved by the local Animal Review Board (Umeå, Sweden) (approval Id: A121-06).

Blood Collection and Serological Methods
Blood samples were collected from rats by cardiac puncture prior to sacrifice. After coagulation and centrifugation at 1400 rpm for 10 minutes at room temperature, serum was collected. Presence of anti-P. acnes IgG was assessed by a western blot procedure where the rat serum functioned as the primary antibody. A total bacterial lysate (1?10 10 bacterial cells dissolved in 400 ml Sample Buffer) was submitted to SDS-PAGE and electroblotted onto PVDF membrane. The filter was blocked by normal rat serum, cut into strips and incubated for 1 h with individual rat serum diluted 1:2000. After washing, the strips were collected into a single container and incubated with goat anti-rat HRP antibody (1:5000) for 1 h, washed and developed with ECC solution (Amersham). CRP levels were determined in an ELISA method (Rat Serum CRP M-1010) according to the manufacturer's instructions (Alpha diagnostics, San Antonio, TX, USA).

Statistical Analysis
Differences between interval variables and in average intensity scores of IgG were tested for by Student's t-test. For small samples (n = 3) a permutation test was used [21]. Due to the small sample size, null hypotheses were rejected for p-values equal to 0.05 at the a = 0.05 significance level. Dependencies between categorical variables were tested for by Fisher's exact test or its extension to larger tables.

Prostatic Infection
Tissue from VP L and DLP in all experimental groups was assessed for the presence of viable P. acnes bacteria. Live bacteria were recovered from the prostate locales of infected animals throughout the duration of the experiment, with decreasing bacterial titers as infection times increased (Fig. 2). 5 days post infection, 6700, 1660 and 4600 CFU were recovered from VP L , and 60500, 215000 and 66000 from DLP. 3 weeks post infection, 10, 0 and 7 CFU were recovered from VP L and 460, 790 and 1360 from DLP of the animals infected with P. acnes type 1A, and 32, 5 and 0 from VP L and 860, 1900 and 6100 from DLP of the animals infected with the P. acnes mixture. 3 months post infection; no bacteria were recovered from VP L , and 0, 10 and 0 from DLP. Bacterial titer counts were significantly higher in DLP than in VP L 5 days post infection (p = 0.05), 3 weeks post infection with type P. acnes 1A (p = 0.05), and 3 weeks post infection with P. acnes mixture (p = 0.05). The possibility that prostate-derived P. acnes isolates would be more potent than P. acnes of type 1A failed to find support in CFU recovery counts 3 weeks post infection (VPL: 1A , mix, p = 0.50, DLP: 1A , mix, p = 0.10). From the rats infected with the P. acnes mix for 3 weeks, 29 recovered bacterial clones, 12 of which from VP L (2 individual rats), and 17 from DLP (3 individual rats), were analyzed for type. Of the VP L isolates, 10/12 were of type 1 and 2/12 were of type 2. Of the DLP isolates, 5/17 were of type 1 and 12/17 were of type 2. Statistical testing for similarity in P. acnes type distributions between VP L and DLP indicated difference (p = 0.00775; (1.53, 137.34) CI of odds ratio).  Genome counts failed to differ between DLP and VP L 5 days post infection (p = 0.45). However, 3 weeks post infection, counts were significantly higher in DLP than in VP L in rats infected with P. acnes type 1A (p = 0.05), as well as in rats infected with P. acnes mix (p = 0.05). The possibility that prostate-derived P. acnes isolates would be more potent than P. acnes of type 1A failed to find support in genome counts 3 weeks post infection (VP L : p = 0.25, DLP: p = 0.2). Genome counts were significantly higher in DLP than in VP L in rats infected for 3 months (p = 0.05). No hypothesis was tested regarding possible difference between DLP and VP L 6 months post infection, due to the small number of rats.

