Induction of Apoptosis in Plasma Cells by B Lymphocyte–Induced Maturation Protein-1 Knockdown

B lymphocyte–induced maturation protein-1 (Blimp-1) is a master regulator crucial for inducing plasmacytic differentiation (1, 2). It is a transcriptional repressor that directly turns off the expression of several genes, including c-MYC (3, 4), CIITA (5), ID3 (6), SPI-B (6), and PAX-5 (7), through interaction with histone deacetylase (HDAC) family proteins (8), Groucho (9), a histone methyltransferase, G9a (10), and an arginine-specific histone methyltransferase, Prmt5 (11). Blimp-1 is present in plasma cells formed in primary and secondary immune responses to either thymus-dependent or thymus-independent antigens (12). Blimp-1 is also expressed in transformed plasma cells (6, 13). Recent results showing that Blimp-1 is required for maintenance of long-lived plasma cells suggest that Blimp-1 regulates a reversible gene expression program and that this protein has potential applications in the therapeutic management of plasma cell–related diseases (14).

Plasma cell tumors in human are classified as multiple myeloma, solitary bone plasmacytoma, or extramedullary plasmacytoma (15). The mechanism(s) of tumorigenesis for multiple myeloma is not clear. Some genetic modifications have been reported to occur during tumor progression [e.g., translocation of genes such as CCND1, CCND3, FGFR3, and MAF to immunoglobulin regulatory regions (16) and rearrangement of c-MYC to a nonimmunoglobulin locus (16)]. Most of the tumor progression takes place in bone marrow where the interaction between multiple myeloma cells and host bone marrow regulates tumor cell growth, survival, and migration (17). Bcl-xL, Bcl-2, and Mcl-1 showed increased levels of expression in multiple myeloma cells, although Mcl-1, rather than Bcl-2 and Bcl-xL, is the main player in promoting multiple myeloma cell survival (18, 19). Previous microarray cluster analysis results revealed that many of the Blimp-1 targets involved in regulating B-cell identity and germinal center function, and in inducing Ig secretion, are regulated normally by Blimp-1 in multiple myeloma, whereas genes controlling proliferation, such as c-MYC, are not subjected to Blimp-1 repression (6), suggesting a role for Blimp-1 in maintaining plasma cells.

Here, we sought to determine whether Blimp-1 must be continuously expressed to maintain survival in human multiple myeloma cells or mouse plasmacytoma cells. Using a lentiviral vector delivering short hairpin RNA (shRNA) specific to human and mouse Blimp-1, we found that Blimp-1 knockdown causes apoptosis but not cell cycle arrest. RNA interference (RNAi) for Blimp-1 resulted in induction of the proapoptotic protein Bim and reduction of the antiapoptotic protein Mcl-1. Additionally, caspase-9 and caspase-3 were activated on reduction of Blimp-1 expression in multiple myeloma lines. Interestingly, we found that Blimp-1 knockdown in multiple myeloma lines could not reprogram the plasma cell phenotype. Accordingly, our results indicate that proteasome inhibitors, the effective therapeutic strategy for multiple myeloma, cause a reduction of Blimp-1 expression that is required for proteasome inhibitor–mediated apoptosis. Lastly, our data imply that IFN regulatory factor 4 (IRF4) might function upstream of Blimp-1 in supporting survival. Our results reveal a novel role for Blimp-1 in maintaining malignant plasma cell survival.

Cell lines and reagents. Cells were grown and maintained as follows: human U266 multiple myeloma cells in RPMI 1640 (Life Technologies) containing 20% fetal bovine serum (FBS; Life Technologies), NCI-H929 (H929) multiple myeloma cells in RPMI 1640 containing 10% FBS and 50 μmol/L β-mercaptoethanol (Life Technologies), and RPMI8226 and IM9 in RPMI 1640 containing 10% FBS. All multiple myeloma cells were grown in medium containing penicillin/streptomycin (100 units/mL; Life Technologies). Mouse plasmacytoma P3X63Ag8 (P3X) cells were grown in RPMI 1640 plus 10% FBS, penicillin/streptomycin, and β-mercaptoethanol (50 μmol/L). 293T and 3T3 cells were maintained in DMEM (Life Technologies) supplemented with 10% FBS and penicillin/streptomycin (100 units/mL). Caspase-3 inhibitor (Z-DEVD-FMK) was purchased from R&D Systems, Inc.

