The significance of a Cripto-1-positive subpopulation of human melanoma cells exhibiting stem cell-like characteristics

Abstract

Cripto-1 (CR-1) protein function differs according to cellular or extracellular expression. In this study, we explore the significance of cell surface CR-1 expression in human melanoma cells. Cell surface CR-1-expressing human melanoma cells (CR1-CS+) were selected by fluorescence-activated cell sorting (FACS) and grown in vitro and in vivo in nude mice to study their growth characteristics. The CR1-CS+ melanoma cells were found to express increased levels of Oct4, MDR-1 and activated c-Src compared with cells lacking this subpopulation (CR1-CS−) or unsorted cells, used as control. CR1-CS+ show reduced proliferation rates and diminished spherical colony formation compared with control cells when cultured in vitro. Orthotopic injections of CR1-CS+ in nude mice formed slow growing tumors with histologic variability across different areas of the CR1-CS+ xenografts. CR-1-expressing cells from first generation CR1-CS+ tumors showed significantly increased tumor-forming rate and aggressiveness following subsequent transplants in nude mice. These data demonstrate that within a heterogeneous melanoma cell population there resides a slow proliferating, cell surface CR-1-expressing subpopulation capable of giving rise to a fast growing, aggressive progeny that may contribute to disease recurrence and progression.

Introduction

A large body of evidence suggests that within the heterogeneous population comprising a melanoma, certain cell types exhibit molecular and functional characteristics similar to stem cells. These putative melanoma stem cells (MSCs) are believed to give rise to a highly plastic, tumor-forming progeny with the potential of assuming adipogenic, chondrogenic, osteogenic and vasculogenic phenotypes capable of drug resistance and metastatic spread.^1,2 There is, however, much debate as to the ideal molecular profile capable of identifying MSCs.

Cripto-1 (CR-1), an epidermal growth factor-related protein, plays a fundamental role for proper signaling of the transforming growth factor (TGF)-β-related morphogen Nodal during normal development as well as during the regulation of self-renewal and pluripotency of mouse and human embryonic stem cells.^3-5 CR-1 has been reported to be broadly expressed at the intracellular and extracellular levels in several types of human cancer tissues, including breast cancer and melanoma.^6,7 Nodal has also been suggested to be responsible, at least in part, for the tumor cell plasticity and aggressive behavior of human melanoma cells.^8,9 However, there has been little study of the role of CR-1 as a cell surface co-receptor for Nodal signaling in human melanoma. Given the significant levels of Nodal expression in melanoma, it would seem logical that CR-1 expression at the cell surface would also be robust; however, in a previous study, melanoma cells were found to express very low levels of cell surface CR-1 in vitro.⁹ Since CR-1 is known to be involved in stem cell maintenance and pluripotency,¹⁰ and because recent studies have detected stem cell markers in CR-1-positive human cancer cells,^11,12 we addressed the hypothesis that the small subpopulation of cell surface CR-1-expressing melanoma cells may exhibit certain stem cell-like characteristics. In this report, we describe the growth characteristics and tumorigenic potential of melanoma cells enriched for cell surface expression of CR-1. The characterization of this subset of melanoma cells selected for cell surface expression of CR-1 could serve as a rationale for further studies exploring CR-1 as a complimentary target in multi-targeted melanoma therapy.

Results

Detection, isolation and in vitro growth characteristics of cell surface Cripto-1-expressing melanoma cells

C8161 and ROS184 human melanoma cell lines were evaluated for cellular CR-1 expression by immunofluorescence cytochemistry (IFC) following methanol fixation to permeabilize cells. Confocal microscopic analysis indicates predominantly intracellular staining as well as rare cell surface expression and few cells with no staining at all (Fig. 1A). Closer examination by fluorescence-activated cell sorting (FACS) analysis of live, non-permeabilized cells shows that approximately 5% of C8161 and 2% of ROS184 human melanoma cells specifically expressed CR-1 protein on the cell surface (Fig. 1B).

