Plakophilin 1 deficiency in prostatic tumours is correlated with immune cell recruitment and controls the up‐regulation of cytokine expression post‐transcriptionally

Plakophilin (PKP1) 1 is a member of the arm‐repeat family of catenins and acts as a structural component of desmosomes, which are important stabilizers of cell–cell adhesion. Besides this, PKP1 also occurs in a non‐junctional, cytoplasmic form contributing to post‐transcriptional regulation of gene expression. Moreover, PKP1 is expressed in the prostate epithelium but its expression is frequently downregulated in prostate cancers with a more aggressive phenotype. This observation may imply a tumour‐suppressive role of PKP1. We found that, in prostatic adenocarcinomas with PKP1 deficiency, the occurrence of T‐cells, B‐cells, macrophages and neutrophils were significantly increased. In a PKP1‐deficient prostatic cancer cell line expressing IL8, these levels were statistically meaningfully reduced upon PKP1 re‐expression. When analysing prostatic PKP1 knockdown cell lines, the mRNA and protein levels of additional cytokines, namely CXCL1 and IL6, were upregulated. The effect was rescued upon re‐expression of a PKP1 RNAi‐resistant form. The corresponding mRNAs were co‐precipitated with cytoplasmic PKP1, indicating that they are components of PKP1‐containing mRNA ribonucleoprotein particles. Moreover, the mRNA half‐lives of CXCL1, IL8 and IL6 were significantly increased in PKP1‐deficient cells, showing that these mRNAs were stabilized by PKP1. In an in vitro migration assay, the higher cytokine concentrations led to higher migration rates of THP1 and PBMC cells. This finding implies that PKP1 loss of expression in vivo correlates with the recruitment of immune cells into the tumour area to set up a tumour‐specific environment. One may speculate that this newly established tumour environment has tumour‐suppressive characteristics and thereby accelerates tumour progression and metastasis.


Introduction
Plakophilin 1 (PKP1) belongs to the PKP protein family comprising three different members PKP1, 2 and 3 that are specifically recruited to desmosomal plaques in a highly cell-type-specific manner where they support desmosome assembly and stability [1,2]. Additionally, PKPs act as crucial regulators of specific signalling programmes and control diverse cellular processes that range from transcription, mRNA abundance, protein synthesis, growth, proliferation and migration to invasion and tumour development [3,4]. Besides a localization in cell-cell contacts such as desmosomes, PKP1 is found both in the cytoplasm and the nucleus and can bind to single-stranded DNA [5,6]. Moreover, PKP1 and/or PKP3 have also been detected as part of mRNA ribonuclein complexes, cytoplasmic stress granules and ribosomal complexes implying modulatory roles in mRNA localization, stability and translation [7]. In fact, PKP1 and PKP3 affect the mRNA stability of desmosomal proteins [8] and PKP1 may serve as a regulator of mRNA translation by promoting eIF4A1 activity [9].
Prostate cancer is one of the most frequent malignancies in men in the Western world [10]. In recent years, the therapy of prostate cancer has been improved, but there is still a strong demand for accurate diagnosis and improved treatment options to decrease mortality. Prostate tumour initiation and progression are a multistep process where prostatic epithelial cells gain new biological capabilities [11]. These changes in cellular behaviour are due to altered gene expression that may be regulated directly by epigenetic alterations, transcription factors or by RNA-binding proteins influencing RNA metabolism and translation [12]. Interestingly, PKP1 expression in prostate cancer is largely reduced in more aggressive cancer types and associated with lymph node metastasis [13,14]. For PKP1, methylation-mediated epigenetic alterations lead to loss of expression [15][16][17]. Hence, PKP1 may have a tumour-suppressive function and PKP1 loss of function may be a critical event during prostate cancer progression.
Over the past years, the tumour-associated stroma comprising different cells that are recruited to the area of solid tumours constitute altogether the tumour environment that has gained strong interest regarding tumour progression and metastasis [11]. Immune cells of the innate and adaptive response such as macrophages, neutrophils or T-cells and B-cells infiltrate the tumour area and thereby may inhibit or, on the contrary, promote tumour progression [18,19]. The dynamic crosstalk between these diverse cell types, through direct cell-cell contact or soluble factors, such as cytokines, creates a specific niche. Cytokines such as CXCL1, IL6 and IL8 comprise a family of lowmolecular-weight proteins involved in host defence, inflammation and tumour immunobiology [20,21]. They may directly have an impact on tumorigenesis by regulating tumour cell growth, invasiveness and metastasis, or act indirectly by exerting modulatory effects on various cells of the tumour microenvironment [22]. Moreover, T-cells and B-cells may also accumulate in organized cellular aggregates, so-called tertiary lymphoid structures (TLS), that develop in non-lymphoid tissues at sites of chronic inflammation including tumours [23,24] and drive the immune response against tumour development and progression [25].
In the present study, we observed that in prostatic adenocarcinomas cell counts for T-cells, B-cells, macrophages and neutrophils were increased in PKP1deficient areas implying a correlation between PKP1 loss of expression and recruitment of immune cells. In in vitro experiments with a PKP1-deficient prostatic cancer cell line, IL8 expression was influenced by PKP1. Moreover, in PKP1 knockdown cell lines, the mRNA and protein level of CXCL1, IL6 and IL8 were increased. The mRNAs of these cytokines were part of PKP1-containing complexes and their mRNA stability was influenced by PKP1. The higher cytokine concentrations in cultures of PKP1-deficient cell lines led to higher migration rates of THP1 and PBMC in vitro.

