|Home | About | Journals | Submit | Contact Us | Français|
Side-population (SP) analysis has been used to identify progenitor cells from normal and malignant tissues as well as revealing tumor cells with increased resistance to radiation and chemotherapy. Despite enhanced chemoresistance, tumor SP cells may still express tumor associated antigens (TAA) which may render them susceptible to elimination by the immune system. In this study, we show that both Hodgkin’s lymphoma (HL) cell lines and primary HL tumor samples contain a distinct SP phenotype. Importantly, while these cells showed increased resistance to gemcitabine, a commonly used drug for the treatment of refractory HL, HL SP cells also expressed higher levels of the TAAs MAGEA4, SSX2, Survivin and NY-ESO-1 which allowed them to be specifically recognized and killed by TAA-specific cytotoxic T lymphocytes. This study suggests that chemoresistant HL SP cells can be targeted by the immune system, providing a rationale for combined chemotherapy and immunotherapy for the treatment of HL.
Hodgkin’s lymphoma (HL) is a common hematological malignancy, for which standard chemotherapy regimens produce long-term, disease-free survival in more than 85% of patients . The remainder of these patients, however, will die of relapse , likely caused by a small number of cells resistant to chemotherapy or radiation treatment that are not eliminated by the endogenous immune system. SP analysis has been used to identify cells with stem/progenitor cell-like characteristics from normal [3, 4] and malignant tissues [5–13]. However, SP cells also express multi-drug transporter proteins, including MDR1 and ABCG2 [14, 15], which not only efflux Hoechst dye but also rapidly reduce the intracellular concentrations and thus the cytotoxicity of many commonly used therapeutic drugs [16–18].
We previously identified tumor cells with primary drug resistance using side-population (SP) analysis . Despite their resistance to chemotherapeutics, SP cells may still be susceptible to immune responses since Englemann and colleagues demonstrated that SP cells from the MCF7 breast cancer cell line, which are both resistant to chemotherapeutic drugs and radiation , express the tumor-associated antigen (TAA) MUC1, which is a target of antigen-specific CTL [20, 21]. Hence, the identification of antigens expressed by SP tumor cells may allow the development of immune-based therapies that target tumor cells resistant to conventional treatment modalities. Our own studies have suggested the clinical relevance of immune targeting of chemoresistant cells, since treatment of relapsed/chemoresistant Epstein-Barr virus (EBV) positive HL with cytotoxic T lymphocytes (CTL) specific for EBV lytic antigens  or latent antigens (e.g. LMP-1 and 2) can induce complete disease responses . Additionally, Till et al showed that infusion of CTL genetically engineered to express an anti-CD20 chimeric antigen receptor could also promote partial and complete responses in patients with relapsed and refractory B cell lymphomas . In other words, adoptive T cell therapy may eradicate chemoresistant tumors provided that the antigenic targets are expressed by the residual cells.
Our current data show the existence of an SP subset in HL which have primary drug resistance, but which remain (like classical Hodgkin’s Reed-Sternberg (HRS) cells) CD30 positive, and express high levels of TAAs following demethylation with decitabine. The ability to target this SP subset by immunotherapy may further improve the response rates in subjects with relapsed and refractory HL.
We obtained four EBV negative HL cell lines L428, HDLM2, L1236 and L540 from DSMZ (Braunschweig, Germany). All were maintained in RPMI 1640 supplemented with 10% fetal bovine serum (FBS) (FBS; Gibco, Carlsbad, CA) and 2 mM L-glutamine (GlutaMAX; Gibco). Primary tumor samples were obtained through a Baylor College of Medicine Institutional Review Board approved protocol. Fresh tissue specimens were prepared on the day of resection by passing the cultured cells multiple times through a 100 micron filter to make a single cell suspension. Cells were then prepared and analyzed as described below.
For SP analysis, HL cell lines and primary HL tumor cells were resuspended at 1×106 cells/ml in pre-warmed RPMI with 2% FBS and 10 mM HEPES (Sigma-Aldrich, St. Louis, MO). Hoechst 33342 dye (Sigma) was added to a final concentration of 5 μg/ml as previously described . Cells were then incubated at 37 C for 105 minutes, washed in ice cold Hank’s buffered salt solution supplemented with 2% FBS and 10 mM HEPES (HBSS+) and then immediately placed on ice. Where indicated, cells were co-incubated with 50 μM Verapamil (Sigma) to block Hoechst efflux. For analysis of SP frequency, a live gate was defined by using the Hoechst red and blue axes to exclude dead cells (PI positive) and debris. For further phenotypic analyses, Hoechst labeled cells were subsequently stained with fluorescent-conjugated antibodies: CD3, CD15, CD19, CD20 and CD30 (all from BD Biosciences, San Jose, CA) for 15 minutes on ice. Cells were then washed again and resuspended in ice cold HBSS+ buffer containing 2 μg/ml propidium iodide (PI; Sigma-Aldrich) for dead cell exclusion. Gates for SP and NSP (bulk tumor cells) were subsequently used to determine expression of surface antibodies using a BD LSRII analyzer (BD Bioscience, San Jose, CA). For SP and NSP isolation, we sorted cells using a high-speed cell sorter (MoFlo, Dako Cytomation, Fort Collins, CO) as previously described .
