Cancer stem cells in multiple myeloma

Abstract

Several key observations providing evidence for the cancer stem cell hypothesis and insights into the unique biology of these cells have come from the study of multiple myeloma. These include evidence that cancer cells may be functionally heterogeneous in spite of their genetic homogeneity and that malignant progenitors share many biological features with normal adult stem cells including drug resistance and regulatory processes governing self-renewal. We review studies that have examined clonogenic cells in multiple myeloma, highlight controversies regarding the cell of origin in multiple myeloma, and discuss potential targeting strategies.

1. Background

Increasing data from several human cancers suggest that neoplastic cells within individual tumors are functionally heterogeneous cells despite their clonal origins. In particular, the potential for long-term proliferation appears to be restricted to subpopulations of “cancer stem cells” functionally defined by their capacity to undergo self-renewal and give rise to differentiated cells that phenotypically recapitulate the original tumor [1-6]. Furthermore, these functional properties have implicated cancer stem cells in mediating disease initiation, relapse and progression. In multiple myeloma, the existence of cancer stem cells has long been proposed since early experiments examining the growth potential of mouse plasma cell tumors were carried out over four decades ago. However, the exact nature of the myeloma stem cell and its relationship to normal plasma cell and B cell development is unclear. We will review the evidence that multiple myeloma consists of heterogeneous cell types and that biologically distinct cancer stem cells exist within the disease. We will also discuss potential strategies that may target myeloma stem cells and prevent disease relapse and progression.

2. Cellular heterogeneity in multiple myeloma

2.1. Functional heterogeneity

Normal plasma cells are terminally differentiated and lack significant replicative potential. Similarly, multiple myeloma plasma cells appear quiescent [7], and early evidence that clonogenic growth potential was not a feature shared by all tumor cells was provided by studies carried out by Bergsagel and Valeriote [8]. These investigators studied a mouse plasma cell tumor, Adj. PC5, derived from Balb/C mice in which chronic peritoneal inflammation was induced by incomplete Freud's adjuvant and inactivated Staphalococci. Building upon the pioneering work by Till and McCullough who demonstrated the existence of multi-potent hematopoietic stem cells [9], these investigators studied tumor cell engraftment within syngeneic recipient animals following intravenous injection. Using spleen colony formation as an indicator of clonogenic growth potential, the tumorigenic cell frequency of Adj. PC5 cells was estimated at 1 in 1000-4000 cells, relatively low compared to mouse models of acute leukemia and aggressive lymphoma. Furthermore, the resulting colonies were capable of engrafting secondary recipients demonstrating that these clonogenic cells could undergo self-renewal. These investigators subsequently developed an in vitro clonogenic assay and confirmed the functional heterogeneity of Adj. PC-5 cells since only a minority of cells was capable of producing tumor colonies [10]. The applicability of these studies to human multiple myeloma was later demonstrated following the development of an in vitro clonogenic assay capable of supporting the growth of human primary tumors by Salmon and Hamburger [11,12]. This assay allowed the clonogenic frequency of a variety of human tumors to be determined, and they estimated that the tumor colony forming potential of primary bone marrow samples derived from multiple myeloma patients was approximately 1 in 100-100,000 cells. Human myeloma colonies could also be dissociated and undergo further colony formation in secondary cultures suggesting that these clonogenic cells had the capacity for self-renewal. Therefore, these studies demonstrate that the clonogenic potential of multiple myeloma cells either in vitro or in vivo is restricted to a minority of tumor cells.

2.2. Phenotypic heterogeneity

Plasma cells normally arise from the differentiation of B cells, and the potential that B cells are involved in the pathogenesis of multiple myeloma has been investigated by a number of groups. The immunoglobulin gene rearrangement and resulting antibody idiotype provides a highly specific means of determining clonal relationships in B cell malignancies that is unique among human cancers. Initial studies utilizing anti-idiotype antibodies generated against the circulating monoclonal immunoglobulin (M protein) produced by myeloma cells identified cells in both the bone marrow and blood [13]. Furthermore, these antibodies identified idiotypes at the cell surface similar to the staining of B cells rather than within the cytoplasm typical of plasma cells, and subsequent phenotyping confirmed that they expressed surface antigens reminiscent of B cells. These clonal relationships were later confirmed by studies demonstrating that the identical immunoglobulin gene sequences or chromosomal abnormalities contained within myeloma plasma cells could be found in B cells isolated from the peripheral blood and bone marrow [14-21]. Additionally, these clonotypic B cells have been implicated as precursor cells in multiple myeloma since they have been found to differentiate into antibody secreting mature plasma cells in vitro [22,23].

