Bone marrow microenvironment–derived signals induce Mcl-1 dependence in multiple myeloma

Long-lived plasma cells can survive in the bone marrow and produce antigen-specific antibodies for decades even in the absence of repeat antigen exposure.¹  Survival of these cells is governed by a family of proteins that includes antiapoptotic members such as Bcl-2, Bcl-x_L, and Mcl-1, as well as proapoptotic members such as Bim. The antiapoptotic proteins bind to and sequester Bim, preventing it from activating apoptotic effector proteins Bax and Bak in the mitochondrial membrane.^2,3  Despite coexpression of multiple antiapoptotic proteins, plasma cells depend primarily on Mcl-1 for survival. Treatment of mice with ABT-737, a compound that inhibits Bcl-2 and Bcl-x_L, results in a significant loss of B cells but has minimal impact on preexisting plasma cells.⁴  Furthermore, conditional deletion of Mcl-1 leads to dramatic reduction in the number of bone marrow plasma cells.^5,6 

As cells undergo malignant transformation, they must overcome a number of proapoptotic signals in order to survive, leaving them even more dependent upon antiapoptotic members of the Bcl-2 family, and thus susceptible to Bcl-2 family inhibitors such as venetoclax (ABT-199), a Bcl-2 specific inhibitor, and navitoclax (ABT-263), a Bcl-2/Bcl-x_L inhibitor.⁷  Despite the numerous changes that occur during transformation, malignant plasma cells tend to remain dependent on Mcl-1 as demonstrated by the use a small-molecule Mcl-1 inhibitor and small interfering RNA (siRNA) knockdown of Mcl-1.^8,9  However, treatment of myeloma patient samples and human myeloma cells lines with venetoclax or ABT-737 revealed that a subset of myeloma is sensitive to these drugs and must therefore be Bcl-2/Bcl-x_L codependent.^9-13 

In addition to preserving Mcl-1 dependence, myeloma also remains highly dependent on the bone marrow microenvironment. Bone marrow stromal cells (BMSCs) provide survival signals through adhesion molecules and secretion of cytokines such as interleukin-6 (IL-6), insulin-like growth factor-1 (IGF-1), and vascular endothelial growth factor. These cytokines contribute to myeloma survival by activating numerous signaling pathways.¹⁴  The tumor necrosis factor family members B cell activating factor (BAFF) and a proliferation-inducing ligand (APRIL) also promote plasma cell survival,^15,16  and along with IL-6 have been reported to increase expression of Mcl-1 and Bcl-x_L.^5,17-23  However, we have previously demonstrated that expression of Mcl-1, Bcl-x_L, and Bcl-2 alone cannot account for Bcl-2 family dependence. Instead, preferential binding of Bim to Mcl-1 vs Bcl-2 or Bcl-x_L correlates with ABT-737 sensitivity more consistently.⁹  We therefore set out to understand how the stromal microenvironment influences myeloma dependence on Bcl-2 family proteins.

RPMI8226 (8226) and Hs-5 cells were purchased from the American Type Culture Collection. KMS11 and KMS12BM were purchased from the Japanese Collection of Research Bioresources Cell Bank. MM.1s was provided by Steven Rosen (City of Hope), KMS18 by Leif Bergsagel (Mayo Clinic), OCI-My5 by Jonathan Keats (TGen), and OPM2 by Nizar Bahlis (University of Calgary). All cell lines were validated using sequencing and phenotypic characterization.

Hs-5 and patient-derived BMSCs were plated 48 hours prior to collection of conditioned media, which was then filtered using a 0.2 micron syringe filter, diluted to the appropriate concentration with complete RPMI, and used to resuspend myeloma cells. Myeloma cells were incubated in conditioned media for 30 minutes prior to addition of drug. Myeloma cells cocultured with either Hs-5 cells or patient-derived BMSCs were added to preplated stromal cells in fresh media (following removal of the conditioned media) and given 30 minutes to attach before treatment. For IL-6 blocking experiments, antibodies were added directly to control and conditioned media at least 30 minutes before cell resuspension.

