|Home | About | Journals | Submit | Contact Us | Français|
Leukemia stem cells (LSC) play a pivotal role in chronic myeloid leukemia (CML) tyrosine kinase inhibitor (TKI) resistance and progression to blast crisis (BC), in part, through alternative splicing of self-renewal and survival genes. To elucidate splice isoform regulators of human BC LSC maintenance, we performed whole transcriptome RNA sequencing; splice isoform-specific qRT-PCR, nanoproteomics, stromal co-culture and BC LSC xenotransplantation analyses. Cumulatively, these studies show that alternative splicing of multiple pro-survival BCL2 family genes promotes malignant transformation of myeloid progenitors into BC LSC that are quiescent in the marrow niche and contribute to therapeutic resistance. Notably, a novel pan-BCL2 inhibitor, sabutoclax, renders marrow niche-resident BC LSC sensitive to TKIs at doses that spare normal progenitors. These findings underscore the importance of alternative BCL2 family splice isoform expression in BC LSC maintenance and suggest that combinatorial inhibition of pro-survival BCL2 family proteins and BCR-ABL may eliminate dormant LSC and obviate resistance.
Human leukemia stem cells (LSC), first described in acute myeloid leukemia (AML) (Lapidot et al., 1994), subvert stem cell properties, such as quiescence, enhanced self-renewal and survival, which render them resistant to conventional therapy (Guzman et al., 2002; Visvader, 2011). Chronic myeloid leukemia (CML) represents an important paradigm for dissecting the molecular evolution of LSC during leukemic progression and the role of LSC in therapeutic resistance because CML was the first malignancy to be targeted with therapy that selectively inhibits the aberrant kinase responsible for CML initiation (Druker et al., 2001). Although BCR-ABL-targeted tyrosine kinase inhibitors (TKIs) eradicate the bulk of BCR-ABL1 expressing cells, they frequently fail to eliminate quiescent, niche-resident LSC that drive relapse (Abe et al., 2008; Barnes and Melo, 2006; Chomel et al., 2011; Corbin et al., 2011) and blast crisis (BC) transformation following TKI discontinuation (Chomel and Turhan, 2011; Cortes et al., 2004; Deininger, 2008; Stuart et al., 2009). Despite improved overall survival (Druker et al., 2006), no curative pharmacologic therapy for CML exists, in part, because the genetic and epigenetic drivers of human BC LSC generation remain to be elucidated.
In human BC CML and in many cases of AML, LSC are enriched within the CD34+CD38+Lin− compartment, which is composed predominantly of granulocyte-macrophage progenitors (GMP) (Eppert et al., 2011; Goardon et al., 2011; Jamieson et al., 2004) with aberrant self-renewal capacity. Serial transplantation experiments show that as few as 1,000 GMP serially transplant human BC CML (Abrahamsson et al., 2009). Moreover, GMP LSC have been identified in transgenic mouse models of both BC CML (Jaiswal et al., 2003) and of AML (Krivtsov et al., 2006) suggesting that malignant transformation of progenitors into LSC, through aberrant acquisition of stem cell properties, is a key driver of leukemic progression.
Evidence from primary patient samples demonstrates that chronic phase (CP) CML is a clonal disorder (Martin et al., 1980) that originates from BCR-ABL (Daley et al., 1990) expressing hematopoietic stem cells (HSC) (Jamieson et al., 2004). Although necessary for CP initiation, BCR-ABL expression is not sufficient to drive BC transformation (Radich et al., 2006). Both mouse transgenic model and xenotransplantation data show that activation of stem cell signaling pathways, including WNT/β-catenin (Abrahamsson et al., 2009; Jamieson et al., 2004; McWeeney et al., 2009; Zhao et al., 2007), Hedgehog (Zhao et al., 2009) and the intrinsic apoptotic pathway regulated by the BCL2 gene family (Jaiswal et al., 2003), promote BC transformation. Malignant transformation of BCR-ABL1 expressing GMP into self-renewing BC LSC (CD34+CD38+Lin−) occurs, in some cases, as a consequence of alternative splicing of GSK3β, a negative regulator of Wnt/β-catenin, Hedgehog signaling and MCL1 (Abrahamsson et al., 2009; Ding et al., 2007). While recent reports reveal that mutations in splicing genes promote progression of myeloid malignancies to acute leukemia (Yoshida et al., 2011), alternative splicing-mediated alterations in the transcriptome may also enable BC transformation in a malignant microenvironment.
