PIM2 kinase has a pivotal role in plasmablast generation and plasma cell survival, opening up novel treatment options in myeloma

Before becoming plasma cells (PCs), the B cells first pass through an immature, proliferative plasmablast (PB) stage. The PBs emerge after major changes in B-cell morphology, epigenome-sustained expression profile, and life span.^1,2 This tightly regulated cell metamorphosis can be hijacked by oncogenic alterations that drive malignant transformation. Although there is a large body of data on normal B-cell differentiation, studies of B-cell malignancies have shown that our understanding of normal B-cell function is incomplete,³ in particular for identifying specific factors and determining their spatiotemporal involvement in the molecular modifications underlying B-cell metamorphosis.

We have reported that the emergence of human PBs is associated with large-scale methylome modifications, with the local acquisition of active epigenetic marks on PC identity genes during a committal step in which some activated B cells differentiate into PBs.⁴ These highly proliferative, activated, committed B cells (hereafter referred to as preplasmablasts [prePBs]) are characterized by the silencing of the interleukin-4 (IL-4)/STAT6 pathway and thus downregulation of CD23 surface marker expression.⁵ By focusing on this feature, a comparative RNA sequencing (RNA-seq) analysis of prePBs allowed us to identify here a striking increase in expression of the gene encoding the proviral integrations of Moloney virus 2 (PIM2) serine/threonine kinase. The 3 members of the PIM kinase family are known to phosphorylate numerous substrates involved in crucial cellular functions.⁶ Dysregulated expression of PIM kinases has been reported in several cancers, including multiple myeloma (MM).^7,8 Surprisingly, PIM2 has never been analyzed in the context of normal B-cell differentiation. In mouse pro–B cells, Pim2 expression depends on growth factor–induced transcription,^9,10 which justifies further research in the context of PC differentiation.

The present study evaluated the effects of PIM2 on PB generation and PC survival. We found that the B-cell commitment activates the transcription of PIM2 in a STAT3-dependent manner, which leads to a strong increase in the protein’s expression level and functional activity. We assessed the biological effects of PIM2 in primary B cells by comparing a normal culture with 2 conditions in which the functions of PIM2 were altered by either stopping gene expression or blocking kinase activity. Along with PIM2-mediated BAD phosphorylation, we detected effects on p21^Cip1, CDC25A, and p27^Kip1. During the final step in differentiation, these effects all promote cell survival and entry into the S phase of the cell cycle. Moreover, PIM2 was found to be strongly expressed in quiescent mature PCs and essential for cell survival. In a malignant setting, PCs are addicted to PIM2, and targeting this kinase was synergistic with BH3 mimetics, offering new treatment options for MM.

Procedures and methods are detailed in the supplemental Methods (available on the Blood Web site).

Human myeloma cell lines were obtained and cultured as described previously.¹¹ Viability was assessed for cells treated with selected compounds: AZD1208 and AZD5991 (Selleck Chemicals). CellTiter-Glo Luminescent Assay kit (Promega) was used as described by the manufacturer. Results were normalized to the samples without treatment, and synergy scores were calculated by using the Bliss method¹² as described by Combès et al.¹³ 

Samples were collected with the approval of the institutional research board from Rennes University Hospital and in accordance with the Declaration of Helsinki. Mononuclear cells were isolated by using Ficoll density gradient centrifugation. Samples were cultured in the presence of 2 ng/mL of IL-6. Cell cytotoxicity was evaluated by using flow cytometry to determine the percentage of viable CD138⁺ cells.

Inhibition of PIM2 expression is enabled by the use of a vivo-morpholino PIM2 splice-switching antisense oligonucleotide (SSO-PIM2) from Gene Tools, LLC. An irrelevant vivo-morpholino standard oligonucleotide was used as control (SSO-CTL). PIM2 messenger RNA (mRNA) knockdown efficiency was evaluated by reverse transcription polymerase chain reaction (RT-PCR) to simultaneously identify full-length (286 pb) and alternatively spliced (176 bp) PIM2 mRNAs on an agarose gel. Quantification of inhibition was also performed by quantitative RT-PCR with exon 2 targeting primers and verified at the protein level by immunoblot.

