Bortezomib Reduces the Tumorigenicity of Multiple Myeloma via Downregulation of Upregulated Targets in Clonogenic Side Population Cells

Side population (SP) cells in cancers, including multiple myeloma, exhibit tumor-initiating characteristics. In the present study, we isolated SP cells from human myeloma cell lines and primary tumors to detect potential therapeutic targets specifically expressed in SP cells. We found that SP cells from myeloma cell lines (RPMI 8226, AMO1, KMS-12-BM, KMS-11) express CD138 and that non-SP cells include a CD138-negative population. Serial transplantation of SP and non-SP cells into NOD/Shi-scid IL-2γnul mice revealed that clonogenic myeloma SP cells are highly tumorigenic and possess a capacity for self-renewal. Gene expression analysis showed that SP cells from five MM cell lines (RPMI 8226, AMO1, KMS-12-BM, KMS-11, JJN3) express genes involved in the cell cycle and mitosis (e.g., CCNB1, CDC25C, CDC2, BIRC5, CENPE, SKA1, AURKB, KIFs, TOP2A, ASPM), polycomb (e.g., EZH2, EPC1) and ubiquitin-proteasome (e.g., UBE2D3, UBE3C, PSMA5) more strongly than do non-SP cells. Moreover, CCNB1, AURKB, EZH2 and PSMA5 were also upregulated in the SPs from eight primary myeloma samples. On that basis, we used an aurora kinase inhibitor (VX-680) and a proteasome inhibitor (bortezomib) with RPMI 8226 and AMO1 cells to determine whether these agents could be used to selectively target the myeloma SP. We found that both these drugs reduced the SP fraction, though bortezomib did so more effectively than VX-680 due to its ability to reduce levels of both phospho-histone H3 (p-hist. H3) and EZH2; VX-680 reduced only p-hist. H3. This is the first report to show that certain oncogenes are specifically expressed in the myeloma SP, and that bortezomib effectively downregulates expression of their products. Our approach may be useful for screening new agents with which to target a cell population possessing strong tumor initiating potential in multiple myeloma.


Introduction
Multiple myeloma (MM) is characterized by the accumulation of a population of malignant plasma cells (10% and the more) within the bone marrow [1,2]. It is the second most frequently occurring hematological disease, affecting mainly elderly individuals [2], and is diagnosed through blood tests (serum protein electrophoresis, serum free kappa/lambda light chain assay), bone marrow examination, urine protein electrophoresis, and X-ray of commonly involved bones. MM is generally responsive to conventional chemotherapy followed by myeloablative doses of alkylating agents and autologous stem cell transplantation [2,3]. However, cytotoxic chemotherapy-based treatment is not curative, and the disease eventually recurs [2,[4][5][6]. This is in part because although currently available anti-MM strategies effectively target the bulk of tumor cells, they do not target the tumor-initiating subpopulation (i.e., cancer stem cells). The obvious need for new approaches to the treatment of MM has provided an incentive for the rapid bench-to-bedside translation of new drug treatments, including the proteasome inhibitor bortezomib, aurora kinase inhibitors and immunomodulatory drugs (IMiDs) such as thalidomide and lenalidomide, as well as novel therapies such as stem cell transplantation [2,[4][5][6][7].
Side population (SP) cells are identified based on their ability to export Hoechst 33342 dye via an ATP-binding cassette (ABC) membrane transporter, which gives these cells a distinct lowstaining pattern with Hoechst 33342 [8][9][10][11][12][13][14]. SP cells exhibit key characteristics of cancer initiating cells, including capacities for differentiation, repopulation, clonogenicity and self-renewal [8][9][10][11][12][13][14][15]. They also express high levels of various members of the ABC transporter family, including ABCB1 (MDR1/P-glycoprotein) and ABCG2 (MXR/BCRP), which in addition to inhibiting Hoechst staining, is responsible for the cells' drug resistance [8,17]. Given that many different types of cancer cells overexpress both of these ABC transporters, it seems reasonable to screen for stem-like fractions among cancer cells based on this characteristic. The SPs from various MM cell lines were recently investigated by Jakubikova and coworkers [18]. They found that MM SP cells exhibit cancer stem cell-like characteristics and that they have greater tumorigenic potential than non-SP (main population; MP) cells. This suggests there is a difference in the gene expression profiles of SP and MP cells. However, there have as yet been no reports of genes specifically expressed in the myeloma SP. In the present study, therefore, we investigated the genes and gene products that were up-or downregulated in myeloma SP cells. Our aim was to identify candidate therapeutic targets expressed within the myeloma SP. For this purpose, we initially endeavored to determine whether myeloma SP cells exhibit cancer-initiating characteristics.

