Stem cell characteristics of transplanted rat mammary clonogens

Rat mammary glands contain a subpopulation of clonogenic epithelial cells with large proliferation and differentiation potentials. When transplanted, the clonogens in monodispersed rat mammary epithelial cell suspensions give rise to either alveolar units (AUs) or ductal units (DUs) depending on the nature of the hormonal milieu in the graft recipient. Clonogens are also the primary cells of origin of mammary cancer following exposure to ionizing radiation or chemical carcinogens. Given the other stem cell characteristics of mammary clonogens, it would be expected that the primary AUs and DUs to which they give rise when grafted and hormonally stimulated (a) would be derived from the same clonogenic cell subpopulation, (b) would contain all of the functionally differentiated cell types of homologous parts of comparably stimulated mammary glands in situ, and (c) would also contain clonogen subpopulations capable when subtransplanted of giving rise to secondary AUs and DUs of similar cell composition. The current experiments were designed to test these expectations. The data are discussed in the context of results of previous studies with this and other experimental models. The results further support the conclusion that rat mammary clonogens are multipotent mammary stem cells.     

Proliferation kinetics of a murine fibrosarcoma during fractionated irradiation

Accelerated growth of tumor clonogens during the course of fractionated irradiation has been considered one of the major causes of radiation treatment failure. Alterations in clonogen growth rate could occur through three basic mechanisms: changes in cell-loss factor, changes in cell-cycle time, and recruitment of previously quiescent cells into the proliferative pool. This study was designed to assess changes in the cell-cycle time of clonogens of a murine fibrosarcoma during fractionated irradiation using an artificial pulmonary micrometastasis model. Lung colonies of various ages (4 h, 1 day, or 4 days) were exposed to single doses of irradiation ranging from 5-13 Gy; the fraction of surviving colonies was used to determine the preirradiation growth kinetics. The growth kinetics during fractionated irradiation was derived from colony-survival data of 4-day-old micrometastasis exposed to single doses or to 2, 5, 9, and 15 fractions separated by 4, 12, or 24-h intervals. The size of dose fractions used ranged from 1.7 to 14 Gy. The estimated clonogen doubling times before irradiation and during overall treatment periods of up to 14 days were 0.71 and 1.1 days, respectively. This significant (P less than 0.0001) increase in the doubling time was most likely a consequence of lengthening of the overall cell-cycle time of the clonogens by radiation-induced division delay. This observation suggests that accelerated growth, when it occurs in some tumors during fractionated treatment, is the result of a decreased cell-loss factor or recruitment of quiescent cells, but not a shortening of the cell-cycle time of the clonogens.     

Factors determining the ability of cytokine-activated killer cells to lyse human ovarian carcinoma targets

Lysis of human ovarian carcinoma cells by natural killer (NK) cells, interferon-alpha activated NK cells (alpha-NK) and lymphokine-activated killers cells (LAK) was studied using both fresh tumor cells and a cell line (HEY) as targets. A clonogenic assay to measure cell kill was more sensitive than a 4-h 51Cr release assay. Both assays showed that single cells were more effectively lysed than were tumor clumps (spheroids). Freshly isolated tumor cells studied in the 51Cr release assay appeared equally susceptible to lysis by LAK cells whether in the form of clumps or single cells, but NK and alpha-NK effectors appeared much less effective in lysing susceptible target cells when they were in clumps. Tumor cells from some patients showed marked resistance to lysis by NK and alpha-NK cells in fractions enriched for clonogenic cells, even when tested in a single cell-suspension, whereas LAK cells were always cytolytic. These data suggest that intrinsic resistance of ovarian carcinoma to lysis by LAKs is unlikely to explain failure of LAK + IL-2 therapy to eradicate tumor in vivo.     

Development of an in vitro colony formation assay for the evaluation of in vivo chemotherapy of a rat brain tumor.