Figure 1. Detection and in vitro growth of cell surface CR-1-expressing melanoma cells. (A) Analysis of immunocytochemistry shows varying degrees of intracellular and cell surface (yellow arrows in insets) staining patterns in CR-1-positive C8161 and ROS184 human melanoma cells (250× original magnification). (B) FACS analysis reveals approximately 5% and 2% cell surface CR-1 expression in C8161 and ROS184, respectively. (C) A purity of > 90% cell surface CR-1-expressing C8161 cells (CR1-CS+) was obtained by cell sorting. (D) In vitro cultures show that the proliferation rate of C8161-CR1-CS+ cells was significantly lower (*p < 0.05) than that of C8161 cells depleted of -CR1-CS+ cells during the cell sorting (C8161-CR1-CS−). (E) Western blot analysis shows no increase in JARID-1B expression in C8161-CR1-CS+ cells compared with C8161-CR1-CS−. (F) Interestingly, C8161-CR1-CS+ cells form spherical colonies when grown in stem cell culture medium for ~2 wk, while C8161-CR1-CS− cells did not.

Based on the higher percentage of cell surface CR-1 expression in C8161 cells, these were selected for further experiments, with the presumption of a greater yield of cell surface CR-1-expressing melanoma cells. The sorted cell surface CR-1-expressing C8161 cells (C8161-CR1-CS+) (Fig. 1C) were then cultured to investigate the in vitro growth characteristics of this subpopulation of human melanoma cells. We found that the sorted C8161-CR1-CS+ cells show significantly reduced growth rates compared with the remainder of the C8161 cell population depleted of the C8161-CR1-CS+ cells, termed C8161-CR1-CS− (Fig. 1D). C8161-CR1-CS− and unsorted C8161 cells show no difference in growth rates as determined by cell counting (data not shown). It has recently been shown that certain slow-cycling melanoma cells express JARID-1B;¹³ however, at least by WB analysis, we were unable to detect an increase in JARID-1B in cell lysates from the slow growing C8161-CR1-CS+ cells (Fig. 1E). Finally, C8161-CR1-CS+ cells were capable of forming spherical colonies, which were evident after approximately 14 d in culture media generally used to maintain stem cells, while control C8161-CR1-CS− cells did not form these characteristic spherical colonies when cultured in the same media during the same time interval (Fig. 1F). Of note, C8161-CR1-CS+ cells did not form spherical colonies and exhibited a similar growth pattern to C8161-CR1-CS− or unsorted C8161 cells when grown for the same time period in regular media routinely used to culture C8161 (data not shown).

C8161-CR1-CS+ cells express Oct4, MDR-1 and P-Src

Slow proliferation rates and spherical colony formation are common characteristics of stem cells. Therefore, the observation of such growth characteristics in C8161-CR1-CS+ cells suggests that this subpopulation of human melanoma cells may possess other stem cell-like characteristics. In fact, a higher level of the stem cell-related transcription factor Oct4 is detected by quantitative real-time PCR (qRT-PCR) (Fig. 2A) and immunofluorescence (Fig. 2B) analyses in the C8161-CR1-CS+ spherical colonies compared with C8161-CR1-CS− cells. The multidrug resistance protein MDR-1, previously shown to associate with melanoma cells displaying stem cell-like characteristics,¹⁴ is also upregulated in C8161-CR1-CS+ cells compared with C8161-CR1-CS− cells as determined by qRT-PCR and IFC (Fig. 2A and B). We also detected increased phosphorylation of Src in C8161-CR1-CS+ compared with C8161-CR1-CS− cells, suggesting increased Src activity in the C8161-CR1-CS+ subpopulation (Fig. 2C) and confirming a previous report describing CR-1-dependant P-Src activity in human melanoma.⁷ No significant difference in expression of either P-Smad2 or Nodal is detected between C8161-CR1-CS+ and C8161-CR1-CS− cells (Fig. 2C). Together, these results suggest that cell surface CR-1 expression segregates with a slow proliferating, spherical colony-forming subpopulation of Oct4-positive melanoma cells expressing increased levels of a drug resistance marker.

