- Top of page
- MATERIAL AND METHODS
- Literature Cited
Implanted human, androgen-independent prostatic carcinoma cells (DU145) into athymic (NCr nu/nu) mice produce diverse tumors on the peritoneal surfaces of many organs. Light and ultrastructural observations show that the mesothelial covering these surfaces are typically microvilli-coated, squamous cells or secretory cuboidal cells. The peritoneal regions colonized by tumors lack mesothelial cells and are covered by actively replicating carcinoma cells that grow as poorly differentiated cell clusters made of cell aggregates to somewhat compact spheroids covered with pleiomorphic microvilli and containing an undifferentiated vascular supply. These xenografts clusters invade the diaphragm and develop into tumors with both a basal solid aspect and an upper region of cribriform morphology. Furthermore, each tumor contains two cell types: (1) a poorly differentiated clear cell type, which grows into intraperitoneal tumors and (2) a large, basophilic cell type, which invades the peritoneal stroma of organs, including of the diaphragm. Anat Rec, 2013. © 2012 Wiley Periodicals, Inc.
Prostate carcinoma is one of the most pervasive and pestilent malignancy encountered in the human males. The frequency and mortality rate of prostate cancer has increased over the past 40 years and is expected to rise steadily in subsequent years (Partin and Coffey, 1994; Turner et al., 1996). In the United States alone an estimated of 218,000 new cases and 32,000 deaths were attributed to prostate cancer in 2010 (Jemal et al., 2010). Males of African descent are almost twice as likely to develop prostate cancer as male Caucasians, while Asian males are half as likely to develop prostate cancer as Caucasian males (Jemal et al., 2010; Cancer Research UK, 2011; Chornokur et al., 2011).
Although prostate cancer in its early stages is responsive to standard treatments, patients with hormone-refractory prostate cancer have an overall median survival of 9–18 months, and no currently available treatment produces a survival advantage (Chen et al., 1998; Feldmann et al., 2000; Jamison et al., 2005). Deaths related to prostate cancer are invariably due to tumor invasion and metastasis to the lymph nodes, lungs, and bone marrow (Turner et al., 1996). Consequently, efforts to improve our understanding of the progression to the invasive and metastatic stages of prostate cancer should enhance the chances for developing meaningful therapeutic approaches.
An intraperitoneal xenograft model has often been employed to assess the invasiveness of human androgen independent prostate carcinoma cells (DU145) into the diaphragm (Turner et al., 1996, 1997; Fizazi and Navone, 2005). This model of invasion is used because of its reproducibility, it is quantifiable, and it correlates strongly with invasion from orthotopic tumor growth (Mamoune et al., 2004). While this model was originally employed to evaluate the metastatic mechanism of prostate tumor in nude mice (Kubota, 1994), the model has recently been employed to elucidate the molecular pathways involved in invasion (Turner et al., 1996, 1997; Chunthapong et al., 2004; Yue et al., 2010; Rybak et al., 2011) as well as the ability of drugs to modulate this invasion (Darnowski et al., 2004). However, in these studies, the accompanying imaging has been performed primarily at the level of light microscopy (LM). In this study, light, scanning, and electron microscopy techniques have been performed to characterize the colonization and invasion of the diaphragm and other peritoneal sites by DU145 cells. This investigation also demonstrates that two types of tumor morphology and invasive cell populations develop from the DU145 cell line originally implanted in nude mice. It is also illustrates that the original epithelium of the peritoneal surfaces is displaced and replaced by the DU145 cells of the growing tumors.
- Top of page
- MATERIAL AND METHODS
- Literature Cited
The mesothelial surfaces in the nude mice are similar to those in humans (Leak and Rahil, 1978; Slater et al., 1989; Jonecko, 1990), other animals including mice (Carr et al., 1969; Watters and Buck, 1972; Schwarz, 1974; Baradi and Rao, 1976; Guggenheim et al., 1979; Hjelle et al., 1989; Gaudio et al., 1990; Ettarh and Carr, 1996; Bot et al., 2003; Michailova and Usunoff, 2006; Yung et al., 2006; Yung and Chan, 2007). The epithelial cells show secretory activity with the apical release of lamellar bodies analogous to those of alveolar or pneumocytes type II (Dobbie and Lloyd, 1989; Dobbie and Anderson, 1996a,b) as well as excrescences and expelled material similar to those observed in the gallbladder epithelium (Gilloteaux et al., 1993). Furthermore, the morphology of the epithelial cells of the mesothelium may vary in accordance with changes in their metabolic activity or following cell injury (Li and Jiang, 1993; Li et al., 1996; Michailova, 1997; Mutsaers, 2002; Yung et al., 2006; Yung and Chan, 2007).