Serology
Aortic blood was collected when rats were sacrificed 5 days, 3 weeks (1A & mix), 3 months, and 6 months post infection. Blood serum from infected and control rats were assessed for presence of P. acnes binding IgG and rated 0-3, where 0 represents no P. acnes binding IgG (saline-instilled controls 5 days post infection), and 1-3 are positive in increasing degree. In the control group, all sera scored 0 at 5 days, 3 weeks, 3 months and 6 months post infection (data not shown). All sera scored 0 in the initial sample, 5 days post infection, but non-zero scores were found in subsequent samples (score distribution shown in figure 3). Tests indicated the highest average score in sera collected 3 months post infection (3 m.3 w: p = 0.04; 3 m.6 m: p = 0.0025). No score difference could be attributed to type of infective agent, 1A or mix, 3 weeks post infection (p = 0.45). Serum CRP was quantified 5 days, 3 weeks, 3 months and 6 months post infection. The infected animals had elevated CRP levels throughout the infection period, exhibiting an average of 1.86, 2.00, 2.13, 1.73 and 1.44 times the values of the control animals 5 days, 3 weeks (1A), 3 weeks (mix), 3 months and 6 months, respectively (Fig. 4). Statistical testing indicated difference between infected and controls in all groups (5d: p = 0.002, 3w 1A: p = 0.00, 3w mix: p = 0.00, 3 m: p = 0.02, 6 m: p = 0.02), but no difference could be attributed to type of infective agent, 1A or mix, 3 weeks post infection (p = 0.947).

Histological Characterization of Inflammation
H&E-stained slides were scored for intensity of inflammation based on counts of infiltrating inflammatory cells, predominantly polymorphonuclear leukocytes in animals infected for 5 days, and lymphocytes in animals infected for 3 weeks, 3 months and 6 months, respectively, (two rats infected for 6 months showed an exception to this pattern and are described further below). Further, inflammation was categorized as focal or diffuse, and volume proportion of inflammatory foci to normal tissue was calculated using a stereological method.

Ventral Prostate Lobe
The left ventral prostate (VP L ) exhibited severe diffuse inflammation 5 days post infection in 5/6 infected animals and moderate diffuse inflammation in 1/6 (Fig. 5A). No inflammation was seen in the lobe contra-lateral to the lobe injected with bacteria (VP R ) (not shown) or in control animals (Fig. 5A). 3 weeks post infection, moderate diffuse inflammation was seen in VP L in 7/12 of the animals infected with P. acnes type 1A, and in 3/6 animals infected with mix of prostate-derived P.acnes isolates. One of the animals infected with P. acnes mix had severe diffuse inflammation in VP L . Minimal diffuse inflammation was present in the VP R in 4/18 of infected animals (not shown). In controls, 3/ 17 had minimal diffuse inflammation in VP L (Fig. 5A), and 4/17 had mild diffuse inflammation in VP R (not shown). 3 months post infection, moderate diffuse inflammation was present in VP L in 3/ 6 ( Fig. 5A) and in VP R in 3/6 of infected animals. One rat (1/6) had severe diffuse inflammation in VP L . In controls, 2/6 animals had moderate diffuse inflammation in both VP L (Fig. 5A) and VP R (not shown). 6 months post infection, there was moderate diffuse inflammation in VP L in 2/9 infected animals (Fig. 5A), and in VP R in 1/9 (not shown). In controls, 2/6 had moderate diffuse inflammation in VP L (Fig. 5A), and 1/6 had moderate diffuse inflammation in VP R (not shown). Statistical analysis indicated significant difference between infected and control groups 5 days post infection (p = 0.002479), 3 weeks post infection (1A) (p = 0.04597), 3 weeks post infection (mix) (p = 0.045), but not 3 months (p = 0.5671) and 6 months post infection (p = 1). In addition, no significant difference could be established in inflammatory intensity between the two P. acnes agents 3 weeks post infection (p = 0.5362).

Visualization of P. acnes in Prostate Tissue
Sequential sections of VP L and DLP from infected animals and controls were stained with H&E and P. acnes-specific immunofluorescence, respectively. Bacteria could be detected by microscopy in DLP up to 3 months and in VP L up to 3 weeks post infection. Bacteria were seen both in stroma and in epithelial glands, and the presence of bacteria co-localized with foci of histological inflammation (Fig. 7). No bacteria were seen in controls.