Figure 2. Expression profile and signaling of C8161-CR1-CS+ cells. (A) Results from qRT-PCR shows increased expression of CR-1, Oct4 and MDR-1 RNA in C8161-CR1-CS+ compared with C8161-CR1-CS− cells. (B) Immunofluorescent cytochemistry also shows increased expression of CR-1, Oct4 and MDR-1 in the spherical colonies formed by C8161-CR1-CS+ compared with C8161-CR1-CS− cells. (C) Western blot analysis shows increased expression of P-Src in C8161-CR1-CS+ compared with C8161-CR1-CS− but no difference in the levels of Nodal or P-Smad2 between these two cell types.

First generation C8161-CR1-CS+ melanoma cells form slow-growing tumors in vivo

We found that ~1.5 × 10⁴ C8161-CR1-CS+ cells give rise to palpable tumors within 7 to 14 d after orthotopic injection in nude mice (Fig. 3A). Injection of a lower number of C8161-CR1-CS+ cells (~50,000) was insufficient to form measurable tumors within the entire 21 d time period designated for the experiment (data not shown). The tumors formed by C8161-CR1-CS+ were significantly smaller compared with tumors formed by either C8161-CR1-CS− or unsorted C8161 cells (p < 0.05 on 14 and 21 d) (Fig. 3A), confirming the comparatively slow growth rate of C8161-CR1-CS+ cells previously observed in vitro. Interestingly, although C8161-CR1-CS− cells formed significantly larger tumors than C8161-CR1-CS+ cells, the C8161-CR1-CS− tumors were still significantly smaller than the tumors formed by unsorted C8161 cells (p = 0.016) (Fig. 3A). Scoring of immunohistochemistry (IHC) staining confirmed higher expression of CR-1 and MDR-1 in tumors formed by C8161-CR1-CS+ compared with C8161-CR-CS− tumors (Fig. 3B and C). Moreover, histological examination of the C8161-CR1-CS+ tumors revealed a heterogeneous mass consisting of areas of large epithelioid cells, spindle-like mesenchymal cells surrounding numerous blood vessels and compact sheets of relatively smaller pleomorphic cells (Fig. 3D). Since serial transplant experiments of cells derived from primary tumors is commonly employed to demonstrate the ability of the isolated cells to propagate tumors in vivo, we proceeded by extracting cells from the harvested xenograft tumors formed by the C8161-CR1-CS+ cells. Western blot analysis of cells extracted from three individual C8161-CR1-CS+ tumors (C8161-CR-1R1, C8161-CR-1R2 and C8161-CR-1R3) confirms that CR-1 was still expressed in these tumor cells (Fig. 4A). Cells from C8161-CR-1R1 were then re-injected into a new set of nude mice. Within the first 14 d, C8161-CR-1R1 cells tended to form tumors at a much greater rate than the tumor-forming rates previously observed for C8161-CR1-CS+ cells, but still smaller than tumor volumes previously observed with C8161-CR1-CS− or unsorted C8161 cells. Between 14 and 21 d, the C8161-CR-1R1 tumors grew rapidly, reaching mean tumor volumes comparable to those previously observed with C8161-CR1-CS− and unsorted C8161 cells (Fig. 4B). Histological examination of these new tumors showed signs of increased proliferation and local aggressive growth, as documented by Ki67 staining of numerous mitotic bodies and evidence of local tissue invasion (Fig. 4C). Results from IHC staining showed continued expression of CR-1 and MDR-1 in the C8161-CR-1R tumors (Fig. 4D).

Figure 3. In vivo growth of C8161-CR1-CS+ cells. (A) Fourteen and 21 d after orthotopic injection in nude mice, the tumor formation rate of C8161-CR1-CS+ cells was significantly lower (*p < 0.05) than that of C8161-CR1-CS− or unsorted C8161 cells. Although C8161-CR1-CS− tumors grew at a significantly greater rate (*p < 0.05) than C8161-CR1-CS+ tumors, at day 21 they remained significantly smaller (*p < 0.05) than tumors formed by unsorted C8161 cells. Immunohistochemistry (B) clearly shows increased staining for CR-1 (brown stain) in C8161-CR1-CS+ compared with C8161-CR1-CS− tumors (200× original magnification). (C) Scoring of the immunohistochemistry results shows a significant increase (*p < 0.05) in the median score (red horizontal bar) for CR-1 and MDR-1 staining in C8161-CR1-CS+ compared with C8161-CR1-CS− tumors. (D) Low power (5× original magnification) of a representative H&E stained section shows how a C8161-CR1-CS+ tumor was composed of areas of spindle, epithelioid and undifferentiated pleomorphic tumor cells (insets 630× original magnification).