Procedures and analysis methods have been described previously.⁵ 

To identify genes regulated specifically during the final commitment step, we used our previously described in vitro differentiation model with human naive B cells generating 4 cell populations on day 6 (D6) (Figure 1A).¹⁴ The RNA-seq data (confirmed by quantitative PCR assays) revealed a striking elevation of PIM2 mRNA expression (but not PIM1 or PIM3 expression) in prePBs (P2/CD23^– population) and PBs (P1 population), relative to B cells diverted from differentiation (P2/CD23⁺ and P3 populations) (Figure 1B; supplemental Figure 1A). In an unsupervised analysis, the results of single-cell quantitative RT-PCRs in P2/CD23^– vs P2/CD23⁺ cell populations ranked PIM2 as the most discriminating gene, ahead of PRDM1 (Figure 1C; supplemental Figure 1B). Analysis of our RNA-seq data sets showed that PIM2 was in the top 150 genes characterizing PBs, relative to noncommitted B-cell populations (supplemental Figure 1C).⁵ The protein levels correlated perfectly with gene expression levels, with strong PIM2 expression in prePBs and PBs (Figure 1B). With regard to human tonsil-derived cell populations, CD19⁺IgD^–CD38^bright PBs expressed higher levels of PIM2 than other B-cell populations (Figure 1D).

To determine whether PIM2 gene expression is coupled to genomic changes during PC differentiation, we analyzed our previously published data sets^4,5 and ENCODE annotations (https://www.encodeproject.org). An assay for transposase-accessible chromatin sequencing (ATAC-seq) showed that the open regions in prePBs and PBs included the PIM2 promoter and an upstream enhancer. This latter overlaps with a region denoted as interacting physically with the PIM2 promoter in the FANTOM5 database¹⁵ and with a superenhancer described in PBs and MM PCs (Figure 1E).¹⁶ The comparative analysis of genome-wide 5hmC marks in naive B cells vs PBs showed that read densities increased in both the identified ATAC-seq regions and the body of the PIM2 sequence; this action testified to the enhancer functions and activation in PBs, as has been shown for PB/PC identity genes (Figure 1E; supplemental Figure 1D).⁴ 

ATAC-seq analysis showed that the time course of promoter opening did not match the gene expression pattern; this disparity suggests that a transcriptional regulator is present in the PIM2 locus (Figure 1E; supplemental Figure 1D). By chromatin immunoprecipitation sequencing of human activated B cells, we identified a BACH2 binding site in the PIM2 promoter.¹⁷ BACH2 inhibition experiments revealed a subsequent increase in PIM2 expression and thus confirmed that BACH2 acts as a repressor (Figure 1F). Furthermore, the stimulation of PBs with IL-10, IL-21, and (to a lesser extent) IL-6 resulted in a sharp induction of PIM2 expression (Figure 1G). Given that all 3 cytokines are known to recruit the STAT3 pathway, we exposed the cells to the STAT3 inhibitor C188-9; this led to the dose-dependent repression of PIM2 expression (Figure 1G-H). In contrast, IL-2 (which recruits STAT5) and interferon-α were weak inducers. Because STAT3 signaling also induces PRDM1/BLIMP1 expression,¹⁸ we wondered if PIM2 was one of the downstream factors of BLIMP1. In fact, under IL-10 (or IL-21 [data not shown]) stimulation, PRDM1 was induced but later than PIM2 expression (supplemental Figure 1E). In addition, siPRDM1 experiments on primary B cells and U266 MM cells showed no modifications in PIM2 expression, unlike the well-known target genes of BLIMP1 (supplemental Figure 1F-G),^19,20 consistent with previous data in pre-PB/PBs from knockout Prdm1 mice.²¹ Finally, by chromatin immunoprecipitation–quantitative PCR on PBs and in RPMI8226 MM cells as well, we confirmed the binding of phosphorylated-STAT3 (pSTAT3) on the PIM2 promoter, as well as on the intron 3 of PIM2 locus (predicted enhancer region) and in the putative enhancer located downstream of the gene (Figure 1E; supplemental Figure 1H). Altogether, these results confirmed that PIM2 respond directly to STAT3 and not via BLIMP1. Interestingly, although PIM2-positive PBs were pSTAT3 positive in the absence of BACH2, PIM2-negative/BACH2-positive CD23⁺ activated B cells (CD23+ aBCs) were also positive for pSTAT3, suggesting that the downregulation of BACH2 is required before STAT3 can induce PIM2 expression (supplemental Figure 1I).