An in vitro colony formation assay for the evaluation of in vivo brain tumor therapy has been developed. When plated, disaggregated cells derived from solid tumors proliferated to form relatively homogeneous colonies after a latency period of 2 to 6 days. Increasing concentrations of fetal calf serum enhanced colony-forming efficiency (CFE) with a plateau between 7 and 16%. Supplementation with either irradiated feeder cells (10(3) to 10(5) cells per dish), or medium conditioned by 1 to 3 days of in vitro incubation with the same cell line, doubled the CFE. The density of tumor cells (untreated or previously treated with chemotherapeutic agents) did not affect the CFE when a minimum of 10(4) total cells (tumor plus feeder) were plated. Therefore, in this system the optimal experimental conditions for evaluating chemotherapy and radiotherapy require incubation of disaggregated tumor cells for 12 days in medium containing 10% of fetal calf serum and enough feeder cells to provide a minimum of 10(4) cells per dish. The CFE for untreated tumors was 18 +/- 10% (+/-S.D.), demonstrating that there is significant biological variation. The assay appeared sensitive, with reproducible results, when applied to individual chemically treated tumors. An estimate of the percentage of clonogenic cells affected by in vivo chemotherapy may be obtained by comparing the CFE of cells from treated and untreated tumors. This assay can measure up to a 5 log(10) cell kill, and it should prove to be valuable in developing more effective regimens for the treatment of solid tumors in animals and man.     

Intestinal crypt clonogens: a new interpretation of radiation survival curve shape and clonogenic cell number

Estimates of the clonogen content (number of microcolony-forming cells) of murine intestinal crypts using microcolony assays show an apparent dependence on the radiation dose used in the assay of clonogen content. Crypt radiation survival curves often show increased curvature beyond that expected on the basis of the conventional linear-quadratic model. A novel form of crypt survival curve shape is proposed based on two contributory mechanisms of crypt killing. Six previously published sets of microcolony data were re-analysed using a dual-kill model, where target cells are killed by two contributory mechanisms, each described by a linear-quadratic function of dose. The data were analysed as two series--high-dose rate and low-dose rate irradiation. The data were fitted to the models using direct maximization of a quasi-likelihood, explicitly allowing for overdispersion. The dual-kill model can reproduce both the apparent dose-dependence of the clonogen estimates and the high-dose curvature of the dose-response curves. For both series of data the model was a significantly better fit to the data than the standard linear-quadratic model, with no evidence of any systematic lack of fit. The parameters of the clonogenic cell component of the model are consistent with other studies that suggest a low clonogen number (somewhat less than five) per crypt. The model implies that there is a secondary mechanism decreasing clonogen survival, and hence increasing clonogen number estimates, at high doses. The mechanisms underlying the modification of the dose-response are unclear, and the implied mechanisms of, for example, slow growth, induced either directly in the surviving cells or indirectly through stromal injury or bystander effects are only speculative. Nevertheless, the model fits the data well, demonstrating that there is greater kill at high doses in these experimental series than would be expected from the conventional linear-quadratic model. This alternative model, or another model with similar behaviour, needs to be considered when analysing in detail and interpreting microcolony data as a function of dose. The implied low number of < or = 5 of these regenerative and relatively radioresistant clonogenic cells is distinct from a similar number of much more radiosensitive precursor stem cells which undergo early apoptosis after doses around 1 Gy.     

Ionizing radiation damage to DNA: molecular aspects

Radioresistant tumor cells are found in tumor specimens from patients in whom radiotherapy has failed or whose tumors have recurred after therapy. This suggests that inherent cellular radioresistance may in part underlie the failure of radiotherapy, and therefore determination of the presence of resistant cells within a tumor might be a useful predictor of response to radiation therapy. Most standard clonogenic assays of radiation response are time-consuming, and alternative assays of radiation response are being sought. In an earlier publication (J. L. Schwartz et al., Int. J. Radiat. Oncol. Biol. Phys. 15, 907-912, 1988), we reported that radioresistant human tumor cells rejoin DNA double-strand breaks, as measured by DNA neutral filter elution (pH 9.6), faster than more sensitive cell lines. To determine whether DNA elution might have potential as a rapid predictive assay, we examined the relationship between the rate of DNA double-strand break rejoining and radiosensitivity in nine first-passage-after-explant squamous cell carcinomas under conditions that minimized the influence of nontumor and nonclonogenic cells. The frequency of DNA double-strand breaks measured 1 h after irradiation with 100 Gy 60Co gamma rays was used as an estimate of relative rejoining rate. This number is a reflection of both the initial DNA double-strand break frequency and the amount of repair that occurs in 1 h. The relative break frequency was compared to radiosensitivity as measured by standard clonogenic survival assays in later passages (p3-p14) of these same cells. A significant relationship (r = 0.61, P less than 0.01) was found between break frequency measured in first-passage cells and radiosensitivity measured in later passages, suggesting that the neutral elution assay as described here has some promise as a relatively rapid assay of the radiosensitivity of human tumor cells.     