Primary MM Samples
SP and MP cells from eight cases of plasma cell myeloma and one case of plasma cell leukemia were collected from Akita University Hospital, Yamamoto Kumiai General Hospital, Akita Kumiai General Hospital and Hiraka General Hospital. This study was conducted with written informed consent of the study participants and the approval of these Institutional Review Boards, according to the Declaration of Helsinki prior to collection of the specimens. Six samples of primary tumor cells were obtained from freshly isolated bone marrow at the time of diagnosis; two other samples (M4 and M8) were obtained after the first melphalanprednisolone treatment showed no chemotherapeutic response. Primary tumors cells were purified from freshly isolated bone marrow of MM patients by Ficoll-Hypaque density sedimentation. Aspirated myeloma cell from bone marrow, and mononuclear cells were harvested in RPMI 1640 (+20%FBS) with 15% DMSO and stocked in Liquid nitrogen tank for cell preservation. For SP analysis, cells were cultured in RPMI-1640 containing 10% heatinactivated FBS, 100 U/mL penicillin, 100 mg/mL streptomycin and 2 mM L-glutamine, and were maintained at 37uC in 5% CO 2 . Information about the patient samples is summarized in Table 1.

Selection and Characterization of Side Population Cells
To identify the SP among MM cells, the cells were resuspended at a concentration 1610 6 cells/mL in Hank's balanced salt solution (HBSS) with 2% FBS and 10 mmol/L HEPES containing 5 mg/mL Hoechst 33342 dye and incubated for 60 min at 37uC with intermittent shaking. As a negative control, SP cells were preincubated with 100 mmol/L verapamil; 50 mmol/L reserpine, an ABC transporter inhibitor. At the end of the incubation, the cells were washed with ice-cold Phosphate Buffered Saline (Dulbecco's PBS without calcium and magnesium, PBS(2)), after which they were centrifuged and resuspended in the HBSS buffer. To gate only viable cells, Propidium iodide solution (PI; Sigma Aldrich, St. Louis, MO; final concentration = 2 mg/mL) was added to the cells. The cells were then filtered through a 70-mm filter to obtain a single cell suspension, which was analyzed and sorted on a Moflo cell sorter (Dako, Glostrup, Denmark). The Hoechst 33342 dye was excited at 357 nm and its fluorescent emission was measured at 402-446 nm and 650-670 nm. After sorting, the SP cell fractions were analyzed using a UV laser flow cytometer (Dako) and found to be .98% pure. All experiments performed at least three times.

Colony-forming Cell Assay
Cellular potential for clonogenicity of SP and MP cells of myeloma cell lines was determined using colony-forming cell (CFC) assay. Briefly, 2610 3 sorted SP and MP cells were plated in duplicate cultures containing one mL RPMI1640 with 1% methylcellulose, 30% FBS, 10-4 M 2-mercaptoethanol (Methocult; Stem Cell Technologies, Vancouver, BC, Canada), plus a cocktail of growth factors (50 ng/mL rh stem cell factor, 10 ng/ mL rh GM-CSF, 10 ng/mL rh IL-3, and 3 U/mL erythropoietin). The cells were plated into 6-well tissue-culture plates, and after 10 days of incubation at 37uC in a $95%humidified atmosphere containing 5% CO2. We used for scoring the number of colonies an inverted microscope with 46, 106 and 206 planar objective followed standard criteria.