Plasmids. The shRNA cassette was first cloned into a pBS/U6 vector (kindly provided by Dr. Yang Shi, Harvard University, Cambridge, MA). The protocol for generating shRNA for Blimp-1, IRF4, or the negative control was essentially as described (20). The human Blimp-1 shRNA contains the sense targeting sequences 5′-GGGACTCCTACGCTTACTTGAA-3′ and 5′-GGGTGCAGCCTTTATGAGTCATC-3′, corresponding to nucleotides 1004 to 1025 and 2254 to 2276, respectively, of the human Blimp-1 coding sequence (Genbank accession no. NM_001198). These sequences were named U6Blimp1-716i and U6Blimp1-1966i, respectively. The mouse Blimp-1 shRNA U6Blimp1-3952i contains a sense targeting sequence located in the 3′-untranslated region, 5′-GGGTGGTTCGGTCCTGTATTCT-3′, nucleotides 3952 to 3973 downstream of the translational start site (Genbank accession no. NM_007548). The human U6IRF4-776i and U6IRF4-1278i contain the sense targeting sequences 5′-GGGAAATCCTCGTGAAGGAGCT-3′ and 5′-GGGCTACGATTTACCAGAACAC-3′, respectively, of the human IRF4 coding sequence (Genbank accession no. NM_002460). The mouse U6IRF4-536i and U6IRF4-1278i contain the sequence 5′-GGGATTATGCCCCTGACCAGTC-3′ and 5′-GGGCTACGAGTTACCTGAACAC-3′, respectively, which targets to mouse IRF4 coding sequence (Genbank accession no. NM_013674). The control shRNA (U6controli) contains scrambled nucleotides, 5′-GGGATATCGGCATATTGCGGCTA-3′, which has no homology to other known mRNAs. The U6-shRNA cassette was removed from the pBS/U6 vector by BamHI digestion and the overhanging sequences from BamHI digestion were filled in and then blunt-end ligation was performed at the PacI site on the FUGW lentiviral vector (21), a generous gift from Dr. Jeremy Luban (Columbia University, New York, NY).

Retroviral and lentiviral transduction. The procedure for preparing retroviral vectors was performed as described (22). For pseudotyped virus, retroviral vector (15 μg), pSV-φ-E-MLV (15 μg), and VSV-G (pMD.G; 15 μg) were transfected into 293T cells. For generation of pseudotyped lentivirus, lentivirus vector (FUGW; 15 μg), Δ8.9 (15 μg), and VSV-G (15 μg) were used for transfection into 293T cells. Target multiple myeloma cells were transduced at a multiplicity of infection (m.o.i.) of 5 to 10 in the presence of 5 μg/mL polybrene (Sigma) and P3X cells were transduced at a m.o.i. of 10 to 15, and the percentage of transduced cells was analyzed by green fluorescent protein (GFP) or yellow fluorescent protein (YFP) expression with fluorescence-activated cell sorting (FACS) analysis 1 day after infection.