Immune cells are recruited to PKP1-deficient tumour areas in prostatic adenocarcinoma
In earlier studies, we noticed that in prostatic adenocarcinoma showing a dedifferentiated state, such as in a high Gleason pattern, PKP1 expression was partially lost [13]. To find out if PKP1 loss correlates with immune cell recruitment in vivo, we re-analysed tissue samples with known heterogeneity in PKP1 expression. Serial sections of such tissue samples together with corresponding tumour-free samples from the same patient were stained with PKP1-and keratin 8/18specific antibodies (Fig. 1). Here, keratin 8/18 staining, an epithelial marker, indicated tumour areas negative for PKP1. These serial sections were counterstained with markers for T-cells (CD3), B-cells (CD19), neutrophils (myeloperoxidase) or macrophages (CD163) (Fig. 2). As a control, all markers were used in parallel on samples with diagnosed prostatitis to verify the specific reaction of the individual antibodies meant as markers for T-cells, B-cells, neutrophils and macrophages.
After a careful inspection in three similar-sized, rectangular areas in prostatic adenocarcinomas, single cells were counted in PKP1-positive or -negative tumour areas as well as in corresponding tumour-free areas. For all four cell types, we noticed an increase in cell counts in PKP1-deficient tumour areas compared not only to PKP1-positive tumour areas but also to tumour-free areas. We also carefully counted the number of TLS found in tumour-free as well as in PKP1positive and PKP1-negative tumour regions by using CD3 staining and accumulation of T-cells as a marker for TLS (Fig. 3A,B). In some areas, individual TLS were partially located in a PKP1-deficient and PKP1positive area. As here, no unambiguous assignment was possible and therefore a new category named border was opened (Fig. 3C). Although the total count of TLS in PKP1-negative tumour areas was higher compared to PKP1-positive ones, the statistical analysis with the Dunnett procedure indicated no significant enrichment of TLS in PKP1-negative tumour areas. Taken together, single immune cells such as T-cells, Bcells, neutrophils and macrophages were recruited to PKP1-negative tumour areas, however, TLS were not accumulating in these tumour areas.

PKP1 influences IL8 expression in a prostatic cancer cell line
Recently, it was reported that IL8 as detected by RNA in situ hybridization (RISH) was expressed in prostate adenocarcinomas and was associated with prostate cancer aggressiveness [26]. Moreover, with this method, the expression of IL8 in tumour cells in vivo and in vitro was validated. To find out if PKP1 may influence IL8 expression, we compared IL8 expression in the prostatic cancer cell lines LNCaP and DU145 (Fig. 4A). Although both cell lines are PKP1 deficient only, DU145 cells showed a high IL8 level. A PKP1-cDNA construct comprising aa235-726 [6] was re-expressed in DU145 cells. The truncated PKP1 form has lost desmosomal targeting capacity and is shown to accumulate in the cytoplasm. The effect on the IL8 level was compared to DU145 control cells (Fig. 4B). Noteworthy, the IL8 level was significantly reduced suggesting that PKP1 influenced the IL8 level.