Cells were treated with U0126, LY294002 (Cell signaling), AZD1480, ruxolitinib (Selleckchem), recombinant human IL-6, anti-IL-6 antibody, and anti-IL-6R antibody (R&D systems). ABT-737 and ABT-199 were generously provided by AbbVie.

Cell death was measured by annexin V-fluorescein isothiocyanate and propidium iodide (PI) staining as previously described.²⁴ 

The following primary antibodies were used for western blot: Bim (EMD Millipore); Mcl-1 (Enzo Life Sciences); Bcl-2 (BD Biosciences); β-actin (Sigma-Aldrich); Bcl-x_L, phospho-Ser69 Bim, phospho-Ser77 Bim, phospho-extracellular signal-related kinase 1/2 (Erk1/2) (Thr202/Tyr204), Erk1/2, phospho-Akt (Ser473), Akt, phospho-STAT3 (Tyr705), STAT3 (Cell Signaling Technology). For coimmunoprecipitation, the following antibodies were used: Mcl-1, Bcl-2 (BD Biosciences); Bcl-x_L (7B2.5)²⁵ ; and Bim (Santa Cruz Biotechnology). Secondary anti-mouse immunoglobulin G–horseradish peroxidase (IgG-HRP), anti-rabbit IgG-HRP (Santa Cruz Biotechnology), anti-hamster IgG-HRP (BD Biosciences) were used for western blotting. CD138-fluorescein isothiocyanate (FITC), CD45-allophycocyanin-Cy7, and CD38-phycoerythrin (BD Biosciences) were used for flow cytometry.

Western blotting was performed using standard techniques as previously described.²⁶ 

Immunoprecipitation experiments were performed using the Exacta-Cruz C Kit (Santa Cruz Biotechnology) following the manufacturer’s instructions as previously described.²⁶ 

siRNAs were purchased from Dharmacon (GE Life Sciences). ON-TARGETplus SMART pool siRNA against STAT3 and the siCONTROL nontargeting siRNA were used per manufacturer’s protocol.

Real time was performed as previously described²⁶  using Applied Biosystems High Capacity cDNA Reverse Transcription Kit (Life Technologies) and amplified using the TaqMan Gene Expression Master Mix (Life Technologies) on the 7500 Fast Real-Time PCR System following the manufacturer’s protocol (Applied Biosystems).

All samples were collected following an Emory University Institutional Review Board–approved protocol as previously described.⁹ 

In order to evaluate the role of stromal cells in Bcl-2 family dependence, we compared the response of patient myeloma cells to ABT-737 either in the presence or absence of stromal cells obtained from the same bone marrow aspirate. Cells codependent on Bcl-2/Bcl-x_L are sensitive to ABT-737–induced cell death, defined by an 50% inhibitory concentration (IC₅₀) of <0.5 μM, whereas cells not dependent on Bcl-2/Bcl-x_L are resistant to ABT-737. Freshly isolated bone marrow aspirates from myeloma patients were divided into 2 fractions, one containing unmanipulated plasma cells and stromal cells, and a second subjected to magnetic bead separation to isolate a pure population of CD138⁺ plasma cells. Both fractions were then treated with increasing concentrations of ABT-737 to determine an IC₅₀ value with and without stromal cells. The presence of stromal cells reduced the degree of Bcl-2/Bcl-x_L dependence in myeloma cells in a large fraction of the samples (Figure 1A). In 14 of the 24 samples tested, the IC₅₀ increased by more than twofold in the presence of stromal cells, increasing in 1 sample by almost 40-fold. Furthermore, in 9 of the ABT-737–sensitive CD138⁺ selected samples, the presence of stroma increased the IC₅₀ to >0.5 μM, indicating that the myeloma was no longer Bcl-2/Bcl-x_L dependent. These data therefore suggest that stroma-derived signals drive loss of Bcl-2/Bcl-x_L codependence and are a potentially important source of resistance to Bcl-2 inhibitors.