We sorted HL cells from the SP and NSP subsets and grew equal numbers in culture media as described above, either with or without treatment with 10 nM gemcitabine (Eli Lily, Indianapolis, IN). For cell viability assays, 5000 sorted SP or NSP cells were cultured with 2×105 gemcitabine-sensitive L1236 feeder cells (for 2% chemoresistant cell frequency). SP and NSP cells were then either cultured without drug or treated with 10 nM gemcitabine (added daily) for 72 hours. Following drug exposure, we washed the cells and plated them in fresh media without gemcitabine for 48 hours, and counted viable cells by trypan blue dye exclusion. To measure cell proliferation, 2000 SP or NSP cells were plated into 96-well plates in triplicate and cultured with or without gemcitabine for 7 days. On day 7, cells were incubated with 5 μCi [3H]-thymidine for 16 hours and then harvested and counted for [3H]-thymidine incorporation using a TriCarb 2500 TR β-counter (Perkin Elmer, Waltham, MA). Results are shown as counts per minute (CPM). To determine whether SP or NSP cells undergo cell death and apoptosis following gemcitabine exposure, we used the fluorochrome inhibitor of caspases (FLICA) (Abd Serotec, Raleigh, NC), which measures apoptosis by detecting active caspases (poly-caspase; caspase-1, -3, -4, -5, -6, -7, -8 and -9) in living cells . Here, tumor cell lines were first cultured with or without 10 nM for 48 hours, then harvested and stained with Hoechst dye, followed by incubation with carboxyfluorescein labeled FLICA for 5 minutes at 37° and 5% CO2, followed washing and resuspension in PBS containing PI. Cells were placed on ice and analyzed for SP and apoptosis by flow cytometry using the LSRII analyzer (BD).
Cell lines and primary HL samples were cultured in the presence of 1 μM 5-aza-2′-deoxycitabine (Decitabine; Sigma-Aldrich) for 48 hours whereas control cells were cultured in media only. After incubation, cells were washed and cultured without decitabine for an additional 5 days to allow for TAA protein expression. Cells were harvested and resuspended in phosphate buffered saline (PBS; Sigma-Aldrich), mounted onto glass slides (Shandon Cytospin 4 Cytocentrifuge; Thermo Scientific, Waltham, MA) and then fixed with 4% paraformaldehyde (Cytofix; BD). Cells were subsequently permeabilized using 0.3% Triton-X-100 (Gibco) for 5 minutes followed by incubation with Digest ALL1 (Zymed, San Francisco, CA) for 10 minutes at 37°C. To block endogenous peroxidase activity, cells were treated with 3% hydrogen peroxide (Sigma). We used Powervision+ kits (ImmunoVision Technologies, Daly City, CA) according the manufacturer’s recommendation for immunohistochemical analysis. Briefly, slides were first incubated in pre-block/diluent for 30 minutes followed by incubation with mouse anti-human MAGEA4 antibody (AbCam, Cambridge, MA) diluted 1:50 in dilutant for 1 hour at room temperature. We used anti-mouse/anti-rabbit horseradish peroxidase (HRP) to detect MAGEA4 positive cells, followed by enzymatic conversion using the chromogenic substrate 3,3 diaminobenzidine (DAB).
To compare expression of ABC transporter genes, RNA was isolated from equal numbers of sorted SP and NSP cells using Qiagen RNeasy Micro Kits (Qiagen, Valencia, CA) and used as template for OneStep RT-PCR (Qiagen) with gene-specific primers to ABCG2, MDR1, MAGEA4, SSX2, NY-ESO-1 and Survivin (purchased from SABiosciences, Frederick, MA) and normalized to expression of the housekeeping gene, GAPDH (SABiosciences). To measure TAA gene expression and the effects of decitabine demethylation on gene expression, cells were cultured with and without decitabine as described above. 48 hours after decitabine exposure, cells were harvested and analyzed for TAA transcripts using One-Step RT-PCR with the above gene-specific primers using conditions specified by the manufacturer. PCR bands were visualized on 1% agarose gels with ethidium bromide (Sigma).