Cytokine expression is deregulated in PKP1 knockdown cells
To understand how PKP1 loss might contribute to immune cell recruitment in prostatic adenocarcinoma, we decided to use the prostatic cell line BPH1 that expressed PKP1 (Fig. 4A) and established different PKP1 knockdown cell lines and characterized the expression profile [27]. When we grouped the altered genes into categories immunological disease, inflammatory disease and inflammatory response were identified with highly significant values above the threshold. We had a closer look into the deregulated genes and noticed that components known to be involved in immunological responses, the cytokines CXCL1, IL8 and IL6 were upregulated with fold changes 3.97, 3.85 and 2.55 respectively. To verify the expression profiling data, qPCR analysis using CXCL1-, IL8-and IL6specific primers was performed. Indeed, in two different PKP1-deficient cell lines, BPH1-shPKP1-2411 and BPH1-shPKP1-2357 cells, the relative mRNA level for CXCL1, IL6 and IL8 was significantly increased (Fig. 5A). Moreover, we quantified the protein concentration of CXCL1, IL6 and IL8 and noticed that  protein levels were also significantly increased in the two different PKP1-decificient cell lines (Fig. 5B). To rule out off-target effects, we performed a rescue experiment. A PKP1-cDNA construct comprising aa235-726 [6] containing no shRNA targeting sites was reexpressed in the knockdown cell line BPH1-shPKP1-2357 (Fig. 5C). Upon stable PKP1 re-expression of a cytoplasmic PKP1 form, the relative CXCL1 level was significantly reduced. Taken together, the knockdown of PKP1 led to higher expression of CXL1, IL6 and IL8 mRNAs. Importantly, higher expression levels of CXCL1, IL6 and IL8 were also observed at the protein level, indicating that PKP1 is physiologically important for the regulation of the expression of the cytokines CXCL1, IL6 and IL8.

Cytokine mRNA levels are regulated by PKP1
We have shown that the non-junctional, cytoplasmic form of PKP1 is part of ribonuclein complexes [7,8]. These complexes also contain mRNA and regulate mRNA stability of desmosomal mRNAs [8,14]. To verify if CXCL1, IL6 and IL8 mRNA are also part of PKP1 complexes, we immunoprecipitated PKP1 from BPH1 cell lysates and tested for the presence of these specific mRNAs by RT-PCR. Indeed, CXCL1, IL6 and IL8 mRNA were detected in PKP1 immunoprecipitate but not the control snoRNA ACA44 (Fig. 6A). Given that the PKP1 knockdown leads to increased mRNA levels of CXCL1, IL6 and IL8, we investigated the influence of PKP1 on the mRNA stability of CXCL1, IL6 and IL8. Transcription was blocked with actinomycin D, and RNA was isolated after 0, 15, 30 or 60 min. The amount of CXCL1, IL6 or IL8 mRNA over time was then measured by quantitative RT-PCR. The half-lives of the specific mRNAs were estimated and compared between the control and PKP1 knockdown cell line (Fig. 6B). Indeed, for all three transcripts, the half-lives were significantly increased in the PKP1 knockdown cell line indicating that PKP1 regulates the mRNA stability of CXCL1, IL6 and IL8.

PKP1 depletion increased the ability of immune cells to migrate in vitro
To test if the higher chemokine levels in PKP1-deficient cells affected the cell migration of immune cells, we applied a transwell cell migration assay. Conditioned medium from the two different PKP1-deficient cell lines, BPH1-shPKP1-2411 and BPH1-shPKP1-2357 cells, and the control cell line was placed in the bottom chamber. THP1 cells or PBMC were filled in the top chambers of the wells and the number of cells that migrated through the transwell filter into the bottom chamber was counted. Indeed, the chemokines produced by PKP1deficient cell lines induced the cell migration of THP1 cells and PBMC (Fig. 7A,B). These data support that PKP1 regulates the cell migration of immune cells in vitro.