XTT Assays
Cell viability assays were performed using an ''In Vitro Toxicology Assay Kit, XTT based'' according to manufacturers' protocol (Sigma-Aldrich Co, St. Louis, MO).
In vivo Transplantation to NOG Mouse 5610 2 -5610 5 SP or MP (from RPMI 8226) cells were injected subcutaneously into the right or left side of the body of 6-8 weekold female NOD/Shi-scid IL-2cnul (NOG) mice (Central Institute for Experimental Animals, Kawasaki, Japan) [27]. Similarly, 5610 5 SP or MP (from AMO1, KMS-12-BM and KMS-11) cells were injected subcutaneously into NOG mice. The protocols for animal experimentation described in this paper were previously approved by the Animal Committee, Akita University (approval number: b-1-2301); The ''Regulation for Animal Experimentation'' of the University were completely adhered to in all subsequent animal experiments.

Gene Expression Analysis
RNA was extracted using an Arcturus PicoPure RNA isolation Kit (Applied Biosystems, Foster City, CA). We analyzed gene expression using a CodeLink TM Bioarray (Applied Microarrays, Inc., Tempe, AZ, USA). The array scanner was a GenePic 4000B (Molecular Devices, Sunnyvale, CA, USA), and the analysis software was Array-Pro Analyzer Version 4.5 (Media Cybernetics, Inc., Bethesda, MD, USA).

Immunofluorescence Staining of p-Hist.H3
Cell lines were treated for 24 h with 0.1-10 mM VX-680 or 1-100 nM bortezomib, then fixed and permeabilized for 15 min with 4% formaldehyde in PBS(2), and blocked for 60 min in 5% normal goat serum in PBS (2). After washing, the cells were incubated for 1 h each at room temperature with anti-p-histone H3 antibody (Cell Signaling) diluted 1:200 in PBS(2) Triton, followed by Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen). After washing again, the cells were counterstained by incubation for 5 min with 1 mg/ml DAPI in PBS (2). Images were obtained using a fluorescence microscope (IX71; Olympus, Tokyo, Japan). The mitotic index was calculated by dividing the number of p-Hist.H3-positive cells by the total number of exposed cells (DAPIpositive cells). At least 100 cells were scored per low-power field, and the cells were counted over three fields. The experiment was performed in triplicate.

Clonogenic MM SP Cell Showing High Tumorigenic Potential and a Self-renewal Capability
To assess the clonogenicity of the SP, we performed CFC assays with SP and MP cells sorted from RPMI 8226, AMO1 and KMS-12-BM lines. After 10 days, sorted SP cells of the cell lines generated 12-14 (RPMI 8226), 5-12 (KMS-12-BM), and 10-16 (AMO1) times more colonies in comparison to MP cells (Fig. S4A). Importantly, the shape and size of the colonies differed significantly between the SP and MP cells. To examine whether SP cell possesses tumorigenic potential, we transplanted RPMI 8226 cells (5610 5 each) into immunodeficient (NOD/Shi-scid IL-2cnul) mice (NOG mice) and found that mice transplanted with SP cells showed much greater tumor development than those transplanted with MP cells (Fig. S4B left panel). Furthermore, SP cells could yield tumors with 5610 2 -1610 5 transplantation into NOG mice ( Fig. S4B right panel), although MP did not. Subcutaneous injection of SP cells of AMO1, KMS-12-BM and KMS-11 into NOG mouse readily reproduced the original tumor in the immunodeficient mice at day 31-44, whereas mice transplanted MP cells did not form tumors (AMO1) or smaller size of tumors (KMS-12-BM and KMS-11) (Fig. S4C). We termed these lesions ''1 st generation'' tumors.
To test for self-renewal of SP and MP cells from the RPMI 8226 and KMS-12-BM cell lines, 1 st generation tumors were removed from NOG mice, after which the cells were subcultured and restained with Hoechst 33342. Then we conducted following serial transplantation experiments. We found that 1 st generation SPs could induce tumors (termed 2 nd generation tumors) but MPs did not induce tumor. 2 nd generation SPs also could yield tumors (termed 3 rd generation tumors). Furthermore, SP was showing the self-renewal capability; that is, they produced both an SP and MP, whereas MP produced only more MP ( Fig. 2A). For instance, after 31 (RPMI 8226) or 57 (KMS-12-BM) days from the original transplantation, cultured SP cells produced significant fractions of both SP and MP cells, whereas MP cells produced only more MP cells (1 st generation). SP cells (1610 5 ) from the 1 st generation were further transplanted into NOG mice and again produced tumors (2 nd generation), which contained both SP and MP cells. SP cells (1610 5 ) from the 2 nd generation were then transplanted into NOG mice, and we confirmed the onset of 3 rd generation tumors. 1610 5 MP cells from the 1 st generation were also transplanted into NOG mice but did not yield 2 nd generation tumors.
In addition, CD138 expression in the 1 st generation of SP and MP cells in RPMI 8226 and KMS-12-BM were examined and found that nearly all SP cells expressed CD138, whereas the MP included both CD138 + and CD138 2 cells (Fig. 2B).