To isolate and expand MAGEA4-specific CTL (MAGE-CTL), we generated monocyte-derived dendritic cells (DC) by isolating CD14+ monocytes from peripheral blood mononuclear cells (PBMC) using magnetic CD14+ bead separation (MACS, Miltenyi, Auburn, CA). Purified monocytes (>80%) were subsequently cultured in DC media (CellGenix, Antioch, IL) supplemented with 1000 U/ml IL-4 and 800 U/ml GM-CSF (both from R&D Systems, Minneapolis, MN). After 5 days, DC were matured in a cocktail containing PGE2 (1 μg/ml), IL-1β (10 ng/ml), IL-6 (100 ng/ml) and TFN-α (10 ng/ml) (all from R&D). On day 7, DC were harvested and pulsed with 2 μg/ml of a MAGEA4 overlapping peptide library (PepMix library, 15mers overlapping by 11 amino acids; JPT Technologies, Berlin, Germany), irradiated (30 Gy) and used to stimulate autologous PBMC at a ratio of 1:10 DC to PBMC on days 0 and 9 of the CTL culture. During the initial stimulation, culture media was supplemented with IL-7 (10 ng/ml), IL-12 (1 ng/ml) and IL-15 (2 ng/ml) (all from R&D). During the second and subsequent stimulations, CTL cultures were stimulated with peptide-pulsed DC in the presence of IL-7 alone. On day 12, culture media was supplemented with 50 U/ml IL-2 (Proleukin; Chiron, Emeryville, CA) to promote cell expansion.
IFN-γ ELIspot assays have been previously described . T cells were plated in triplicate and MAGEA4 or an irrelevant peptide library added. Plates were then developed and IFN-γ spot number determined by an independent, blinded observer (ZellNet Consulting, Inc., NY, NY).
To determine whether CTL could recognize and eliminate CTA expressing SP cells, we added decitabine to 1 μM concentration daily for two days to 5×105 HDLM2 tumor cells in triplicate. After drug treatment, the tumor cells were washed and grown in fresh media without drug, and co-cultured with 1×106 MAGE-A4-specific, HLA-A*02 matched or non-matched CTL. After 5 days, the co-cultures were analyzed for SP frequency and cell viability, using trypan blue exclusion. We reported the results as the percentage viable cells in the drug treated versus untreated tumor SP cells. Cell death following co-culture assays was also measured by PI positivity by flow cytometry.
All in vitro data are presented as means plus or minus standard deviation. Where applicable, 2 tailed student t tests were used to determine the statistical significance differences between samples. A p value of ≤.05 was considered significant.
To determine whether HL contains a chemoresistant SP subset, we stained four HL cell lines (HDLM2, L428, L1236 and L540) with Hoechst 33342 dye and analyzed them by flow cytometry. In two of the cell lines, HDLM2 and L428, we detected a distinct SP phenotype forming 0.1% and 0.48% of the total cells, respectively (Figure 1a). To confirm that the SP phenotype was due to transporter dependent mechanisms, we co-incubated the cells with Verapamil, which abolished Hoechst efflux and SP formation (Figure 1b) indicating that this phenotype was not attributable to non-specific staining (Figure 1b). Two of the four HL cell lines, L1236 and L540, lacked SP cells (Figure 1c).
We next sought to determine whether primary HL samples contained SP cells, and whether these cells expressed phenotypic markers characteristic of HRS cells (CD30+CD15+CD20−) . We analyzed 6 primary lymph node biopsies from HL patients and found that all (100%) samples contained a distinct SP phenotype (range = 0.02%–6.5%; mean = 1.7%) (Figure 2a). Fromm and colleagues demonstrated that HRS cells can be detected by flow cytometry using a panel of surface markers (CD30+, CD15dim CD20− and CD45−), and that HRS cells were often found in HRS-T cell rosettes [28, 29]. To determine if SP cells possessed a similar HRS phenotype, we stained with CD30, CD3, CD15, CD19 and CD20 antibodies following Hoechst labeling and compared expression of these antigens on gated SP and NSP (bulk) viable cells. We found a higher proportion of CD30+CD3+CD19−CD20− cells in the SP (mean 24.2%±18%; range 0.0% to 49.0%) fraction compared to NSP cells (mean 8.5%±8.5%; range 1.2% to 22.4%) (p=0.03) (Figures 2b and 2c). In this sample set, we were unable to detect a significant cell population that co-expressed both CD30 and CD15 (not shown). These data indicate that cells isolated from lymph node biopsies from patients diagnosed with HL contain SP cells with an HRS phenotype.