Discussion
In PKP1-deficient areas of prostatic adenocarcinomas, cell counts for T-cells, B-cells, macrophages and neutrophils were increased implying a correlation between PKP1 loss of expression and recruitment of immune cells. In vitro, in a PKP1-deficient prostate cancer cell line, the expression of the chemokine IL8 was    (B). (A) By immunoprecipitation using lysates from BPH1 wild-type cells with PKP1-specific antibodies PKP1 was enriched, compared to control (con) and immunoprecipitated (IP). In corresponding samples, the occurrence of specific mRNAs was determined after RNA isolation by RT-PCR. Note that IL6, IL8, CXCL1 and PKP2 but not ACA44 mRNA could be detected in total lysates (Load, Post) and the PKP1 immunoprecipitate (IP). (B) mRNA stability measurements for CXCL1, IL6 and IL8 were performed in BPH1-shPKP1-2357 (KD2) and negative control BPH1-shLac (con) after actinomycin D treatment. RNA was isolated after 0, 15, 30 and 60 min and the amount of specific mRNA was detected by quantitative RT-PCR. The data of seven independent biological replicates were normalized to 0 h time point and the mRNA half-live was determined. Error bars show SD. Statistical comparisons between the two experimental groups were analysed using the unpaired Student's t-test. *P < 0.05; **P < 0.01. Note that the differences in mRNA half-lives are significant for CXCL1, IL6 and IL8 mRNA. regulated by PKP1. Moreover, in PKP1 knockdown cell lines, the mRNA and protein level of CXCL1, IL6 and IL8 were increased. The mRNAs of these cytokines were part of PKP1-containing complexes and their mRNA stability was influenced by PKP1. The higher cytokine concentrations in cultures of PKP1-deficient cell lines led to higher migration rates of THP1 and PBMC in vitro.
Our finding that in PKP1-deficient prostatic tumour areas significantly higher counts of B-cells, T-cells, neutrophils and macrophages were observed is in line with numerous studies exploring evidence regarding immune infiltration as a potential mediator and indicator of aggressive prostate cancer, reviewed by Strasner and Karin [18]. From these studies, using mostly cell biological methods such as immunohistochemistry, it is suggested that growing tumours can induce the recruitment of immune cells into the prostate microenvironment and initiate a reciprocal interaction that promotes disease progression. Here, especially tumourinfiltrating T-cells and macrophages were found to be pro-tumorigenic.
However, all these studies, including this one, are impaired by the complexity of infiltrating immune cells. A newly developed bioinformatic approach using the CIBERSORT program allows differentiating among 22 tumour-infiltrating immune cells [28]. By using this method, it was shown that five different immune cell types accumulated in prostate cancer tissue. In accordance with our data, these were B-cells, M1/M2 macrophages and neutrophils. Interestingly, a higher infiltration of M1 macrophages and neutrophils significantly reduced the survival of prostate cancer patients [29]. It has to be mentioned that depending on the data sets analysed other results may be obtained with this method [30][31][32]. Moreover, this bioinformatic analysis does not consider the occurrence of tertiary lymphoid organs that consist of Tcells, B-cells and other cells and their specific function in tumour progression [23].
Increased numbers of immune cells are also found in both acute and chronic prostatic inflammation [33]. Chronic inflammation has been designated as an enabling characteristic of cancer development [11] and has an impact at each stage of tumour developmentfrom tumour initiation and promotion to progression and metastasis [34]. Therefore, it would be interesting to analyse if PKP1 deficiency is also occurring in inflammation-associated regions of prostate atrophy (PIA) or prostatic intraepithelial neoplasia (PIN).
Intra-tumoral cytokine expression of, for example IL6 or IL8 has been already reported [26,35] and this expression is proposed as a key regulator of infiltrating immune cell recruitment. It is well documented that cytokines such as IL6 or IL8 may be secreted by prostate epithelial cancer cells and thereby may promote prostate cancer growth [20,26,35,36]. Indeed, we were able to verify IL8 expression in a prostate cancer cell line and noticed that IL8 expression decreased upon re-expression of PKP1, suggesting that PKP1 is involved in regulating its expression. The expression of cytokines is regulated on the one hand by transcription factors such as androgen receptor (AR) or erythroblast transformation-specific-related gene (ERG) [26,35]. Especially, AR plays a central role in prostate cancer initiation and progression, and therefore androgen ablation therapy is in clinical use for several decades [37][38][39]. Here, we, however, were able to show that PKP1 is part of an mRNA ribonucleoprotein particle that contains the mRNAs of CXCL1, IL8 or IL6. This observation is in line with earlier findings on the occurrence of PKP1/3 mRNPs in cytoplasmic extracts [7]. Moreover, in this study, we present evidence that PKP1 deficiency stabilizes the mRNAs of CXCL1, IL8 and IL6. An effect of PKP1 on mRNA half-lives has been already reported, although here the mRNAs of desmosomal proteins were destabilized [8]. One may assume that additional factors in these PKP1-containing mRNPs may trigger the destabilization or stabilization of specific mRNAs. Indeed, the regulation of cytokine expression is an excellent example of the complex network of cis-acting sequence elements and trans-acting factors controlling mRNA stability [40]. Some trans-acting factors (e.g. TTP or HuR) may bind to the same sequence elements (e.g. AU-rich elements), compete for binding and once bound may have opposite effects on mRNA stability [41]. In addition, mRNAs with AU-rich elements are prone to accumulate in the so-called P-bodies, storage sites for translationally repressed mRNAs and inactive mRNA decay enzymes [42,43]. PKP1 loss may lead to a cytokine mRNA release from these sites of longterm storage and the mRNAs are now ready to be translated leading to the accumulation of the protein [44].
Besides analysing the occurrence of specific immune cells in the prostatic tumour environment, it is also important to search if there is an immune cell-specific correlation with the expression of certain proteins. It has been shown that PTEN deficiency in prostate cancer is associated with an immunosuppressive tumour environment [45]. Likewise, it is intriguing to assume that PKP1 loss is also linked to an immunosuppressive state in prostate cancer, and determining PKP1 status may help to indicate which patients might benefit from immunotherapies.