Taken as a whole, our results show that PB differentiation is associated with a sharp elevation of PIM2 kinase expression; this elevation requires the removal of BACH2 repression and STAT3 activation (Figure 1I).

To determine the functional involvement of PIM2 in the committal step and obtain further differentiated cells on D6, we focused our study on resulting PBs and CD23⁺ post-activated B cells (hereafter referred to as CD23⁺ aBCs), which both differed markedly in their PIM2 expression (Figure 2A). In subsequent experiments, we compared a normal culture setting with conditions in which PIM2 was inhibited on day 4 (D4) by exposure to the small-molecule pan-PIM inhibitor AZD1208 (hereafter referred as PIMi) or by specifically blocking PIM2 gene expression using a novel antisense RNA strategy based on a morpholino SSO (Figure 2B). Cells cultured with SSO-PIM2 exhibited a dose-dependent decrease in full-length PIM2 mRNA, the expression of a weaker band corresponding to alternatively spliced mRNAs lacking exon 2, as well as PIM2 protein expression (Figure 2C). At the selected dose of 2 µM, PIM2 expression was almost completely abrogated in PBs, without affecting PIM1 expression (Figure 2D; supplemental Figure 2B). Although exposure to SSO-PIM2 did not affect the expression of key transcription factors involved in B-cell differentiation (supplemental Figure 2C), PIM2 knockdown led to a sharp decrease in PB generation (Figure 2E-F; supplemental Figure 2D). In contrast, CD23⁺ aBCs were not affected. These results suggest that PIM2 kinase activity sustains the PB commitment.

A comparison of gene expression in prePBs vs noncommitted B-cell populations evidenced the downregulation of SMAD3 in the former population. Accordingly, prePBs upregulated CDC25A, CCND2, and CDK6 expression (supplemental Figure 3A), reentered the cell cycle, and gave rise to PBs.⁴ At D6, two days after inhibition of PIM2, the S phase decreased, and cells accumulated in G0/G1 (Figure 3A; supplemental Figure 3B). In parallel, the number of generated PBs decreased (Figure 2E-F), regardless of effects on cell death as shown by the QVD-OPH condition (supplemental Figure 3B-C). Moreover, after short exposure to PIMi (12 hours), the decrease of S phase was already observable in the absence of an effect on apoptosis (supplemental Figure 3B-D). Altogether, these results suggest that PIM2 promotes the G1/S transition independently of the cell death induction. We next performed cell cycle synchronization experiments in the XG21 MM cell line. As expected, unlike the control conditions, in which cells reached the S phase 6 hours after release, PIM2 inhibition was associated with blockage of the cells in G1 (supplemental Figure 3E).

The phosphatase CDC25A is required for the G1/S transition.²² CDC25A gene expression was significantly greater in PBs than in CD23⁺ aBCs, as well as the protein expression (Figure 3B). Chemical inhibition of CDC25A first led to a dramatic decrease in the proportion of S-phase cells and then, ultimately, to a strong reduction in the number of generated PBs (supplemental Figure 3F-G). Interestingly, PIM2 inhibition led to a decrease of CDC25A protein level in PBs without affecting CDC25A expression (Figure 3B; supplemental Figure 3H). At D4, the addition of PIMi to cycloheximide-treated cells accelerated the degradation of CDC25A (Figure 3C; supplemental Figure 3I), whereas in the presence of MG-132, cells maintained CDC25A expression (Figure 3D).

