Critical Role for Cap-Independent c-MYC Translation in Progression of Multiple Myeloma

Multiple myeloma cells can protect themselves from ER stress–induced death by restraining mTORC1 activity to limit unnecessary protein translation (1–5). Indeed, loss-of-function alleles of mTOR confer susceptibility to plasmacytoma development in mice (6). mTORC1 activates so-called “cap-dependent” translation by phosphorylating 4EBP-1, which liberates eIF-4E, allowing eIF-4E to interact with the cap structure at the 5′ end of transcripts.

Although inhibition of mTOR protects against ER stress, key oncoproteins must be translated in multiple myeloma cells to maintain viability. One of these is c-myc, a factor in multiple myeloma progression (7–9). During periods of mTOR inhibition, c-myc can be alternatively translated through a cap-independent mechanism, mediated by an internal ribosome entry site (IRES; refs. 10–12). IRESes are sequences in the 5′UTR of transcripts that directly recruit the 40S ribosomal subunit for translation initiation in a process facilitated by RNA-binding proteins termed IRES trans-activating factors (ITAF; refs. 11–14). The myc IRES is located in exon 1 and relies on several ITAFs for optimal function (14, 15). One of these is hnRNP A1 (16) which is a mandatory requirement for myc IRES activity. Since IRES-dependent translation does not require mTOR activity, there is no risk of enhanced global translation and ER stress.

Although prior work with cultured human lines (10), suggests a role for hnRNP A1 in multiple myeloma, in vivo supporting data have been lacking. Thus, in this study, we focused on a myc-driven murine model of multiple myeloma. The results provide support for myc IRES activity and hnRNP A1 as key determinants of multiple myeloma progression.

The IL6Myc-1 is a continuously cultured multiple myeloma line derived from an IL6Myc transgenic mouse (17), maintained in Iscove's modified Dulbecco's medium with 10% FBS. The pRF reporter was a gift from Dr. A. Willis. The myc IRES was cloned into pRF as described (16) to obtain pRmF. The p27 leader was subcloned upstream of the myc ORF as previously described (18) to generate the p27-myc ORF construct. Editing of the A1 gene used two separate guide RNAs for targeting the mouse A1 gene with the sequences: taccgtcatgtctaagtccg and tacctcggacttagacatga which were inserted into LentiCRISPRv2 (Addgene). Cells were transfected with the packaged lentiviruses. Single cell clones were screened for A1 gene deletion. The antibody used to identify A1 protein expression was purchased from Santa Cruz (antibody 4B10:sc-32301). Primary multiple myeloma cells were purified from bone marrow aspirates as previously described (10).

Polysome analysis was performed as described (18, 19). Briefly, after separation of extracts on a 15% to 50% sucrose gradient, fractions were collected and UV absorbance at 254 nm measured to differentiate monosome from polysome fractions. Associated myc RNA or actin RNA was quantified by qt-PCR.

As reported (20), the pSH CAT, containing the CAT reporter under the control of the A1 promoter was transfected into RPMI8226 cells with overexpression constructs of myc, MAF, or IRF-4, or shRNA targeting these transcription factors. CAT assay read-outs were assayed after 24 hours as described (20).

Dicistronic reporter constructs were transfected using Effectene Reagent (Qiagen) and normalized for transfection efficiency by co-transfection with pSVβGal. Transfection efficiencies were 5% to 15%. After 18 hours, cells were harvested, followed by detection of Renilla luciferase, Firefly luciferase, and β-Galactosidase activities (16). All luciferase activity is normalized to the luciferase values obtained for pRF in the absence of treatments, which is designated as a value of ‘1’.

All animal studies have been conducted in accordance with and with the approval of the Greater Los Angeles VA Healthcare Center and Medical College of Wisconsin. NOD/SCID mice were injected subcutaneously (SQ) with 1.5 × 10⁶ IL6Myc-1 cells. For IV challenge, B cell–deficient NOD/SCID mice were conditioned with irradiation (3 Gy) 4 hours before injecting 1.5 × 10⁶ luciferase-expressing cells IV. IV-challenged mice (6/group) were randomly divided into groups to receive either DMSO or inhibitor starting on day 7. After sacrifice, bones were imaged ex vivo using microCT (IVIS SpectrumCT). The scan was performed at 50 kV with 1 mA and reconstructed to a voxel size of 50 μm. Three-dimensional parameters BV/TV (bone volume over total volume) were analyzed with the help of the BoneJ2 (v.7.0.7) plugin of imageJ (v.1.53c).

Total RNA was purified from multiple myeloma cells by Qiagen RNeasy Kit and cDNA was generated with high capacity cDNA Reverse Transcription Kit (Applied Biosystems). MAX exon 5 alternative splicing was assayed using primers designed to constitutive exons flanking alternative exons (21). The primers were MAX (85F), 5′-tcagtcccatcactccaagg-3′; MAX (85R), 5′-gcacttgacctcgccttct-3′; Reverse primers were ³²P-end labeled and PCR reactions were amplified for 22 cycles and resolved by denaturing PAGE and imaged.