Figure 4. Tumor-forming ability of CR-1-expressing cells extracted from C8161-CR1-CS+ tumors. Cells were extracted from first generation C8161-CR1-CS+ tumors and injected into new nude mice to determine conserved tumor-forming ability. (A) Western blot analysis confirmed the expression of CR-1 in lysates from three different second generation tumors (C8161-CR1-R1, -R2 and -R3) (H9 = human ES cell lysate as positive control). (B) Tumor formation rate of C8161-CR1-R1 was significantly greater (p < 0.05) than the tumor-formation rate of first generation C8161-CR1-CS+ previously observed. (C) Histological examination of the C8161-CR1-R1 tumors revealed evidence of increased proliferation, as shown by increased Ki67 expression (brown stain) and local invasion between adjacent muscle tissue (arrows in H&E stain) (630× original magnification). (D) Immunohistochemistry staining confirms the expression of CR-1 and MDR-1 (brown stain) in C8161-CR-1R1 tumor (630× original magnification).

Discussion

While CR-1 has previously been described in human melanoma,^7,9 little emphasis was placed on defining the cellular localization of CR-1 expression. This distinction is critical, because CR-1 can have different biological effects depending on its localization.^5,15,16 For instance, CR-1 can localize to the cell surface and act as a co-receptor for Nodal signaling,⁵ can localize to specific intracellular compartments and affect protein processing,¹⁵ or can even be cleaved from the cell surface and exert paracrine effects on endothelial cells.¹⁶ In this study, we show that cell surface expression of CR-1 is restricted to a subpopulation of melanoma cells. Since CR-1 generally functions as a cell surface co-receptor for Nodal,⁵ a growth factor known to be expressed in melanoma cells,⁹ we were surprised to observe such low levels (< 5%) of cell surface expression of CR-1 when Nodal has been reportedly observed at a higher frequency (~25%) in the same cells.¹⁷ However, because we observe no difference in the expression levels of Nodal or P-Smad2 (second messenger for Nodal signaling) between C8161-CR1-CS+ and C8161-CR1-CS−, we postulate that Nodal may be functioning independently of CR-1, assuming that, as in other systems,¹⁸ co-receptor activity at the cell surface, and not intracellularly localized CR-1, regulates Nodal activity in melanoma cells. Since we find higher levels of activated Src in C8161-CR1-CS+ cells, it is possible that CR-1 also signals independently of Nodal in melanoma cells, confirming our previous study showing increased CR-1-dependant Src signaling in melanoma.⁷ Given that increased Src activity is traditionally associated with activated cell growth and proliferation, the observed increase in Src phosphorylation appears to contradict the slow proliferation rate seen in C8161-CR1-CS+ cells. However, Src signaling has also been shown to play an important role during maintenance of self-renewal of slow-cycling mammalian embryonic stem cells,¹⁹ suggesting that Src has distinct context-dependent functions that would therefore support slow-growing melanoma cells.

CR-1 is often included in molecular marker panels to identify stem cells;³ for this reason we examined whether CR-1 expression could actually represent a subpopulation of melanoma cells with more stem cell-like features. In vitro culture of isolated C8161-CR1-CS+ show that these melanoma cells have a significantly slower growth rate compared with control cells. In fact, it has been proposed that cancer cells with stem cell-like characteristics can give rise to a slow-cycling progeny. In this regard, a subpopulation of slow-cycling, stem cell-like melanoma cells have previously been shown to express JARID-1B in some melanoma cell lines,¹³ raising the possibility that slow-growing C8161-CR1-CS+ cells might also express this marker. However, WB analysis does not reveal any increase in JARID-1B levels in C8161-CR1-CS+ cells, suggesting that JARID-1B and CR-1 may be expressed in different subsets of slow-cycling melanoma cells. This is not surprising given the heterogeneous nature of the population comprising a melanoma. Western blot analysis of isolated C8161-CR1-CS+ cells also revealed that these cells express the stem cell-related transcription factor Oct4. This is in accordance with a previously described association between Oct4 and CR-1 in other cellular systems¹² and suggests a potential connection between CR-1-expressing cancer cells and cancer cells with stem cell characteristics.