Patients and tissue samples
In total, 27 formaldehyde-fixed, paraffin-embedded sections were applied in this study [13]. Thirteen of 27 contained prostatic adenocarcinoma, 12 were tumour free and in two samples prostatitis was diagnosed. For 11 tumour samples, the corresponding tumour-free section from the same patient was used. The mean age was 64.8 years and ranged from 60 to 68 years. Gleason scores in prostatic adenocarcinoma were above 7 and for all samples, a heterogenous PKP1 expression was reported [13]. Tissue samples were provided by the tissue bank of the National Center for Tumour Diseases (NCT, Heidelberg, Germany) in accordance with the regulations of the tissue bank and the approval of the ethics committee of Heidelberg University. The regulations follow the standards set by the Declaration of Helsinki and include a written patients' consent on the use of tissue samples for research purposes.

Cell lines and isolation of PBMCs
Cancer cell lines used (LNCaP and DU145) were described earlier [13] and further information on the two cell lines may also be found in the catalogue of the American Tissue Culture Collection (ATCC, Manassas, VA, USA). DU145 cells were transduced with the pLenti plasmid 6.2/V5-dest PKP1-cDNA construct containing aa235-726 [6] or an empty vector as control according to the transduction protocol of the manufacturer, and cells were selected with 10 lgÁmL À1 blasticidin (Invitrogen, Thermo Fisher Scientific, Karlsruhe, Germany). The cell lines BPH1-shPKP1-2411 (KD1) and BPH1-shPKP1-2357 (KD2) show a stable PKP1 knockdown, and the control cell line BPH1-shLac (con) and BPH1 cells have been already described [27]. For a rescue experiment, an RNAi-resistant PKP1-cDNA construct containing aa235-726 [6] including a C-terminal myc-tag [13] was expressed in the BPH1-shPKP1-2357 (KD2) cell line using the pLenti plasmid 6.2/V5-dest according to the transduction protocol of the manufacturer and cells were selected with 10 lgÁmL À1 blasticidin (Invitrogen, Thermo Fisher Scientific) and single-cell clones obtained by limited dilution [13]. THP-1 cells [46] were cultivated in RPMI-1640 medium. All cell lines were authenticated by multiplex cell line authentication and regularly tested for mycoplasma contamination [47,48]. Peripheral mononuclear blood cells (PBMC) were isolated from human whole blood by density gradient centrifugation with Lympho-Paque solution (Genaxxon Bioscience, Ulm, Germany) using Leucosep tubes (Greiner Bio-One, Frickenhausen, Germany). Separated blood cells were sedimented and suspended in ACK lysing buffer (Gibco, Life Technology Cooperation, Grand Island, NY, USA) to remove red blood cells. PBMC were then kept in RPMI-1640 medium until use.