Cyclin-dependent kinase (CDK) inhibitor 1B (p27^Kip1, encoded by the CDKN1B gene) prevents progression to G1.²³ This effect is linked to the location of p27^Kip1 in the nucleus; cells that enter the cell cycle exhibit a cyclin D2–mediated nuclear export of p27^Kip1.²⁴ CDKN1B expression was significantly higher in PBs than in CD23⁺ aBCs, whereas the p27^Kip1 protein expression was lower and mainly located in the nucleus in CD23⁺ aBCs (Figure 3E). In addition, we detected the threonine 198-phopshorylated [p(Thr198)] form of p27^Kip1, predominantly expressed in PBs compared with CD23⁺ aBCs. These results prompted us to assess the half-life of p27^Kip1 in PBs vs CD23⁺ aBCs. Cycloheximide-treated PBs were more susceptible to p27^Kip1 proteasome-mediated degradation, as shown after the addition of MG-132 (Figure 3F-G). Interestingly, in highly proliferative D4 B cells, endogenous immunoprecipitation of p27^Kip1 revealed the binding of PIM2, and short treatment by PIMi resulted in a time-dependent decrease of p(Thr198)-p27^Kip1 (Figure 3H). PBs generated after PIM2 inhibition showed an increase in the global level of p27^Kip1 expression, independently to CDKN1B expression, particularly in the nucleus (Figure 3I; supplemental Figure 3J-K). In cycloheximide-treated PBs, the addition of PIMi stabilized p27^Kip1 (Figure 3J; supplemental Figure 3L). Finally, PIMi-treated PBs showed a decrease of p(Thr198)-p27^Kip1 as well as a decrease in poly-ubiquitinylated forms of p27^Kip1 in the presence of MG-132 (Figure 3K-L; supplemental Figure 3M).

Overall, PIM2 expression during PB differentiation promotes the G1/S transition in prePBs by stabilizing CDC25A phosphatase and increasing the degradation of cytoplasmic p27^Kip1 (Figure 3M).

The most prominent changes during PB commitment are related to cell death/survival.⁴ Indeed, in culture, the majority of B cells were dead on D6 (Figure 4A). The pan-caspase inhibitor QVD-OPH rescued the cells from death with the exception of the PBs; the latter seemed to be protected from apoptosis and despite an increase in CASP3 gene expression (Figure 4A-B). Investigations of the proteins produced by the various cell subsets in our differentiation model showed that PBs contained higher levels of antiapoptotic phosphorylated BAD (pBAD) and X-linked inhibitor of apoptosis protein (XIAP) (Figure 4C-D).^9,10 Consistent with this finding, immunoprecipitation of PIM2 revealed a binding on BAD most likely required before phosphorylation (Figure 4E). In accordance with this finding and despite the elevation in levels of procaspase 3, PBs displayed low levels of caspase 3 autocatalytic activity and thus did not produce many active p19/p17 forms [as highlighted by lower levels of cleaved poly(ADP-ribose) polymerase (PARP)] as P2/CD23⁺ cells did (Figure 4F).²⁵ Overall, PBs maintain caspase 9 in an inactivated state, and XIAP protein remains protected from degradation (Figure 4C-F).²⁶ XIAP is thought to have a key role in PBs by inhibiting caspase 3 activation. Indeed, immunoprecipitation of caspase 3 and XIAP revealed the binding of XIAP on p19/p17 fragments of caspase 3 (Figure 4G), which impairs their catalytic activity on procaspase 3.^27,28 Lastly, the P2/CD23⁺ cells’ complete degradation of PARP (Figure 4F) showed that terminal B-cell differentiation is associated with apoptosis if cells are not otherwise protected by the PB commitment. In PBs, PIM2 inhibition induced an increase in activated caspase 3 levels and thus a decrease in cell viability (Figure 4H-I; supplemental Figure 4A). This effect was driven by: (1) the strong mitochondrial depolarization due to the inhibition of pBAD; (2) the production of cleaved caspase 9; (3) the decrease in XIAP; (4) the detection of the p17 fragment of caspase 3; and (5) PARP cleavage (Figure 4J-L). Lastly, addition of the caspase 3 inhibitor Q-DEVD-OPH rescued the PBs from cell death (supplemental Figure 4B).

Taken as a whole, these data show that PIM2 counteracts the activation of caspase 3 in PBs and thus promotes cell survival.

