SEARCH

SEARCH BY CITATION

Keywords:

  • CD133;
  • Human;
  • Ovarian cancer;
  • Cancer stem cells;
  • Tumor-initiating cells

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Evidence is accumulating that solid tumors contain a rare phenotypically distinct population of cells, termed cancer stem cells (CSC), which give rise to and maintain the bulk of the tumor. These CSC are thought to be resistant to current chemotherapeutic strategies due to their intrinsic stem-like properties and thus may provide the principal driving force behind recurrent tumor growth. Given the high frequency of recurrent metastasis associated with human ovarian cancer, we sought to determine whether primary human ovarian tumors contain populations of cells with enhanced tumor-initiating capacity, a characteristic of CSC. Using an in vivo serial transplantation model, we show that primary uncultured human ovarian tumors can be reliably propagated in NOD/SCID mice, generating heterogeneous tumors that maintain the histological integrity of the parental tumor. The observed frequency of tumor engraftment suggests only certain subpopulations of ovarian tumor cells have the capacity to recapitulate tumor growth. Further profiling of human ovarian tumors for expression of candidate CSC surface markers indicated consistent expression of CD133. To determine whether CD133 expression could define a tumor-initiating cell population in primary human ovarian tumors, fluorescence-activated cell sorting (FACS) methods were employed. Injection of sorted CD133+ and CD133 cell populations into NOD/SCID mice established that tumor-derived CD133+ cells have an increased tumorigenic capacity and are capable of recapitulating the original heterogeneous tumor. Our data indicate that CD133 expression defines a NOD/SCID tumor initiating subpopulation of cells in human ovarian cancer that may be an important target for new chemotherapeutic strategies aimed at eliminating ovarian cancer. STEM CELLS 2009;27:2875–2883


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Ovarian cancer is the most lethal form of gynecological cancer, due in part to the lack of a reliable early detection method resulting in the majority of patients being diagnosed at an advanced stage of disease. For these patients, surgery alone is insufficient given the propensity of this tumor type to metastasize. Unfortunately, most affected women will develop recurrent disease that is less effectively treated with current chemotherapies.

Recently, specific subpopulations of cells with stem-like properties, termed cancer stem cells (CSC), have been identified in solid tumors of various origins [1–12]. Given the intrinsic properties of these CSC, including self-renewal capacity and increased chemoresistance, it is thought that they may represent the principal driving force behind solid tumor growth and metastasis.

To date, the isolation of these putative tumorigenic CSC has been achieved based on the expression of specific cell surface markers that distinguish these rare cell subpopulations from the main tumor bulk. Either alone or in combination, these markers have been successful in identifying tumorigenic populations in primary human breast (CD44+/CD24−/low/Lin [1]; CD44+/CD24−/low [8]), brain (CD133+ [10]), pancreatic (CD44+/CD24+/ESA+ [6]; CD133+/CXCR4+ [5]), colon (CD133+ [7, 9]; CD44+/ESA+/CD166+ [3]), prostate (CD44+2βmath image/CD133+ [2]), lung (CD133+ [4]), head and neck (CD44+ [12]) and skin (CD133+/ABCG2+ [11]) cancer.

Several studies have indicated that CSC may also perpetuate ovarian tumor growth and metastasis. Bapat and colleagues [13] showed that single clones isolated from the ascites of an ovarian cancer patient possessed stem-like properties and were both clonogenic in vitro and tumorigenic in vivo. Also, in mouse ovarian cancer cell lines, a side population (SP) of cells was identified and was tumorigenic in vivo [14]. Similarly, Baba and colleagues [15] identified tumorigenic populations of cells from human ovarian cancer cell lines based on CD133 expression. In a recent study using primary human ovarian tumor tissue, tumor cells were propagated as anchorage-independent spheroids in specialized media conditions originally designed for maintaining neuronal stem cells [16]. Using this methodology, a rare stem-like population of CD44+/CD117+ cells was identified as an ovarian cancer-initiating population capable of recapitulating histologically similar tumors in vivo. Subsequently, using similar methodologies, CD44+/MyD88+ cells were found to be an ovarian tumor cell population with some stem cell-like properties [17]. Although these studies are highly informative, the propagation of cells under such media conditions may not be ideal, as the in vitro culture environment may influence surface antigen expression based upon the media components present. Such a discrepancy in marker expression using different dissociation methodologies has already been described in human breast cancer studies with respect to ESA (EpCAM) expression [1, 8]. Additionally, Ince and colleagues [18] used differential culture methods to isolate two separate cell populations from transformed mammary epithelial cells that had been derived from the same initial source. The two populations had differential tumorigenic potential following injection into immunocompromised mice and generated histologically different tumors with distinct metastatic potential.