Radiogenic initiation of thyroid cancer: a common cellular event

A transplantation system for clonogenic cells in rat thyroid was used, which allows quantitative evaluation of both the acute cytotoxicity and the late neoplastic effects of ionizing radiation at the cellular level in vivo. We have obtained direct experimental evidence that radiogenic initiation of neoplasia in vivo is a common cellular event, and that cell number influences the expression of initiation. Ten per cent of those graft sites which had received 26 clonogens surviving 5 Gy developed carcinomas, while 4 per cent of those which received 26 unirradiated clonogens developed carcinomas. By comparison, 26 per cent of the sites that were inoculated with 411 surviving irradiated clonogens developed carcinomas while none of the 38 transplant sites that received 411 unirradiated clonogens developed carcinomas. Total tumour incidence (carcinomas plus adenomas) followed the same pattern.     

Survival curves for normal-tissue clonogens: a comparison of assessments using in vitro, transplantation, or in situ techniques

A survey of survival curves in the literature, for clonogenic cells (clonogens) in normal tissues, highlights the following features: the sensitivity of some human and dog clonogens apparently is greater than that of their counterparts in mice and sheep, assessed in vitro. However, this should be interpreted with caution because of the possibility of cell selection and the ability to modify sensitivity markedly in some systems by variations in growth conditions; extrapolation numbers are in general higher when assessed in vivo than in vitro. This is due partly to the lack of measurements of repair of potentially-lethal damage using many assays in vitro. This feature increases the extrapolation number when measured using transplantation assays in vivo; epithelial clonogens in vivo demonstrate a remarkable similarity in sensitivity between tissues. The range is similar for clonogens assayed in situ or by transplantation, and this argues against the possibility that a resistant subpopulation may be selected in most assays in situ. It is emphasized from the comparisons that caution must be exercised in extrapolating results, obtained for clonogens assayed in vitro or by transplantation in vivo, to the situation in situ.     

The potential value of the neutral comet assay and the expression of genes associated with DNA damage in assessing the radiosensitivity of tumor cells

The assessment of tumor radiosensitivity would be particularly useful in optimizing the radiation dose during radiotherapy. Therefore, the degree of correlation between radiation-induced DNA damage, as measured by the alkaline and the neutral comet assays, and the clonogenic survival of different human tumor cells was studied. Further, tumor radiosensitivity was compared with the expression of genes associated with the cellular response to radiation damage. Five different human tumor cell lines were chosen and the radiosensitivity of these cells was established by clonogenic assay. Alkaline and neutral comet assays were performed in γ-irradiated cells (2-8Gy; either acute or fractionated). Quantitative PCR was performed to evaluate the expression of DNA damage response genes in control and irradiated cells. The relative radiosensitivity of the cell lines assessed by the extent of DNA damage (neutral comet assay) immediately after irradiation (4Gy or 6Gy) was in agreement with radiosensitivity pattern obtained by the clonogenic assay. The survival fraction of irradiated cells showed a better correlation with the magnitude of DNA damage measured by the neutral comet assay (r=-0.9; P<0.05; 6Gy) than evaluated by alkaline comet assay (r=-0.73; P<0.05; 6Gy). Further, a significant correlation between the clonogenic survival and DNA damage was observed in cells exposed to fractionated doses of radiation. Of 15 genes investigated in the gene expression study, HSP70, KU80 and RAD51 all showed significant positive correlations (r=0.9; P<0.05) with tumor radiosensitivity. Our study clearly demonstrated that the neutral comet assay was better than alkaline comet assay for assessment of radiosensitivities of tumor cells after acute or fractionated doses of irradiation.     

Cytological and functional analysis of inflammatory infiltrates in human malignant tumors: III. Further functional investigations using cultured autochthonous tumor cell lines and freeze-thawed infiltrating inflammatory cells

Short-term monolayer cultures, dominated by cells with malignant characteristics, were established from human tumors displaying an unusually strong host-inflammatory response. Upon repeated testings in the ⁵¹Cr release cytotoxicity assay, blood leukocytes were frequently cytotoxic (a) to autologous and allogeneic tumor cells, without any apparent restriction as to tumor origin or HLA type, and (b) to the so-called natural killer (NK) target cells. The anti-tumor cytotoxicity disappeared with time. The in situ inflammatory cells, freshly isolated or recovered from the deep freeze, did not display any type of cytotoxic activity. Nor were they notably suppressive to either type of blood leukocyte cytotoxic activity in the ⁵¹Cr release assay. Cytological analysis demonstrated that the “large granular lymphocytes” (LGL), known to be largely responsible for the NK activity in man, were prominent in the blood but not in the inflammatory infiltrate. These preliminary observations suggest that lack of cytotoxic activity in situ correlates with the absence of effector cells in the inflammatory infiltrate.     