As with p27^Kip1, nuclear p21^Cip1 is a CDK inhibitor and inhibits entry into the S phase of the cell cycle.²⁹ PB commitment was associated with a significant increase in transcription of CDKN1A (which encodes p21^Cip1), and levels of p21^Cip1 were markedly higher in PBs than in CD23⁺ aBCs (Figure 5A). Surprisingly, p21^Cip1 was essentially expressed in the cytoplasm, and endogenous immunoprecipitation experiments yielded a multiprotein complex in which p21^Cip1 interacted with PIM2 and heat shock protein 90β (HSP90β) (Figure 5B-C). Moreover, PIM2 inhibition led to a dramatic decrease in cytoplasmic p21^Cip1 without effect on either CDKN1A expression or nucleus relocalization (Figure 5D; supplemental Figure 5A-B). The addition of PIMi to cycloheximide-treated cells accelerated the degradation of p21^Cip1 (Figure 5E). Treatment with MG-132 induced a strong accumulation of p21^Cip1, and the addition of PIMi was associated with elevated poly-ubiquitinylated p21^Cip1 (Figure 5F-G). XG21 cells presented similar results (supplemental Figure 5C-F). Collectively, these results show that PIM2 stabilizes p21^Cip1 in PBs by promoting its binding to a cytoplasmic multiprotein complex and, thus, preventing proteasome-mediated degradation of p21^Cip1.

This PIM2-induced cytoplasmic stabilization of p21^Cip1 prompted us to question the latter’s role in PB commitment. As described previously for other cell types,^30,31 we hypothesized that p21^Cip1 interacts with procaspase 3. In fluorescence resonance energy transfer (FRET)³² experiments on XG21 cells, we detected a p21^Cip1/procaspase 3 complex whose formation was abrogated by pretreatment of the cells with PIMi (Figure 5H). Immunoprecipitation experiments confirmed the presence of this complex in PBs (Figure 5I; supplemental Figure 5G). To determine whether p21^Cip1 mediates apoptosis inhibition, small interfering RNA was used to downregulate CDKN1A expression (supplemental Figure 5H). On D6, the generated PBs were found to contain p17-forms of caspase 3 and cleaved PARP, confirming that apoptosis had taken place (Figure 5J; supplemental Figure 5I). Treatment with the small-molecule p21^Cip1 inhibitor UC2288 was associated with a dose-dependent decrease in cell viability, which was restored by treatment with the caspase 3 inhibitor Q-DEVD-OPH (Figure 5K; supplemental Figure 5J-K).

Taken as a whole, these findings show that in addition to its mitochondrial effects, PIM2 inhibits the execution of apoptosis by stabilizing cytoplasmic p21^Cip1, allowing the latter to bind to procaspase 3, and thus inhibiting the production of activated caspase 3 (Figure 5L).

PC maturation is a continuum that runs from PBs to fully mature, long-lived PCs such as those residing in the bone marrow (BM) niche, in which cell survival depends on various extrinsic factors.³³ A comparison of various tonsil-derived B-cell populations, PBs, and BM PCs revealed that the level of PIM2 expression increased along the maturation pathway (Figure 6A). On paraffin-embedded BM biopsy specimens, staining experiments revealed that virtually all BM PCs (detected either with CD138 or B-cell maturation antigen markers) were strongly positive for PIM2 (Figure 6B; supplemental Figure 6A-B).

To characterize the regulation and functional contribution of PIM2 in mature PCs, we adapted our cell culture model so that it produced nonproliferating CD138⁺ PCs (supplemental Figure 6C-F), in line with reports on an early PC (ePC) phenotype.^34,35 As was the case in BM PCs, high levels of PIM2 mRNA and protein expression were observed in ePCs (Figure 6C). This expression was driven by IL-6 via STAT3 signaling and by the supernatant from a mesenchymal stromal cell culture (Figure 6D). As expected, the addition of anti–IL-6 and anti–IL-6 receptor antibodies to the cell culture prevented the induction of PIM2. In this context, we observed clusters of CD138⁺/PIM2⁺ PCs in some areas on BM biopsy specimens, close to CD41⁺ megakaryocytes, cells known to produce large amounts of IL-6 and participating in PC BM niche (Figure 6E).³³ 

Treatment of ePCs with PIMi induced an elevation of activated caspase 3 levels, and immunoblot experiments confirmed the increase in mitochondrial depolarization and the execution of apoptosis (Figure 6F-G). PIMi treatment also led to a significant decrease in p21^Cip1 levels (Figure 6G). Interestingly, ePCs (as in some BM PCs) only expressed p21^Cip1 in the cytoplasm (Figure 6H; supplemental Figure 6G), and FRET experiments revealed that PIM2 interacted with p21^Cip1 as well as p21^Cip1 with caspase 3 (Figure 6I).