The present study using an in vivo model of direct propagation of uncultured human primary ovarian tumors in NOD/SCID mice provides definitive evidence that a subpopulation of cells in these malignancies possesses a greater capacity to form tumors. We provide a putative CSC marker profiling of primary human ovarian tumors and human ovarian tumors propagated in vivo in the absence of spheroid formation. Taken together, our results provide evidence that CD133 expression defines a population of tumorigenic cells that may contribute to human ovarian tumor growth and recurrent disease.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Primary Human Ovarian Tumor Cell Isolation and Xenograft Procedure

All primary human ovarian tumor biopsies were collected in accordance with the policies of the Massachusetts General Hospital (MGH) Institutional Review Board. All isolated ovarian tumors were pathologically characterized (supporting information Table 1). Tumor samples were minced with scalpels to 2 mm3 pieces and incubated with agitation in HBSS (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) containing collagenase Type II (800 U/ml; Worthington Biomedical Corporation, Lakewood, NJ, http://www.worthington- biochem.com) and 0.0025% DNase (Sigma-Aldrich) at 37°C for 30 minutes. After agitation, digested tumor material was further disaggregated by mechanical disruption in a Stomacher (Seward Laboratory Systems Inc., Bohemia, NY, http://www.seward.co.uk) for 10 minutes. Following filtration through a 100 μm cell strainer, single cells were washed three times in HBSS containing 2% fetal bovine serum and 1 mM EDTA (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). After incubation in ACK lysing buffer (Biowhittaker, Lonza, Walkersville, MD, http://www.lonza.com), viable tumor-derived cells were isolated following centrifugation over Ficoll-Paque (GE Healthcare Bio-sciences AB, Uppsala, Sweden, http://www.gehealthcare.com). Endothelial and hematopoietic cells were removed from the suspension by magnetic bead depletion using CD31- and CD45-conjugated microbeads (Miltenyi Biotec Inc., Auburn, CA, http://www.miltenyibiotec.com). The remaining ovarian tumor cells were serially diluted, resuspended in 1:1 PBS/Matrigel (BD Biosciences, San Jose, CA, http://www.bdbiosciences.com) and injected subcutaneously (s.c.) into 4- to 6-week-old female NOD/SCID mice (The Jackson Laboratory, Bar Harbor, ME, http://www.jax.org). Control mice were simultaneously injected with 1:1 PBS/Matrigel only. Mice were monitored biweekly for tumor formation.

In Vivo Serial Transplantation of Human Ovarian Tumor Cells

All mouse studies adhered to protocols reviewed and approved by the MGH Institutional Animal Care and Use Committee. Mice bearing tumors of 1 cm diameter generated following injection of primary human ovarian tumor cells were euthanized by CO2 inhalation. Tumor explants were excised aseptically, and viable cells were isolated as described above. H-2Kd+ mouse cells were eliminated by magnetic bead based depletion, and serial dilutions of H-2Kd-depleted tumor derived cells were resuspended in 1:1 PBS/Matrigel and injected s.c. into recipient female NOD/SCID mice. Tumor development was assessed biweekly by visual inspection and palpation, and any tumor explant that formed was processed and used as a source of human tumor derived cells in subsequent serial transplantation experiments.

Flow Cytometry

To examine expression of putative tumor-initiating cell markers, single-cell suspensions from primary human or transplanted ovarian tumors were isolated as outlined. Following incubation with FcR blocking reagent (Miltenyi Biotec), tumor cells were stained with anti-CD24, anti-CD133, anti-CD117, anti-EpCAM (phycoerythrin (PE)-conjugated; Miltenyi Biotec), and/or anti-CD44 (allophycocyanin (APC)-conjugated; BD Biosciences). Respective IgG isotype controls for each fluorophore were included as negative controls. Non-viable cells were excluded using the LIVE/DEAD Fixable Dead Cell Stain kit (Invitrogen). For primary tumors, CD31+ and CD45+ cells were excluded using FITC-conjugated CD31 and CD45 antibodies. Similarly, for xenograft tumors, H-2Kd+ cells were eliminated using a FITC-conjugated H-2Kd antibody. After washing, cells were fixed by incubation in paraformaldehyde (4%) for 60 minutes and analyzed using a LSRII (BD Biosciences) within 24 hours. Data were analyzed using FlowJo software (Version 8.2).

For sorting, ovarian transplanted tumors were dissociated to viable single-cell suspensions and stained with anti-CD133 (PE-conjugated) as described above. Discrimination and exclusion of dead cells and H-2Kd+ cells were achieved as outlined, and respective IgG isotype was included as negative control. CD133+ and CD133 cell populations were isolated using a FACSAria flow cytometer (BD Biosciences). An aliquot of each sorted population was resorted post collection to confirm population purity. Serial dilutions of the isolated cell populations were resuspended in 1:1 PBS/Matrigel and injected s.c. into female NOD/SCID mice.

CD133 Immunohistochemistry

Immunohistochemistry was carried out on formalin-fixed, paraffin-embedded tissue pieces from primary human ovarian tumors and xenotransplanted tumors. Sections of 5 μm thickness were dewaxed in xylene and rehydrated through serial dilutions of ethanol and distilled water. Antigen retrieval was carried out in a pressure cooker using 10 mM Na citrate (pH 6.0). After blocking endogenous peroxidase activity and applying nonspecific background blocking serum, sections were incubated with CD133 (C24B9; Cell Signaling Technology, Inc., Danvers, MA, http://www.cellsignal.com) or Rabbit Negative Control (Biogenex, San Ramon, CA, http://www.biogenex.com) at 1:100 dilution overnight at 4°C. Elite Vectastain ABC systems (Vector Laboratories, Inc., Burlingame, CA, http://www.vectorlabs.com) and DAB substrate chromogen (Sigma-Aldrich) were used for detection, followed by counterstaining with hematoxylin. Sections were visualized and images were captured using a Nikon Eclipse TE2000-S microscope and Spot Software (Diagnostic Instruments, Inc., MI, http://www.diaginc.com).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Human Ovarian Tumor Cells Can Be Engrafted In Vivo