To corroborate our patient sample findings, we cultured a panel of 7 human myeloma cell lines with the Hs-5 stromal cell line and treated the cells with ABT-737. Using Mcl-1 siRNA knockdown, coimmunoprecipitation, and ABT-737 sensitivity, we have previously shown that MM.1s, 8226, and KMS18 are codependent on Bcl-2/Bcl-x_L and Mcl-1, whereas OPM2 and U266 are primarily Mcl-1 dependent, dying with only high concentrations of ABT-737.⁹  KMS12BM and OCI-My5 are highly sensitive to ABT-199 and thus Bcl-2 dependent (supplemental Figure 1, available on the Blood Web site). In all 7 cell lines, the presence of stromal cells inhibited apoptosis induced by ABT-737 or ABT-199 (Figure 1B-C). Stromal cells had little to no effect on bortezomib- and arsenic trioxide–treated cells, arguing that stromal cell protection is specific to the mechanism of action of ABT-737 (Figure 1B). These results are consistent with previous studies demonstrating that the effectiveness of bortezomib and arsenic trioxide are independent of Bcl-2 family dependence or target Mcl-1 by upregulating Noxa.^26-28  Culturing the cell lines with Hs-5 conditioned media resulted in equivalent levels of protection from ABT-737 compared with Hs-5 cells, suggesting that a secreted factor is contributing to loss of Bcl-2/Bcl-x_L dependence. We obtained similar results when culturing MM.1s with stromal cells and conditioned media from 4 different myeloma patient samples (supplemental Figure 2).

IL-6 is a known stromal-derived plasma cell survival cytokine. We therefore hypothesized that IL-6 may be a component of the conditioned media responsible for reducing Bcl-2/Bcl-x_L dependence and ABT-737 sensitivity. Indeed, addition of 10 ng/mL recombinant human IL-6 protected cells from ABT-737 much like conditioned media and stromal cells (Figure 2A). Conversely, blocking IL-6 with a neutralizing antibody reversed the protective effects of Hs-5 conditioned media (Figure 2B). When cells are cultured in 0.5% conditioned media, reversing IL-6–induced loss of Bcl-2/Bcl-x_L dependence can be achieved either with a neutralizing antibody or by antagonizing the IL-6 receptor. However, in the presence of 50% conditioned media or stromal cells, the combination of IL-6 neutralization and IL-6 receptor blockade was more effective than either alone (Figure 2C). Normal BMSCs were also able to reduce Bcl-2/Bcl-x_L dependence and rescue MM.1s cells from ABT-737–induced apoptosis, confirming that these results are not an artifact of using an immortalized stromal cell line or changes in the stroma of myeloma patients (Figure 2D). Similar to the findings with Hs-5, IL-6 neutralization partially reversed the protective effect of freshly isolated stromal cells. Finally, freshly isolated plasma cells from myeloma patients behaved similarly to MM.1s. IL-6 reduced the cell death with ABT-737 in isolated plasma cells while an IL-6 neutralizing antibody partially reversed the protective effect of coculture with stromal cells (Figure 2E). A second stromal-derived cytokine, IGF-1, had little to no effect on ABT-737–induced apoptosis, suggesting IL-6 is unique in its ability to inhibit Bcl-2/Bcl-x_L dependence (supplemental Figure 3).

IL-6 has previously been reported to affect the expression of antiapoptotic Bcl-2 family members such as Mcl-1, Bcl-x_L, and Bcl-2.^17,19-23  To determine how IL-6 regulates these proteins in KMS18 and MM.1s, we examined changes in protein expression with IL-6, conditioned media, and Hs-5 cells. None of the 3 conditions altered expression of Bim, Bcl-2, Bcl-x_L, or Mcl-1 in MM.1s (Figure 3A). However, we did observe increased Mcl-1 protein expression with these treatments in KMS18 (Figure 3B). To better understand IL-6 induced changes, we treated 5 cell lines with IL-6 and quantitated protein expression of Bcl-2, Bcl-x_L, and Mcl-1 by western blot. IL-6 exposure resulted in a statistically significant increase in Mcl-1 protein in only 1 of the 5 cell lines (KMS18; Figure 3C-D). Bcl-2 protein decreased by less than twofold in 2 of the 5 cell lines, whereas Bcl-x_L protein levels did not change in any of the cell lines. We also measured the effects of IL-6 on messenger RNA (mRNA) levels (Figure 3E). In KMS18 (Mcl-1) and MM.1s (Bcl-2), IL-6 induced changes in mRNA reflected the changes we observed at the protein level. Interestingly, similar changes in Mcl-1 and Bcl-2 mRNA were present in additional cell lines, and Bcl-x_L mRNA also decreased with IL-6 treatment in all 5 lines (supplemental Figure 4A); however, these did not translate into significant protein level changes (Figure 3C). This suggests that although IL-6 signaling may be similar in these lines, the impact of transcriptional changes is limited, although the lack of change in Bcl-2 and Bcl-x_L protein may be related to the longer half-life of these proteins.^29,30  We confirmed that IL-6 is highly active in all 5 cell lines by demonstrating significant upregulation of SOCS3 RNA, a known STAT3 target (supplemental Figure 4B).³¹ 