Taken as a whole, our data show that PIM2 is still required in quiescent PCs to protect cells from caspase 3 activation; PIM2 acts at various levels and notably stabilizes cytoplasmic p21^Cip1.

Although the involvement of PIM2 in MM has been recognized for more than a decade, our present findings prompted us to further investigate the status of PIM2 in malignant PCs. Data from the DepMap Project showed that mRNA expression of PIM2 (but not of PIM1) was elevated in MM cells compared with other tumor cells, whereas CRISPR screening revealed a specific high PIM2 dependency score for malignant PCs (Figure 7A-B; supplemental Figure 7A). BM biopsy results from patients with MM revealed intense, widespread expression of PIM2 (Figure 7C). Data from the University of Arkansas Total Therapy 2 (TT2) MM cohort showed the strongest PIM2 expression levels in the high-risk molecular subgroups (supplemental Figure 7B).³⁶ An analysis of data from 765 patients in the multiple myeloma research fundation CoMMpass Study cohort showed that PIM2 expression was significantly elevated in high-risk cytogenetic subgroups and low in low-risk subgroups (supplemental Figure 7C).^37,38 In addition, high PIM2 expression patients compared with the PIM2^low group had a worse survival probability (Figure 7D). This is in line with small conditional RNA-seq data on malignant PCs from patients resistant to the carfilzomib-daratumumab combination therapy, which presented higher PIM2 expression (supplemental Figure 7D).³⁹ Likewise, we observed a significant increase in PIM2 expression in cells from an in-house cohort of 14 patients with disease progression soon after first-line treatment (supplemental Figure 7E). Collectively, these data suggest that the severity of MM is associated with PIM2 expression levels.

Drug resistance is the greatest challenge in MM. It might be possible to overcome this obstacle by acting at the mitochondrial level and lowering the threshold for triggering mitochondrial apoptosis.⁴⁰ Venetoclax is reportedly effective in BCL2-translocated MM patients.⁴¹ Furthermore, myeloid cell leukemia-1 (MCL1) is also an attractive target in MM^41,42 because of its essential effects on PC survival via inactivation of the pro-apoptotic multidomain of BAX and BAK effectors.⁴³ However, in contrast to BCL2 and BCL-xL, MCL1 activity is not antagonized by BAD.⁴⁴ Interestingly, data from the TT2 MM cohort showed that the expression levels of MCL1 and PIM2 are significantly correlated (supplemental Figure 7F). We hypothesized that targeting both PIM2 and MCL1 might have synergistic effects on mitochondria depolarization because the 2 proteins act at different levels (Figure 7E). To this end, we first evaluated the respective effects of PIMi,⁴⁵ an MCL1 inhibitor (AZD5991, referred to hereafter as MCL1i)⁴⁶ alone and then when combined. The XG7 and RPMI8226 cell lines were sensitive to MCL1i but the U266 cells were not (supplemental Figure 7G). Low-dose treatment with PIMi greatly enhanced the response to MCL1i in XG7 and RPMI8226 cells (Figure 7F; supplemental Figure 7H). Interestingly, U266 cells (resistant to PIMi alone and MCL1i alone) became sensitive when the drugs were combined (Figure 7G). Lastly, PIMi and MCL1i had synergistic effects on primary MM cells from 6 patients without effect on CD138^– cells (Figure 7H).

Taken as a whole, these results confirmed that the mitochondrial control of PIM2 is similar in normal and malignant PCs; this finding opens up novel treatment options in MM.

The processes by which committed B cells and nascent PBs switch from one molecular program to another had not previously been fully characterized. Here, our experiments with human primary B cells showed that PIM2 kinase is a pivotal factor for the generation and maintenance of PB and PC populations. In particular, our results show how committed B cells reenter the cell cycle after their proliferation burst and, under subsequently restricted culture conditions, express high levels of PIM2 and thus decrease mitochondrial apoptosis.