Previous studies have utilized the xenograft transplantation model for propagating human tumors in vivo [1, 9]. We tested whether primary human ovarian tumors representative of three histological ovarian subtypes (serous (SOC), clear cell (COC), and endometrioid (EnOC)) could engraft in immunocompromised mice. Primary human ovarian tumors obtained at the time of initial cytoreductive surgery prior to any chemo- or radiotherapy were dissociated and injected s.c. into female NOD/SCID mice. Although mice were initially injected both intraperitoneally (i.p.) and subcutaneously (s.c.), no mice injected i.p. developed ascites. Some mice did have evidence of microscopic human ovarian tumors deposited throughout the peritoneal cavity, although this was only noted at necropsy. Therefore, the s.c. route of injection was used for xenograft tumors to permit adequate monitoring of tumor formation. A total of eleven human ovarian tumor xenografts (designated as first transplant tumors) were established following injection of 106 cells at an engraftment success rate of 85% (supporting information Table 2). The tumor formation rate of the initial xenografts varied between the different tumor subtypes. Importantly, the histological integrity of the original patient tumors was retained by the first transplant tumors (supporting information Fig. 1).

To assess potential differential tumorigenicity within heterogeneous ovarian tumor derived cells, dissociated tumor cells from eight primary human serous carcinomas were injected s.c. into NOD/SCID mice at serial dilutions ranging from 106 to 103 cells (supporting information Table 2). Tumor formation was monitored biweekly for 8 months post-injection, with tumors harvested at an approximate volume of 400 mm3. The majority of mice injected with 1 × 106 cells formed tumors, and the frequency of tumor formation decreased as fewer cells were injected, with no tumor formation detected following injection of less than 1 × 104 cells. These data suggest that not all cells isolated from an ovarian serous tumor have the capacity to recapitulate tumor growth in NOD/SCID mice in vivo. Similar analyses in clear cell and endometrioid ovarian tumors were limited by an insufficient number of available tumors. Our preliminary results suggest that tumor formation can occur following injection of as few as 1 × 104 cells derived from clear cell and endometrioid tumor subtypes (data not shown). These data provide evidence supporting the idea that primary human ovarian tumors contain a subpopulation of cells that are capable of initiating tumor formation in NOD/SCID mice.

Human Ovarian Xenografts Can Be Serially Transplanted In Vivo

We next sought to determine whether human ovarian tumors could be serially transplanted in NOD/SCID mice. Successful propagation of human tumor explants by serial injection of heterogeneous tumor derived cells would suggest the presence of a self-renewing cell population in the heterogeneous tumor. Our results showed that human ovarian serous and clear cell tumors could be transplanted multiple times in NOD/SCID mice (Table 1). Significantly, the tumors generated following each serial transplant maintained the same histopathological phenotype as the human primary tumor used in the initial injection (supporting information Figs. 1 and 2). Our analyses also determined that serial injection of ovarian tumor derived cells resulted in a decrease in the time to tumor formation with each round of serial transplantation despite similar or fewer numbers of cells being injected. This phenomenon persisted up to six iterations, at which point no increase in tumor formation rate was noted, which is likely due to a maximum threshold for time to onset of tumor formation being reached. We have carried out these serial transplantation studies with twelve different primary human ovarian tumors including nine serous carcinomas and three clear cell carcinomas. Although we did observe successful engraftment of a primary endometrioid tumor, it could not be propagated by serial transplantation in vivo. In all serous and clear cell transplantation studies, we have observed consistent results with respect to maintenance of tumor histopathology over the course of serial transplantation (data not shown).

Table 1. Serial transplantation of primary human ovarian serous and clear cell tumors in vivo
inline image

Cell Surface Marker Analyses of Primary and Serially Transplanted Human Ovarian Tumors

Numerous studies of various solid tumor types of distinct origin have identified cell surface marker signatures that distinguish those specific subsets of cells that have an increased potential to form tumors in vivo [1–12, 16, 17]. We analyzed the expression profile of a subset of candidate CSC markers in both primary and serially transplanted human ovarian tumors. These included CD44, CD24, CD133, CD117, and EpCAM. In primary human serous tumors, both CD24 and EpCAM were consistently expressed in a majority of tumor-derived cells (Table 2). In contrast, the frequency of both CD44+ and CD133+ cells was much lower in the analyzed populations, ranging from 4.8% to 28.5% (median = 11.4%) for CD44 and from 0.3-35% (median = 8.9%) for CD133. The observed range of CD133 expression was consistent with previous studies [19].

