A basal cell defect promotes budding of prostatic intraepithelial neoplasia

The normal prostate gland is a simple secretory epithelium containing a basal cell population [which can be detected by analyzing for the presence of high-molecular-weight cytokeratin (HMWCK)] harboring stem cells (Bonkhoff, 1996; Bonkhoff and Remberger, 1996; Bostwick, 1996a,b) and a luminal cell population [detected by racemase (AMACR) staining] that secretes PSA (also known as KLK3) (Thomson and Marker, 2006). During normal glandular development, extracellular matrix (ECM)–cell-receptor interaction provides contextual cues and a developmental ‘morphogenesis checkpoint’ for ordered repopulation (Brown, 2011). Cell divisions parallel to the basal cell surface maintain proximity to the ECM and control mitotic spindle orientation during epithelial morphogenesis and repair (Xia et al., 2015).

During the early stages of prostate cancer progression, a defective glandular structure forms, called high-grade prostatic intraepithelial neoplasia (HG-PIN), which is defined by focal loss or attenuation of the basal cell layer and ECM (Nagle et al., 1994), and loss of both integrin α6β4 expression and its the corresponding ECM ligand laminin-332 (i.e. laminin comprising the α3Aβ3γ2 chains) (Cress et al., 1995; Davis et al., 2001; Hao et al., 1996; Nagle et al., 1995; Pontes-Junior et al., 2009). HG-PIN often contains a mixture of basal and luminal cell markers, consistent with a loss of normal contextual cues and mitotic spindle misorientation (Bonkhoff and Remberger, 1996) as seen during regrowth of prostate following castration and androgen reintroduction (Verhagen et al., 1988). HG-PIN has genomic instability (Haffner et al., 2016; Iwata et al., 2010; Mosquera et al., 2009, 2008; Nagle et al., 1992; Petein et al., 1991) and is a precursor of invasive prostate cancer (Bonkhoff and Remberger, 1996; Bostwick, 1996a,b; Bostwick et al., 1996; Haggman et al., 1997; Montironi et al., 1996a,b; Montironi and Schulman, 1996).

Here, we report a new three-dimensional (3D) HG-PIN-type model using two different isogenic human prostate epithelial cell lines, called RWPE-1 (Bello et al., 1997; Roh et al., 2008; Webber et al., 1997) and PrEC 11220, with a stable modification to deplete β4 integrin expression. The model provides a means to test consequences of basal cell defects in human HG-PIN progression.

Normal human prostate glands contain an ordered basal and luminal cell distribution, as shown in Fig. 1A,B. Integrin α6β4 is found within the basal cell layer (Fig. 1C) and is required for anchoring basal cells to laminin-332 ECM through the hemidesmosome (Nagle et al., 1995; Pulkkinen and Uitto, 1998; Wilhelmsen et al., 2006). In contrast, HG-PIN (Fig. 1A,B) contains cells with enlarged nuclei and prominent nucleoli that proliferate within the lumen, enlarging the glands, resulting in continuity gaps (Nagle and Cress, 2011). In these gaps, laminin-332 and α6β4 integrin – essential requirements for functional hemidesmosomes – are absent; HG-PIN and cancer lesions are known to lack basal cells and laminin-332 deposition, becoming exposed to laminin-511 (i.e. laminin comprising the α5β1γ1 chains) within the muscle stroma (Davis et al., 2001; Nagle and Cress, 2011; Nagle et al., 1995). Defective β4 integrin function results in defective laminin-332 assembly (Yurchenco, 2015) and laminin-511 is a known potent morphogen essential for embryonic development (Ekblom et al., 1998). We observed extensive budding of cell clusters through β4 integrin gaps and into the stroma in HG-PIN (Fig. 1D, asterisk). Hence, loss of α6β4 integrin is associated with abnormal outgrowth of the epithelium in human HG-PIN.