Our molecular characterization provided strong evidence of the regulation of PIM2 gene expression and illustrates the status of PIM2 as a true PC-identity gene recruited at the same time as PRDM1 and IRF4, once BACH2 repression has been released. In prePBs and PBs, both IL-10 and IL-21 cytokines drive PIM2 expression via pSTAT3 signaling and (in PCs) by IL-6, which establishes a direct link between growth factor–induced transcription and a kinase-dependent pathway that promotes survival and cell cycle reentry.

The cell cycle provides windows of opportunity for chromatin remodeling and thus gene activation.⁴⁷ Committed B cells increase their expression of CDC25A, and in parallel PIM2 helps to stabilize the CDC25A protein and thus prompts cells to enter the S phase and progress to the PB state independently of effects on apoptosis.²² Interestingly, promotion of the cell cycle by PIM2 is accentuated by the degradation of cytoplasmic p27^Kip1, which prevents the latter from accumulating or returning to the nucleus. Indeed, the effect of PIM2 on p27^Kip1 is mediated by the phosphorylation of threonine 198 residue, consistent with the known cytoplasmic localization of p(Thr198)-p27^Kip1 and its binding to 14-3-3 leading to its proteasomal degradation.⁴⁸ In contrast, in PCs, p27^Kip1 accumulated in the nucleus, independently of high PIM2 expression, to halt the cell cycle and so without being degraded by cytoplasmic PIM2.

Caspases 9 and 3 become resistant to activation by apoptotic stimuli when B cells differentiate into PBs, and the apoptosis of short-lived PCs in vitro and in vivo is largely independent of the key apoptotic caspases.⁴⁹ Our present results provide insights into the mechanisms whereby B-cell death is avoided, despite intense expression of cell death–related genes in committed cells,⁴ allowing PB generation and survival. Production of pBAD helps to maintain the integrity of mitochondria, as evidenced by increased XIAP expression and the overall inhibition of caspase 3 self-activation. Knockdown and dose-dependent experiments have shown that this mitochondrial adaptation is mediated and made possible by high levels of PIM2 expression and its binding activity to BAD; this is in line with data from constitutive PIM2 expression models, with the detection of pBAD, its binding to 14-3-3, and inhibition of BAD/BCL-XL binding.^10,50

One of the ways in which PIM2 regulates apoptosis during the differentiation of B cells into PBs involves restricting p21^Cip1 to the cytoplasm. During this step, cells progress from G1 to S and proliferate; hence, the cytoplasmic expression of p21^Cip1 makes sense because its presence in the nucleus blocks cells in G1.⁵¹ However, further investigations on quiescent PC populations show that p21^Cip1 stays in the cytoplasm, which sheds light on its pleotropic functions. In PBs, PIM2 stabilizes cytoplasmic p21^Cip1 by promoting the formation of a complex with heat shock protein 90β and thus reducing the proteasomal degradation of p21^Cip1. Interestingly, p21^Cip1 also binds to procaspase 3 in a PIM2-dependent manner, and FRET experiments showed that the addition of PIMi abrogated binding of p21^Cip1 to procaspase 3. In PBs, PIM2 acts by promoting the formation of a multiprotein complex, which inhibits the caspase 3 activation. This effect complements the production of pBAD and the inhibition of mitochondrial depolarization and shows that PIM2 acts in several ways to ensure the survival of PBs.

Entry into cell cycle quiescence accompanies the final phenotypic maturation of PCs and requires CDK inhibitors regulated by IL-6.⁵² By modifying the culture conditions, we obtained fully mature PCs that strongly expressed PIM2 and absolutely required IL-6. At the cellular level, the antiapoptotic effects of PIM2 were characterized by maintenance of mitochondrial integrity and promotion of caspase 3/p21^Cip1 binding. Although it is now accepted that viability is not an intrinsic property of BM PCs but is conferred on the cells by their immediate environment,⁵³ our present observations of marked expression of PIM2 in BM PCs suggest that this kinase has a crucial role in long-term cell survival and might be involved in the functional activity of the PC niche through IL-6–dependent crosstalk between the cell and its microenvironment.⁵⁴ 