RNA isolation and reverse transcription-quantitative PCR analysis. Total RNA was isolated on an RNeasy spin column (Qiagen). Total RNA (1 μg) was used for cDNA synthesis by the Omniscript Reverse Transcriptase kit (Qiagen) performed according to the manufacturer's instructions. cDNA was used for reverse transcription-quantitative PCR (RT-QPCR) analysis with the Taqman Universal PCR Master Mix in an ABI/Prism 7000 sequence detection system (Applied Biosystems). The Taqman primer sets used in this study are the following: PRKG1 (assay ID: Hs00183512_ml) PRDM1 (assay ID: Hs00153357_ml), PPIA (assay ID: Hs99999904_ml), PU.1 (assay ID: Hs00231368_m1), BCL-6 (assay ID: Hs00153368_m1), CIITA (assay ID: Hs00172106_m1), XBP-1 (assay ID: Hs00231936_ml), and PAX-5 (assay ID: Hs00172003_m1). The primer sets used for RT-QPCR with the SYBR green method are as follows: c-MYC, 5′-GCCACGTCTCCACACATCAG-3′ and 5′-TCTTGGCAGCAGGATAGTCCTT-3′; SPI-B, 5′-GGCCACACTTCAGCTGTCTGTA-3′ and 5′-AGGAGCCCCCTCTGAATCAG-3′; ID3, 5′-CCCTGGACCCCCTGATG-3′ and 5′-TTTGTCGTTGGAGATGACAAGTTC-3′; and MTA3, 5′-GGAGAGCCTGTGAGAGCTGCTA-3′ and 5′-CATATTAGGTGGGCCCCAAGA-3′. The primer sets used for semiquantitative reverse transcription-PCR (RT-PCR) are as follows: glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-GGGCGCCTGGTCACCAGGGCTG-3′ and 5′-GGGGCCATCCACAGTCTTCTG-3′; SPI-B, 5′-CACGGAGAACTTCGCT-3′ and 5′-CTTGGCGTAGTTTCGG-3′; XBP-1, 5′-CGCTGAGGAGGAAACTGAA-3′ and 5′-GGGAGGCTGGTAAGGAACT-3′; and ID3, 5′-TGTGCTGCCTGTCGGAACG-3′ and 5′-ACAAGTTCCGGAGTGAGCT-3′.

FACS analysis. Apoptotic and dead cells were determined by Annexin V-phycoerythrin (PE) or Annexin V-allophycocyanin (APC) plus 7-aminoactinomycin D staining according to the protocol provided by the manufacturer (BD PharMingen). Cell cycle phase was determined by propidium iodide (BD PharMingen) staining according to the manufacturer's protocol. The flow cytometry protocol determining surface marker staining was essentially as described (6). The following antibodies used in this study were from BD PharMingen: PE-conjugated anti-human CD138/syndecan-1 (clone DL-101), PE-conjugated anti-human CD38 (clone HB7), PE-conjugated anti-human CD19 (HIB19), APC-conjugated anti-human CD20 (clone 2H7), PE-conjugated anti-human MHCII (HLADR, clone TU36), and PE-conjugated anti-human CD54 (clone HA58). Experimental results were analyzed by a FACSCalibur or FACSCanto machine (Becton Dickinson) with CellQuest or FCS Express 3.0 software.

Immunoblotting and ELISA. Nuclear extracts and total cell lysates were performed as described (3). Nuclear extracts (10–20 μg) or total cell lysates (20 μg) were subjected to SDS-PAGE and immunoblotting as described using antibodies specific to Blimp-1 (6): IRF4 (M17, Santa Cruz Biotechnology), Bcl-2 (Ab-1, Calbiochem), Bcl-xL (2H12, Santa Cruz Biotechnology), Mcl-1 (S-19, Santa Cruz Biotechnology), Bim (AB17003, Chemicon International, Inc.), procaspase-4 (4B9, MBL), α-tubulin (Sigma), or actin (Sigma). Blots were then hybridized with one of the following horseradish peroxidase–conjugated secondary antibodies: rabbit anti-goat IgG (Sigma), goat anti-mouse IgG (Promega), or goat anti-rabbit IgG (Sigma). The immunoreactive proteins were detected by the enhanced chemiluminescence system (Amersham Biosciences) according to the manufacturer's instructions, and the chemiluminescent immunocomplexes were scanned and detected by the Fujifilm LAS-3000 system. The ELISA was essentially following the previous protocol (23).