The shift in expression from Bcl-2 to Mcl-1 could potentially explain the IL-6–induced resistance to ABT-737 in KMS18 because ABT-737 does not inhibit Mcl-1. However, we have previously shown that expression levels alone cannot account for Bcl-2 family dependence.⁹  Despite coexpression of multiple antiapoptotic proteins, Bim preferentially binds to certain antiapoptotics in a cell-specific manner as demonstrated by coimmunoprecipitation. In MM.1s, Bim binding to Bcl-2 and Bcl-xL is greater, making the cells Bcl-2/Bcl-xL dependent and ABT-737 sensitive, whereas in U266, Bim is almost exclusively bound to Mcl-1, making it Mcl-1 dependent and insensitive to ABT-737.⁹  We performed similar coimmunoprecipitation experiments in KMS18 and MM.1s in the presence or absence of IL-6 to determine whether IL-6 altered the pattern of Bim binding. Mcl-1, Bcl-x_L, and Bcl-2 were immunoprecipitated from control cells or after 24 hours with IL-6, and the Bim associated with each antiapoptotic was quantitated by western blot. In both cell lines, the proportion of Bim bound to Mcl-1 increased at the expense of Bcl-2 associated Bim (Figure 3F). Therefore, in addition to decreasing Bcl-2/Bcl-x_L dependence, IL-6 simultaneously increases Mcl-1 dependence by altering Bim priming, decreasing the amount of Bim available to be released from Bcl-2 and Bcl-x_L while increasing the proportion bound to Mcl-1. Importantly, in MM.1s the change in Bim priming occurred without changes in the amount of coprecipitating Mcl-1, Bcl-x_L, or Bcl-2, suggesting that the change in Bim binding is because of posttranslational modification.

Although our data in 4 of 5 cell lines suggest that the changes in Bim binding and Mcl-1 dependence with IL-6 are not mediated by changes in Mcl-1 expression, this cannot be ruled out in KMS18, where a significant increase in Mcl-1 is observed following IL-6 treatment. Because the change in Mcl-1 expression occurs at both the mRNA and protein level we examined the role of STAT3 in IL-6–induced changes in KMS18 using siRNA silencing. Consistent with a role for STAT3-mediated transcription, silencing blocked the upregulation of Mcl-1 mRNA (Figure 4A) and protein (Figure 4B), but only partially reversed IL-6 induced protection from ABT-737 (Figure 4C), suggesting that in KMS18 IL-6 stimulation results in transcription independent events such as posttranslational modification that contribute to Bcl-2/Bcl-x_L dependence. Efficient STAT3 knockdown was confirmed by western blot (Figure 4B) and transcription of SOCS3 (Figure 4A). STAT3 silencing had no effect in MM.1s suggesting a transcription independent mechanism in these cells (supplemental Figure 5). STAT3 silencing also had no effect on the basal levels of Mcl-1 and did not result in any cell death on its own, suggesting that STAT3 may be responsible only for IL-6–induced upregulation of Mcl-1 in KMS18.