Normal stem cells are known to express high levels of ATP-binding cassette (ABC) transporters, which contribute to stem cell resistance to drugs and toxins.²⁰ It is thought that a stem cell-like subpopulation in melanoma may also exhibit increased drug resistance and may, therefore, be responsible for failure of chemotherapy, disease recurrence and metastatic spread.¹⁴ Human melanomas express many types of drug-resistant molecules, and although some reports have suggested the use of these molecular markers for selecting melanoma cells with stem cell-like properties,²¹ there is still a lack of consensus on whether drug-resistant proteins and other proposed markers can consistently identify a putative melanoma stem cell population.²² A comprehensive analysis of drug resistance molecules that may be associated with CR-1 expression is beyond the scope of this study; however, we did initiate a preliminary analysis of the expression of one drug resistance protein, MDR-1 (P-glycoprotein), which has been proposed to segregate with melanoma cells exhibiting stem cell-like properties.¹⁴ We show in vitro by IFC and qRT-PCR analysis that C8161-CR1-CS+ cells express higher levels of MDR-1, further supporting the potential stem cell-like characteristics of cell surface CR-1-expressing melanoma cells. We also show in vivo that tumors formed by C8161-CR-1-CS+ or by cells extracted from the first generation CR-1 tumors that MDR-1 continues to be robustly expressed, suggesting the potential increase in drug resistance of these tumors.

When C8161-CR1-CS+ cells were injected orthotopically into nude mice, the tumors that resulted were histologically heterogeneous, suggesting that CR-1 may be expressed by cells with an increased pluripotency. CR-1-expressing cells extracted from the C8161-CR1-CS+ tumors show increased tumor growth rate and aggressive behavior when re-injected into nude mice compared with the growth rate of the initial first generation C8161-CR1-CS+ tumors. A major factor contributing to poor clinical outcome is the inability to eradicate all cancer cells; therefore, one possibility is that CR-1-expressing cells may represent a residual subpopulation with chemoresistant properties that can contribute to rapid tumor recurrence. However, it is also possible that priming of the re-injected CR-1-expressing cells by growth factors or other components of the initial host microenvironment may also have contributed, in part, to the more rapid and aggressive growth seen in the subsequent transplants.

Currently, treatment of melanoma is hampered by poor therapeutic response rates and disease relapse. This may be ascribed to the presence of a stem cell-like subpopulation of melanoma cells capable of giving rise to a progeny with potential for disease recurrence and aggressive growth. The identification of molecular markers that can reliably detect and eradicate this subpopulation of melanoma cells is critical. Our data suggest that CR-1 could represent one such marker with potential value in detecting a melanoma subpopulation with stem cell-like characteristics, including slow growth and expression of a drug resistance marker. Based on our observations, it is possible that patients with CR-1-positive melanoma may go on to develop advanced stage disease that could potentially benefit from early aggressive intervention. Future studies are required to confirm the potential relationship between CR-1 and chemoresistance. Given the lack of evidence in the literature supporting any potential germline or somatic mutations as possible cause(s) for deregulated CR-1 expression in cancer cells, pharmacogenetic approaches for the detection and targeting of CR-1-expressing cells seems unlikely at this time. Based on our current data, targeting CR-1 should predominantly affect CR-1-expressing cancer cells, since these are most likely to be dependent on its biological effect. Thus, this approach could potentially suppress stem cell-like subpopulations and confer chemosensitivity, thereby improving patient survival.