Because of the association of the SP phenotype with the presence of multi-drug resistance proteins, we examined whether the SP cells of HL cell lines expressed MDR1 and ABCG2. The cell lines L428 and HDLM2 were sorted into SP and NSP fractions and RT-PCR used to measure expression of these two transporters. As anticipated, we observed increased levels of ABCG2 and MDR1 in SP cells compared to NSP cells (Figure 3a). Since the SP cells from other hematologic malignancies are more resistant to commonly used chemotherapeutic agents than the NSP subset [30, 31], we assessed the gemcitabine resistance of HL SP cells, since this agent is widely used to treat relapsed HL. We first cultured SP positive cell lines (HDLM2 and L428) and SP negative cell lines (L1236 and L540) with and without gemcitabine and evaluated cell viability after 7 days. The viability of the SP lines HDLM2 and L428 were 81%±22% and 39%±6% respectively. These results were compared to the observed viability of gemcitabine treated HL cell lines which lack SP cells; the viability of each was substantially lower (L1236 7%±2% and L540 14%±2%), indicating that cell lines containing SP cells are more resistant to f gemcitabine than cell lines without (Figure 3b). We compared the SP cell frequency in HDLM2 and L428 before and after exposure to gemcitabine and found a 20 and 60-fold enrichment, respectively (Figure 3c). We then sorted SP and NSP cells from L428 and found only 5% of NSP tumor cells survived gemcitabine treatment compared to 20% of SP cells (Figure 4a), while analysis of DNA synthesis showed higher thymidine incorporation in SP cells than NSP cells after drug exposure (Figure 4b). These effects were not due to Hoechst dye toxicity, as both SP and NSP HL cells had equivalent survival and thymidine incorporation after exposure to Hoechst dye in the absence of gemcitabine (Figure 4c). Enrichment of SP cells and reduction in survival of NSP cells is likely due to greater apoptosis in the NSP (bulk) tumor fraction, since the fluorochrome-labeled inhibitors of caspases (FLICA) assay  showed greater FLICA staining in the NSP fraction (6.7%) compared to the SP fraction (3.9%) 48 hrs after gemcitabine exposure (Figure 4d). Together, these data indicate that SP cells have increased tolerance to gemcitabine compared to bulk tumor (NSP) cells and support the potential of HL SP cells to contribute to disease resurgence following chemotherapy.
We compared expression of HL-associated tumor antigens on SP and NSP cells, examining MAGEA4, SSX2, survivin and NY-ESO-1 [32, 33]. Although none of these antigens were highly expressed in SP or NSP from any cell lines by RT-PCR (not shown) treatment with the demethylating agent decitabine (5-aza-2′-deoxycitidine) induced MAGEA4 protein expression in both HDLM2 and L428 (Figure 5a) and increased RNA expression of SSX2, MAGE-A4, NY-ESO-1 and survivin (Figure 5b). L428 and HLDM2 were sorted into SP and NSP fractions following decitabine exposure and analyzed by quantitative PCR (Q-PCR), where SSX2, MAGEA4, and survivin expression were increased more prominently in the SP fraction compared to bulk tumor cells (Figure 5c). MAGEA4 protein expression in SP and NSP tumor cells was subsequently confirmed by immunohistochemistry demonstrating that demethylation by decitabine treatment can induce TAA expression in chemoresistant SP cells (Figure 5d).
To discover whether chemoresistant SP cells could be targeted by T cells directed to the TAA expressed by the SP cells, we generated MAGEA4-specific CTLs, that responded in an ELIspot assay to MAGEA4 peptides (PepMix), but not to a control peptide library (Aspf16) (Figure 6a). MAGEA4-specific CTLs from HLA-A2+ normal donors could kill decitabine treated HLA-A2+ HDLM2 tumor cells, while CTLs from HLA-A2− donors could not, (mean tumor cell viability: 24% versus 74%, p=0.01) (Figure 6b). In addition, co-culture experiments using MAGEA4-specific CTL with the HDLM2 cell line showed increased killing via PI staining when used in combination with decitabine, and negligible killing without demethylation treatment (Figure 6c). Because SP cells express elevated levels of MAGEA4 following decitabine treatment compared to NSP, we then compared the susceptibility of SP cells and NSP cells to elimination by MAGEA4-specific CTLs. There was no change in SP frequency when HDLM2 were exposed to decitabine or CTLs alone (Figure 7). However, the combination of decitabine treated (HLA-A2+) HDLM2 tumor cells and HLA-A2 matched MAGE-specific CTL dramatically reduced SP cell frequency (p<0.05) (Figures 7a and 7b), showing that MAGEA4-specific CTLs can kill chemoresistant SP HL tumor cells.