2.3. Multiple myeloma cell of origin

Analysis of the immunoglobulin gene sequence itself has provided insights into the stage of normal B cell development that gives rise to multiple myeloma. Early B cell development is marked by VDJ rearrangement within the immunoglobulin heavy-chain gene [24]. Following antigenic stimulation during the late stages of the primary and secondary immune responses, further antibody diversification is generated by somatic hypermutation within the germinal center. The B cell progeny which have high antigen avidity undergo extensive proliferation and clonal expansion whereas those with poor specificity undergo apoptosis. After this selection process, somatic hypermutation ceases and memory B cells and plasma cells lack the ability for further antibody diversification. Sequencing of the immunoglobulin genes from myeloma plasma cells has demonstrated the presence of extensive somatic hypermutation without intraclonal variation [25-27]. This suggests that the disease arises from a post-germinal center compartment, and subsequent studies of circulating clonotypic cells have revealed that they express CD19 and CD27 consistent with a memory B cell phenotype [28,29].

3. Cancer stem cells in multiple myeloma

Cellular heterogeneity is a feature of most human tumors, and the association between distinct phenotypes and specific functional growth properties has been highlighted by the identification of cancer stem cells in an increasing number of human malignancies. These studies have relied upon the engraftment of human tumors in immunodeficient mice, and the first model to support xenotransplantation of primary human multiple myeloma specimens was developed by Yaccoby and Epstein using SCID mice implanted with human fetal bone fragments (SCID-hu mice) [30]. In this model, the majority of primary multiple myeloma bone marrow specimens engraft when injected directly into the human bone fragment, and the several clinical aspects of multiple myeloma including circulating M protein, hypercalcemia, and resorption of the human bone are observed. Subsequent studies demonstrated that growth within the human bone fragment of SCID-hu mice was restricted to plasma cells defined by a CD38^highCD45^neg surface phenotype, whereas cell fractions containing CD19⁺ B cells were incapable of growth [31]. In addition, the capacity of plasma cells to self-renew was demonstrated by engrafting secondary SCID-hu recipient mice.

In contrast to these studies, Pilarski and Belch demonstrated that cells from the peripheral blood of patients with late-stage disease or G-CSF mobilized peripheral blood from patients with minimal disease were capable of engrafting immunodeficient NOD/SCID mice lacking any human elements following intraosseous or intracardiac injection [32]. Engrafting animals demonstrated circulating M protein as well as lytic bone lesions typical of myeloma, and tumor cells could be transplanted into secondary recipients indicative of self-renewal. Furthermore, engrafting tumor cells consisted of both CD38^hi plasma cells as well as CD19⁺ B cells, but the capacity of each of these cell types to achieve primary or secondary engraftment was not determined. A subsequent study examining a single patient with progressive disease found that circulating clonotypic cells expressing the B cell marker CD19 and lacking the plasma cell surface antigen CD138 could engraft NOD/SCID mice and give rise to both CD19⁺ and CD138⁺ tumor cells [33]. These results suggest that clonotypic B cells are myelomagenic in NOD/SCID mice.

Our group studied the engraftment of primary myeloma specimens in NOD/SCID mice and found that plasma cells defined by surface expression of CD138 were incapable of engraftment following intravenous injection [34]. In contrast, cells lacking CD138 were able to engraft mice and give rise to mature CD138⁺ plasma cells functionally capable of producing circulating M protein. Using an in vitro clonogenic assay in which CD138^neg, but not CD138⁺ cells, gave rise to myeloma colonies similar to our in vivo results, we found that clonogenic myeloma cells expressed the B cell surface antigens, CD45, CD19, CD20, and CD22. Therefore, clonogeic myeloma growth appears to be restricted to phenotypic B cells.