All the mice implanted with DU145 cells possessed highly disseminated tumors on multiple parietal and visceral peritoneal surfaces. SEM and TEM of tumor bearing mice demonstrate the uniform morphology of the cobblestone-like, microvilli-coated epithelial cells of the mesothelium have been displaced by DU145 through epithelial damage (Groothuis et al., 1988; VanderWal et al., 1997), amoeboid characteristics of the tumor cells (Koga et al., 1980; Nakamura, 1994), cell surface modifications (Abercrombie and Ambrose, 1962; Abercrombie, 1967), lytic enzymes, or increased proliferation and movement mediated through paracrine activities of the adjacent tumor or stromal cells (Vo and Khan, 2011) such as hepatocyte growth factor (HGF) (Nishimura et al., 1998, 1999; Mamoune et al., 2004; Oosterhoff et al., 2005; Cai et al., 2008; Wang et al., 2010). A similar process of invasion was reported in the case of ovarian tumor cells in vitro (Iwanicki et al., 2011; Kenny et al., 2011). The presence of stellate-shaped microvilli is unique because they are seen only in some rare human seminomas (Tannenbaum, 1979).
LM of random hematoxylin and eosin stained cross sections of the majority of tumors reveal a mixed architecture (Fig. 1A–C). The obvious question is how did this morphological dichotomous tumor arise from a single tumor cell population?
The DU145 cell line is an androgen independent cell line derived from prostate carcinoma cells of a brain metastatic site of a 69-year-old, male, Caucasian patient (Stone et al., 1978). LM of the DU145 cell line shows pleiomorphic cells with at least two distinct phenotypes, elongated and polygonal (Peehl, 1994) and a small population of stem cells (Pfeiffer and Schalken, 2010). After serial passage of these cells through basement membrane-coated membranes, Chunthapong et al. (2004) were able to isolate a poorly invasive population with epithelial-like morphology (DU145-E) and a second, highly invasive population with elongated, fibroblastic-like morphology (DU145-F). These DU145-E cells possessed many filopodial extensions from the cell surface and many prominent infoldings or plicae and long, microvillar projections, while the DU145-F cells were more elongated, exhibited few if any cytoplasmic extensions, possessed filopodia extending from the leading edges of migrating cells and exhibited membranous folds and short microvilli on their cell surfaces.
E-cadherin is an important regulator of invasion in prostatic carcinomas (Morton et al., 1993; Lu, et al., 2003). DU145-E cells express a similar amount of E-cadherin as the parental DU145 cells, while the DU145-F cells express <0.1-fold of E-cadherin than the parental DU145 or DU145-E populations (Chunthapong et al., 2004). The levels of two E-cadherin-associated proteins, β-catenin and p120cas, also decreased, while the protein levels of cytokeratin 18 (an epithelial marker) and vimentin (a mesenchymal marker) remained constant in both populations and higher extracellular levels of pro-matrix metalloproteinase-2 (pro-MMP-2) were found in the DU 145-F population. This decreased expression of E-cadherin and increased vimentin expression was confirmed in DU145 cells forming subcutaneous tumors in SCID mice (Luo et al., 2006; Zhao et al., 2011) and suggest the DU145-F cells have undergone an epithelial-mesenchymal transition (EMT) to promote invasion and metastasis. Cancer cells may pass through EMTs to differing extents, with some cells retaining many epithelial traits while acquiring some mesenchymal ones and other cells shedding all vestiges of their epithelial origin and becoming fully mesenchymal (Kalluri and Weinberg, 2009).
As mentioned previously, the epithelial cells of the intact mesothelial surfaces (Figs. 2A–C, 3A–E, 4A,B, 8C,D) are displaced and replaced fibroblast-like tumor cells with very few microvilli (Fig. 5A,B). This process has been described in general by McCandless et al. (1997). The morphology of these cells closely resembled the DU145-F population of the parental DU145 cell line whose level of E-cadherin was decreased 10-fold while their extracellular levels of metalloproteinases were elevated. These fibroblast–like cells are believed to have undergone an EMT in an effort to enhance their migratory capacity, invasiveness, and to elevate their resistance to apoptosis (Kalluri and Weinberg, 2009). Paradoxically, these EMT-derived migratory cancer cells typically establish secondary colonies at distant sites that resemble the primary tumor from which they arose (Zeisberg et al., 2005; Coulson-Thomas et al., 2010). This observation suggests that the metastasizing cancer cells may shed their mesenchymal phenotype via a mesenchymal-epithelial transition (MET) during the course of secondary tumor formation (Zeisberg et al., 2005). If the fibroblast-like cells seen in our images maintain some of their epithelial and mesenchymal properties like the DU145-F population of the parental cell line, a MET is not too difficult to conceive.
