Clinical significance of cytoplasmic PML and CRM1 expression: association with poor prognosis

Although previous studies have shown a correlation between the loss of PML expression and prostate cancer progression,^{24, 25} the role of cytoplasmic PML in this process is still unclear. To examine this, the association of CRM1 and PML with cancer-specific mortality and metastasis in a cohort of patients with prostate cancer was investigated by immunohistochemical staining of prostate cancer tissue microarray (TMA). Both proteins were variously located within the nuclei of benign prostate tissue (Figures 2a and c) and in the nuclei and cytoplasm of prostate cancer tissue and were more abundant in the cytoplasm of tumours compared with normal prostate tissue (Figures 2b, d and Supplementary Table 2). Neither nuclear or cytoplasmic expression of PML, nor the expression of cytoplasmic CRM1 protein alone was found to be predictive of reduced disease-specific survival or metastasis (Supplementary Table 2). To test the hypothesis that PML is transported by CRM1, the capacity of PML/CRM1 expression to predict clinical outcome was assessed. Eighty percent (40/50) of patients exhibited either complete loss of PML and CRM1 in tumours or simultaneous expression of cytoplasmic PML and CRM1. This patient group showed reduced cancer-specific survival (χ²=6.47, P=0.011) over a 10-year term (Figure 2e), but not significantly increased metastasis over the same term (χ²=3.108, P=0.078). A Cox proportional hazards regression model demonstrated that simultaneous expression of cytoplasmic PML/CRM1 predicted reduced disease-specific survival that was independent of prostate-specific antigen levels (>4ng/ml) and the Gleason score (>7; hazard ratio=0.264, P=0.009, 95% confidence interval=0.097–0.720).

Open in a separate window

Figure 2

Immunohistochemical localization of PML and CRM1 in prostate cancer patients' TMA. (a) Lack of cytoplasmic PML and (c) CRM1 in benign tissue. (b) Nuclear and cytoplasmic localization of PML and (d) CRM1 in cancer. (e) Kaplan–Meier plot. Scale bars as indicated.

The cytoplasmic fraction of PML promotes cell invasion and EMT

Although, PML and CRM1 do not appear to significantly correlate with metastasis in prostate cancer, the cytoplasmic localization of PML as a result of CRM1 nuclear export may be an early event, which has a role in tumour invasion of adjacent tissues. To investigate the role of cytoplasmic PML on cell migration and invasion, we generated three PMLI haemagglutinin (HA)-tagged mutant constructs: PMLI wild-type HA, PMLI ΔNLS-HA (without the NLS) and PMLI ΔNES-HA (without the NES). The constructs were stably expressed in DU145 and PC3 prostate cancer cell lines using a lentiviral expression system (Figure 3a). Their protein expression pattern was investigated by IB and IF (Figures 3a and b), and the effect of expression on cell migration and invasion determined using wound healing (Scratch assay), well cell migration and well cell invasion assays. Interestingly, DU145 and PC3 cells expressing PMLI ΔNLS exhibited an increased migration and invasive capacities in comparison to cells expressing the empty vector, PMLWT-HA and PMLΔNES constructs (Figures 3c–h and Supplementary Figures 2a and b). Furthermore, knockdown of endogenous PML in DU145 and PC3 resulted in a decreased cell migration. Treatment of these cells with Leptomycin B also affected cell migration (Supplementary Figure 2c). Finally, CRM1 knockdown in DU145-expressing PMLI ΔNLS did not significantly affect cell migration (Supplementary Figure 3d). The increased cell migration and invasion observed in the presence of PMLI ΔNLS was not caused by increased proliferation (Supplementary Figure 3e). These results suggest that the cytoplasmic fraction of PML promotes the migration and invasive capacities of prostate cancer cells.

Open in a separate window

Figure 3

The migration of DU145 and PC3 cells is promoted by cytoplasmic PML. (a) Schematic representation of PMLI mutant constructs (PMLI WT-HA, PMLI ΔNLS-HA and PML ΔNES-HA) and their protein expression profiles as shown by immunoblotting from DU145 and PC3 whole-cell extracts and using an anti-HA antibody. (b) Immunofluorescence images demonstrating stable expression of PMLI mutants in DU145 and PC3 cells using an anti-HA antibody. Scale bar=10μm. (c, d) Wound healing assay (Scratch assay) using DU145 and PC3 cells expressing PMLI mutant constructs and the corresponding data quantifying gap closure at time points 0 and 24h following scratching. Analysis of variance (ANOVA): ****P<0.0001 and ***P=0.0001. Scale bar=200μm. (e, f) Well cell migration assay using DU145 (left panel) and PC3 cells expressing PMLI mutant constructs and the data relating to the number of migrated cells after 24h of culture. ANOVA: **P=0.0018 and ANOVA: ***P=0.0002. (g, h) Cell invasion assay using DU145 (left panel) and PC3 cells (right panel) expressing PMLI mutant constructs and the data relating to the number of migrated cells after 24h of culture. ANOVA: *P=0.0108 and ANOVA: *P=0.0304.