MM SPs Showing Cell Cycle Regulation
We examined cell proliferation ability as well as cell cycle analysis of SP and MP cells for RPMI 8226 cell line. We first confirmed that the cell cycle patterns of Hoechst-stained RPMI 8226 cells did not differ from untreated control cells (Fig. 3A). Next we performed cell cycle analysis as shown in Fig. 2B. We then found that the cell cycle patterns of SP and MP cells clearly differed: SP cells showed cell cycle progression (G 1 -S-G 2 /M), whereas both CD138 + and CD138 2 MP cells showed cell cycle arrest (CD138 + MP: S/G 2 arrest; CD138 2 MP: G 1 /S arrest) (Fig. 3B). Moreover, the cell cycle patterns of other myeloma cell lines (AMO1, KMS-12-BM and KMS-11) were similar to those of RPMI 8226 cells: SP cells showed cell cycle progression and MP cells showed G 1 /S (CD138 2 MP) or S/G 2 (CD138 + MP) arrest ( Fig. 3C-F). Collectively then, our findings indicate that MM cell lines are CD138 + , have a capacity for have high tumorigenic potential with higher proliferation abilities than MP cells.

Detection of Candidate Genes for Targeting MM SP Cells
Because MM SP cells have a capacity for have high tumorigenic potential, to therapeutically target MM cells, it will be important to identify genes specifically expressed in the MM SP. To detect genes differently expressed between SP and MP cells, we carried out a genome wide gene expression analysis using RPMI 8226 and AMO1 cells. Our criteria of cut off the gene expression analysis of SP/MP are up regulated genes: .2.0 or downregulated genes : ,0.5, respectively. This analysis revealed 41 genes that were commonly upregulated in SP cells and two genes that were commonly downregulated; no commonly up-or downregulated microRNAs were matched to the criteria (Table S1AB).
Up regulated are genes coding for proteins associated with cell cycle and mitosis (e.g. CCNB1, CDC2, CDC25C, CDC20, CLK1, BIRC5, CENPE, SKA1, TPX2, NUSAP1, ASPM, TOP2A, AURKB, KIF15, KIF11, KIF2C, GTSE1), polycomb (e.g. EPC1, EZH2), ubiquitin-proteasome related genes (e.g. UBE3C, UBE2D3) [28][29][30]. To confirm the gene expression levels, we conducted Real time quantitative PCR for twelve candidate genes of these genes  (Fig. 4A). We found that in at least four of the five lines, ten genes were significantly more strongly expressed in SP than MP cells, and SP cells from three lines showed significantly stronger BIRC5 expression. Moreover, Western blot analysis confirmed that cell cycle and mitosis-, proteasome-and polycomb-related proteins were all more strongly expressed in SP than MP cells from the RPMI 8226 and AMO1 lines (Fig. 4B). In addition, commonly downregulate genes (AVEN and CDC42P3) in RPMI 8226 and AMO1 (Table S1B) are also validated for the five cell lines. However, because there was no significant difference between SP and MP (data not shown), we did not further analysis for these genes.
We also separately analyzed aforementioned genes in SP, CD138 + MP and CD138 2 MP cells from the RPMI 8226 and AMO1 lines. We found that the expression levels of the eight genes were significantly higher in SP cells than in either CD138 + or CD138 2 MP cells (Fig. S5). Similarly, expression of these genes was higher in MM SP cells than in CD138 + or CD138 2 MP cells. This suggests that comparison of gene expression between CD138 + SP and CD138 + MP cells might be a useful way to evaluate genes specifically expressed in primary MM SP cells.