Because CML becomes increasingly refractory to TKIs during progression to BC (Karbasian Esfahani et al., 2006; Sawyers et al., 2002), understanding the epigenetic mechanisms that drive BC LSC maintenance and contribute to therapeutic resistance is essential. In addition, several studies suggest that LSC quiescence induction by the stem cell niche is a major component of therapeutic resistance (Barnes and Melo, 2006; Corbin et al., 2011; Forsberg et al., 2010; Holyoake et al., 1999; Saito et al., 2010). Although, recent evidence shows that increased expression of BCL2 family members contributes to CML pathogenesis (Aichberger et al., 2005; Dai et al., 2004; Tauchi et al., 2003), the precise nature of BCL2 splice isoform usage had not been examined even though a number of isoforms have antithetical functions (Akgul et al., 2004).
Pro-survival BCL2 family genes contribute to leukemogenesis (Beverly and Varmus, 2009), CML progression (Jaiswal et al., 2003), TKI resistance (Aichberger et al., 2005; Horita et al., 2000; Jaiswal et al., 2003; Konopleva et al., 2002; Sanchez-Garcia and Grutz, 1995) and hematopoietic stem and progenitor cell survival (Domen and Weissman, 2003; Milyavsky et al., 2010) by direct inhibition of MOMP. Expression of BCL2 family genes has also been linked to bone marrow niche-dependent TKI resistance in vitro (Bewry et al., 2008). However, whether pro-survival BCL2 family gene splice isoform expression promotes human BC LSC maintenance has not been elucidated. Moreover, the role of niche-dependent BCL2 family gene expression has not been delineated in the context of BC LSC quiescence induction and TKI resistance in vivo. Thus we compared BCL2 family expression in FACS-purified CML progenitors from normal, CP and BC patients and in BC LSC engrafted in different hematopoietic niches. We also investigated whether BC LSC could be targeted with a novel pan-BCL2 inhibitor, sabutoclax, capable of inhibiting BCL2, MCL1, BFL1, and BCLXL. Finally, the capacity of pan-BCL2 inhibition to overcome niche-dependent TKI resistance was assessed both in vitro and in BC LSC xenograft models as a paradigm for understanding the potential utility of sabutoclax in the sensitization of quiescent CSCs to anti-proliferative agents in a broad array of malignancies.
Although several studies have linked BCL2 gene upregulation with CML progression, most have focused on BCR-ABL-expressing cell lines (Amarante-Mendes et al., 1998; Gesbert and Griffin, 2000; Sanchez-Garcia and Grutz, 1995) or bulk CD34+ cells (Aichberger et al., 2005; Horita et al., 2000; Radich et al., 2006) rather than self-renewing human BC LSC (CD34+CD38+Lin−) that promote BC transformation. While many BCL2 family genes encode splice variants with both pro-apoptotic and anti-apoptotic functions (Moore et al., 2010), relatively little is known about the pattern of BCL2 family gene isoform expression in human BC LSC. Therefore, we utilized splice-isoform specific qRT-PCR and whole transcriptome RNA sequencing (RNASeq) to analyze BCL2 family isoform expression in FACS-purified progenitors from primary human normal (n=6), CP (n=8) and BC (n=9) samples (Supplementary Table 1). Notably, BC LSC expressed significantly higher levels of BCR-ABL and pro-survival BCL2L, MCL1L, BCLXL and BFL1L splice isoforms than CP progenitors (Fig. 1a), as well as higher BCL2L, BCLXL and BFL1L than normal progenitors (Supplementary Fig. 1a-b). Both qRT-PCR and RNASeq revealed a relative abundance of anti-apoptotic MCL1-long compared with pro-apoptotic short isoforms in BC LSC (Fig. 1b-c and Supplementary Fig. 1a-b). These data suggest that pro-survival BCL2 family gene isoforms are globally upregulated during CML BC transformation.