The expression of PML constructs also influenced the cellular morphology of DU145 and PC3 prostate cancer cells. Notably, cells expressing the PMLΔNLS-HA construct exhibited a mesenchymal phenotype, which was characterized by the presence of individual elongated cells. In contrast, DU145 and PC3 cells expressing PML ΔNES-HA exhibited a more epithelial phenotype. The cells expressing the wild-type construct presented an intermediate phenotype, which was probably a result of its dual nuclear and cytoplasmic expression (Figures 4a and b). To confirm these observations, we performed IF on these cells using antibodies that recognize key markers of epithelial cells (E-Cadherin) and mesenchymal cells (Vimentin and and N-Cadherin). Lower E-cadherin expression and increased expression of Vimentin and N-Cadherin were observed in cells expressing PMLΔNLS. On the contrary, cells expressing PML ΔNES-HA, expressed higher levels of E-Cadherin, but a lower expression of Vimentin and N-cadherin. Some of the cells that expressed PMLWT-HA exhibited a higher expression of E-Cadherin, whereas others expressed higher levels of Vimentin and N-Cadherin. To confirm these observations, we performed IB of lysates from cells expressing the empty vector, PML WT-HA, PML ΔNLS-HA and PML ΔNES using antibodies recognizing key markers of epithelial and mesenchymal phenotypes (Figures 4c and d). PMLΔNLS cells expressed lower levels of epithelial markers and higher levels of mesenchymal markers (Figures 4c and d). PMLΔNES expressed higher levels of epithelial markers and lower levels of mesenchymal markers. These results demonstrate that cytoplasmic PML promotes EMT, whereas nuclear PML promotes an epithelial phenotype.

Open in a separate window

Figure 4

Epithelial–mesenchymal transition (EMT) is induced by cytoplasmic PML. (a, b) Micrographs of immunofluorescences on DU145 and PC3 cells expressing PML mutant constructs using antibodies to E-Cadherin (red), Vimentin (green) or N-Cadherin (green). Scale bar=20μm. (c, d) Immunoblotting analysis of whole-cell extracts from DU145 and PC3 cells expressing PML mutant constructs using antibodies against molecules expressed by epithelial cells (E-Cadherin, β-Catenin, ZO-1 and Claudin1) and mesenchymal cells (Vimentin, N-Cadherin, SNAIL, SLUG and ZEB1). β-Actin represents the loading control.

Cell lines, growth conditions, cell fractionation and cell proliferation assay

DU145 and PC3 prostate cancer cell lines were purchased from American Type Culture Collection (Manassas, VA, USA) and cultured in Eagle's Minimum Essential Medium (BE12-6621, Lonza, Basel, Switzerland) and F-12k Nut Mix (1x; 21127-022, Gibco Life Technologies, Waltham, MA, USA) respectively. The media were supplemented with 10% (v/v) fetal calf serum and 1% (w/v) L-glutamine (Lonza). Cells were incubated at 37°C in 5% (v/v) CO₂ and 100% (v/v) humidity. For the relevant experiments, cells were treated for 48h with 3μM of TGFβR1 inhibitor SB431542 (1614, Tocris Bioscience, Moorend Farm Avenue, Bristol, UK) and for 20h with 15ng/ml of Leptomycin B. TGF-β1 was purchased from PeproTech (London, UK) (100-22) and used at 10ng/ml. Cell proliferation was quantified using CyQUANT NF Cell Proliferation Assay Kit (C35006, Invitrogen, Waltham, MA, USA).

Retroviral expression of PML constructs

The PMLI gene fragment was obtained by PCR amplification from pcDNA3.1-PMLI (a kind gift from Dr Mounira Chelbi-Alix, University PARIS V) using forward primer SpeI/PML F′: 5′-CATCTAACTAGTATGGAGAGCCTGCACCCGCCC-3′ and reverse primer PML R′/BamHI 5′-CATCTAGGATCCTCAGCTCTGCTCGGAGGCC-3′ (Eurofins MWG Operon). The fragment obtained was cloned into the pLKOpuro vector using the SpeI and BamHI restriction sites (SHC001, Sigma, St Louis, MO, USA). An HA-tag was introduced in the C-terminal end of PMLI by PCR using QuikChange Site-Directed Mutagenesis Kit (200518, Stratagene, Cedar Creek, TX, USA) and the following primers: HA-tag-sense 5′-CCTCCCAGCAGAGCTACCCTATCGATGTTCCAGATTACGCTTGAGGATCCACCGG-3′ and HA-tag-antisense 5′-CCGGTGGATCCTCAAGCGTAATCTGGAACATCGTATGGGTAGCTCTGCTGGGAGG-3′. The deletions of PMLI NLS and PMLI NES were also performed by PCR using QuikChange Site-Directed Mutagenesis Kit using ΔNLS-sense 5′-GCCCCAGGAAGGTCGGGAAGGAGGCAAG-3′ ΔNLS-antisense 5′-CTTGCCTCCTTCCCGACCTTCCTGGGGC-3′ or ΔNES-sense 5′-ACATTAACAGGCTGTGGGAAGTGCCCGGGGC-3′ ΔNES-antisense 5′-GCCCCGGGCACTTCCCACAGCCTGTTAATGT-3′. The lentiviral particles used to infect DU145 and PC3 with pLKO-PML constructs were produced according to the manufacturer's recommendations. The lentiviral packaging mix was also purchased from Sigma (SHP001).