Figure 8A shows a cluster of pale-staining DU145 cells surrounding a basophilic cell (a putative DU145 cancer stem cell) proximal to the mesothelial surface and suggests the interesting possibility that, in this model, successful metastasis may require the presence of both fibroblast-like cells (DU145-F) and one or more tumor stem cells. This could help explain why usually <100 tumors are formed from the 5 × 106 DU145 cells that are injected. If indeed, DU145 stem cells represent 0.01% of the population and the plating efficiency of these in forming holoclones in vitro is 35% (Pfeiffer and Schalken, 2010), one would expect a maximum of 175 colonies (5,000,000 × 0.0001 × 0.35 = 175). If it is assumed that the plating efficiency in vivo is only 1/3 as effective as in vitro, a maximum of 60 colonies (5,000,000 × 0.0001 × 0.12 = 60) is expected from an injection of 5 × 106 DU145 cells. Regardless of the plating efficiency, the basal regions of the resulting tumor surrounding the putative stem cell strongly resemble the holoclones produced in the DU145 experiments (Barrandon and Green, 1987). Subsequently, these basophilic cells develop into small nests, clusters, and then nodules, which invade the submesothelial region and then the muscle stroma. Conversely, the paraclones are highly irregular in shape and contain more flattened and scattered cells that resemble the morphology of the upper portions of the tumors that protrude into the peritoneal cavity and are composed of less intimately associated cells with crypts that announce a future adenocarcinomatous architecture.
The dichotomous morphology in the most advanced tumors results from a process akin to prostatic intraepithelial neoplasia (PIN) (McNeal, 1969; Bostwick et al., 1993; Helpap and Riede, 1995; Bostwick and Qian, 2004), studied morphologically by Kastendieck and Altenähr (1976) and recently reviewed by several authors (Montironi et al., 2000; Shin et al., 2000; Latour et al., 2008; Epstein, 2009; Dema et al., 2010; Shah et al., 2010). Prostatic lesions show isolated voids that appear as “sieve holes” within the tumors and evolve into a “cribriform acinar prostate carcinoma” through programmed cell death (Nagao et al., 2003; Shah et al., 2010). These areas can then evolve into adenoid or adenocarcinomatous aspect (Epstein, 2009; Dema et al., 2010) by increasing their number and enlargement of the voids (Figs. 1B, 7C,D, 10A,E) and have occurred (Fig. 10A,B,D) as the result of entotic deaths (Brouwer et al., 1984; Overholtzer et al., 2007). As noted, these cribriform structures are initially distributed in the upper regions of the DU145 carcinomas. They have been further described and discussed in other studies (Gilloteaux et al., 2012, submitted for publication).
Tumor growth and metastasis are dependent on the tumor's supply of blood vessels (Folkman, 1995). In this study, LM show blood-containing capillaries lined by endothelial cells (Figs. 7B,D, 9C–E) as well as blood-containing capillaries that appear to be lined by tumor cells without organized endothelium (Figs. 7C, 9B,E). SEM images (Figs. 6D, 7A) suggest these capillaries arise from a tumor cells that strongly resemble the DU145-F population. The chain-like structure of tumor cells that runs from the upper left to the lower right of Fig. 6D gives rise to a capillary-like structure in Fig. 7A. To explain these observations one needs to examine the process of tumor vascularization.
The architecture of tumor vasculature is different from normal vasculature because of its incomplete endothelial lining (Steinberg et al., 1990; McDonald and Foss, 2000). Tumor tissues progressively become hypoxic and necrotic due to rapid proliferation and insufficient blood supply and must constantly make new vasculature (Furuya et al., 2005). Angiogenesis is one method of developing vasculature with angiogenin expression being elevated in prostate cancer cells and their stem cells (Katona et al., 2005). Analysis of the tumor vascular bed of human prostate carcinomas by quantifying microvessel density count, proliferating capillary index, proliferating tumor versus endothelial cell index; and microvessel pericyte coverage index (MPI) indicates that prostate tumors are not very angiogenic and yet only one-third of the vasculature within these tumors is covered by pericytes (Eberhard et al., 2000). These observations suggest the majority of prostate tumor vasculature is being generated by a process other than angiogenesis.
Other mechanisms of tumor vascularization include vessel co-option, intussusception, recruitment of endothelial precursor cells (EPCs) and vasculogenic mimicry (VM) (Paulis et al., 2010). VM describes the ability of tumor cells to express multiple cellular phenotypes, gain endothelial characteristics, and form vascular-like networks (Maniotis et al., 1999; Folberg and Maniotis, 2004). In prostate tumor cells, VM involves cooperative interactions of distinct phenotypic subpopulations (Sharma et al., 2002) and higher expression of the basement membrane extracellular matrix components laminin5γ2, and metalloproteinase (MMP)-1, −2, −9, and −14 [membrane type (MT)1-MMP], which act cooperatively to form tubular networks of tumor cells, without endothelial cells or fibroblasts (Seftor et al., 2001; Kaminski et al., 2006). In this study, VM appears to be driven by the DU145-F population alone or in combination with the DU145 stem cell population (Jones et al., 2012). The DU145-F cell population fits the morphological profile and the genotypic profile in that it possesses both epithelial and mesenchymal markers, expresses EphA2, MT1-MMP and produces elevated extracellular levels of pro-MMP-2.