Table 2. Range and median cell surface marker expression in primary and xenotransplanted human ovarian tumors
inline image

To determine if these expression phenotypes were maintained in serially transplanted tumors, we next profiled CSC marker expression on cells derived from transplanted human ovarian serous tumors (Table 2). We detected no expression of CD117 on cells derived from any analyzed primary or serially transplanted tumor. In parallel experiments, CD117+ cells were detected in a population of CD34+ cells from human umbilical cord blood (data not shown), indicating that the lack of detectable CD117 expression in the analyzed ovarian tumors was not due to technical limitations. CD44, CD24, and EpCAM were expressed with wide-ranging positivity across all serous transplanted tumors assayed (CD44, 0-80.7%; CD24, 30-95%; EpCAM, 43-100%) whereas CD133 was consistently expressed at relatively low levels ranging from 0.2-12.5% (median = 3%) in serous tumors.

Our efforts to carry out CSC marker profiling of primary human clear cell tumors have been restricted by limited patient samples due to the infrequent occurrence of this histological subtype. We have, however, been able to evaluate marker expression in clear cell-derived tumor xenografts that have been propagated and expanded in vivo through serial transplantation in NOD/SCID mice. In these analyses, CD24 and EpCAM were again consistently expressed in a majority of the tumor-derived cells (Table 2). The frequency of CD44 expression ranged from 0.3-73% (median = 9%). Analyses of CD133 expression in cells derived from serially transplanted clear cell tumors determined that the frequency of CD133+ cells ranged from 65.6-94% (median = 86%).

CD133 Displays a Distinct Expression Pattern in Ovarian Serous and Clear Cell Tumors

Immunohistochemistry was used to further investigate the observed differential expression pattern of CD133 in serous and clear cell ovarian tumors. As shown in Figure 1A, CD133 expression in primary human ovarian tumors is localized to the epithelial surface of the tumor cells. As expected, the stroma was negative for CD133, with the majority of staining restricted to the epithelial edges of gland-like tumor structures. CD133 positivity in clear cell tumors (COC2) was widespread with a dense staining pattern distinctly localized to the membrane in contrast to the serous tumors where the staining pattern ranged from weakly sporadic (SOC4) to more densely distributed in focal areas of tumor tissue (SOC16). The overall intensity and degree of CD133 expression in analyzed clear cell tumors (2/2) were notably higher than that in the ovarian serous tumors (8/8), consistent with our flow cytometry data.

thumbnail image

Figure 1. CD133 expression in primary and serially transplanted human ovarian tumors. (A) Immunohistochemical analyses of CD133 in primary human ovarian serous (SOC4 and SOC16) and clear cell (COC2) tumors. Human fetal gut (117d gestation) is included as a positive control for CD133 staining (arrow heads indicate positively stained cells). IgG control antibody is included as negative control. H&E staining of the same sections is also shown. Scale bar, 50 μm. (B) CD133 immunostaining in serially transplanted serous (SOC16) and clear cell (COC2) tumors. CD133 staining in serous tumors is predominantly membranous, with some cytosolic positivity present, whereas clear cell transplants show distinct membrane staining. No difference in staining intensity is evident across the transplants. Scale bar, 50 μm. Abbreviations: COC, clear cell; H&E, hematoxylin and eosin; IgG, immunoglobulin; SOC, serous.

Download figure to PowerPoint

Having shown that in vivo serial transplantation of ovarian tumors potentially selects for cells with an increased tumorigenic capacity, we next examined whether expression of CD133 was altered across serially transplanted tumors. As shown in Figure 1B, the staining pattern of CD133 in both serous and clear cell tumors was relatively unchanged across consecutively transplanted tumors. These data also correlated with similar CD133 mRNA transcript levels in these transplanted tumors (data not shown).

CD133+ Cells Have an Increased Tumorigenic Capacity and Can Recapitulate the Original Heterogeneous Tumor Cell Phenotype

In order to determine the tumorigenic capacity of CD133 expressing cells, serous and clear cell derived transplanted tumors were dissociated to single-cell suspensions and sorted by FACS into CD133+ and CD133 fractions. Initial CD133 positivity in the analyzed tumor derived cell populations ranged from <3% (SOC16) to >65% (COC2 (>87%) and COC3 (>65%)). Purity of these sorted populations as assessed by post sort flow cytometry was >98% and >99% for CD133+ and CD133 fractions, respectively. Serial dilutions of each fraction were injected into NOD/SCID mice, and tumor formation was evaluated over 8 months postinjection.

