Introduction

Multiple myeloma is characterized by the clonal expansion of malignant plasma cells that results in anemia, renal insufficiency, and bone disease [1]. Although plasma cells phenotypically characterize the disease, recent studies have suggested that these cells lack significant proliferative capacity and instead, arise from clonogenic cells that resemble memory B cells [2–5]. However, these results are far from conclusive as other reports suggest that some or all malignant plasma cells have tumorigenic and self-renewal properties [6, 7]. The precise reasons for these disparate experimental findings are unclear, but they bring to light recurring issues regarding cancer stem cells in several human diseases, namely, inconsistencies between reports that describe their phenotypes. For example, differing stem cell phenotypes have been reported in acute lymphocytic leukemia and colorectal, pancreatic, bladder, brain, and breast cancers [8–22]. In this review, we describe the experimental approaches that have been used to identify tumorigenic cells in multiple myeloma, discuss potential factors that may contribute to the conflicting data and speculate on how advancements in the general understanding of myeloma may ultimately resolve these controversies.

Phenotypic identification of cancer stem cells

Cancer stem cells have been identified in many human cancers, and several general approaches have been employed to identify markers that distinguish them from bulk tumor cells. In myeloid leukemias and brain tumor, two of the earliest cancers examined, the surface antigens expressed by normal hematopoietic and neural stem cells were found to enrich for clonogenic tumor cells [19, 23–25]. These findings have suggested that human cancers may arise from normal stem cells and retain a cellular hierarchy with self-renewing cancer stem cells giving rise to differentiated cells that ultimately make up the majority of the tumor but lack significant long-term proliferative capacity. Furthermore, cancer stem cells may be isolated using markers expressed by their normal counterparts.

This approach has not been useful in most solid tumors since few surface antigens that mark normal stem cells are known. In these tissues, normal stem cells can be identified by their histological location, such as bulge cells in the skin or within crypts in the gastrointestinal tract or by intracellular antigens that cannot be used to isolate viable cells. Instead, the identification of surface antigens expressed by solid tumor stem cells has been somewhat empiric, such as the use of CD44 and CD24 in breast cancer [21]. These antigens were previously noted to be expressed in breast cancer and hypothesized to play a role in cell motility, metastasis, and disease progression [26, 27]. However, little existing data suggested that they would be differentially expressed by breast cancer stem cells. Nevertheless, this pioneering study by Al Hajj et al. was the first to prospectively identify human solid tumor stem cells, and CD44 expression has been subsequently found to enrich for clonogenic cells in colorectal, pancreatic, and head and neck squamous cell carcinomas [14, 15, 28].

Although no universal stem cell antigen currently exists, several properties common to normal stem cells from different tissues may also be used to identify tumorigenic cells. Normal stem cells are highly resistant to toxic injury because of multiple cellular processes that include high expression of membrane-bound drug transporters and intracellular detoxification enzymes [29, 30]. Tumor regrowth following treatment suggests that clonogenic cells are resistant to therapy, and flow cytometric assays based on these drug resistance mechanisms, such as the side population assay and measurement of intracellular aldehyde dehydrogenase (ALDH) activity have been able to identify clonogenic cells in several cancers [5, 31–33].

Functional assessment of cancer stem cell activity

Following the identification of putative cancer stem cells, it is necessary to evaluate their functional properties. In vitro, the examination of colony formation in semisolid media, such as soft-agar or methylcellulose, has long been used to evaluate clonogenic growth potential. Since colony formation using either of these assays may arise from either stem cells or self-limited progenitors, long-term proliferative potential may serve as a surrogate for self-renewal. Therefore, the ability to form secondary colonies or spheres through serial rounds of replating may aid in distinguishing functionally primitive populations [4].

Since these in vitro methods may not account for cell-extrinsic factors that influence cancer stem cell function, in vivo assays have emerged as the gold-standard to evaluate tumorigenic potential. The development of immunodeficient strains of mice have provided a means of overcoming xenografting barriers and allows the growth of human tumor cells to be assessed. NOD/SCID mice that lack B and T cells have been utilized in both the initial studies identifying clonogenic stem cells in acute myeloid leukemia and subsequently, in many other human cancers [23, 24]. The phenotypic analysis of formed tumors allows the differentiation capacity of injected cells to be assessed. Moreover, self-renewal potential may be demonstrated through serial transplantation.