Discussion
We have established a novel rat model to investigate the pathogenic features of P. acnes infections of the prostate gland. Accumulating evidence for frequent presence of P. acnes in human diseased prostate tissue, combined with the widely accepted hypothesis that prostatic inflammation is a significant etiologic factor in both prostate cancer and benign prostatic hyperplasia (BPH), a characterization of the properties of this specific bacterial infection is highly relevant to the understanding of disease development. Instillation of 5?10 7 CFU P. acnes into the prostate gland did not provoke any overt signs of disease, a fact that demonstrates the low direct pathogenicity of this bacterium. The infection initially caused a strong acute histological inflammation in both VP and DLP, and the inflammatory patterns were different in the two lobes; in DLP the pattern of inflammation was focal but in VP it was diffuse. In DLP the initial acute inflammation was succeeded by chronic inflammation that persisted in approximately 1/3 of the animals examined 3 months post infection. Bacterial cells were associated with inflammatory foci, and live bacteria could be retrieved from 1 out of 3 animals 3 months post infection. In VP, on the other hand, the initial acute state evolved into low grade inflammation. At an early stage, 3 weeks post infection, bacteria were only seen in a minority of animals, and only low numbers of live bacteria could be cultivated. The differences between lobes may suggest that the DLP is innately more susceptible for bacterial infections than is the VP. Similar conclusions have been drawn in studies of rodents experimentally infected with E. coli [22,23]. Recent research suggests that clonal subpopulations of P. acnes carry specific virulence traits [24]. Studies have reported that P. acnes isolates derived from prostate are genetically and biochemically distinct from skin isolates, and that isolates derived from malignant prostate tissue are predominantly of type 2 [13,17]. We infected rats with either prostate-or non-prostate derived P. acnes to observe possible differences in infectious or inflammatory properties between the bacteria. In addition, the prostate derived agent contained a mixture of 2 strains of biotype 1, and 2 strains of biotype 2. There were no differences between prostate and nonprostate isolates regarding severity of inflammation or remaining bacterial load 3 weeks post infection. However, the types distributed differently in prostate lobes; type 2 was more prone to persist in DLP. An interpretation of this result is that a wide range of P. acnes strains have capacity to exert the inflammatory effects observed at this time point, and, in addition, that properties specific to prostate-derived strains may have impact on locale tropism and infectivity. In animals infected with a mixture of isolates, only type 2 isolates could be recovered 3 months post infection. Although the number of recovered bacterial clones, 10 CFU, is too limited to support extensive conclusions regarding increased fitness for type 2 biotype in prostate infections, the result does not rule out this intriguing theory. The mean serum CRP levels of uninfected rats ranged between 280-360 mg/ml during the 6 months time span of the experiment. These values are in line with CRP levels reported for other rat strains [25]. The P. acnes infection caused an increase in serum CRP throughout the studied period, with a peak value of 627 mg/ml 3 weeks post infection. Elevated CRP levels, when prostatic inflammation has declined, are not typical for an acute-phase reaction [25]. The explanations may involve secondary infections, permanent tissue damage or intrinsic properties of this particular rat breed. The humoral immune response against P. acnes was maximal 3 months post infection, when all infected animals had detectable IgG, and a majority of them high titers. 6 months post infection, half the population had lost detectable P. acnes specific IgG, and only 10% had high titers. Sprague-Dawley rats are not prone to spontaneous development of prostatitis. The DLP is reported to stay free of histological inflammation up to one year of age [26]. Other studies report a 16% frequency of spontaneous prostatic inflammation in VP at 6 months of age [27]. Our results support these earlier reports in that the DLPs from our controls were free of inflammation throughout the study time up to 9-10 (3-4+6) months. VPs in controls exhibited spontaneous inflammation debuting at approximately 4 months of age, and at 9-10 months of age one third of animals were inflamed. Further studies are required to describe the molecular differences between prostate lobes responsible for this histological pattern. Recently, several mouse prostate infection models with uro-pathogenic E. coli have been published [23,28,29,30]. Interestingly, the E. coli mouse models generate evidence of infection-dependant pre-cancerous tissue transformations [29], and reactive hyperplasia as well as increased epithelial proliferation [23]. Given the similarities in the prostatic inflammatory responses to P. acnes in our data and to E. coli infections reported in these studies, such adverse complications upon P. acnes infections of prostate may not be ruled out. Further studies are needed to assess the consequences of this chronic infection at cellular and tissue level. While it is beyond the scope of the present study to resolve the cellular identity of the inflammatory infiltrates, the observed aggregate of PMNs surrounding solid structures in the glandular lumen of DLP 6 months post infection (Fig. 6) may yet be of interest. Corpora amylacea (CA) inclusions in prostate glands are known to mainly consist of amyloid forms of proteins originating from neutrophil granules as reported by us and others [31,32]. We suggest that figure 6 captures an early stage in the formation of CA, where neutrophiles have clustered, allowing their granule content to aggregate into a solid structure.
Conclusions P. acnes generate a self-limiting chronic inflammation in the prostate of rats. Thus, it is a potentially useful model system for general studies on long-term effects of infectious inflammation on prostate health. Upon P. acnes challenge, the prostate lobes differ in inflammatory response and bacterial clearance, a finding that is interesting to translate into a human system where predominantly the peripheral zone is affected by chronic inflammation.