Tumor formation data for CD133+, CD133, and bulk tumor cell populations compiled from both serous and clear cell subtypes indicate that CD133+ cells have an increased tumorigenic capacity as compared to their CD133 and bulk population counterparts (Table 3). For comparison of tumor formation and latency, specific FACS experiments for both serous and clear cell subtypes are shown in Table 4. For serous carcinoma, the SOC16-derived CD133+ fraction was capable of generating tumors following injection of as few as 500 cells whereas only 1 × 105 CD133 cells generated tumors in the same time period. Tumor formation in the lower dilutions for CD133 cells was either significantly delayed or undetected within the 8-month experimental time frame. In subsequent analyses, the tumor generated following injection of 1 × 104 SOC16-derived CD133+ cells was profiled and resorted into CD133+ and CD133 populations (SOC16-104; Table 4 and Fig. 2). The CD133+ fraction of this tumor accounted for 15.7% of the cells present in the unsorted tumor cell population suggesting that a heterogeneous tumor phenotype was recapitulated following injection of a highly purified fraction of CD133 expressing cells. This tumor was resorted, and the resulting CD133+ and CD133 fractions were injected. Tumor formation was observed in mice injected with as few as 100 CD133+ cells. In contrast, a tumor with similar latency was detected only in the mouse injected with 1 × 105 CD133 cells. Flow cytometric analyses of the tumors generated following these injections of CD133+ and CD133 cells determined that CD133+ cells comprised a similar percentage of the total tumor cell population (Fig. 2, Resulting Tumors). The tumors formed from both the injected CD133+ and CD133 cells maintained equivalent serous histological phenotypes (Fig. 2, Histology). In similar analyses of clear cell tumors (COC2 and COC3), injection of only 500 CD133+ cells was required for tumor formation whereas no tumors were generated following injection of fewer than 1 × 104 CD133 cells (Table 4). Taken together, these data from serous and clear cell tumors indicate that CD133 expression defines a population of cells with an increased NOD/SCID tumorigenic capacity in primary human ovarian malignancies.

Table 3. Overall tumor formation capacity of CD133+, CD133, and bulk (H-2Kd-) populations from serially diluted transplanted ovarian tumor cells in vivo
inline image
Table 4. Tumorigenic capacity and latency of CD133+, CD133 and bulk (H-2Kd-) ovarian transplanted tumor cells in vivo
inline image
thumbnail image

Figure 2. Sorted CD133+ and CD133 cells generate histologically similar tumors that contain CD133+ populations. Parent CD133+ tumor (SOC16-104), containing 15.7% CD133+ cells (Pre-Sort), was sorted with high efficiency (Post-Sort), and the isolated CD133+ and CD133 cell fractions were injected subcutaneously into NOD/SCID mice. CD133 expression in the tumors generated following injection of each fraction was analyzed (Resulting Tumors). The histopathology of the tumors was assessed by hematoxylin and eosin staining (Histology: upper panel, CD133+ cell-derived tumor; lower panel, CD133 cell-derived tumor). Scale bar, 50 μm.

Download figure to PowerPoint

Previous studies investigating the existence of CSC populations in human ovarian cancer have identified CD44+/CD117+ and CD44+/MyD88+ as subpopulations of cells with stem cell-like features [16, 17]. The lack of CD117+ expression in both primary and transplanted human ovarian tumors propagated in our in vivo model system, and the unavailability of suitable antibody for live cell MyD88 expression analysis by flow cytometry prevented us from isolating these previously identified populations. Given that CD44 has been postulated to be a CSC marker in ovarian tumors, we sought to determine any correlation in expression between CD133 and CD44 by combined marker analysis using flow cytometry. As shown in Figure 3, the analyzed serous (SOC6, SOC8) and clear cell (COC2, COC3) transplanted tumors had overlapping CD133+ and CD44+ expression, with the frequency of the CD133+/CD44+ subpopulations ranging from 0-4% (median = 0.34%) in serous transplanted tumors (n = 8) and from 1.55-79% (median = 30.7%) in clear cell transplanted tumors (n = 10). In both subtypes analyzed, neither population completely encompassed the other. The frequency of CD133+/CD44+ subpopulations in the serous transplanted tumors was consistent with data derived from independent primary human ovarian serous tumors (SOC17, 0.34%; SOC18, 1.03%, data not shown).

thumbnail image

Figure 3. CD133+/CD44+ cells are present in human serous and clear cell ovarian tumors. NOD/SCID xenotransplanted serous (SOC6, SOC8) and clear cell (COC2, COC3) tumors were excised, digested, and analyzed by flow cytometry to test for dual marker expression. Combined marker analyses for CD133 and CD44 in transplanted tumors indicate subpopulations of cells double positive for both markers, with median expression of 0.34% and 30.7% in serous (n = 8) and clear cell (n = 10) transplanted tumors, respectively. All four possible populations generated from this staining regimen (CD133+/CD44+, CD133+/CD44, CD133/CD44+, CD133/CD44) were detected with varying consistency in both histological subtypes. Percentages reported in upper right quadrants indicate the percentage of CD133+/CD44+ cells (or respective IgG controls). Abbreviations: COC, clear cell; IgG, immunoglobulin; SOC, serous.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

To date, studies aimed at investigating the existence of CSC populations, also referred to as tumor-initiating cells, in ovarian cancer have relied upon established cell lines or improved spheroid culture assays [15–17]. Although anchorage-independent growth of colonies precludes plastic-induced differentiation, it remains possible that growth factors present in the culture medium may influence cell surface marker expression [1, 8, 20]. Here, we describe a NOD/SCID xenotransplantation system that has allowed us to propagate and expand primary human epithelial ovarian cancer cells in vivo in the absence of exposure to in vitro culture conditions. Time to tumor formation for initial primary human tumor xenografts ranged from 2-8 months, depending on the histological subtype of the tumor, with clear cell ovarian tumors having an accelerated tumor formation rate. The engraftment rate of primary human ovarian serous tumors using this in vivo model clearly indicates that only a small fraction of cells within these malignancies have the capacity to form tumors in NOD/SCID mice. The successful serial transplantation of human ovarian tumors in this in vivo system supports the hypothesis that these tumors contain a population of cells that are capable of self-renewal. Importantly, the histological integrity of the tumors is maintained, even after serial transplantation, consistent with the premise that subpopulations of human ovarian tumor cells can recapitulate a heterogeneous tumor phenotype (supporting information Figs. 1 and 2).