Correlation between the clonogenic initial slope and the response of polykaryon-forming units: the behavior of strains defective in XRCC5 and ATM and the heritability of small variations in radioresponse

The polykaryon-forming unit (PFU) assay measures the survival of multiple cycles of DNA synthesis after exposure to ionizing radiation, and it is known that there is a strong correlation between the slope of the PFU dose-response curve and the clonogenic initial slope. This suggests that DNA lesions expressed in clonogens are also important in PFU. Cells having a mutation in XRCC5 (also known as Ku80; strain xrs-6) and ATM (strain AT5BIVA) were hypersensitive in the PFU assay and in clonogens, while a strain of xrs-6 cells transfected with hamster wild-type XRCC5 cDNA displayed wild-type resistance in both assays. These data suggest that the DNA double-strand break (DSB) is an important lesion in PFU, although the relative radioresistance of PFU compared to clonogens indicates differential DSB toxicity. We propose that this results from the absence of cytokinesis-related loss of DNA fragments. Small variations in the radioresponse of PFU were observed between CHO K1 cell substrains, such that the xrs parental substrain RR-CHOK1 (carrying wild-type XRCC5) was more sensitive than an independent K1 substrain (E-CHOK1). Somatic hybridization showed that this variation is heritable and that the resistant E phenotype is dominant. In RR-CHOK1 cells there was a biphasic PFU radioresponse, which suggests that there may be transient expression at a locus selectively affecting PFU sensitivity.     

A hypo-osmotic medium to disaggregate tumor cell clumps into viable and clonogenic single cells for the human tumor stem cell clonogenic assay

A hypo-osmolar medium and tissue processing technique is described which is useful for disaggregation of residual human tumor cell clumps persisting after mechanical or enzymatic treatment of solid tumors and malignant effusions. The addition of the hypo-osmolar procedure to the standard methods for disaggregation increased the viable single cell yield in solid tumors by 47% and in malignant effusions by 67%. In 5 of the 26 solid tumor specimens tested in the human tumor stem cell assay, clonogenic single cells were obtained with the hypo-osmolar procedure, whereas no growth was observed using standard methods. Overall, the success rate for clonogenicity increased from 46% to 65% for the 26 solid tumors, with the major improvement occurring in ovarian cancer. Clonogenicity was obtained in 80% of malignant effusions both by standard methods and the hypo-osmolar techniques. The increased total yield of clonogenic cells obtained with this procedure enhances the opportunity for experimental versatility and in vitro drug testing.     

Probing lethal damage expression in cytochalasin B-induced polykaryons by radiation quality

The polykaryon-forming unit (PFU) cell survival assay is based on the postirradiation flow cytometric analysis of the DNA content accumulated in high-ploidy cells (polykaryons) induced by the cytokinesis inhibitor cytochalasin B and can provide a meaningful measure of cell radiosensitivity. In this assay, cell survival is defined as the ability to form a polykaryon of a given ploidy after irradiation. The slope of the polykaryon dose response has been shown to be highly correlated with the initial slope of the clonogenic survival curves after gamma irradiation, which implies a common subset of lethal lesions. We reported previously on an apoptotic mode of cell death in the polykaryon system and on the heritability of small variations in polykaryon radioresponse. We now show that exposure of PFUs to a given dose of alpha particles results in a greater reduction in the proportion of cells able to reach at least 16C when compared to the same dose of low-LET radiation. This reduction is less than that observed in the low-dose (alpha term) region of the clonogenic curve. On the basis of published LET-dependent spectra of radiation-induced DNA damage, we suggest that this behavior reflects a differential expression of lethal damage that can be probed by varying the LET of the radiation and that base damages contributing additional complexity to clustered DNA lesions may be more deleterious in PFUs than in clonogens.     