Generation of ABL-MYC(Ψ2) virus and plasmacytoma in BALB/c mice. Helper virus-free ABL-MYC(Ψ2) stocks were obtained from the cultured supernatant of a packaging line as described (24). The Ψ2 packaging cell line (kindly provided by Dr. J. Frederic Mushinski, National Cancer Institute, Bethesda, MD) was cultured in DMEM with 10% FCS and 1 mmol/L sodium pyruvate (Invitrogen). ABL-MYC(Ψ2) virus titers were determined by NIH3T3 cell transformation assays by counting the number of foci formation after 2 weeks as previously reported (24). Eight-week-old BALB/c mice (purchased from BioLASCO Taiwan Co.) were injected with pristane (0.5 mL; Sigma) 2 days before i.p. injection with ABL-MYC(Ψ2) titer (2 × 10⁵ foci/mL). Ascites and tumors from mice were collected ∼14 days after ABL-MYC(Ψ2) injection and then subjected to immunoblot analysis of Blimp-1 and actin expression. In parallel, tumor cells were cultured in RPMI 1640 containing 10% FBS, 50 μmol/L β-mercaptoethanol, 1 mmol/L sodium pyruvate, 100 μmol/L nonessential amino acids, and murine interleukin-6 (IL-6; 5 ng/mL; PeproTech) at a density of 10⁶/mL. After ∼2 weeks, the surviving, nonadhering cells were isolated and subcultured to develop into an IL-6–independent line. Lentiviral vectors carrying shRNA for Blimp-1 or control sequences were used for transduction in established plasmacytoma lines at a m.o.i. of 10 to 15 in the presence of 5 μg/mL polybrene.

Caspase activity assay. Cells were harvested 3 and 4 days after lentiviral infection. Activity of caspase-3, caspase-8, and caspase-9 was assayed according to the manufacturer's instructions (Calbiochem). We used the following colorimetric peptide substrates in this assay: caspase-3–specific Ac-DEVD-pNA, caspase-8–specific Ac-IETD-pNA, and caspase-9–specific LEHD-pNA. The free chromophore pNA light emission was quantified using a microtiter plate reader (SpectraMax M2) at 405 nm.

Blimp-1 shRNA causes apoptosis in plasma cell lines. We first used lentiviral vector to deliver the expression of shRNA for Blimp-1 to determine the importance of continuous Blimp-1 expression. Two designed shRNAs against human Blimp-1 (1004i and 2254i) and one shRNA against mouse Blimp-1 (3952i) were generated. The efficiency of endogenous Blimp-1 expression knockdown after lentiviral transduction in human or mouse plasma cell lines was monitored by immunoblot (Fig. 1A). Human U6Blimp1-1004i seemed to be more effective than U6Blimp1-2254i (Fig. 1A). A lentiviral vector carrying either the U6 promoter alone or the U6 promoter plus shRNA with scrambled sequences (U6controli) was used as controls. Notably, we found enhanced cell death in multiple myeloma lines, including H929 and U266, on knockdown of Blimp-1 expression (Fig. 1B). This is a specific effect because U6controli caused almost identical cell survival to U6 (Fig. 1B). The effect of mouse Blimp-1 shRNA, U6Blimp1-3952i, on reducing cell survival was also observed in the P3X plasmacytoma cell line (data not shown). Annexin V staining of GFP-positive cells showed that Blimp-1 RNAi-mediated cell death is apoptotic (Fig. 1C). Similar results were also observed in human RPMI8226 cells (data not shown). Taken together, we found that reduction of Blimp-1 expression via shRNA-mediated knockdown results in apoptotic cell death in human multiple myeloma lines and in a mouse plasmacytoma cell line.