One interesting feature of our serial transplantation assays is the increased tumor formation rate with successive transplant of tumors generated from fewer tumor-derived cells (Table 1). It is likely that this is due to the selective pressure inherent in the xenotransplantation assay which selects for the most tumorigenic cells and deselects less tumorigenic cells. A potential limitation of in vivo mouse systems [21] is that they may in fact select for tumor cells most capable of adapting to the mouse microenvironment regardless of the inherent tumorigenicity of those cells. Additionally, recent data suggest that the use of NOD/SCID mice in studies aimed at identifying CSC in solid tumors may result in an underestimation of the percentage of tumorigenic cell subpopulations due to residual immunoreactivity in the animals [22]. Although these caveats must be taken into consideration when utilizing such models, the results obtained from our analyses of both primary tumors and tumor explants propagated through an in vivo serial transplantation system in the absence of in vitro culturing suggest the existence of CSC-like cells in human ovarian tumors.

Previous studies in various solid tumor types have utilized specific cell surface markers to isolate the relevant tumorigenic cell population from the heterogeneous tumor cell mix [1–12, 16, 17]. In an effort to ascertain the potential, if any, of these markers to isolate similar tumorigenic populations responsible for ovarian tumor growth, the expression profiles of CD44, CD24, CD133, CD117, and EpCAM were examined in both primary and serially transplanted human ovarian tumors. From these analyses, CD24, CD44, and EpCAM expression patterns were found to be wide-ranging, from moderately to highly expressed. Our finding that CD117, a marker previously determined to define ovarian tumorigenic cells isolated from spheroid cultures [16], was not expressed in any analyzed tumor may reflect a differential effect of the cell selection methods utilized or induction of CD117 expression based on the media components present. Although the wide range of expression of CD24, CD44, and EpCAM in the analyzed ovarian tumors is insufficient to exclude any of these proteins as a marker of a tumorigenic CSC population, CD133 was more consistently expressed in both serous and clear cell tumors, although the degree of CD133 positivity was much greater in clear cell tumors. This difference was corroborated by immunohistochemical analyses of CD133 expression in both primary and serially transplanted ovarian serous and clear cell tumors. We did not observe any notable changes in CD133 expression as indicated by flow cytometry (data not shown) and immunohistochemical studies (Fig. 1) across serial transplants derived from the same primary tumors despite the decreased time to tumor formation. It is possible that CD133+ cells are maintained at a predetermined threshold within these tumors to allow for both the differentiation and expansion of CD133 daughter cells and the retention of a pool of slow-cycling, stem cell-like tumor initiating cells. This hypothesis is supported by data from our cell-sorting analyses. In one such experiment, CD133+ cells comprised 15.7% of the total SOC16 tumor cell population. Injection of a highly purified SOC16-derived CD133+ fraction (>98% purity) into NOD/SCID mice gave rise to heterogeneous tumors in which the CD133+ cells comprised 17.8% of the total tumor cell population (Fig. 2), suggesting that the injected cells can generate CD133 daughter cells and are maintained at an equivalent frequency in the resulting heterogeneous tumor.

CD133 has been identified as a marker of both normal stem cells in several organs and of tumorigenic populations in many tumor types including brain [10], pancreas [5], colon [7, 9], lung [4], and prostate [2]. More recent evidence derived from studies of established human ovarian cancer cell lines indicates that CD133 expression defines a tumorigenic population in these lines, but no information regarding the relative tumorigenicity of CD133+ cells derived from primary human ovarian tumors was reported [15].

In initial pilot experiments, we used a magnetic bead based separation method to isolate fractions enriched for CD133+ or CD133 cells from primary human ovarian tumors propagated in our NOD/SCID mouse system. Preliminary data suggested that these enriched fractions had differential tumor-initiating capacity (data not shown). To extend these analyses, we generated highly purified CD133+ and CD133 cell fractions from serially transplanted human serous and clear cell tumors by FACS, injected equivalent numbers of cells from each isolated fraction into NOD/SCID mice and monitored tumor formation in the injected animals. In all cases, tumors were generated following injection of CD133+ cells (Tables 3 and 4). Strikingly, tumors were readily detectable following injection of as few as 100-500 CD133+ cells whereas the cell numbers required for tumor formation mediated by injection of CD133 cells were orders of magnitude higher. Our data clearly indicate that CD133+ cells have an increased tumorigenic capacity relative to their CD133 counterparts in this in vivo model system and are consistent with the previously reported results from established human ovarian cancer cell lines [15]. Our data, however, are in contrast to a recent report [23] indicating that CD133+ cells from human ovarian ascites are non-tumorigenic. This contradiction in analyses of primary patient samples may be explained by differences in either the source material used to generate the analyzed CD133+ cell populations (solid tumors vs. ascites) or the method used to expand the tumor-derived material prior to sorting (in vivo vs. in vitro propagation).