GPI/AMF inhibition blocks the development of the metastatic phenotype of mature multi-cellular tumor spheroids

Epithelial–mesenchymal transition (EMT) and cellular invasiveness are two pivotal processes for the development of metastatic tumor phenotypes. The metastatic profile of non-metastatic MCF-7 cells growing as multi-cellular tumor microspheroids (MCTSs) was analyzed by determining the contents of the EMT, invasive and migratory proteins, as well as their migration and invasiveness potential and capacity to secrete active cytokines such as the glucose phosphate isomerase/AMF (GPI/AMF). As for the control, the same analysis was also performed in MCF-7 and MDA-MB-231 (highly metastatic, MDA) monolayer cells, and in stage IIIB and IV human metastatic breast biopsies. The proliferative cell layers (PRL) of mature MCF-7 MCTSs, MDA monolayer cells and metastatic biopsies exhibited increased cellular contents (2–15 times) of EMT (β-catenin, SNAIL), migratory (vimentin, cytokeratin, and fibronectin) and invasive (MMP-1, VEGF) proteins versus MCF-7 monolayer cells, quiescent cell layers of mature MCF-7 MCTS and non-metastatic breast biopsies. The increase in metastatic proteins correlated with substantially elevated cellular abilities for migration (18-times) and invasiveness (13-times) and with the higher level (6-times) of the cytokine GPI/AMF in the extracellular medium of PRL, as compared to MCF-7 monolayer cells. Interestingly, the addition of the GPI/AMF inhibitors erythrose-4-phosphate or 6-phosphogluconate at micromolar doses significantly decreased its extracellular activity (> 80%), with a concomitant diminution in the metastatic protein content and migratory tumor cell capacity, and with no inhibitory effect on tumor lactate production or toxicity on 3T3 mouse fibroblasts. The present findings provide new insights into the discovery of metabolic inhibitors to be used as complementary therapy against metastatic and aggressive tumors.     

26S proteasome activity is down-regulated in lung cancer stem-like cells propagated in vitro

Cancer stem cells (CSCs) are a small subset of cancer cells capable of self-renewal and tumor maintenance. Eradicating cancer stem cells, the root of tumor origin and recurrence, has emerged as one promising approach to improve lung cancer survival. Cancer stem cells are reported to reside in the side population (SP) of cultured lung cancer cells. We report here the coexistence of a distinct population of non-SP (NSP) cells that have equivalent self-renewal capacity compared to SP cells in a lung tumor sphere assay. Compared with the corresponding cells in monolayer cultures, lung tumor spheres, formed from human non-small cell lung carcinoma cell lines A549 or H1299, showed marked morphologic differences and increased expression of the stem cell markers CD133 and OCT3/4. Lung tumor spheres also exhibited increased tumorigenic potential as only 10,000 lung tumor sphere cells were required to produce xenografts tumors in nude mice, whereas the same number of monolayer cells failed to induce tumors. We also demonstrate that lung tumor spheres showed decreased 26S proteasome activity compared to monolayer. By using the ZsGreen-cODC (C-terminal sequence that directs degradation of Ornithine Decarboxylase) reporter assay in NSCLC cell lines, only less than 1% monolayer cultures were ZsGreen positive indicating low 26S proteasome, whereas lung tumor sphere showed increased numbers of ZsGreen-positive cells, suggesting the enrichment of CSCs in sphere cultures.     

Three tumor sensitivity tests evaluated with mouse tumors

The predictive value of three types of tumor sensitivity tests was evaluated using mouse tumors. Sensitivities of osteosarcoma C22LR, Lewis lung and M2661 carcinoma were determined for the following drugs: DNA interacting or alkylating agent (doxorubicin, cisplatin, 1,3-bis(2-chloroethyl)-1-nitrosourea, melphalan), antimetabolite (5-fluorouracil, methotrexate) and microtubule inhibitor (vinblastine, vincristine). Volume measurements of the subcutaneously growing tumors after treatment with the same drugs were considered to be the traditional reference system with which the results of the in vitro clonogenic assay, the labeled precursor incorporation assay and the subrenal capsule assay were compared. Results obtained with the in vitro clonogenic assay were highly reproducible. With the 1-h drug exposure technique the predictive accuracy was 71%. This result is in the same range as those found by others for human tumors. Predictive accuracy after continuous drug exposure was only 25%. Vinblastine, vincristine and cisplatin caused no inhibition of labeled precursor incorporation. However, the assay is too unreliable to use, due to the extreme variability when used with the other drugs. From 31 consecutively performed duplicate tests in the subrenal capsule assay, nine showed opposite results. This degree of disagreement between duplicate test results was considered too high to make reliable predictions of tumor sensitivity with this assay.     