Inhibition of Blimp-1 expression causes apoptosis in plasmacytomas. We next tested whether the above-mentioned Blimp-1 knockdown-mediated apoptosis in cell lines could be recapitulated in a newly established plasmacytoma line derived from a freshly isolated plasmacytoma. For this purpose, we generated plasmacytomas induced by inoculation of a murine retrovirus that encodes the v-abl and c-myc oncogenes, ABL-MYC(Ψ2), into BALB/c mice (24, 25). As previously reported, superficial, mesenteric connective tissue tumors and ascites developed in nearly all mice injected with ABL-MYC(Ψ2) i.p. (25). Immunoblot analysis showed expression of Blimp-1 protein in freshly isolated tumor and ascites but not in spleen from mice injected with ABL-MYC(Ψ2) (Fig. 2A). The isolated tumor cells were further established as a stromal cell–independent, IL-6–independent CD138, the plasma cell surface marker, positive and IgM-secreting line after ∼2 months of culture in vitro (data not shown). We then verified the expression of Blimp-1 and showed the efficiency of decreasing Blimp-1 expression by U6Blimp1-3952i in the newly established plasmacytoma line (Fig. 2B). Consistently, we also found an increased proportion of Annexin V–positive apoptotic cells in the newly established plasmacytoma line after various days of U6Blimp1-3952i transduction (analyzed from gated CD138-positive and GFP-positive cells; Fig. 2C). After 4 days, the production of IgM from U6Blimp1-3952i–transduced cells was also largely reduced compared with that from U6controli-infected cells (Fig. 2D). Thus, knockdown of Blimp-1 causes apoptosis in a newly established plasmacytoma line.

Knockdown of Blimp-1 expression cannot reprogram the plasma cell phenotype. Previous results using a conditional knockout strategy to delete prdm1, the gene encoding Blimp-1, after the formation of plasma cells in mice showed reduced numbers of antigen-specific plasma cells from bone marrow (14). This observation suggests two possible models for the role of Blimp-1 in maintaining plasma cells in the bone marrow. Blimp-1 may regulate cell survival or, alternatively, may maintain the plasma cell phenotype. We therefore tested whether Blimp-1 knockdown could reprogram the multiple myeloma cell surface phenotype and gene expression, in addition to triggering apoptosis. We found that expression of plasma cell surface markers, such as CD138 and CD38, remained similar after 4 days of U6Blimp1-1004i or U6 transduction (Fig. 3A). The levels of other surface markers, CD54, CD19, CD20, and MHC class II, also remained unchanged (Fig. 3A). We also examined the mRNA levels of known Blimp-1 target genes, such as PAX-5, XBP-1, SPI-B, ID3, PU.1, BCL-6, c-MYC, and CIITA, in this setting. The levels of PAX-5 were too low to be accurately estimated by RT-QPCR; however, the mRNA levels of the majority of target genes tested remained unchanged, including SPI-B, ID3, PU.1, XBP-1, and c-MYC, suggesting that Blimp-1 knockdown could not globally alter the Blimp-1–dependent transcription program (Fig. 3B and C). However, we observed a dramatic increase in CIITA and BCL-6 mRNA expression by U6Blimp1-1004i in H929 (Fig. 3C), indicating that a small subset of Blimp-1 target genes could be derepressed in this cell context.

Changes of antiapoptotic and apoptotic protein expression on inhibition of Blimp-1 expression. The above results suggested that Blimp-1 is not involved in maintaining the plasma cell phenotype. We therefore determined the role of Blimp-1 in regulating cell survival by examining the mechanism of Blimp-1 RNAi-mediated apoptosis. We first examined the cell cycle status. We found that Blimp-1 knockdown in H929 cells did not dramatically affect cell cycle progression; the proportion of cells in G₁, S, and G₂-M phases were practically identical between U6Blimp1-1004i–transduced and U6controli-transduced cells at day 2 (Fig. 4A,, top). After 4 days of transduction, the proportion of cells in each phase were similar between the two groups, although more cells accumulated in sub-G₁ phase in the U6Blimp1-1004i–transduced cells (Fig. 4A , bottom). Similar results were found in U266 cells (data not shown).