Because BCR-ABL induces BCL2 family gene expression in CML cell lines (Aichberger et al., 2005; Horita et al., 2000; Sanchez-Garcia and Grutz, 1995), we examined whether pro-survival BCL2 family overexpression coincided with BCR-ABL amplification in sorted CML progenitors. A striking correlation was observed between BCR-ABL and BCLXL levels in CML progenitors, which was confirmed in lentiviral BCR-ABL-transduced progenitors (Fig. 1d), suggesting that increased BCLXL expression is driven by BCR-ABL amplification in BC LSC, as previously reported (Aichberger et al., 2005; Horita et al., 2000; Sanchez-Garcia and Grutz, 1995). Expression of other pro-survival BCL2 family gene isoforms did not correlate with BCR-ABL, indicating that upregulation occurs through BCR-ABL-independent mechanisms. Consistent with qRT-PCR results, an increase in BCL2 and MCL1 proteins was detected by FACS analysis in BC LSC compared with CP progenitors (Fig. 1e and Supplementary Fig. 1c). Notably, BCL2 protein expression was higher in serially transplantable CD34+CD38+Lin− BC LSC than normal or CP CD34+CD38−Lin− and CD34+CD38+Lin− cells (Supplementary Fig. 1d). Moreover, increased expression of both BCL2 transcript levels and protein correlated with expansion of CD123+ GMP BC LSC (Supplementary Fig. 1e-f) suggesting that BCL2 overexpression portends CML progression. In addition to the increased pro-survival BCL2 family gene expression detected by RNA Seq (Supplementary Fig. 1g), an apoptosis qRT-PCR array demonstrated that BC LSC harbored distinct expression patterns of pro-death BCL2 family genes as well as TP53 and TNF superfamily receptors, such as FAS and other components of the extrinsic apoptotic machinery compared with normal progenitors (Supplementary Table 2). To gain further insight into the role of survival regulators in BC transformation, RNASeq analysis was performed on FACS-purified CD34+CD38+Lin− normal, CP and BC samples (Supplementary Table 1). Both heat map (Fig. 1f) and unsupervised principal component (Fig. 1g) analyses revealed that survival-related gene expression distinguished BC LSC from CP as well as TKI-treated and normal progenitor samples. Together these data suggest that a distinct survival gene signature predicts LSC generation and BC transformation.
Previous research demonstrated a link between BCL2 family member expression and arrest of cells in G0/G1 of the cell cycle (Zinkel et al., 2006) (Vairo et al., 1996). In T and B cells of BCL2 transgenic mice, higher BCL2 expression correlated with a higher G0/G1 fraction, lower S phase fraction and decreased BrdU incorporation (O'Reilly et al., 1997a; O'Reilly et al., 1997b; O'Reilly et al., 1996). Moreover, enforced BCL2 expression was recently shown to restore quiescence of progenitors in a mouse model of myelodysplastic syndrome (Slape et al., 2012). Seminal studies also show that quiescent LSC are TKI resistant (Barnes and Melo, 2006; Bewry et al., 2008; Holyoake et al., 1999; Saito et al., 2010).
To analyze the capacity of various hematopoietic niches to maintain dormant LSC, human BC CD34+ cells, labeled with a membrane-bound fluorescent dye, DiR, which is retained by non-dividing cells, were transplanted into neonatal RAG2−/−γ −/−c mice (Abrahamsson et al., 2009). Within 10 weeks, transplanted mice developed BC CML typified by myeloid sarcoma formation as well as robust liver, spleen, blood and bone marrow engraftment (Supplementary Fig. 2 and Supplementary Table 3). Notably, FACS analysis revealed that marrow-engrafted BC LSC harbored higher levels of DiR fluorescence than those in other niches (Fig. 2a), corresponding to a distinct population of G0 (Ki67low7-AADlow) progenitors (Fig. 2b-c) in the marrow. Confocal fluorescence microscopic and immunohistochemical analyses revealed dormant pHis-H3−Ki-67low human CD45+CD34+CD38+ cells adjacent to the marrow endosteal region (Fig. 2d-e and Supplementary Fig. 3a-b), as previously reported in AML LSC xenograft models (Saito et al., 2010). Moreover, FACS analysis revealed that CD34+CD38+CD123+CD45RA+Lin− (granulocyte-macrophage progenitor; GMP) BC LSC, previously shown to harbor the greatest serial transplantation potential, were more prevalent in the marrow than other hematopoietic niches (Supplementary Fig. 3c). In addition, cell cycle FACS analysis revealed that proportion of quiescent BC LSC was enriched in the marrow compared to the splenic niche (Supplementary Fig. 3d-e).