Adaptive immune recall over the lifetime of an individual is mediated by memory B and T cells. Although relatively differentiated and incapable of maturation into multiple tissue types, both memory B and T cells are thought to share biologic properties with normal stem cells that contribute to their long-term persistence. Comparative genetic analysis of memory B and T cells have demonstrated that these cells highly express genes critical for the self-renewal of hematopoietic and embryonic stem cells [35,36]. Furthermore, memory T cells have been found to undergo asymmetrical cell division, a process generally thought to be restricted to true stem cells [37]. Although asymmetric cell division has not been demonstrated for memory B cells, careful studies examining the kinetics of memory B cells in response to specific antigen exposure coupled with the maintenance of their specific phenotype and long-term proliferative capacity strongly suggests that the process of self-renewal maintains life-long antigen recognition [38-41]. Other human cancers in which stem cells have been well defined include acute myeloid leukemia and brain tumors. Interestingly, cancer stem cells in these two diseases phenotypically resemble normal hematopoietic and neural stem cells, respectively, suggesting that human cancers may arise from cells that normally have the capacity for self-renewal [4,42,43]. The finding that clonotypic myeloma B cells phenotypically resemble normal memory B cells and evidence that these cells are capable of self-renewal during the maintenance of long-term immunologic memory suggests that myeloma arises from this compartment. We recently further studied the CD138^neg clonogenic myeloma cells and found that in vitro colony formation was mediated by cells that co-express CD19 and CD27 surface antigens that define memory B cells [44,45]. We also found that CD19⁺CD27⁺CD138^neg cells isolated from the peripheral blood of multiple myeloma patients were able to engraft NOD/SCID mice and generate mature CD138⁺ myeloma plasma cells. Moreover, CD19⁺ B cells isolated from engrafted animals had the ability to recapitulate disease in secondary recipients. Therefore, our studies suggest that clonogenic myeloma cells are not mature CD138⁺ plasma cells but instead resemble memory B cells that have the capacity to undergo self-renewal and give rise to differentiated tumor cells.

The discrepancies between these reports have generated controversy surrounding the phenotype of cancer stem cells in multiple myeloma. However, similar inconsistencies have emerged in studies of clonogenic cell populations in other human cancers, such as colon and pancreatic carcinomas [5,6,46-48]. These results are likely to be influenced by the nature of the specific functional assay used to study clonogenic myeloma growth. The bone marrow microenvironment has been found to play an important role in the proliferation, survival, and drug resistance of multiple myeloma plasma cells, and the presence of fetal human bone fragments within SCID-hu mice may contribute to the engraftment of mature plasma cells in this setting. However, it is unclear whether the engraftment of myeloma plasma cells in these animals fully establishes their role as the myeloma stem cell since SCID-hu mice can engraft with acute myeloid leukemic blasts expressing markers of lineage commitment that are though to lack long-term proliferative potential [49]. It is also possible that the precise methods used to isolate candidate cell populations or the biological heterogeneity among patients at different clinical stages of disease progression may have also influenced the observed results. Therefore, a comparison of clonogenic assays using precisely defined cellular subsets and primary samples derived from clinically similar patient populations may better define the stem cell in myeloma.

4. Therapeutic implications of myeloma cancer stem cells

4.1. Drug resistance

A variety of clinical approaches are capable of producing responses that include corticosteroids and conventional cytotoxic chemotherapeutic agents administered at normal levels or at high-doses in combination with stem cell rescue [50]. Furthermore, several novel agents with unique mechanisms of action of been recently approved for clinical use [50]. This wide range of therapeutic options has steadily improved the ability to induce clinical responses detected as decreased plasma cells in the bone marrow or circulating M protein. However, multiple myeloma remains incurable for the vast majority of patients suggesting that cancer stem cells with the growth capacity to mediate relapse are relatively resistant to these clinical strategies. Several studies have found that circulating clonotypic B cells may persist following systemic treatment and their frequency increases during clinical relapse [51-54]. These results suggest that these cells are drug resistant and mediate tumor regrowth, and clonotypic B cells have been found to express the multi-drug transporter P-glycoprotein [55].