Although NOD/SCID mice are most frequently used to demonstrate in vivo clonogenic growth potential, recent studies have suggested that significant xenografting barriers persist in these animals that may skew the engraftment capacity of different tumor cell populations. For example, initial studies in acute myeloid leukemia found that CD34+CD38neg tumor cells could engraft NOD/SCID mice. However, a recent report suggested that CD34+CD38+ leukemic cells could also be transplanted if NOD/SCID mice were further treated with an antibody against natural killer cells in addition to the radiation normally used for conditioning [34]. Another study using a more severely immunodeficient mouse strain (NOD/SCID/IL2γreceptorko) that also lacks natural killer cells found that a high proportion of human melanoma cells were capable of engraftment unlike previous reports using NOD/SCID mice [35–37]. Furthermore, no specific phenotypic population of engrafting cells could be identified suggesting that at least in melanoma, the relationship between distinct phenotypes and functional capacities may not be firmly linked.

It is likely that some of the differences between reports describing the phenotype of putative cancer stem cells may reflect the distinct requirements for cellular growth and expansion within each of these assay systems. In the following sections, we will focus on the identification of clonogenic cells in multiple myeloma and the potential role that the various assays utilized to study their functional properties may have played in discrepancies regarding their phenotype.

Phenotypic heterogeneity in multiple myeloma

Following antigen exposure, normal naïve B cells with immunoglobulin V(D)J gene rearrangements engage in germinal center reactions where they undergo class switch recombination and somatic hypermutation. The immunoglobulin gene sequences in multiple myeloma plasma cells are somatically hypermutated and remain constant throughout the clinical course suggesting that the disease arises from a postgerminal center B cell [38, 39]. Several studies have found that multiple myeloma patients harbor phenotypic B cells expressing the immunoglobulin gene sequence and idiotype unique to the individual myeloma clone [40–44]. Therefore, multiple myeloma may arise from these clonotypic B cells and recapitulate aspects of normal plasma cell development. The ability to induce differentiation of clonotypic B cells into plasma cells in vitro provides support for this theory [45, 46]. These findings imply that clonotypic B cells may be involved in the human disease process but offer no definitive proof that B cells represent the proliferating tumor compartment.

Clonogenic growth of multiple myeloma

The clonogenic potential in multiple myeloma has been examined using both in vitro and in vivo assays, and a comparison of the merits of these approaches is summarized in Fig. 1. The first successful in vitro system capable of growing human myeloma colonies was described by Hamburger and Salmon and later used soft agar along with a feeder of either human erythrocytes or mouse spleen cells [47]. This system suggested that the clonogenic frequency of clinical myeloma specimens was 0.001% to 0.1% of all tumor cells. These results were confirmed in our studies utilizing methylcellulose supplemented with lymphocyte conditioned media as a source of growth factors [4]. Furthermore, we found that myeloma plasma cells characterized by surface expression of CD138 were incapable of significant clonogenic growth, but that CD138neg cells expressing typical B cell surface antigens produced tumor colonies that could be serially passaged. More recently, a novel in vitro 3D stromal culture system has been developed that recapitulates both cellular and extracellular features of the bone marrow, and tumor growth in this assay also appears to arise from clonotypic B cells [48].

Fig. 1
figure 1

A comparison of in vitro and in vivo cancer stem cell functional assays

Full size image

The first in vivo assay capable of supporting the growth of human myeloma was developed by Yaccoby and Epstein using a SCID mouse implanted with human fetal bone fragments to create a humanized microenvironment (SCID-hu) [6, 7]. Using this model, mature CD38++CD45neg plasma cells generated disease that included circulating M protein, hypercalcemia, and resorption of the human bone fragment. In contrast, CD38negCD45pos peripheral blood B cells were unable to engraft suggesting that only mature plasma cells are clonogenic. In a subsequent study, Pilarski et al. found that G-CSF mobilized peripheral blood specimens resulted in lytic bone lesions and malignant plasma cells in the bone marrow following intracardiac or direct intraosseous injection into NOD/SCID mice [2]. Since the injected cells did not contain phenotypic plasma cells, the authors concluded that clonotypic B cells were responsible for tumor formation in this model. The same group subsequently demonstrated that clonotypic B cells isolated from an advanced myeloma patient could generate disease in NOD/SCID mice [3]. Similarly, we recently found that CD138+ plasma cells failed to engraft NOD/SCID mice following tail vein injection [4, 5]. However, peripheral blood cells lacking CD138 and expressing the memory B cell markers CD19 and CD27 were able to serially engraft mice and give rise to clonotypic CD138+ plasma cells functionally capable of producing circulating M protein. Therefore, these studies suggest that clonotypic B cells, rather than CD138+ plasma cells, are clonogenic in vivo.