Our results revealed a hitherto unexpected role of PIM2 kinase in terminal B-cell differentiation and, more generally, the biology of the PCs. The pleotropic effects of PIM2 kinase enable the cell to modify its regulation of apoptosis and to promote survival when faced with protein secretion overload. The data from the DepMap Project show that malignant PCs are “addicted” to PIM2, and elevated PIM2 expression in the most severe cases of MM suggests that PIM2 is linked to clinical severity. Small-molecule inhibitors of PIM have produced some encouraging results in preclinical models and recently in MM patients with relapsed and/or refractory disease.⁵⁵ Indeed, and despite a relative low objective response rate (partial response or better), in patients who reached the recommended dose of treatment, PIM447 drug exhibited single-agent antitumor activity with a significant improvement of the median progression-free survival and the rate of clinical benefit. However, BM examination of patients failed to reveal a clinically significant reduction in malignant PCs, suggesting that PIM447 exerted antitumor responses via cytostatic rather than apoptotic mechanisms.⁵⁵ Taking into account that MM tumor cells increase their dependence on MCL1 during the course of the disease,⁴¹ we could speculate that MCL1 could counteract the induction of cell death by PIMi. In the context of the PIM447 clinical trial,⁵⁵ antiproliferative effects could appear at low doses, whereas cell death would require dose escalation, possibly with unacceptable toxicities, to counterbalance MCL1 dependency. Given that the action of MCL1 on the mitochondrial membrane is independent of BAD, dual pharmacologic inhibition of PIM2 and MCL1 is highly synergistic; low doses of PIMi significantly increased the efficacy of the anti-MCL1 drug. Interestingly, this drug combination can reverse the primary resistance of U266 cells to MCL1i and PIMi.

In conclusion, our present results showed that the final transcriptional modifications in B cells committed to differentiation into PBs are associated with the expression of a set of crucial genes, including PIM2. The newly established adaptive status of the generated PBs is maintained in mature PCs. Lastly, we showed that the antiapoptotic effects PIM2 kinase should be further investigated in the context of novel treatment strategies in MM.

Cell sorting was performed at the Biosit Flow Cytometry and the CytomeTRI cell-sorting facilities (UMS6480 Biosit). The authors are indebted to the Centre de Ressources Biologiques–Santé (BB-0033-00056; http://www.crbsante-rennes.com) at Rennes University Medical Center for its help with processing biological samples, and they also thank several clinicians and patients for providing samples for research purposes. The authors thank the MMRF (Multiple Myeloma Research Foundation) for sharing biological and clinical data for patients enrolled in the CoMMpass study via the MMRF genomics portal.

M.H. has received a doctoral fellowship from FHU CAMIn, Ligue Contre le Cancer/Comité d’Ile et Vilaine. A.P. received a fellowship from Région Bretagne and Ligue Nationale contre le Cancer. Immunofluorescence studies were performed at the Microscopy Rennes imaging center (MRic) and in the H2P2 facility, both of which are members of the UMS 6480 Biosit and the French national France-BioImaging infrastructure network funded by the French Research Agency (ANR-10-INBS-04). Funding support for this article was provided by the an internal grant from the Hematology Laboratory, university hospital of Rennes, and la Ligue contre le Cancer Région Grand Ouest.

Contribution: M.H. designed and conducted the experiments and wrote the paper; G.C. designed the cell differentiation model and performed cytometry and cell sorting; F.C. performed genomic analysis; E.A. performed chromatin immunoprecipitation–quantitative PCR STAT3 experiments; J.D. performed drug combination experiments; A.P. performed single-cell quantitative PCRs; C.D. helped with BACH2 assays and discussed results; S.M. performed CDC25A immunoblotting; G.B. performed FRET experiments; R.V. performed immunostaining; F.L.-G. provided BM biopsy specimens; A.M. designed morpholino SSO; O.D. provided MM samples; K.T. discussed results; L.D. provided the antisense strategy; J.M. provided cell lines and original data; and T.F. designed and supervised the experiments and wrote the paper.

Conflict-of-interest disclosure: M.H., G.C., A.M., L.D., J.M., and T.F. have applied for a European patent on the use of an SSO to modulate PIM2 expression. The remaining authors declare no competing financial interests.

Correspondence: Thierry Fest, INSERM U1236, Faculté de Médecine, 2 avenue du Pr Léon Bernard, CS 34317, 35043 Rennes Cedex, France; e-mail: [email protected].