Transwell cell migration assay
Conditioned medium from epithelial cell lines was placed in 24-well culture dishes, and transwell filter inserts (Greiner Bio-One) were inserted [49]. Here, filters with 8 lm pore size for THP-1 cells and filters with 3 lm pore size for PBMC were used. A total of 100 000 cells in the cell culture medium were filled into the upper reservoir and incubated for 3 h in a cell incubator. Then, the inserts were removed, and the plate and cells that had moved into the lower reservoir were collected and counted in a Neubauer chamber.

Immunofluorescence and counting of immune cells
Serial sections of formalin-fixed, paraffin-embedded tissue sections were first deparaffinized, and antigen retrieval was performed as described previously in a steamer with 0.1 M Tris-HCl buffer containing 5% urea, pH 9.0 [13]. Sections were then treated with 0.2% Triton X-100 in PBS, and blocked with 2% milk powder and 5% donkey serum in PBS. For double localization markers for immune cells were incubated with either PKP1-or Keratin 8/18-specific polyvalent sera. An incubation step with DAPI (1 lgÁmL À1 , #18860, Serva for Electrophoresis, Heidelberg, Germany) was included to stain nuclei. To reduce unspecific background, an incubation step in Vector True View (SP8400; Biozol Diagnostics, Eching, Germany) was included. The specimens were then inspected at a fluorescence microscope (Axioscan, Zeiss, Oberkochen, Germany), and images were taken. For quantification, three random PKP1-positive andnegative areas (1.69 mm 2 ) were identified in prostatic adenocarcinoma and the number of single T-cells, B-cells, neutrophils and macrophages were manually counted. In parallel, in tumour-free samples, three similar-sized squares of these cells were also counted. For counting TLS, the whole image was inspected and the counted TLS were categorized in PKP1 positive, negative or being at the border of both.

Cytokine ELISA
To measure the concentration of CXCL1, IL6 and IL8 in the supernatant of cultured cell lines, the following kits were used: Human CXCL1 (GRO alpha) Human Sim-pleStep ELISA Kit (ab190805), Human IL-6 (interleukin-6) SimpleStep ELISA Kit (ab178013) and Human IL-8 (interleukin-8) SimpleStep ELISA Kit (ab214030). All procedures were performed according to the manufacturer's protocol and performed in technical and biological triplicate (Abcam, Cambridge, UK).

RNA stability assay
For mRNA stability measurements, the cells were treated with 5 lgÁmL À1 actinomycin D (Serva Electrophoresis) and incubated for 15 min, 30 min or 1 h in an incubator at 37°C. RNA isolation and quantitative, real-time PCR were performed as described above. mRNA half-lives were calculated assuming a first-order decay rate. Curves were fitted by linear regression, and mRNA half-lives were calculated as follows: t 1/2 = ln (2)/k [8].
Immunoprecipitation of proteins and proteinassociated RNA For immunoprecipitation analysis, magnetic beads conjugated with secondary protein A beads (both DYNAL Magnetic Beads, Invitrogen) were used. Immunoprecipitation of proteins (IP) has been described in detail [8,51]. Incubation of lysates with beads only was used as a negative control for unspecific binding. RNA was isolated with Trizol (Ambion, Life Technologies, Darmstadt, Germany), transcribed into cDNA (High Capacity cDNA kit; Applied Biosystems) and then further used for RT-PCR. Specific primers were used as mentioned above for quantitative, real-time PCR, and primers for controls have been described in [51] [8].

Gel electrophoresis and western blot
Tissues lysates were generated and analysed by western blot as previously described [13]. Light emission was recorded by a photomultiplier in a gel documentation system (Amersham Imager 680, Cytiva, Marlborough, MA, USA).

Statistical analysis
Values are representative of at least three independent biological experiments including technical replicates. Data are shown as mean AE SD. Statistical comparisons between experimental groups were analysed using unpaired Student's t-test or oneway analysis of variance (ANOVA) with pairwise post hoc comparison according to the Holm-Sidak procedure; a Pvalue < 0.05 was considered to be significant (*P < 0.05; **P < 0.01; ***P < 0.001). The analysis was performed with SIGMAPLOT 13 (Systat Software GmbH, Erkrath, Germany).