It is likely that differences between functional assays contribute to the discrepancies in reported cancer stem cell phenotypes. Although the capacity for self-renewal may be cell intrinsic, external factors within the stem cell niche also regulate this process [49]. Little data exists regarding the role or existence of the niche in regulating cancer stem cells, but the tumor microenvironment has emerged as a major focus of myeloma biology and serves as a therapeutic target in the disease [50]. Therefore, differences in extrinsic factors within each clonogenic assay may contribute to the ability of specific cell types to home, survive, and proliferate in vivo. For instance, the human fetal bone fragments used in the SCID-hu mice have the ability to support plasma cells but may lack factors required for the growth of human B cells. On the other hand, the bone marrow in NOD/SCID mice may not initially support mature plasma cells, but the engraftment of B cells may induce changes that subsequently allow plasma cell survival. The site of injection may also play a role in tumorigenic potential, and a variety of methods (e.g., intravenous, intracardiac, or intraosseous) have been used to assess myeloma growth, and it is possible that differences in cell trafficking also play a role in determining which cells can engraft. Other experimental differences that may contribute to discrepancies in the stem cell phenotype include distinct methods used to isolate specific cell populations, such as positive or negative selection or the derivation of tumor cells from the blood or bone marrow.

Like most human cancers, multiple myeloma displays a wide clinical biology that may also contribute to the reported differences in stem cell phenotypes. An individual patient’s stage of disease and previous therapy may impact the biology of tumorigenic cells. Moreover, several recurrent genetic alterations have been described in multiple myeloma [51], and it is possible that multiple myeloma represents a number of biologically distinct diseases each containing different initiating cells. For example, the t(4;14) chromosomal translocations may carry an especially poor prognosis, and it is possible that tumorigenic cells in these cases significantly differs from those carrying other genetic abnormalities.

Future directions in multiple myeloma stem cell research

The most definitive identification of myeloma stem cells would involve assessing the tumorigenic potential of candidate cell populations though syngeneic transplantation studies. Obviously, human studies of this nature cannot be carried out, but several mouse models of myeloma have been generated and may be useful in evaluating the clonogenic potential of specific cell populations. Several animal models displaying plasmacytosis have been derived from the aberrant expression of c-Myc. In one model, the coexpression of Bcl-XL and c-Myc in B cells results in polyclonal plasma cell expansion that later progresses to monoclonal plasmablastic malignancies [52]. In a separate model, c-Myc overexpression in postgerminal center B cells results in plasma cell expansion, serum monoclonal immunoglobulin, and deposition of the M protein in renal glomeruli [53]. Overexpression of the transcription factor Xbp-1, which is required for the differentiation of B cells into plasma cells in B cells, also results in the plasmacytosis and lytic bone lesions [54]. Despite recapitulation of some aspects of human disease, deregulation of c-Myc is thought to be a late event in human myeloma and specific genetic lesions that result in the overexpression of Xbp-1 have not been described. Therefore, the true fidelity of these models is largely unknown. As the genetic events associated with myeloma are better understood, it is possible that these will provide the basis for even better models to study myeloma stem cell biology. For example, recurrent chromosomal translocations that bring genes such as CYCLIN D1, FGFR3, MMSET, and c-MAF under the regulation of immunoglobulin enhancer elements are thought to represent disease initiating events since these can be found in monoclonal gammopathy of unknown significance (MGUS) [51], and transgenic models that include these abnormalities may be able to better recapitulate the full spectrum of human disease.

A major question in cancer stem cell biology remains as to what role they play in disease progression. In chronic myeloid leukemia, studies have suggested that the transition from chronic phase to blast crisis is mediated by changes in Wnt/β-catenin signaling [55]. Interestingly, these changes appear to occur not within the originating cell responsible for chronic phase, but rather within a phenotypically more mature progenitor. Thus, it is possible that cancer progression is driven by genetic or epigenetic events that confer self-renewal to more rapidly proliferating but previously self-limited progenitor compartments and that a “shift” in the clinically relevant stem cell occurs. Disease progression in multiple myeloma has been associated with several recurrent genetic events including amplifications of chromosome 1q, mutations in RAS and inactivation of p53 [56–59]. Therefore, it is possible that these lesions can be used to mark or track specific cellular compartments and provide evidence for their involvement during disease progression.

Conclusions

The true nature and phenotype of the cancer stem cell in multiple myeloma remains unclear and controversial. However, improvements in understanding the capacities and limitations of the clonogenic assays used to assess their functional properties, development of novel animal models of the disease, and incorporation of the growing understanding of myeloma biology may provide insights into the true nature of the cell responsible for clonogenic growth. The precise identification of the cancer stem cell in multiple myeloma may allow for development of novel therapeutic strategies that inhibit tumor regrowth, delay clinical relapse, and improve long-term outcome such as overall survival. Moreover, definite understanding of how or if the myeloma stem cell may be biologically distinct when driven by specific genetic lesions or over the course of human disease is likely to be highly relevant for many other cancers.