To examine the capacity of TKIs to eliminate quiescent self-renewing BC LSC, RAG2−/−γc−/− mice were transplanted with human BC CD34+ cells and treated orally with dasatinib, a potent BCR-ABL-targeted TKI (Supplementary Fig. 4a). Transplantation resulted in robust engraftment of human CD45+ (Fig. 3a-b and Supplementary Fig. 2a-b) and BC LSC (CD34+CD38+Lin−) cells in medullary and extramedullary microenvironments (Fig. 3b-d and Supplementary Fig. 3). Although dasatinib treatment (50 mg/kg) significantly reduced the CD45+ leukemic burden compared with vehicle treated controls (Fig. 3e and Supplementary Fig. 4b-d), a dasatinib-resistant BC LSC population persisted in the marrow (Fig. 3c, 3f and Supplementary Fig. 4e). Following dasatinib treatment, nanoproteomic analysis of FACS-purified marrow-derived BC LSC revealed a significant reduction in phosphorylation of CRKL, a direct substrate of the BCR-ABL kinase (Fig. 3g-h) (Goldman and Brender, 2000), indicative of adequate BCR-ABL kinase inhibition. However, cell cycle FACS analysis demonstrated an increase in quiescence (Fig. 3i-j) suggesting that quiescent BC LSC are resistant to BCR-ABL kinase inhibition and enriched in the marrow niche thereby providing a reservoir for relapse.
Because BCL2 overexpression has been linked, in mouse transgenic models and cell lines, to apoptosis and TKI-resistance (Amarante-Mendes et al., 1998; Domen and Weissman, 2003; Konopleva et al., 2002), we hypothesized that pro-survival BCL2 family gene expression is enhanced in marrow engrafted BC LSC and that they harbor greater TKI resistance than those in other niches. Comparative apoptosis qRT-PCR array analysis performed on FACS-purified CD45+CD34+CD38+Lin− cells revealed that while BCLX, BFL1 and BCLW were not differentially expressed, BCL2 was significantly upregulated in marrow compared with spleen (Fig. 4a-b) as was the expression of the pro-survival isoforms of MCL1 and BFL1 (Supplementary Fig. 5a-b), thereby favoring BC LSC survival. Similarly, RNA Seq revealed increased BCL2 and decreased BIM expression in marrow-engrafted BC LSC compared to BC LSC before transplantation (Supplementary Fig. 5c). To further support these findings, gene set enrichment analysis (GSEA) of RNA Seq data demonstrated that cell cycle checkpoint and cell cycle arrest genes were upregulated in FACS purified BC LSC compared with their normal counterparts (Supplementary Figure 5d). Finally, BCL2 protein expression was significantly higher in marrow engrafted BC LSC than in non-LSC (human CD45+CD34− cells) in the same niche, and correlated with a decreased sensitivity to dasatinib treatment (Supplementary Fig. 5e-f). Thus, marrow niche-resident BC LSC express high levels of pro-survival BCL2 family gene isoform expression leading to enhanced TKI resistance.
Both immunohistochemical and confocal fluorescence microscopic analysis demonstrated that human BCL2 and MCL1 protein expression (Fig. 4c) co-localized with human CD34 and CD38 expressing cells in the marrow endosteal niche (Fig. 4d). Interestingly, BCL2 and MCL1 expressing human BC CD34+ cells were enriched in the femoral epiphysis, a preferential site for homing, proliferation and survival of human leukemia cells following xenotransplantation (Supplementary Fig. 5g) (Ninomiya et al., 2007). Dasatinib treatment increased BCL2 and MCL1 expression and reduced Ki67 (Supplementary Fig. 5h-i), consistent with FACS analyses showing an increase in the proportion of quiescent BC LSC following TKI treatment (Fig. 3i-j). Although TKIs effectively eliminate LSC in extramedullary microenvironments, they fail to eradicate quiescent, BCL2- and MCL1-expressing BC LSC from the marrow niche.