In this study, we have found that both HL cell lines and primary HL tumor cells contain a minor side-population of tumor cells that have increased resistance to gemcitabine, but that also express tumor associated antigens which render them susceptible to killing by tumor-specific CTLs following demethylation with decitabine. While this primary drug resistant SP subset may therefore contribute to resurgent disease, a combination of chemotherapy and targeted immunotherapy may be able to reduce tumor burden and decrease the risk of relapse.
An SP subset of cells has been identified in a variety of hematologic malignancies and solid tumors [9, 30, 34–37], but had not yet been reported in HL. We detected SP cells in two HL cell lines, HLDM2 and L428, which conferred primary drug resistance to the cell lines compared to those lacking SP cells. Importantly, we also detected SP cells in primary HL biopsy samples that possessed an HRS phenotype, including increased expression of CD30, CD3 (T cell rosettes [28, 29]) and lack of CD20 (Figure 2). Because not all SP cells were CD30+, it is likely that SP cells found in these biopsies include normal B and T lymphocytes in addition to malignant cells, and that definitive identification of HRS cells with SP properties will need to be confirmed by clonal analyses, However, these results are suggestive that a malignant chemoresistant cell type exists that could contribute to relapse following disease treatment and may be targetable by antigen-specific CTL.
In other malignancies, SP cell resistance to lipophilic anti-neoplastic drugs, such as doxorubicin and mitoxantrone, can be attributed to the expression of the multi-drug transporter proteins including ABCG2 and MDR1 [5, 6]. More recent studies, however, have indicated that malignant SP cells may be more generally resistant to cytotoxic drugs, such as gemcitabine, independent of their excretion by transporter proteins [8,33] and may also resist radiotherapy , This may help explain the resistance of refractory HL to these commonly used treatment modalities [39, 40]. Because HL SP cells have lower apoptosis rates following exposure to gemcitabine compared to NSP cells, and can proliferate even in the presence of the drug, such chemotherapy exposure can simply enrich for drug resistant SP cells (Figures 3 and and4).4). Although we do not know the mechanisms for this transporter-protein independent resistance in HL-SP cells study of SP cells in pancreatic  and gastrointestinal derived cell lines , show the cells express CEACAM6 (carcinoembryonic antigen-related cell adhesion molecule 6), which modulates Akt-associated anti-apoptotic activity, which would provide gemcitabine resistance . Additionally, Bleau and colleagues found that PI3K/Akt regulates ABCG2 expression in SP cells in malignant gliomas, suggesting that malignant SP cells reside in a distinct biological state that renders them insensitive to radiotherapy and a variety of chemotherapy agents .
The potential benefits of CTL for chemoresistant hematological malignancy is supported by the long-term cures obtained following adoptive transfer of donor leukocyte infusions (DLI) to relapsed patients after allogeneic stem cell transplant [44, 45]. Predictably safe and effective immunotherapy, however, requires identification of specific target antigens that reside within the chemoresistant population. We have now shown that HL SP cells from cell lines have enhanced expression of TAAs, including MAGEA4, SSX2, NY-ESO-1 and survivin compared to NSP cells, although such expression was only seen at consistently high levels following decitabine demethylation (Figure 5). Due to technical difficulties in culturing and isolating sufficient numbers of viable SP cells from primary samples following demethylation, we were unable to determine whether this effect is also relevant to HRS cells in vivo. However, other studies have shown that HL cells have increased CpG methylation , and global hypomethylation is commonly seen in leukemias , so that demethylating treatments may be a generally applicable tool for enhancing TAA expression in lymphoma and leukemia if the malignant cells are to be made susceptible to immunotherapies [48–54]. Since multiple TAA are upregulated within chemoresistant SP cells, it should be possible to target several antigens simultaneously, thereby reducing the risk of tumor immune escape .
In summary, we have identified a distinct SP subset in HL cell lines and primary tumor biopsies that are resistant to chemotherapy, but can be eradicated with TAA-specific CTL following decitabine demethylation. Combination therapeutic strategies that use conventional chemotherapy to debulk tumor burden, followed by novel drugs such as histone deacetylation (HDAC) inhibitors and T cell immunotherapy may eliminate residual chemoresistant tumor cells and help prevent disease relapse.
We would like to acknowledge Christopher Threeton and Tatiana Gotsolva for flow cytometry and cell sorting technical assistance. This work was supported by NIH SPORE Grant P50CA126752.