We then looked at the expression of antiapoptotic or proapoptotic proteins involved in regulating apoptosis. Caspase activity assays showed that U6Blimp1-1004i caused activation of caspase-9 at day 4 of transduction as well as the activation of the downstream effector caspase-3 (Fig. 4B). Notably, we did not observe the activation of caspase-8 on reduction of Blimp-1 (Fig. 4B), suggesting that Blimp-1 knockdown-mediated apoptosis in plasma cells results from intrinsic apoptotic pathways (26). Accordantly, we observed decreased expression of procaspase-4, due to protein cleavage (27), by U6Blimp1-1004i (Fig. 4D). Moreover, Blimp-1 RNAi-induced apoptosis could be blocked by caspase-3 inhibitor, Z-DEVD-FMK (Fig. 4C,, bottom). Expression of various antiapoptotic and proapoptotic Bcl-2 protein family on inhibition of Blimp-1 expression was monitored by immunoblot analysis. In multiple myeloma cells, the levels of Bcl-2, Bcl-xL, and Mcl-1 are elevated (19, 28). We therefore examined the expression of these proteins at day 4 of U6Blimp1-1004i transduction in H929 cells and showed that Mcl-1 was reduced and Bim was increased, whereas Bcl-2 and Bcl-xL remained unchanged (Fig. 4D). Thus, our collective data suggested that Blimp-1 knockdown-mediated apoptosis works through activation of an intrinsic apoptotic pathway as well as through the down-regulation of the antiapoptotic protein Mcl-1 and the induction of the proapoptotic protein Bim.

Proteasome inhibitor–induced apoptosis involves reduced expression of Blimp-1. To examine whether the current effective agents for multiple myeloma therapy affect Blimp-1 expression, we treated various cell lines with the proteasome inhibitors MG132 and bortezomib. Bortezomib (Velcade, formerly known as PS-341) is approved for treatment of multiple myeloma patients who have relapses from prior therapies (29). Treatment of MG132 at 1 μmol/L or of bortezomib at 10 nmol/L induced significant cell death in H929, U266, RPMI8226, and IM9 cells (data not shown), as previously reported (30). Both proteasome inhibitors caused ∼50% cell death in H929 cells after 24 h of treatment (data not shown). Notably, we found that Blimp-1 expression was reduced in H929 cells after treatment with proteasome inhibitors (Fig. 5A). Likewise, proteasome inhibitors caused reduction of Blimp-1 mRNA expression in multiple myeloma cells (Fig. 5B).

To test the causal relationship of down-regulation of Blimp-1 expression in proteasome inhibitor–mediated apoptosis, we ectopically expressed Blimp-1 using a bicistronic retrovirus expressing Blimp-1 and YFP (pGC-HABlimp-YFP; ref. 5). We observed a reduced proportion of apoptotic cells on MG132 treatment in pGC-HABlimp-YFP–transduced U266 cells compared with the control pGC-YFP–transduced cells (Fig. 5C), supporting the notion that Blimp-1 may be required for maintaining multiple myeloma cell survival. Because Blimp-1 has a survival role in multiple myeloma cells, we next examined whether Blimp-1 RNAi could sensitize multiple myeloma cells to proteasome inhibitors. We found that H929 cells transduced with U6Blimp1-1004i for 24 h showed increased apoptosis at early phase, ∼16 h, of 10 nmol/L bortezomib treatment (Fig. 5D). However, likely due to the strong cytotoxicity of bortezomib, we did not observe the enhanced cell death in U6Blimp1-1004i–infected H929 cells at later phase of drug treatment (data not shown).