We compared the effects of several anti-myeloma agents on distinct cellular compartments and found that CD138^neg clonogenic precursors are relatively resistant to most of these agents at concentrations that inhibit mature CD138⁺ plasma cells [44]. These agents included glucocorticoids, alkylators, the thalidomide analogue lenalidomide, the proteosome inhibitor bortezomib that are currently used in the treatment of myeloma patients, and it is likely that these results explain the why these agents are effective at tumor debulking but lack curative potential as single agents. The cellular processes that mediate drug resistance by cancer stem cells are poorly understood, but normal stem cells are similarly resistant to a wide variety of toxic compounds through multiple mechanisms. Interestingly, several flow cytometric assays with the capacity to identify both normal and cancer stem cells, such as the side population and Aldeflour assays, measure distinct cellular properties that can confer drug resistance [56,57]. We found that these assays can identify memory B cells within the circulation of multiple myeloma patients clonally related to the malignant plasma cells [44]. Therefore, the drug resistance exhibited by myeloma stem cells may be mediated by similar processes that protect normal stem cells.

4.2. Targeting self-renewal pathways

Cancer stem cells have been defined by two critical properties, the capacity to undergo self-renewal and give rise to differentiated tumor cells [58]. Similar to the shared properties that may promote drug resistance, it is likely that the pathways regulating the self-renewal of normal stem cells play a similar role in cancer stem cells. Highly conserved signaling pathways, such as Notch, Wnt, and Hedgehog regulate stem cell fate decisions during normal development, and each of these is active in a wide range of human cancers, including multiple myeloma [59,60]. We investigated the role of Hedgehog signaling in multiple myeloma, and found that key regulators of Hedgehog signal transduction were over-expressed in both multiple myeloma cell lines and clinical specimens [61]. Furthermore, Hedgehog pathway activation appeared to be increased in the CD138^neg cells we previously identified as myeloma cancer stem cells. We examined the functional role of Hedgehog signaling on multiple myeloma stem cells and found that pathway activation by Hedgehog ligand induced the expansion of less differentiated CD138^neg cells, whereas pathway inhibition using a monoclonal neutralizing antibody against Hedgehog ligands or the naturally occurring small molecule inhibitor cyclopamine limited subsequent clonogenic growth. The inhibition of clonogenic growth was coupled with the accelerated differentiation of these cells into mature CD138⁺ cells suggesting that Hedgehog signaling plays a critical role in dictating multiple myeloma stem cell fate decisions. Therefore, conserved stem cell regulatory processes, such as developmental signaling pathways may provide therapeutic targets within drug resistant cancer stem cells.

4.3. Immunological targeting of myeloma stem cells

Monoclonal antibodies have emerged as effective agents for the treatment of a wide variety of human cancers. In particular, the humanized antibody rituximab can improve clinical outcomes in a wide variety of B cell malignancies [62,63], and findings that myeloma stem cells phenotypically resemble normal B cells have suggested that it may be able to target these cells. Examining primary myeloma specimens, we found that treatment with rituximab in combination with complement to induce antibody-dependent cellular lysis limited tumor colony formation in vitro [44]. Rituximab has been clinically studied in multiple myeloma and demonstrated limited effects on outcomes in a small single-armed clinical trial [64]. However, the disparity between these clinical results and our in vitro data may be explained by recently correlative laboratory studies suggesting that rituximab may bind to myeloma stem cells, but lacks the capacity to eliminate these cells in vivo (C.A. Huff, personal communication).