Detection of increased pro-survival BCL2 isoforms in primary BC samples together with enhanced BCL2 and MCL1 expression in marrow-engrafted BC LSC, particularly following dasatinib treatment (Supplementary Fig. 4), provided the impetus for testing the LSC inhibitory capacity of an optically pure novel derivative of apogossypol, sabutoclax, which inhibits all pro-survival BCL2 family proteins (Wei et al., 2009; Wei et al., 2010) (Fig. 5a). Sabutoclax treatment increased apoptosis of BC LSC in a dose-dependent manner in vitro, as measured by cleaved capase-3 and propidium iodide staining (Fig. 5b). Because BC LSC were TKI-resistant in the marrow niche, the anti-LSC efficacy of sabutoclax was tested in a genetically engineered SL/M2 stromal co-culture system that secretes human SCF, IL-3 and G-CSF and supports long-term survival of self-renewing BC LSC (Hogge et al., 1996) (Supplementary Fig. 6a). Despite induction of pro-survival BCL2 family gene expression in BC LSC supportive stromal co-cultures (Supplementary Fig. 6b), sabutoclax reduced LSC survival and LSC colony forming capacity (Fig. 5c-d and Supplementary Fig. 6a) at doses that spared normal progenitors (Fig. 5c-d and Supplementary Table 4). Moreover, lentiviral-mediated short-hairpin RNA knockdown of BCL2 reduced colony-forming capacity of BC LSC but not normal progenitors (Fig. 5e-f). However, BCL2 knockdown did not completely abrogate BC LSC colony formation suggesting that inhibition of multiple BCL2 family proteins, including MCL1, is required to eradicate BC LSC in supportive niches.
To further assess the role of BCL2 in BC LSC survival, ABT-737, a potent BCL2 and BCLXL inhibitor, was utilized in parallel stromal co-culture experiments. Fluorescence polarization assays (FPA) demonstrated that sabutoclax and ABT-737 dissociate a BIM-peptide from BCL2 and BCLXL at nanomolar concentrations. However, only sabutoclax effectively displaces BIM from MCL1 and BFL1 (Supplementary Fig. 6c-d and Supplementary Table 4). Because ABT-737 resistance is associated with increased MCL1 and BFL1 expression (Vogler et al., 2009a; Yecies et al., 2010) and both qRT-PCR and transcriptome data showed that BC LSC express multiple BCL2 family members, including MCL1 and BFL1 (Supplementary Fig. 5a-b), the anti-LSC efficacy of sabutoclax and ABT-737 were compared. Sabutoclax reduced BC LSC survival more than ABT-737 at all doses tested in stromal co-cultures (Supplementary Fig. 6f-g and Supplementary Table 4), even though activity looked comparable in stroma-independent K562 cells (Supplementary Fig. 6h), thereby underscoring the importance of the niche in BCL2 family member induction. Hence, eradication of niche-dependent BC LSC is predicated on inhibition of multiple BCL2 family proteins, including MCL1 and BFL1.
To examine the necessity of pro-survival BCL2 family expression for BC LSC maintenance, we tested the efficacy of sabutoclax at inhibiting BC LSC survival in the marrow compared with the splenic niche (Fig. 6a). In BC CD34+ cell engrafted mice, FACS analysis revealed that sabutoclax (5 mg/kg) reduced LSC burden (Fig. 6b and Supplementary Table 5) commensurate with a decrement in human BCL2 and MCL1 expressing cells in the marrow (Fig. 6c-d). Moreover, sabutoclax treatment increased G2/S (Fig. 6e) and TUNEL+ apoptotic cells (Fig. 7a), indicative of both cell cycle and apoptosis induction. Consistent with in vitro results, no significant reduction was observed in normal progenitor engraftment in the marrow following sabutoclax treatment (Supplementary Fig. 7b-d) suggesting that a reasonable therapeutic index exists between BC LSC and normal HSC.