Real Time Quantitative PCR of Candidate Genes in SPs from Primary MM Samples
Data obtained from our MM cell lines (Fig. 1) suggest that primary MM SP cells might also be identified based on CD138 positivity. We attempted to collect SP cells from eight primary samples (Table 1, Fig. 5A). The cells were then stained with both Hoechst 33342 and CD138-PE, and CD138 + SP and MP cells were collected and subjected to RNA extraction. We successfully sorted CD138 + SP cells from the eight samples, collecting from 1,500 to 50,000 SP and MP cells from each sample. The percentages of CD138 + SP cells ranged from 0.15% to 3.25%, while the percentages of CD138 2 SP cells ranged from 0.37% to 2.93%. Addition of reserpine (50 mM) resulted in a 40% to 100% reduction in the size of the SP.
We performed Real time quantitative PCR analysis for four candidate genes of cell cycle and mitosis (CCNB1 and AURKB), polycomb (EZH2) and ubiquitin-proteasome related (PSMA5) using the eight primary samples. Almost all cases showed upregulation of these genes examined: We found that seven cases (87.5%) showed significantly higher expressions of CCNB1, AURKB and PSMA5 than MP cells and six cases (75%) of SP showed significantly higher expression of EZH2 than MP cells (Fig. 5B).
These results demonstrate that the SP of CD138 + primary MM cells also shows significantly greater expression of cell cycle and mitosis (CCNB1 and AURKB)-, polycomb (EZH2)-and proteasome (PSMA5)-related genes than the MP. These results suggest that upregulated genes of CD138 + SP cells might be identical between cell lines and primary samples.

Effects of Aurora kinase Inhibitor (VX-680) for Targeting MM SP Cells
We initially conducted XTT assays with cells exposed to various concentrations of VX-680 to determine the IC 50 (RPMI 8226:6.5 mM and AMO1:2 mM) (24 h exposure, data not shown).
In subsequent experiments, cells were exposed to VX-680 for 24 or 48 h, as indicated. Among the VX-680-treated cells, large numbers were in G 2 /M phase and smaller numbers were in G 1 /S phase, which suggests G 2 /M arrest may have occurred in cells exposed to VX-680 (Fig. 6A).
Expression of Aurora B during the G 2 /M phase transition is tightly coordinated with histone H3 phosphorylation [34,35]. We therefore immunostained for phospho-histone H3 (p-Hist.H3), an M-phase-specific marker, to determine whether there was a reduction in the number of M-phase cells. As   expected, VX-680 induced cell cycle arrest at G 2 /M phase, with a corresponding reduction in the number of M-phase cells (Fig. 6B). The western blot in Fig. 6C shows the concentrationdependent reduction in p-Hist.H3 levels elicited by VX-680. Collectively, these results ( Fig. 6A-C) suggest that VX-680 increases numbers of G 2 cells while reducing the numbers of mitotic cells. We next examined expression of the polycomb proteins EZH2 in RPMI 8226 and AMO1 cells treated with different concentrations of VX-680. Cells exposed to VX-680 showed no reduction in the expression of these proteins, which suggests VX-680 may reduce SP cell proliferation by inhibiting their G 2 -M transition. Consistent with that idea, we found that VX-680 concentration-dependently reduced the %SP among RPMI 8226 and AMO1 cells (Fig. 6D). VX-680 thus appears to effectively reduce SP cell numbers by inhibiting cell proliferation. Further we examined effects of VX-680 for clonogenic abilities of SP against RPMI 8226 and AMO1 in presence (1 mM or 10 mM). Colonies of SP were slightly reduced in the presence of VX-680 but not yield significance (Fig. 6E).