Given the importance of the glandular structure for maintaining homeostasis (Brown, 2011; Chandramouly et al., 2007; Wilhelmsen et al., 2006; Xia et al., 2015) and the role of the basal cell stem cell compartment (Bonkhoff and Remberger, 1996; Bostwick, 1996b; Hudson, 2004; Karthaus et al., 2014; Schmelz et al., 2005), we generated a 3D model of prostate glands using two different primarily diploid prostate epithelial cell lines, RWPE-1 and PrEC 11220. RWPE-1 is a stable non-tumorigenic human cell line derived from a histologically normal adult human prostate (Bello et al., 1997; Roh et al., 2008; Webber et al., 1997) and is used as a normal prostate cell line in genomic and differentiation studies (Wang et al., 2011; Webber et al., 1997). We note that RWPE-1 was isolated from a donor undergoing cryoprostatectomy and might carry abnormalities. The cell line was immortalized by HPV-18 and has characteristics of an intermediate cell type (Verhagen et al., 1992) expressing both basal and luminal cell cytokeratin, androgen receptor and PSA in response to androgen (Bello et al., 1997). For comparison to RWPE-1 cells, we used another normal human prostate cell line (PrEC 11220), immortalized by hTERT, to test spheroid formation (Dalrymple et al., 2005; Salmon et al., 2000). We found both cell lines formed compact multicellular spheroids as represented by RWPE-1 (Fig. 2A) and observed a dramatic cell-spinning phenotype as spheroids formed from a single cell (Movies 1 and 2). The rotational motion was similar to mammary epithelial acini that coordinate rotational movement with laminin matrix assembly (Wang et al., 2013). Most (65%) of the RWPE-1 spheroids contained a lumen after 10 days, as measured by presence of β-catenin on the cell surface and absence of DAPI staining nuclei in the interstices of the spheroid (Fig. 2B,C). Positive detection of four different molecular markers [laminin-332 production and organization, α6β4 integrin expression, HMWCK and p63 (also known as TP63)] confirmed the presence of basal cells within spheroids created by both RWPE-1 and PrEC 11220 cell lines (Fig. 2D). Spheroid production was seen in both cell lines and, hence, was independent of the immortalization method. The RWPE-1 cells were superior in forming a continuous and well-circumscribed layer of laminin-332 and continuous β4 integrin distribution as compared to the PrEC 11220 cells (Fig. 2D). The PrEC 11220 cells contained a punctate yet circumscribed distribution of β4 integrin, consistent with the dynamic processes expected when the basal lamina forms as directed by β4 integrin clustering on basal cells (Yurchenco, 2015). Taken together, the data suggest that both cell lines assemble laminin-332 and β4 integrin on the spheroid surface. The punctate distribution observed with PrEC 11220 spheroids suggests that it could be useful for tracking dynamic β4-integrin-directed assembly of the basal lamina. In contrast, RWPE-1 spheroids assembled a robust and tight laminin-332 and β4 integrin layer, which could be used as a stringent test for determining the influence of contextual signals on the invasive budding in HG-PIN. RWPE-1 spheroids contained functional luminal cells as increased PSA secretion could be detected in response to dihydrotestosterone (DHT) treatment compared to untreated controls (Fig. 2E,F).

We next measured the accuracy of mitotic spindle orientation in basal cells from RWPE-1 spheroids. The RWPE-1 spheroids were used as these have a highly ordered laminin-332 layer as a reference point. Mitotic spindles and spindle poles were visualized by staining for α-tubulin and the centriole protein CEP135; spindle orientation in basal cells was measured relative to laminin-332 deposition (Fig. 2G; Movie 3). Approximately 67% of mitotic spindles in basal cells were oriented parallel to or within 30° of the laminin-332 layer resulting in new cells that were side-by-side (Fig. 2H). The ability to measure the distribution of centrosomes and spindle orientation will allow testing of genetic, signaling and contextual determinants within basal cells that promote invasive human HG-PIN.

We generated two stable RWPE-1 cell lines expressing one of two different short hairpin RNAs (shRNAs) to silence β4 integrin (sh-β4-1 and sh-β4-2) expression to test the transition to a HG-PIN phenotype. The RWPE-1 spheroids were used as these had a highly ordered laminin-332 layer that serves as a stringent test for determining the influence of β4 integrin depletion. Western blotting confirmed efficient depletion of β4 integrin (Fig. 3A). Spheroids made with β4-integrin-expressing cells displayed a continuous layer of laminin-332 co-distributing with β4 integrin [Fig. 3B,C, wild-type (WT)]. In contrast, depletion of β4 integrin resulted in a discontinuous layer of laminin-332 (Fig. 3C, sh-β4, arrowheads). As expected, β4 integrin depletion did not affect the abundance of laminin-332 (Dowling et al., 1996; Georges-Labouesse et al., 1996), as confirmed by western blotting (Fig. 3D). Approximately 65% of β4-integrin-depleted spheroids contained a discontinuous laminin-332 layer as compared to 19% of β4-integrin-containing control spheroids (Fig. 3E). β4 integrin depletion did not alter the ability of cells to form spheroids or produce laminin-332 but altered the organization of laminin-332, creating a discontinuous layer. We speculate that a discontinuous laminin-332 layer will disrupt normal glandular homeostasis and give rise to spindle misorientation to facilitate genomic instability, tissue disorganization, metastasis and expansion of cancer stem cell compartments (reviewed in Pease and Tirnauer, 2011). Therefore, this model can be used to determine whether mechanisms that disrupt laminin-332 continuity [whether it be defects of the essential adhesion receptor (β4 integrin), defective production of laminin-332 by the stroma or loss of normal basal cell layer by luminal expansion] will stimulate the HG-PIN process.

Given that discontinuities in the basement membrane can induce cell invasion and dissemination (Nguyen-Ngoc et al., 2012), we used live imaging of β4-integrin-depleted spheroids to investigate phenotypic changes. Strikingly, we observed prominent budding in β4-integrin-depleted spheroids in regions of a discontinuous laminin-332 layer that was not observed in controls (Fig. 4A; Movies 4–6). The budding spheroids contained α6 integrin on their surfaces (Fig. 4B), which in the absence of the β4 integrin subunit, is the α6β1 integrin. On day 10, budding of the epithelial cells was observed in β4-integrin-depleted spheroids (Fig. 4C) at a significantly higher level than in WT RWPE-1 spheroids (Fig. 4D). The loss of β4 integrin expression and persistence of α6β1 expression is consistent with the switching of the heterodimer composition during prostate cancer progression (Cress et al., 1995).