In our studies, CD133+ cell populations derived from both human serous and clear cell ovarian tumors had increased tumor-initiating capacity despite a significant difference in CD133 expression levels as determined by flow cytometry and immunohistochemistry. The potential biological relevance of this disparity is unclear. It is possible that the higher percentage of CD133+ cells in human clear cell tumors is responsible for the rapid tumor formation we consistently observe in our in vivo system. It is interesting to note that clear cell ovarian carcinomas are often clinically resistant to conventional platinum-based chemotherapy [24]. Whether that also reflects the presence of a highly abundant drug-resistant CSC population in these tumors remains to be determined.

One unexpected finding in our analyses is the observation that tumors that developed following injection of sorted CD133+ (>98% purity) and CD133 (>99% purity) cell fractions were both heterogeneous with respect to CD133 expression and in fact had similar levels of CD133+ cells (Fig. 2). One likely possibility is that the injected CD133 cell fraction was contaminated with potentially 0.1-1% CD133+ cells that initiated tumor formation in the injected animals. This is consistent with the fact that injection of 100,000 CD133 cells in this experiment was required to generate a tumor with the same latency as a tumor that developed following injection of only 100 CD133+ cells (Table 4). Still, we cannot completely rule out the possibility that both CD133+ and CD133 populations derived from human ovarian tumors are tumorigenic, as has previously been postulated for glioblastomas [25] and metastatic colon tumors [26], or that each population can generate both CD133+ and CD133 cells as reported in human gliomas [27].

Our data demonstrate that CD133+ ovarian tumor derived cell fractions are enriched for tumor initiating cells. The level of enrichment is modest, however, suggesting that the definition of a rare tumor-initiating cell subpopulation in ovarian cancer will require further refinement of cell surface marker profiling strategies that extend beyond purification based on differential CD133 expression alone. To this end, we have used combined marker profiling of ovarian transplanted tumors to detect the presence of a CD133+/CD44+ subpopulation in both serous and clear cell ovarian tumors generated in our in vivo system (Fig. 3). Subsequent analyses of the relative tumorigenicity of this fraction require a large amount of tumor material propagated in our NOD/SCID model. We are currently working to expand our system to generate sufficient quantities of material for these analyses. Direct sorting of primary tumor material to isolate this CD133+/CD44+ subpopulation is similarly restricted by limited availability of tissue.

Continued functional and phenotypic studies of such cell populations isolated based on overlapping expression of CD133 and other CSC markers will provide insights into both new and highly relevant molecular targets for therapeutic intervention and the eventual development of more intelligent treatment strategies for ovarian cancer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

The authors thank Laura Prickett and Kat Folz-Donahue at the HSCI/MGH Flow Cytometry Core Facility and Michael Waring and Andrew Cosgrove at the Ragon Institute (MGH/MIT/Harvard) for help with cell purification; Dr. Esther Oliva for analyzing tumor pathology; Dr. Borja Saez for providing CD34+ cells; and colleagues in the VCRB and Scadden laboratories for critical review of the manuscript.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

This work was supported by an anonymous gift to the Harvard Stem Cell Institute and funding from the Dana Farber/Harvard Cancer Center Ovarian SPORE (RF/BRR), Marsha Rivkin Center for Ovarian Cancer Research (RF), The Julie Fund (RF), Advanced Medical Research Foundation (BRR), Ovarian Cancer Research Fund (BRR), Vincent Memorial Research Fund (BRR), Ovarian Cancer and Education Awareness Network (BRR), and a gift from Robert Lynn v. d. Luft and Christa v. d. Luft (BRR).