To quantify the TKI-sensitizing effects of sabutoclax in the presence of human BC LSC-supportive cytokines not present in mouse marrow, human BC LSC from sabutoclax or vehicle treated mice were FACS-sorted into SL/M2 stromal co-cultures in the presence of dasatinib (Supplementary Fig. 7a). In this ex vivo assay, sabutoclax pre-treated progenitors were more sensitive to dasatinib than vehicle pre-treated controls (Supplementary Fig. 7b). To further examine the synergistic effects of sabutoclax and dasatinib, BC LSC-engrafted mice were treated with lower dose sabutoclax, dasatinib or the combination followed by FACS-mediated LSC analysis. While lower dose dasatinib and sabutoclax alone had no significant effect on marrow BC LSC engraftment, combination treatment significantly reduced marrow LSC survival (Fig. 7e and Supplementary Table 6). These results suggest that sabutoclax sensitizes quiescent, BCL2 and MCL1-expressing BC LSC to dasatinib-mediated cell death. Finally, the capacity of combined treatment to eradicate self-renewing BC LSC was assessed by transplanting treated marrow into secondary recipients and monitoring survival time. Mice transplanted with combination-treated marrow had a significant survival advantage compared to those that received dasatinib-treated marrow (Fig. 7f). Sabutoclax-mediated TKI sensitization was dose (Supplementary Table 6) and route of administration dependent, with greater bioavailability provided by intravenous dosing as shown by pharmacokinetic studies (Supplementary Fig. 7c). More clinically applicable intravenous dosing resulted in a significant reduction in BC LSC following combination sabutoclax and dasatinib (Supplementary Fig. 7d) at doses that spared normal hematopoietic progenitors (Fig. 7 c-d). Overall, our data demonstrate that dasatinib alone, while effective at reducing bulk leukemic cell burden, does not eradicate marrow niche-resident BC LSC. In contrast, combined dasatinib and sabutoclax therapy significantly inhibits both primary and serial LSC engraftment (graphical summary), indicative of abrogation of both TKI-resistance and BC LSC self-renewal.
Malignant transformation of human myeloid progenitors into BC LSC through alternative splicing represents a novel molecular mechanism driving CML BC transformation and therapeutic resistance. By analyzing FACS-sorted serially transplantable CD34+CD38+lin− cells from primary patient samples, we show that BC LSC harbor increased expression of multiple pro-survival BCL2 family genes compared to both CP and normal progenitors. This pro-survival gene expression is further upregulated upon co-culture with human LSC supportive cytokine-secreting bone marrow stroma and upon engraftment in the bone marrow niche. These data are consistent with previous reports demonstrating increased BCL2 family expression in CML cells (Aichberger et al., 2005; Horita et al., 2000; Sanchez-Garcia and Grutz, 1995) and upregulation via niche-dependent signals (Bewry et al., 2008). However, our study is unique in that we show pro survival-BCL2 family splice isoform upregulation in self-renewing BC LSC and that niche-dependent BCL2 family expression is associated with TKI resistance in vivo. This study represents the first whole transcriptome and splice isoform specific qRT-PCR-based elucidation of isoform-specific BCL2 family gene expression signatures in CML LSC, which is important given that the BCL2 family is spliced into variants with antithetical functions (Akgul et al., 2004; Bingle et al., 2000) and has potential clinical significance with regard to predicting leukemic progression.
In a robust RAG2−/−γc−/− xenograft model of human BC CML, we demonstrate that BC LSC are protected from TKI-mediated cell death when engrafted in the marrow microenvironment compared with extramedullary hematopoietic niches suggesting that LSC are subject to marrow-specific cytoprotection (graphical summary) independent of BCR-ABL as demonstrated by nanoproteomic phospho-CRKL analysis. Although dasatinib treatment effectively reduces leukemic burden in engrafted mice, it does not fully eliminate BC LSC as evidenced by the fact that mice serially transplanted with dasatinib-treated bone marrow quickly develop BC CML. These data add to previous findings that CML BC LSC also depend on BCR-ABL-independent survival mechanisms (Corbin et al., 2011). Our findings expand on this concept by identifying pro-survival BCL2 family isoform expression as an important niche-specific survival mechanism and molecular target for CML BC LSC sensitization to TKI therapy. While lentiviral BCR-ABL transduction experiments suggest that BCLXL expression is BCR-ABL dependent, our in vivo studies suggest that marrow microenvironmental cues promote splice isoform switching that favors expression of multiple pro-survival BCL2 family splice isoforms in BC LSC thereby providing the impetus for elucidating these extrinsic factors in future studies.