Taken together, our data indicate that loss of β4 integrin expression generates laminin-332 continuity gaps through which basal cells exit the spheroid. The depletion of β4 integrin expression in normal prostate basal cells creates a new HG-PIN-type model that can be used to determine which microenvironmental or contextual cues, and somatic mutations, promote HG-PIN progression.

Formalin-fixed de-identified human cancer tissues were stained using a Discovery XT Automated Immunostainer (Ventana Medical Systems, Inc., Tucson, AZ) in a core support service within the UA Cancer Center. The use of de-identified human tissue was in accordance with the guidelines and approval of the University of Arizona Institutional Review Board. Informed consent was obtained for all tissue donors and that all clinical investigations have been conducted according to the principles expressed in the Declaration of Helsinki. Samples were imaged using an Olympus BX40 system (Southwest Precision Instruments, Tucson, Arizona, USA) with a 4× (NA 0.13) and 40× objective (NA 0.75).

The lentiviral vectors pGFP-C-shLenti shRNA-29 (Addgene, TL312080) containing a GFP reporter were used. Cells were infected with lentivirus containing the scrambled shRNA (Addgene, TR30022) as control, and shRNA hairpin against β4 integrin. Sequences of shRNAs are 5′-GUACAGCGAUGACGUUCUACGCUCUCCAU-3′ (sh-β4-1) and 5′-CCGUAUUGCGACUAUGAGAUGAAGGUGUG-3′ (sh-β4-2). Transfected cells were selected in 500 ng/ml puromycin and sorted by flow cytometry [BD Flow Machine FACSAria III; UACC Flow Cytometry shared resource (FCSR)] using the GFP signal.

RWPE-1 cells were obtained from the American Type Culture Collection (ATCC, CRL­11609TM), and were maintained at 37°C in a 5% CO₂ atmosphere, and cultured in Iscove's modified Dulbecco's medium (IMDM; Corning, cat. no. 10-016-CV), supplemented with 10% fetal bovine serum containing detectable levels of testosterone and corticosteroids (FBS, Seradigm, Lot 294R13), 100 IU penicillin, 100 μg/ml streptomycin and 0.25 μg/ml amphotericin (MP Biomedicals, cat. no. 1674049). RWPE-1 identity was verified by assessing the allelic signature of 15 different genetic markers. Immortalized human PrEC 11220 cells were generously provided by John Isaacs (The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University, Baltimore, MD) (Dalrymple et al., 2005; Salmon et al., 2000). These cells were immortalized by sufficient basal hTERT expression. The cells have a normal human male karyotype and grow in KSFM (Life Technologies, ref. 17005-075), 100 IU penicillin, 100 μg/ml streptomycin and 0.25 μg/ml amphotericin (MP Biomedicals, cat. no. 1674049).

For 3D culture, the RWPE-1 cells were maintained as described previously (Tyson et al., 2007). For DHT treatment, the medium was supplemented with 5 nM DHT (Sigma-Aldrich, D-073). Growth-factor-reduced matrigel (BD Biosciences, lot no.5173014) was used from a single lot with protein concentrations between 10 and 11 mg/ml. 3D indirect immunofluorescence microscopy was performed as described previously (Debnath et al., 2002; Klebba et al., 2013). Statistical analysis was performed using Graphpad Prism 6. Antibodies used were: anti-α-tubulin FITC-conjugated DM1a (1:100; Sigma F2168), anti-laminin-332 (1:200; Abcam ab14509), anti-ZO1 (1:200; Life Technologies ZO1-1A12), anti-β-catenin (1:100; Cell Signaling 9562), anti-β4 integrin (1:200; EPR 8558), anti-HMWCK (1:200; DAKO 34βE12), anti-p63 (1:100; Biorbyt orb214808), anti-PSA (1:100; Cell Signaling D6B1), anti-CEP135 (1:100; Abcam ab196809), and anti-α6 integrin (1:100; J1B5). For live-cell imaging, cells were grown on eight-well tissue culture Lab-Tek chambered coverglass (Thermo Fisher Scientific, cat. no. 155411) for 3D imaging. Cells were imaged as above under optimal growth conditions. Images were captured every 15 min for 2–3 days.

Cell extracts were produced by lysing cells in cold RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% v/v Triton X-100, 1% w/v sodium deoxycholate, 0.1% SDS) as described previously (Klebba et al., 2013). Antibodies used included anti-β4 integrin (1:1000; EPR 8558), anti-laminin-332 (1:1000; Abcam ab14509), anti-α-tubulin (1:1000; Sigma DM1a) and IRDye 800CW secondary antibodies (1:1500; Li-Cor Biosciences).

The gift of the normal human prostate cell line (PrEC 11220) by John T. Isaacs is gratefully acknowledged. We thank the staff of the TACMASR for the tissue processing and staining and the FCSR service for cell sorting, Dan Buster for assistance with image analysis and Cynthia S. Rubenstein for assistance with 3D culture and time-lapse microscopy. We thank William L. Harryman for the editing assistance.