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information
  • 1
    Al-Hajj M, Wicha MS, Benito-Hernandez A et al. Prospective identification of tumorigenic breast cancer cells. Proc Natl Acad Sci USA 2003; 100: 39833988.
  • 2
    Collins AT, Berry PA, Hyde C et al. Prospective identification of tumorigenic prostate cancer stem cells. Cancer Res 2005; 65: 1094610951.
  • 3
    Dalerba P, Dylla SJ, Park IK et al. Phenotypic characterization of human colorectal cancer stem cells. Proc Natl Acad Sci USA 2007; 104: 1015810163.
  • 4
    Eramo A, Lotti F, Sette G et al. Identification and expansion of the tumorigenic lung cancer stem cell population. Cell Death Differ 2008; 15: 504514.
  • 5
    Hermann PC, Huber SL, Herrler T et al. Distinct populations of cancer stem cells determine tumor growth and metastatic activity in human pancreatic cancer. Cell Stem Cell 2007; 1: 313323.
  • 6
    Li C, Heidt DG, Dalerba P et al. Identification of pancreatic cancer stem cells. Cancer Res 2007; 67: 10301037.
  • 7
    O'Brien CA, Pollett A, Gallinger S et al. A human colon cancer cell capable of initiating tumour growth in immunodeficient mice. Nature 2007; 445: 106110.
  • 8
    Ponti D, Costa A, Zaffaroni N et al. Isolation and in vitro propagation of tumorigenic breast cancer cells with stem/progenitor cell properties. Cancer Res 2005; 65: 55065511.
  • 9
    Ricci-Vitiani L, Lombardi DG, Pilozzi E et al. Identification and expansion of human colon-cancer-initiating cells. Nature 2007; 445: 111115.
  • 10
    Singh SK, Hawkins C, Clarke ID et al. Identification of human brain tumour initiating cells. Nature 2004; 432: 396401.
  • 11
    Monzani E, Facchetti F, Galmozzi E et al. Melanoma contains CD133 and ABCG2 positive cells with enhanced tumourigenic potential. Eur J Cancer 2007; 43: 935946.
  • 12
    Prince ME, Sivanandan R, Kaczorowski A et al. Identification of a subpopulation of cells with cancer stem cell properties in head and neck squamous cell carcinoma. Proc Natl Acad Sci USA 2007; 104: 973978.
  • 13
    Bapat SA, Mali AM, Koppikar CB et al. Stem and progenitor-like cells contribute to the aggressive behavior of human epithelial ovarian cancer. Cancer Res 2005; 65: 30253029.
  • 14
    Szotek PP, Pieretti-Vanmarcke R, Masiakos PT et al. Ovarian cancer side population defines cells with stem cell-like characteristics and Mullerian Inhibiting Substance responsiveness. Proc Natl Acad Sci USA 2006; 103: 1115411159.
  • 15
    Baba T, Convery PA, Matsumura N et al. Epigenetic regulation of CD133 and tumorigenicity of CD133+ ovarian cancer cells. Oncogene 2009; 28: 209218.
  • 16
    Zhang S, Balch C, Chan MW et al. Identification and characterization of ovarian cancer-initiating cells from primary human tumors. Cancer Res 2008; 68: 43114320.
  • 17
    Alvero AB, Chen R, Fu HH et al. Molecular phenotyping of human ovarian cancer stem cells unravels the mechanisms for repair and chemoresistance. Cell Cycle 2009; 8: 158166.
  • 18
    Ince TA, Richardson AL, Bell GW et al. Transformation of different human breast epithelial cell types leads to distinct tumor phenotypes. Cancer Cell 2007; 12: 160170.
  • 19
    Ferrandina G, Bonanno G, Pierelli L et al. Expression of CD133–1 and CD133–2 in ovarian cancer. Int J Gynecol Cancer 2008; 18: 506514.
  • 20
    Vermeulen L, Sprick MR, Kemper K et al. Cancer stem cells—Old concepts, new insights. Cell Death Differ 2008; 15: 947958.
  • 21
    Clarke MF, Dick JE, Dirks PB et al. Cancer stem cells—Perspectives on current status and future directions: AACR Workshop on cancer stem cells. Cancer Res 2006; 66: 93399344.
  • 22
    Quintana E, Shackleton M, Sabel MS et al. Efficient tumour formation by single human melanoma cells. Nature 2008; 456: 593598.
  • 23
    Kusumbe AP, Mali AM, Bapat SA. CD133-Expressing Stem Cells Associated with Ovarian Metastases Establish an Endothelial Hierarchy and Contribute to Tumor Vasculature. Stem Cells 2009; 27: 498508.
  • 24
    Pectasides D, Fountzilas G, Aravantinos G et al. Advanced stage clear-cell epithelial ovarian cancer: The Hellenic Cooperative Oncology Group experience. Gynecol Oncol 2006; 102: 285291.
  • 25
    Beier D, Hau P, Proescholdt M et al. CD133(+) and CD133(−) glioblastoma-derived cancer stem cells show differential growth characteristics and molecular profiles. Cancer Res 2007; 67: 40104015.
  • 26
    Shmelkov SV, Butler JM, Hooper AT et al. CD133 expression is not restricted to stem cells, and both CD133+ and CD133- metastatic colon cancer cells initiate tumors. J Clin Invest 2008; 118: 21112120.
  • 27
    Wang J, Sakariassen PO, Tsinkalovsky O et al. CD133 negative glioma cells form tumors in nude rats and give rise to CD133 positive cells. Int J Cancer 2008; 122: 761768.

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  9. References
  10. Supporting Information

Additional supporting information available online.

FilenameFormatSizeDescription
STEM_236_sm_suppinfofigure1.tif39965KSupplementary Figure 1. Hematoxylin and eosin stained sections of primary serous (SOC1-6), clear cell (COC1-2) and endometrioid (EnOC1) tumors and their respective first transplant tumors. First transplant (xenograft) tumors were generated by s.c. injection of 1x106 bulk primary tumors cells into NOD/SCID mice. The histology of the parent primary tumor is maintained by the xenografted tumors formed in vivo. Magnification, 20X; scale bar, 50μm.
STEM_236_sm_suppinfofigure2.tif21684KSupplementary Figure 2. Hematoxylin and eosin stained sections of serially transplanted serous (SOC1) and clear cell (COC1) tumors, which show that the histology of the tumors is retained over multiple transplants in NOD/SCID mice. Magnification, 20X; scale bar, 50μm.
STEM_236_sm_suppinfotable1.doc45KSupplementary Table 1. Patient age and tumor subtype information
STEM_236_sm_suppinfotable2.doc45KSupplementary Table 2. In vivo tumor formation capacity of serially-diluted human primary ovarian serous tumor cells

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.