Knockdown of IRF4 expression leads to apoptosis and reduced Blimp-1 expression. Interestingly, when examining the levels of Blimp-1 expression on proteasome inhibitor treatment, we also observed that the mRNA (data not shown) and protein levels (Fig. 5A) of another plasma cell transcription factor, IRF4, were significantly reduced. Because Blimp-1 is regulated by IRF4 (31), we examined whether IRF4 also regulates multiple myeloma cell survival by expressing shRNA against human and mouse IRF4 using a lentiviral vector in plasma cell lines. We found human U6IRF4-1278i to be more effective than U6IRF4-776i, and mouse U6IRF4-1278i to be more effective than U6IRF4-536i (Fig. 6A), for repressing IRF4 expression. Indeed, either human or mouse U6IRF4-1278i–transduced cells showed reduced levels of Blimp-1 (Fig. 6A). However, we did not observe reduced IRF4 expression in the reciprocal experiment using RNAi for Blimp-1 (data not shown). These results confirm that IRF4 may act upstream of Blimp-1 expression in plasma cells (31). Notably, we found that knockdown of IRF4 expression also caused cell apoptosis in H929, U266 multiple myeloma cells, and mouse P3X plasmacytoma cells (Fig. 6B and C) and that, for human or mouse U6IRF4-1278i, a high percentage of cell death correlated with a high efficiency of IRF4 and Blimp-1 knockdown (Fig. 6B). Together, these data suggested that IRF4 may regulate plasma cell survival through the regulation of Blimp-1.

Blimp-1 functions as a master regulator controlling a cascade of gene expression during plasma cell differentiation (6). The role of continuous Blimp-1 expression in plasma cells or in their neoplastic counterparts is not clear. Conditional deletion of the Blimp-1 gene, prdm1, after the formation of plasma cells in mice showed a reduced number of antigen-specific plasma cells from bone marrow (14). This observation implies that Blimp-1 may either regulate plasma cell survival or maintain plasma cell phenotype. Our data favor the former possibility. Failure to reprogram plasma cell fate by depletion of Blimp-1 expression is probably due to continuous suppression of PAX-5 because PAX-5 has been shown to repress plasma cell differentiation (32, 33). However, the mechanism of failure of reprogramming of plasma cell fate might also be due to a lack of the Blimp-1–independent coactivator/corepressor or an irreversible epigenetic modification. Although exogenous expression of BCL-6 has been shown to reprogram plasma cell fate in H929 and U266 cells (34), it is important to note that this was in an MTA3-dependent manner. It is therefore likely that our observed lack of plasma cell reprogramming in the presence of increased BCL-6 expression is linked to the fact that we also observed no concomitant change in MTA3 expression (Fig. 3C). The mode of action of Blimp-1 repression is through interaction with transcriptional repressors, including HDAC1/2 (8), Groucho (9), the histone methyltransferase G9a (10), and the arginine-specific histone methyltransferase Prmt5 (11). Our results showed that the majority of Blimp-1 target genes remained repressed whereas CIITA and BCL-6 were significantly derepressed on Blimp-1 knockdown in H929 cells, suggesting differential mode of action by which Blimp-1 regulates expression of target genes in plasma cells.

The expression of the plasma cell–specific transcription factors IRF4, Blimp-1, and XBP-1 has been confirmed in multiple myeloma cells (1). The role of XBP-1 in maintaining myeloma cell survival has been shown, as evidenced by the fact that depletion or inactivation of XBP-1 in myeloma cells leads to increased apoptosis in response to endoplasmic reticulum stress (35). IRF4 and Blimp-1 have overlapping expression patterns in a subset of centrocytes in the germinal center and in plasma cells (12). Recent gene expression profiling data revealed that the molecular signature of stimulated Irf4^−/− splenic B cells resembles that of prdm1^−/− B cells (31). Irf4-deficient mice have severe defects in mature T-cell and B-cell function and lack memory B cells and plasma cells (36). A recent study using conditional deletion of Irf4 in germinal center B cells indicated that IRF4 may be involved not only in plasmacytic differentiation but also in class switch recombination (37). In this study, we observed that IRF4 knockdown leads to the reduction of Blimp-1 (Fig. 6A), confirming the finding that IRF4 controls the expression of prdm1 (31). We also observed that IRF4 knockdown causes apoptosis in transformed plasma cells, consistent with the notion that Blimp-1 works downstream of IRF4 in supporting plasma cell survival. The mechanism underlying IRF4 regulation of plasma cell survival and prdm1 expression remains to be examined.