Having determined that IL-6 induces Mcl-1 dependence through both transcriptionally dependent and independent mechanisms, we then examined the contribution of the signaling pathways activated by IL-6. Binding of IL-6 to its receptor activates the JAK/STAT, Ras/MAPK, and phosphatidylinositol 3-kinase (PI3K)/Akt pathways.¹⁴  We treated KMS18 and MM.1s with inhibitors of JAK (AZD1480, ruxolitinib), MEK (U0126), and PI3K (LY294002) and measured apoptosis in the presence of ABT-737 and IL-6. Not surprisingly, JAK inhibition completely reversed the protective effect of IL-6, most likely by blocking the initial activation event and therefore all downstream pathways (Figure 5A). The MEK inhibitor was as effective as JAK inhibition in MM.1s, suggesting that the Ras/MAPK pathway is vital to rescuing these cells from ABT-737. However, in KMS18, MEK inhibition only restored ∼50% of the apoptosis induced by ABT-737. The discrepancy between these 2 lines may reflect underlying differences in the biology of the cell lines. Of note, neither ruxolitinib, U0126, nor LY294002 induced any cell death when given alone, but U0126 did sensitize both cell lines to ABT-737 such that we observed increased apoptosis with the combination compared with ABT-737 alone. In fact MEK inhibition increased the sensitivity of OPM2 and U266 to ABT-737, although overall they remained resistant when compared with KMS18 and MM.1s (supplemental Figure 6). Moreover, MEK inhibition also increased the sensitivity of all 4 cell lines to the Bcl-2 specific inhibitor ABT-199 (supplemental Figure 6). These results would suggest that MEK inhibition can enhance Bcl-2 dependence in myeloma even in the absence of IL-6. PI3K inhibition had minimal effect on IL-6–mediated protection. We confirmed the effectiveness of these inhibitors by phospho-western blot (supplemental Figure 7).

We also examined the effects of IL-6 and signaling inhibitors on Bcl-2 family members. We used a modification of the standard SDS-PAGE system known as PhosTag, which allows identification of phosphorylated proteins without phospho-specific antibodies.³²  Interestingly, in MM.1s a large fraction of Bim is constitutively phosphorylated. IL-6 induces additional phosphorylation of Bim but has no effect on Mcl-1, Bcl-x_L, and Bcl-2 (supplemental Figure 8A). Bim is known to be phosphorylated by Erk1/2.^33-35  Phosphorylation of serine 69 and serine 77 were detected in both MM.1s and KMS18 upon IL-6 treatment and could be inhibited with U0126 as well as AZD1480 (Figure 5B-C; supplemental Figures 7 and 8B). As expected for a target of Erk kinase activity, phosphorylation of Bim correlated with phosphorylation of Erk. IL-6–induced upregulation of Mcl-1 in KMS18 was not blocked by U0126 but was blocked by ruxolitinib (Figure 5C-D). Treatment with IGF-1 potently induced phospho-Akt but had no effect on Bim serine 69 or Erk phosphorylation, consistent with its limited ability to prevent ABT-737–induced apoptosis, and further suggesting that the PI3K/Akt pathway has no impact on Bcl-2/Bcl-x_L dependence (supplemental Figure 9). Together these data suggest that IL-6 controls Bim binding in part through an MEK-dependent mechanism that is independent of changes in Mcl-1 expression.

The factors responsible for determining which antiapoptotic Bim binds to have yet to be identified. After observing both a shift in Bim binding to Mcl-1 and Bim phosphorylation with IL-6 stimulation, we hypothesized that the 2 processes may be related. We therefore performed Bim coimmunoprecipitation with both IL-6 and U0126. In 4 cell lines (KMS18, MM.1s, KMS12BM, and OCI-My5) U0126 increased the amount of Bcl-2 bound to Bim (Figure 6). Interestingly, this occurred with and without IL-6 stimulation, possibly because of low levels of constitutive Ras signaling, either a result of FGFR3 activation in KMS18, the activating Ras mutation in MM.1s, or signals from growth factors present in the culture media. Consistent with this possibility, we routinely observe small background amounts of serine 69 phosphorylated Bim by western in the absence of IL-6 stimulation. In addition to the change in the Bim binding pattern, we also detected an increase in the total amount of Bim coprecipitating with the antiapoptotics. This increase occurred despite similar amounts of Bim in the whole cell lysate. The increased Bim and its binding to Bcl-2 and Bcl-x_L may account for the enhanced cell death seen with ABT-737 in the presence of U0126. We subjected the same lysates of KMS18 and MM1.s to coimmunoprecipitation of Mcl-1, Bcl-x_L, and Bcl-2 and found similar results (supplemental Figure 10). As expected, IL-6 increased the amount of Bim associated with Mcl-1 whereas U0126 increased the amount of Bim coprecipitating with Bcl-2.