Both cell cycle and immunofluorescence analysis, demonstrate that quiescent CML BC LSC engraft the marrow niche and are enriched in the endosteal region, consistent with previous AML xenograft studies (Saito et al., 2010). Moreover, immunohistochemical analyses show that endosteal niche resident BC LSC express pro-survival BCL2 and MCL1. Strikingly, dasatinib treatment does not eliminate quiescent bone marrow BC LSC. These quiescent BC LSC harbor enhanced engraftment potential (Barnes and Melo, 2006), which may explain why mice serially transplanted with dasatinib-treated marrow still develop BC CML.
Notably, BC LSC in stromal co-culture and in the marrow are sensitive to a novel pan-BCL2 inhibitor, sabutoclax, in a dose dependent manner (Wei et al., 2009; Wei et al., 2010). Sabutoclax also sensitizes marrow-niche BC LSC to TKI treatment suggesting that marrow-specific TKI protection is predicated, at least in part, on BCL2 family expression in the niche and can be overcome with a pan-inhibitor. Also, unlike dasatinib, sabutoclax targets quiescent self-renewing LSC. This is further evidenced by our observation that sabutoclax combined with dasatinib significantly improves survival of serially transplanted mice.
While BCL2 inhibition has been previously explored in CML, most studies have focused on CML cell lines (Kuroda et al., 2006; Meng et al., 2007) or CD34+ cells grown in culture (Mak et al., 2011) rather than self-renewing CML BC LSC in selective niches. Moreover, published reports do not address the potential antithetical roles of BCL2 family splice isoforms or the role of the microenvironment in promoting LSC survival. Treatment with ABT-737 (Kuroda et al., 2006; Mak et al., 2011), a potent BCL2 and BCLXL inhibitor, does not inhibit MCL1L or BFL1(Oltersdorf et al., 2005; Wei et al., 2010), which accelerate leukemogenesis (Beverly and Varmus, 2009), mediate resistance (Chen et al., 2007; Vogler et al., 2009a; Yecies et al., 2010) and are upregulated in CML progenitors during progression from CP to BC. Because inhibition of both subfamilies of pro-survival BCL2 family proteins is necessary for apoptosis initiation (Vogler et al., 2009b), inhibition strategies that include MCL1 would be expected to be more successful than those that target BCL2 alone (Placzek et al., 2010). Recently, paired-end DNA sequencing analysis revealed an intronic deletion polymorphism in the pro-apoptotic gene BIM (BCL2-like 11), which generated a splice isoform lacking the BH3 domain and preventing BIM-induced apoptosis in response to TKI therapy (Ng et al., 2012). Thus, pan-BCL2 inhibition may prove to be more effective at targeting TKI resistant BC LSC that naturally express multiple BCL2 family proteins in response to niche-dependent stimuli in vivo.
BCL2 family genes are regulated in a wide variety of hematologic malignancies (Beverly and Varmus, 2009; Reed, 2008) and solid tumors (Placzek et al., 2010). Moreover, CSC identified in several tumor types (Hermann et al., 2010), could conceivably rely on expression of multiple pro-survival BCL2 family isoforms, making them candidates for panBCL2 inhibition as a vital addition to combination CSC eradication therapy. Our findings may also have relevance for the elimination of therapeutically recalcitrant solid tumor CSC where metastasis and survival in the metastatic niche are mediated by pro-survival BCL2 family expression (Mehlen and Puisieux, 2006). Thus, panBCL2 inhibition with sabutoclax could provide an important new component of combination therapies that target a broad array of CSC residing in protective niches.
Additional details and methods can be found in the supplementary experimental procedures section.
Normal cord blood and adult peripheral blood samples were purchased from All Cells. CML samples were obtained from consenting patients at the University of California San Diego, Stanford University, the University of Toronto Health Network, MD Anderson and the University of Bologna according to Institutional Review Board approved protocols. CD34+ cells were initially purified by magnetic bead separation (MACS; Miltenyi, Bergisch Gladbach, Germany) followed by FACS progenitor purification using human-specific CD34 and CD38 antibodies as previously described(Abrahamsson et al., 2009; Jamieson et al., 2004). Peripheral blood mononuclear cells (PBMC) were extracted from peripheral blood following Ficoll density centrifugation, CD34+ selected, stained with fluorescent conjugated antibodies, and analyzed and purified using a FACS Aria and Flowjo software as described previously(Abrahamsson et al., 2009; Jamieson et al., 2004).