Initial slope of radiation survival curves is characteristic of the origin of primary and established cultures of human tumor cells and fibroblasts

The published survival curves of 110 human tumor cell lines and 147 nontransformed human fibroblast strains have been reanalyzed using three different statistical methods: the single hit multitarget model, the linear-quadratic model, and the mean inactivation dose. The 110 tumor cell lines were classified in two ways: (a) into three categories defined by clinical radiocurability criteria, and (b) into seven categories based on histopathology. The 147 fibroblast strains were divided into eight genetic groups. Differences in the radiosensitivities of both the tumor cell and fibroblast groups could be demonstrated only by parameters that describe the slopes of the initial part of the survival curves. The capacity of the survival level to identify significant differences between groups was dose dependent over the range 1 to 6 Gy. This relationship showed a bell-shaped curve with a maximum at 1.5 Gy for the tumor cell lines and 3 Gy for the fibroblasts. Values for intrinsic radiosensitivity for a number of groups of tumors have also been obtained by primary culture of tumor cells. These values are strictly comparable to those obtained by clonogenic methods. This confirms that intrinsic radiosensitivity is a determinant of the response of tumor cells to radiotherapy and suggests that tissue culture methods may be used as a predictive assay.     

Cell Density Impacts Epigenetic Regulation of Cytokine-Induced E-Selectin Gene Expression in Vascular Endothelium

Growing evidence suggests that the phenotype of endothelial cells during angiogenesis differs from that of quiescent endothelial cells, although little is known regarding the difference in the susceptibility to inflammation between both the conditions. Here, we assessed the inflammatory response in sparse and confluent endothelial cell monolayers. To obtain sparse and confluent monolayers, human umbilical vein endothelial cells were seeded at a density of 7.3×10³ cells/cm² and 29.2×10³ cells/cm², respectively, followed by culturing for 36 h and stimulation with tumor necrosis factor α. The levels of tumor necrosis factor α-induced E-selectin protein and mRNA expression were higher in the confluent monolayer than in the sparse monolayer. The phosphorylation of c-jun N-terminal kinase and p38 mitogen-activated protein kinase or nuclear factor-κB activation was not involved in this phenomenon. A chromatin immunoprecipitation assay of the E-selectin promoter using an anti-acetyl-histone H3 antibody showed that the E-selectin promoter was highly and specifically acetylated in the confluent monolayer after tumor necrosis factor α activation. Furthermore, chromatin accessibility real-time PCR showed that the chromatin accessibility at the E-selectin promoter was higher in the confluent monolayer than in the sparse monolayer. Our data suggest that the inflammatory response may change during blood vessel maturation via epigenetic mechanisms that affect the accessibility of chromatin.     

The Human Melanoma Side Population Displays Molecular and Functional Characteristics of Enriched Chemoresistance and Tumorigenesis

Melanoma remains the most lethal skin cancer, mainly because of high resistance to therapy. Side population (SP) cells are found in many types of cancer and are usually enriched in therapy-resistant as well as tumorigenic cells. Here, we identified a Hoechst dye-effluxing SP in a large series of human melanoma samples representing different progression phases. The SP size did not change with disease stage but was correlated with the prognostic “Breslow’s depth” in the primary (cutaneous) tumors. When injected into immunodeficient mice, the SP generated larger tumors than the bulk “main population” (MP) melanoma cells in two consecutive generations, and showed tumorigenic capacity at lower cell numbers than the MP. In addition, the SP reconstituted the heterogeneous composition of the human A375 melanoma cell line, and its clonogenic activity was 2.5-fold higher than that of the MP. Gene-expression analysis revealed upregulated expression in the melanoma SP (versus the MP) of genes associated with chemoresistance and anti-apoptosis. Consistent with these molecular characteristics, the SP increased in proportion when A375 cells were exposed to the melanoma standard chemotherapeutic agent dacarbazine, and to the aggravating condition of hypoxia. In addition, the SP showed enhanced expression of genes related to cell invasion and migration, as well as to putative (melanoma) cancer stem cells (CSC) including ABCB1 and JARID1B. ABCB1 immunoreactivity was detected in a number of tumor cells in human melanomas, and in particular in clusters at the invasive front of the primary tumors. Together, our findings support that the human melanoma SP is enriched in tumorigenic and chemoresistant capacity, considered key characteristics of CSC. The melanoma SP may therefore represent an interesting therapeutic target.