Although myeloma is typically thought of as Mcl-1 dependent, and potent Mcl-1 inhibitors such as S63845 are likely to be very effective in this disease,⁸  a distinct subset of myeloma may be primarily Bcl-2 dependent and potentially resistant to Mcl-1 inhibitors. A number of studies have described the Bcl-2 dependence of myeloma characterized by the t(11;14) translocation.^9-13  These results are supported by an ongoing phase 1 trial of venetoclax in multiple myeloma that has reported an objective response rate of 40% among patients with t(11;14).³⁶  However, given that 60% of t(11;14) patients failed to respond to venetoclax, the presence of t(11;14) is clearly not sufficient to dictate Bcl-2 dependence. In addition, cyclin D1 has not been reported to affect Bcl-2 family expression or Bim phosphorylation. These observations suggest that although t(11;14) is associated with venetoclax sensitivity, it may not play a direct role in Bcl-2 dependence. Given the uncertainty surrounding factors regulating Bcl-2 dependence in multiple myeloma, we investigated the effects of bone marrow stroma and IL-6 on this process and have demonstrated that IL-6 regulates Bcl-2 family dependence. More importantly, our data provide a scientific rationale for combining IL-6 signaling inhibitors with Bcl-2 inhibitors in multiple myeloma.

Stroma and stroma-derived cytokines have been previously demonstrated to influence Bcl-2 family dependence in other cell types. In chronic lymphocytic leukemia, costimulation with CD154 and IL-4 resulted in increased resistance to ABT-737 through increased expression of Bcl-x_L, BCL2A1, and NOXA, although Bim binding to the antiapoptotics was unchanged.³⁷  Neither CD154 nor IL-4 induce MEK activation, which may account for the absence of change in Bim binding. Alternatively, specific cytokine and cell type interactions may regulate Bcl-2 family dependence through different mechanisms.

Plasma cells are supported by BMSC secretion of crucial cytokines such as IL-6, IGF-1, and vascular endothelial growth factor, as well as adhesion-related signals. Of these cytokines, IL-6 is likely to play a dominant role.³⁸  Indeed, exogenous IL-6, but not IGF-1, was able to recapitulate the increased Mcl-1 dependence induced by stromal cells and conditioned media while an IL-6 neutralizing antibody restored Bcl-2/Bcl-x_L dependence in cell line experiments. However, in experiments with freshly isolated myeloma patient cells, exogenous IL-6 was not as effective as stromal cells, and IL-6 neutralizing antibody did not completely restore the ABT-737–induced apoptosis in the presence of stroma. These results may have been because of an inability to fully block IL-6. In experiments with MM.1s, we observed additive effects of combined IL-6 neutralization and IL-6 receptor blockade with 50% conditioned media and Hs-5 cells. Alternatively, IL-6 may not be the sole factor regulating Bcl-2 family dependence and a signal from direct cell–cell contact between plasma cells and stromal cells may be contributing as well. Integrin mediated signals, CD28/CD80/CD86, and BAFF/B cell maturation antigen also promote plasma cell survival.^14-16,39 