Normal or CML CD34+ cells were stained with mouse anti-human BCL2 (Dako) monoclonal antibody and analyzed by FACS. Quantitative RT-PCR to detect BCL2, MCL1, BCLX and BFL1 isoforms in FACS-sorted normal versus CML progenitors was performed with SYBR GreenER two-step qRT-PCR Kit (Invitrogen). Quantitative BCL2 isoform and apoptosis gene analysis was also performed in FACS-sorted normal and CML progenitors by whole transcriptome RNA sequencing.
BCL2 genes were also analyzed in engrafted CML cells. Briefly, 20,000-50,000 CD34+CD38+lin− cells were FACS-sorted from engrafted tissues and analyzed using isoform-specific qRT-PCR as above or using an RT-PCR apoptosis-pathway OpenArray “nanoplate” (Invitrogen). BCL2 protein was also measured in engrafted tissue cells as described above.
20,000-50,000 hematopoietic progenitor cells were sorted from the indicated cell populations using FACS, total RNA was isolated and cDNA was synthesized as described previously(Abrahamsson et al., 2009; Jamieson et al., 2004). Quantitative PCR (qRT-PCR) was performed in duplicate on an iCycler using SYBR GreenER Super Mix (Invitrogen, Carlsbad, California), 5ng of template mRNA, and 0.4mM of each forward and reverse primer. Splice isoform-specific primers were designed for BCL2, MCL1, BCLX, and BFL1 and isoform specificity was confirmed by sequencing of each PCR product. mRNA levels for each transcript were normalized to HPRT and compared using the delta-delta CT method.
Normal or CML CD34+ cells were selected and plated on confluent, mitomycin-C treated SL/M2 cells along with different doses of BI-97C1 (sabutoclax). After 1 week of culture, human progenitor cells were quantified by FACS and cells were plated in methylcellulose for colony forming assays. Colonies were scored after 2 additional weeks in culture.
BCL2 mRNA expression was silenced using shBCL2 encoding SMARTVector 2.0 lentiviral particles (Thermo-Dharmacon, #SK-003307). Efficiency of shBCL2 and control lentiviral vectors was tested by transduction of 293T and K562 cell-lines. Knockdown of ~50% of BCL2 transcripts was confirmed by qRT-PCR. Cells transduced with lentiviral shBCL2 and shControl were FACS-sorted into Methocult media (20-50 cells per well of a 96-well plate, 5-10 wells per condition), total colonies were counted for each condition after 2 weeks of culture and BCL2 knockdown was measured in the colonies.
Statistical analyses were performed with the aid of Microsoft Excel, SAS version 9.2 and Graphpad Prism software as indicated in the figure legends.
We thank Dennis Carson for his continuous advice and mentorship, Dennis Young for expert assistance on all FACS experiments, Ida Deichaite for excellent assistance with material transfer and industry relations, Jennifer Black and Rusty Wall for their help with sample preparation and mouse experiments, Isabel Newton for advice and help with DiR studies, Derrick Duarte for assistance with IHC studies, Jerry Wu for assistance with FACS cell cycle experiments and Kimberly Wilson for assistance with grant and manuscript preparation and submission. This work was generously supported by the Ratner Family Fund and by California Institute for Regenerative Medicine (CIRM) grants (#RN2-00910-1; #RS1-00228-1; #TR2-01789; #DR1-01430). DJG was supported by the CIRM UCSD Stem Cell Training Grant II and the UCSD Cancer Training Grant. This work was also funded by the National Cancer Institute (NCI, #CA-55164) and National Institute of Health (NIH, #CA-149668), and supported by the Ontario Institute for Cancer Research, through generous support from the Ontario Ministry of Research and Innovation and the Cancer Stem Cell Consortium with funding from the Government of Canada through Genome Canada and the Ontario Genomics Institute (OGI-047), and through the Canadian Institute of Health Research (CSC-105367). Drs. Reed, Wei and Pellecchia are co-inventors on Sabutoclax and related compounds, licensed by the SBMRI to Oncothyreon (Seattle).