Materials and Methods

Human melanoma cells, immunofluorescence cytochemistry, cell sorting and qRT-PCR

The human melanoma cell lines C8161 and ROS184 were used and cultured as previously described.^7,9 These cell lines were authenticated at the Molecular Diagnostic/HLA Typing Core at Lurie Children’s Hospital of Chicago (C8161) and INT-Fondazione Pascale (ROS184). Immunofluorescence cytochemistry was performed as previously reported¹⁷ using the following primary antibodies: anti-CR-1 (600401997, Rockland Immunochemicals); anti-Oct4 (ab19857, Abcam); anti-MDR-1 (sc555110, S. Cruz) and images captured using Zeis Meta510 confocal microscope. Cells were fixed in methanol in order to determine the full range of cellular expression of CR-1 within the population of the melanoma cell lines analyzed. For FACS sorting and cytometric analysis, PE-conjugated mouse anti-Cripto-1 (FAB2772P, R&D Systems) was used to more accurately quantify and sort live cell surface CR-1-expressing melanoma cells (CR1-CS+), as previously described.¹² To detect CR-1, MDR-1 and Oct4 mRNA expression in CR1-CS+ cells, qRT-PCR was performed as previously described,²³ using specific primers (Applied Biosystems, Grand Island, NY: Oct4- Hs00999632_g1; CR-1- Hs02339499_g1; MDR-1- Hs00184500_m1).

In vitro growth of Cripto-1-positive cells and western blotting

To evaluate in vitro proliferation rate and growth characteristics, CR1-CS+ melanoma cells were enriched by FACS and cultured in medium used to maintain embryonic stem cells, as this has been shown to favor the selection and maintenance of melanoma cells with stem cell-like characteristics.²⁴ Every 24 h for 4 d, cell cultures were harvested with EDTA, and cells were counted from triplicate wells to obtain mean cell numbers. Lysates were also obtained from these cells for WB analysis of CR-1 (Rockland Immunochemicals), Oct4 (Abcam), JARID-1B (22260002, Novus), Src (05184, Upstate), P-Src^(Y418) (4460G, Invitrogen), P-Smad2 (44244G, Invitrogen), Smad2 (5339, Cell Signaling) and Nodal (sc28913, S. Cruz), as previously described.^7,11,13

In vivo growth of Cripto-1-positive cells

To determine tumorigenicity and growth rates of these cells in vivo, the sorted CR1-CS+ cells were injected subcutaneously in groups of 6–8-wk-old female nude mice (n = 5 mice/group) in order to establish an orthotopic xenograft model, as previously described.⁹ Tumor volumes [(width)² × (length)/2] were regularly measured using appropriate calipers and growth curves plotted for comparison of growth rates between CR1-CS+ cells with tumors formed by melanoma cells depleted of the CR1-CS+ subpopulation (CR1-CS−) or unsorted parental melanoma cells. Tumors formed by CR1-CS+, CR1-CS− and unsorted melanoma cells were harvested from the nude mice after 21 d, examined histologically and stained for CR-1 (Rockland Immunochemicals) and MDR-1 (S. Cruz) by IHC. A portion of the CR1-CS+ tumors was used to harvest tumor cells as previously described.²⁵ Cells dissociated from CR1-CS+ tumors were then cultured in vitro for 24–48 h, as described above, to allow separation of dead cells and other cellular debris. A portion of the cells harvested from the CR1-CS+ tumors was again analyzed for CR-1 expression by WB, while another portion was re-injected (CR-1R) orthotopically into a new set of 6–8-wk-old female nude mice (n = 6 mice/group) to evaluate tumor growth and histologic characteristics of the transplanted CR-1R cells. All animals were housed in the fully accredited Lurie Children’s Hospital of Chicago Research Center animal facility, and all experiments were performed in accordance with current animal protocol, approved by the Center’s Institutional Animal Care and Use Committee.

Statistical analysis

Mann-Whitney U test was used for statistical analysis of data when appropriate. Data was considered statistically significant when p < 0.05.

Glossary

Abbreviations:

CR-1

Cripto-1

MSCs

melanoma stem cells

TGF

transforming growth factor

C8161-CR1-CS+

cell surface CR-1-expressing C8161 cells

C8161-CR1-CS−

C8161 depleted of cell surface CR-1-expressing cells

ABC

ATP-binding cassette transporters

IFC

immunofluorescence cytochemistry

FACS

fluorescence-activated cell sorting

qRT-PCR

quantitative real-time PCR

IHC

immunohistochemistry

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.