The mechanism that controls the continuous expression of Blimp-1 in plasma cells is unclear, but one possibility is through IL-6 signaling. IL-6 secreted from bone marrow plays an important role in promoting normal plasma cell and multiple myeloma cell growth/survival in the bone marrow milieu (38). In accord with these data, IL-6 and several IL-6 downstream signaling molecules, including signal transducers and activators of transcription 3 (STAT3) and activator protein-1, have been shown to induce Blimp-1 transcription (6, 39, 40). Our finding that Mcl-1 is reduced and Bim is induced on Blimp-1 knockdown seems to fit with the IL-6–mediated signaling scenario because blockade of IL-6 signaling down-regulates Mcl-1 and up-regulates Bim (41). In addition to regulating Mcl-1 and Bim, Blimp-1 suppresses another antiapoptotic Bcl-2 family protein, A1 (6, 42). However, A1 is an unlikely player in regulating plasma cell survival because it is strongly repressed in normal and malignant plasma cells (43). The mechanism that leads to up-regulation of Bim and down-regulation of Mcl-1 on Blimp-1 knockdown awaits further identification.

Proteasome inhibitors induce significant antitumor activity and apoptosis in multiple myeloma through multiple mechanisms, including inhibition of IL-6–triggered phosphorylation of the extracellular signal-regulated kinases 1/2, STAT3, and Akt (30), activation of mitochondrial cytochrome c release, activation of the caspase-9 and Fas/caspase-8–dependent apoptotic pathways (44), and inhibition of the activity of nuclear factor-κB (45). The molecular targets of proteasome inhibitor–mediated apoptosis in multiple myeloma, however, are not fully understood, although it has been shown that proteasome inhibitors, including MG132 and bortezomib, cause apoptosis by disrupting the unfolded protein response via suppression of the generation of the spliced form of XBP-1 in J558 myeloma cells (35). In this study, we found that IRF4 and Blimp-1 mRNA and protein expression is reduced on proteasome inhibitor treatment (Fig. 5), suggesting that factor(s) dependent on proteasome activity may regulate the expression of both genes in multiple myeloma. Additionally, Blimp-1 shRNA-transduced cells are more sensitive to bortezomib treatment (Fig. 5D), which may provide a clue for future therapeutic design.

Multiple myeloma is a largely incurable plasma cell malignancy that accounts for ∼20% of deaths from hematologic malignancy and ∼2% of deaths from cancer in the United States (46). The main obstacle for therapeutic elimination of multiple myeloma cells arises from the complex bone marrow microenvironment that provides growth factors and cell survival signals. Here, we have provided several lines of evidence showing that plasma cells rely on Blimp-1 for survival but not for maintaining plasma cell phenotype. Our results may help delineate future therapeutic regimens for plasma cell–related tumorigenesis, including multiple myeloma.

Grant support: National Science Council grants 93-2320-B-001-052, 94-2320-B-001-026, and 96-2320-B-001-013-MY3 and Academia Sinica intramural grant.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Drs. Kathryn Calame, Chung-Leung Li, and Wei Hsu for critical review of the manuscript; Pei-Chuan Lee and Wen-Wen Chen for excellent technical assistance; Dr. Shie-Liang Hsieh (National Yang-Ming University and Taipei Veterans General Hospital) for the gift of bortezomib; and Dr. J. Frederic Mushinski for advice and for the Ψ2 packaging cell line.