The graft versus host reaction generated during allogeneic bone marrow transplantation is associated with long-term remissions and suggests that immune based approaches may be able to eliminate multiple myeloma cancer stem cells [65,66]. The morbidity and mortality associated with allogeneic transplantation has limited the use of this procedure use in the majority of multiple myeloma patients, but its curative potential has generated great interest in developing tumor specific immunotherapeutic approaches. The ideal target antigen in multiple myeloma is unknown, but the high degree of tumor specificity associated with the M protein idiotype has led to its use in clinical studies. One such study involving vaccination with purified M protein led to a reduction in the levels of circulating clonotypic B cells in 4 of 6 evaluable patients [67]. Little effect was observed on the M protein levels indicating that this approach had little impact upon overall disease burden. However, the prolonged life-span of multiple myeloma plasma cells may have prevented detection of clinical benefit during the 30 week evaluation period even if clonogenic multiple myeloma stem cells had been targeted. Furthermore, several small clinical trials have utilized dendritic cells pulsed with M protein as a vaccination strategy either alone or following high-dose therapy [68-71]. These studies clearly indicate that anti-idiotype immunity can be generated using these approaches, but the capacity to target myeloma stem cells has yet to be addressed. The embryonic stem cell-associated antigen SOX2 may represent another potential antigen expressed by multiple myeloma stem cells [72]. SOX2 was identified during the examination of tumor antigens serologically recognized by patients with monoclonal gammopathy of unknown significance but absent in those with multiple myeloma. Intriguingly, SOX2 is expressed by clonogenic myeloma cells and stimulation of anti-SOX2 immunity limited the clonogenic growth of primary myeloma samples in vitro. Autologous lymphocytes infiltrating tumors have been found to recognize cancer cells in several diseases. In multiple myeloma the bone marrow represents the tumor microenvironment and this compartment has been found to have a significantly greater frequency of T cells recognizing myeloma antigens compared to the peripheral blood. Following in vitro activation and expansion using anti-CD3/CD28 beads, marrow derived T cells have been found to be capable of inhibiting both CD138⁺ and CD138^neg cells in vitro suggesting that they are active against both mature tumor cells and cancer stem cells [73].

5. Persistent questions and future directions

Cancer stem cells have been identified in an increasing number of human malignancies, but many questions remain. For example, the precise relationship between cancer stem cells and their normal counterparts is unknown. In myeloid leukemias and brain tumors shared antigenic profiles suggest that cancer stem cells arise from the transformation of normal adult stem cells. Similarly, our data suggest that clonogenic multiple myeloma cells phenotypically resemble normal memory B cells that may be self-renewing. However, it is unclear whether all human cancers arise from normal self-renewing cells since studies in chronic myeloid leukemia and model systems of AML have demonstrated that specific genetic or epigenetic alterations may induce self-renewal in more differentiated cells that normally have limited replicative potential [74,75]. Furthermore, it is unclear what role cancer stem cells play during disease progression or whether the functional cellular hierarchy characterizing tumors at initial diagnosis is maintained during this process.

The processes that regulate cancer stem cells are also poorly understood. In several cancers, specific genetic lesions carry prognostic significance. This suggests that these alterations directly influence the biology of clonogenic cells which dictate long-term outcomes, but the functional consequences of specific lesions on cancer stem cells are just beginning to emerge [76,77]. Extrinsic regulation of normal adult stem cells is mediated by the stem cell niche that maintains quiescence and self-renewal potential [78]. But there is little data suggesting that cancer stem cells similarly require or are regulated by the microenvironment. Furthermore, several signaling pathways that regulate normal stem cells are active in cancer stem cells, but their potential to serve as anti-cancer targets is unknown.

In most human cancers studied to date, stem cells have been found to be relatively rare and restricted to phenotypically defined subpopulations of tumor cells. However, recent findings demonstrate that the cells responsible for in vivo clonogenic growth of murine tumors in syngeneic recipients may be less phenotypically restricted and relatively frequent [79]. Therefore, the identification and characterization of cancer stem cells may be influenced by xenotransplantation of human cancers in immunodeficient mice. It is likely that this issue as well as the relevance of cancer stem cells will be more definitively resolved through carefully executed and analyzed therapeutic trials and correlative laboratory studies.

Multiple myeloma may provide the ideal setting for investigating these questions. Several advances have been made in understanding basic myeloma biology, such as the role of the microenvironment in therapeutic resistance and identification of specific genetic lesions that can dictate long-term outcomes or are associated with disease progression [80-83]. Additionally, ready access to primary clinical samples along the entire disease continuum that spans asymptomatic monoclonal gammopathy of unknown significance to aggressive myeloma allows clonogenic cell populations to be serially investigated within individual patients. Therefore, the integration of disease-specific biology with more definitive studies defining the multiple myeloma stem cell may provide insights that are applicable to many human tumors.