Although IL-6 regulates transcription of some Bcl-2 family members, expression alone does not account for sensitivity to Bcl-2 inhibitors. We therefore examined posttranslational modifications of Bim and its effects on binding to Mcl-1, Bcl-x_L, and Bcl-2. ERK-mediated phosphorylation of Bim has been reported to result in proteasomal degradation of Bim.^33,34,40,41  However, we did not observe any significant reduction in Bim protein upon IL-6 stimulation. In our experiments, IL-6 stimulation and Bim phosphorylation did correlate with a redistribution of Bim from Bcl-2 and Bcl-x_L to Mcl-1. MEK inhibition blocked Bim phosphorylation, increased association of Bim with Bcl-2 and Bcl-x_L, and sensitized cells to ABT-737 and ABT-199. Together, these results indicate that posttranslational modifications may selectively affect the binding affinity of Bim to antiapoptotics, and thus Bcl-2 family dependence. ABT-737 and MEK inhibitors have been reported to act synergistically in other cell types, although through different mechanisms. MEK inhibitors enhance cell death with ABT-737 by reducing Mcl-1 expression and Bcl-2 phosphorylation in acute myeloid leukemia,^42,43  while in B-cell acute lymphocytic leukemia, MEK inhibition results in increased binding of Bim to both Bcl-x_L and Mcl-1.⁴⁴  Treatment of KRAS or BRAF mutant cells derived from solid tumors with an MEK inhibitor also increased Bim levels and binding to Bcl-x_L as well as Mcl-1 and synergized with ABT-737 or navitoclax, suggesting this may be a common mechanism of evading apoptosis.^45,46  Similar activating Ras and Raf mutations are common in myeloma and could result in increased Bim phosphorylation and Mcl-1 dependence as we have demonstrated here.

The insight into IL-6–mediated regulation of the Bcl-2 family in multiple myeloma provided by our data has a number of clinical implications. In light of the early clinical results with venetoclax in myeloma, the relationship between Bcl-2 dependence and the bone marrow microenvironment is very heterogeneous. Patients who achieved a sustained complete response with single agent venetoclax must be highly Bcl-2 dependent despite signals from the microenvironment. However, these patients represent only a small fraction of the overall population, and the majority of patients, including many with t(11;14), are likely to be either partially Bcl-2 dependent or Mcl-1 dependent. In this group of patients, IL-6 signaling may play an important role in regulating Bcl-2 family dependence. We have demonstrated that signaling from bone marrow microenvironment–derived IL-6 and the Ras/MAPK pathway induce Mcl-1 dependence while simultaneously decreasing dependence on Bcl-2/Bcl-x_L in multiple myeloma through both transcriptional and posttranslational mechanisms (Figure 7), representing a potential source of resistance to Bcl-2 inhibitors. Although responses in myeloma to the IL-6 inhibitor siltuximab as a single agent have been limited,⁴⁷  IL-6 inhibition may have had an unappreciated effect on cells resulting in increased Bcl-2 dependence, both by decreasing Mcl-1 expression and shifting binding of Bim to Bcl-2. Similarly, some of the responses seen in early studies of the MEK inhibitor trametinib in myeloma patients both with and without activating Ras pathway mutations may be related to inhibition of IL-6–mediated Bim phosphorylation.⁴⁸  By enhancing Bcl-2 dependence in myeloma, inhibitors of IL-6 signaling such as anti-IL-6 or IL-6R antibodies, JAK inhibitors, or MEK inhibitors may act synergistically with venetoclax and therefore warrant additional study in combination trials.

The authors would like to thank all the patients who agreed to donate research specimens for this study.

This work was supported by grants from the National Institutes of Health, National Cancer Institute (R01 CA192844 and 5 P30CA138292). L.H.B. is supported by the TJ Martell Foundation and is a Georgia Research Alliance Cancer Scholar.

Contribution: V.A.G., S.M.M., and L.H.B. participated in conceptualization; V.A.G., S.M.M., and J.E.C.-P. designed and performed experiments; A.K.N., J.L.K., and S.L. provided patient samples; V.A.G. prepared the manuscript; S.M.M., J.E.C.-P., A.K.N., J.L.K., S.L., and L.H.B. assisted with review and editing of the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lawrence H. Boise, Department of Hematology and Medical Oncology, Winship Cancer Institute of Emory University, Emory University School of Medicine, 1365 Clifton Rd NE, Room C4012, Atlanta, GA 30322; e-mail: lboise@emory.edu.