Stable and transient transfection of cell lines

The transient knockdown of CRM1 and TGFβRI was carried out using an CRM1- or TGFβRI-specific small interfering RNA molecule (CRM1 small interfering RNA (h), sc-35116, Santa Cruz Biotechnology, Inc; TGFβRI small interfering RNA (h), sc-40222, Santa Cruz Biotechnology, Inc) and Interferin transfection reagent (Polyplus, Woodley, Reading, UK) following the manufacturer's recommendations. PML lentiviral plasmids short hairpin RNA control and PML short hairpin RNA 1 and 2 were purchased from Sigma (TRCN0000003866, TRCN0000003869 and SHC001).

Immunoblotting and immunoprecipitation

The experimental procedures were performed as described previously.³⁰ For IB, the cells lysed in 1 × solution containing 50mM Tris-HCl (pH 6.8), 100mM dithiothreitol, 2% (w/v) SDS, 0.1% (w/v) bromophenol blue and 10% (v/v) glycerol, and loaded onto Tris/glycine SDS–polyacrylamide gels for electrophoresis. The proteins were transferred onto Amersham Hybond-P PVDF membranes (GE Healthcare, Chalfont St Giles, Buckinghamshire, UK). Membranes were blocked with 10% (w/v) Marvel milk powder/tris-buffered saline solution with 0.1% (v/v) Tween-20. Following washes in tris-buffered saline solution with 0.1% (v/v) Tween-20, membranes were incubated with primary antibodies (in blocking solution) overnight at 4°C followed by washing and incubation with secondary antibodies for 1h at room temperature before visualization using Rapid Step ECL reagent (Calbiochem, San Diego, CA, USA) and a CCD camera (Fujifilm, Martins Way, Bedford, UK). For IP, the cells were lysed in IP buffer (0.05M Tris, pH 7.4, 0.15M NaCl, 0.5% (v/v) Triton X-100 and 0.001M EDTA). The extracts were pre-cleared for 1h at 4°C using Protein G Plus–Agarose beads (IP08, Calbiochem) and corresponding IgG. The PML antibody was incubated with agarose beads in IP buffer for 2h before adding the pre-cleared extracts for overnight IP. The beads were washed four times in IP buffer and suspended in lysis buffer, after which samples were analysed by mass spectrometry and IB.

Immunohistochemistry (IHC) and immunofluorescence

Paraffin-embedded human prostate cancer TMAs were purchased from US Biomax (Rockville, MD, USA). The sections were treated and stained as previously reported.³⁰ For IHC, 3 3' diaminobenzidine staining was performed using Dako Real Detection Systems LSAB⁺ (K5001, Agilent Technologies, Santa Clara, CA, USA).

IHC for PML and CRM1 expression in patient's prostate cancer samples

Liquid chromatography-matrix-assisted laser desorption/ionization time-of-flight mass spectrometry identification of PML-interacting partners

Following IP, identification of the potential interacting partner proteins was carried out by LC-MALDI mass spectrometry. Samples were tryptically digested and fractionated by nano-LC (C₁₈ PepMap100, 150mm × 75μm; 5μm particle size) before spotting in 10s fractions over a 384-spot target plate. Analysis was carried out on a Bruker UltrafleXtreme MALDI-TOFTOF mass spectrometer (Bruker Daltonics, Banner La, Coventry, UK) in reflectron-positive mode over a m/z range of 650–3500. Tandem mass spectrometry fragmentation of the top 10 precursors per spot was carried out and results exported using Biotools 3.1 software (Bruker Daltonics) for searching via Mascot 2.1 against the human Swissprot database.

This work was funded by the John and Lucille van Geest Foundation and Cancer and Polio Research Fund. We thank Mrs Anne Schneider, Mr Stephen Reeder, Mr Chris Nolan and Dr Claire Hawkes for their assistance in this study.