Effects of a Proteasome Inhibitor (bortezomib) Targeting MM SP Cells
The IC 50 for bortezomib (48 h exposure) was 7 nM in RPMI 8226 cells and 4 nM in AMO1 cells (data not shown), and apoptosis assays showed that bortezomib concentration-dependently increased the incidence of apoptosis (Fig. 7A). We further examined its cell cycle effects. Bortezomib-treated cells included high numbers of G 2 /M phase cells and fewer G 1 /S phase cells (Fig. 7B), which suggests bortezomib induces G 2 /M arrest. This was confirmed by immunostaining for p-Hist.H3, which showed that bortezomib induced cell cycle arrest at G 2 /M phase, with a reduction in the numbers of M-phase cells (Fig. 7C). Moreover western blot analysis showed that bortezomib concentrationdependently reduced levels of p-Hist.H3, and EZH2, whereas dexamethasone did not (Fig. 7D). Thus bortezomib increases the number of G 2 cells while decreasing the number of mitotic cells.
To determine the degree to which bortezomib reduced numbers of SP cells, we assessed the %SP after exposing cells to bortezomib for 24 or 48 h (Fig. 6E). We found that bortezomib markedly reduced the %SP among both RPMI 8226 and AMO1 cells after 48 h, though no reduction was seen after 24 h. After 48 h of treatment with 1 or 10 nM bortezomib, SP cell numbers were reduced by -99% among RPMI 8226 cells, and from 84% (1 nM) to 90% (10 nM) among AMO1 cells (Fig. 7E). These results suggest that bortezomib reduces SP cell numbers by inhibiting proliferation through inhibition of G 2 -M transition, in vitro. The effects of bortezomib on clonogenicity were examined using the RPMI 8226 and AMO1 SPs in the presence (1 nM or 10 nM) or absence (DMSO) of bortezomib (Fig. 7F right panel). SP cell colonies were significantly reduced in the presence of bortezomib. However there is no reduction of SP colonies in the presence of dexamethasome (Fig. 7F left panel). The results demonstrate that bortezomib can reduce the clonogenicity of SP among RPMI 8226 and AMO1 cells. This might be because bortezomib has a broader range of targets than VX-680; for example, bortezomib reduced expression not only of p-Hist.H3 but also EZH2.

Discussion
SP cells are an enriched source of cancer-initiating cells with stem-like properties and have been identified in a variety of cancers, including MM [8][9][10][11][12][13][14][15][16][17][18]. The SP cells from the myeloma cell lines we examined showed clonogenicity and strong tumorigenicity. Moreover, when serially transplanted, myeloma SP cells yielded tumors in every generation, confirming that they possess a capacity for self-renewal. This population thus meets the criteria for cancer initiating cells [15,16]. But although the myeloma SP likely contains true myeloma initiating cells, this population nonetheless appears heterogeneous. For example, we showed that myeloma SP cells express CD138 antigen, as previously reported [18]; however, CD138 is not a useful marker of myeloma-initiating cells because approximately 70-80% of MP cells also express the antigen. Further study will be required to identify specific antigens able to serve as markers of myeloma initiating cells within the SP from both cell lines and CD138 + primary samples.
Our aim in this study was to identify genes and gene products that were expressed specifically in the myeloma SP and so could serve as targets for anti-myeloma therapies. The genes we identified can be categorized into three groups: cell cycle and mitosis-related, polycomb-related, and proteasome-related. In the following, we will discuss the functions of representative genes and their products from these groups.
Aurora kinases (Aurora A, B and C) belong to a highly conserved family of mitotic serine/threonine kinases [36][37][38][39]. Among them, Aurora-B is a ''chromosomal passenger'' protein essential for the chromosome congression that regulates chromosome segregation through the control of microtubule-kinetochore attachment and cytokinesis [37][38][39][40]. Expression of Aurora B peaks during G 2 /M phase of the cell cycle, and its aberrant expression and activation occurs in a wide range of human tumors, leading to aneuploidy and tumorigenesis [6,39,40]. In addition, overexpression of kinase-inactive Aurora-B disrupts kinetochore-microtubule interactions, cleavage furrow formation and cytokinesis, leading to polyploidy in myeloma cells [6,7,40]. It is therefore conceivable that an aurora kinase inhibitor such as VX-680 would reduce numbers of myeloma SP cells. In our study, AURKB was strongly expressed in SP cells from both myeloma cell lines and primary myeloma samples. When we assessed the ability of VX-680 to target myeloma SP cells, we found that it effectively reduced SP numbers in myeloma cell lines. However VX-680 did not completely eliminate the population, suggesting that use of an aurora kinase inhibitor alone would be therapeutically insufficient. Instead, its use in combination with one or more agents targeting other molecules would be required for complete elimination of the myeloma SP.
EZH2 and EPC1 are two polycomb-related genes upregulated in the myeloma SP. EZH2 is a component of PRC2 (polycomb repressive complex 2) and represents one of two classes of polycomb-group (PcG) proteins. It is known that polycomb proteins play crucial roles in aggressive tumor growth and cancer stem cell maintenance [41], and that the EPC1-E2F6 complex interacts with EZH2 to contribute to cell proliferation [42]. It is thus plausible that interaction between EPC1 and EZH2 within myeloma SP cells contributes to their proliferation as well as to cancer stem cell maintenance. Consistent with that idea, EZH2 is reportedly overexpressed in aggressive MM [6,40]. In the present study, we for the first time demonstrated that the proteasome inhibitor bortezomib reduces both p-Hist.H3 and EZH2. By contrast, aurora kinase inhibition alone did not reduce EZH2 enough to diminish the clonogenicity of SP cells in vitro. This suggests bortezomib reduces myeloma SP cell clonogenicity by acting on both Aurora B and EZH2.
The naturally occurring and synthetic inhibitors of the ubiquitin-proteasome pathway include lactcystin, peptide aldehydes and boronic acid peptide [43]. The dipeptide boronic acid analogue bortezomib was the first proteasome inhibitor to be used as an anticancer agent in a clinical setting, and thanks to its potent, highly selective and reversible inhibition of proteasome activity [43], bortezomib administration is now the standard of care for newly diagnosed MM patients [5,32,33]. We found that the proteasome subunit gene PSMA5 is upregulated in myeloma SP cells. The 26S proteasome consists of the cylindrical 20S proteasome, which is the proteolytic core particle, and the 19S regulator [44]. Structurally, the 20S proteasome consists of four stacked rings, each with seven distinct subunits. Its two outer rings are identical and composed entirely of a subunits (called PSMAs) [44,45]. The 20S proteasome is the major proteolytic enzyme complex involved in intracellular protein degradation. Its broad substrate spectrum includes cell cycle regulators, signaling molecules, tumor suppressors and transcription factors [44,45]. PSMA5 encodes one of the a subunits of the 20S proteasome and Zhu and coworkers recently demonstrated that knocking down proteasome genes, including PSMA5, could negatively impact the inhibitory effects of bortezomib on growth [46], suggesting PSMA5 transcription could be a key target for anti-myeloma SP therapy.
In summary, we found that MM SP cells exhibit strong tumorigenicity in vivo, and that they more strongly express cell cycle-and mitosis-related, polycomb-related, and ubiquitinproteasome-related genes than do non-SP cells. Moreover, bortezomib has the ability to downregulate expression of these targets. Our approach could be useful for screening new agents with which to target a cell population possessing strong tumor initiating potential in MM.