As described (22), cells were lysed and extracts cleared by centrifugation and immunoprecipitated with anti–eIF-4E or control IgG. Associated RNA was subjected to qRT-PCR assays, which were performed using TaqMan qPCR technology in the 7900HT system (Applied Biosystems; Myc TaqMan primer reference Mm00487804-m1).

Total RNA was extracted using RNeasy kit (Qiagen) following the manufacturer's instructions. The cDNA library preparation and RNA sequencing (RNA-seq) were performed by the UCLA Center for Genomics & Bioinformatics, as previously described (23). Briefly, the cDNA library was prepared using KAPA Stranded mRNA-Seq Kit (Roche). Reads were obtained by Illumina HiSeq 3000 sequencer for Single End 50 bp. Partek Flow Genomic Analysis software was used for bioinformatics analysis including reads alignment, gene feature annotation, and differential gene expression analysis. Reads were aligned to the mouse (mm10) reference genome using STAR-2.7.8a aligner. Ensembl Transcripts released 102 gtf was used for gene feature annotation. The differential gene expressions were examined using DESeq2 algorithm. QIAGEN Ingenuity Pathway Analysis software was used for pathway enrichment analysis of differentially expressed genes. The RNA-seq data was deposited in NCBI's Gene Expression Omnibus and are accessible through GSE185925 accession number.

Unless otherwise described, the Student t test was used to determine P values.

We first mined a public database (24–26) of RNA expression in patients with plasma cell dyscrasias. As previously described (7–9), c-myc expression increased as multiple myeloma disease activity increased (mean fold increase vs. normal plasma cells shown below Fig. 1A). A1 expression also increased significantly as disease activity increased (Fig. 1A; P < 0.0001 by ANOVA). In a post hoc analysis (Tukey–Kramer test, Supplementary Fig. S1A), statistically significant differences were identified between multiple myeloma diagnoses (smoldering, at diagnosis, or at relapse) versus normal or monoclonal gammopathy of undetermined significance (MGUS) diagnoses. In contrast, myc expression continues to increase during the course of tumor progression. This suggests that A1 dysregulation may be critical for early stages of myelomagenesis but not later stages of progression where continued myc dysregulation occurs. In a second database [Multiple Myeloma Research Foundation (MMRF) CoMMpass cohort], evaluating symptomatic patients, c-myc expression also correlated with A1 expression (Fig. 1B). In addition, A1 expression correlated with decreased survival (Supplementary Fig. S1B). Figure 1C (upper) demonstrates that A1 protein expression is also present in primary multiple myeloma cells which was comparable with the RPMI8226 multiple myeloma cell line. A1 immunoprecipitated from these multiple myeloma cells is bound to the myc IRES (Fig. 1C, lower), suggesting that it can function as a myc ITAF in primary cells.

To identify mechanisms of A1 RNA expression, we targeted the multiple myeloma–associated transcription factors myc, IRF4, and MAF in A1 reporter assays (Fig. 1D). At 24 hours after shRNA was added to silence myc, IRF4, or MAF, there was minimal cytotoxicity [7 ± 3% for myc, 5 ± 2% for IRF4, and 6 ± 1% for MAF knock downs (n = 3)]. In the RPMI8226 multiple myeloma cell line, knockdown of myc or IRF4 resulted in decreased reporter expression (left). There was also a significant increase when myc was ectopically overexpressed (Fig. 1D, right). Although IRF4 overexpression also resulted in enhanced A1 reporter expression, this did not reach statistical significance. In contrast, there was no effect of MAF knockdown or overexpression. Western blot experiments (Fig. 1E) supported the importance of myc as an A1 inducer. We successfully knocked down myc and IRF-4 in the RPMI8226 cell line (left). As stated above, both knockdowns resulted in minimal cytotoxicity at 24 or 48 hours (<10%) although, by 72 hours, there was excessive death (>40%) attesting to the previously reported (8, 27) myc and IRF4 addiction in multiple myeloma. C-myc knockdown successfully inhibited A1 expression. Myc knockdown in MM1.S cells (Fig. 1E, right) also inhibited A1 expression further supporting the notion that c-myc drives A1 expression. The data with IRF4 is not as clear. In an autoregulatory circuit, IRF4 induces myc and myc induces IRF4 (27) and our immunoblot assay (Fig. 1E) is consistent with this autoregulatory loop. However, IRF4 knockdown, although clearly inhibiting A1 reporter expression (Fig. 1D), in contrast, increased A1 protein expression (Fig. 1E). It is possible that an IRF4 knockdown results in reduced A1 promoter activity but the loss of IRF4 may also derepress factors required for posttranscriptional processes resulting in overall accumulation of A1 at the protein level. The explanation for this dichotomy will require additional future experimentation. Nevertheless, although the role of IRF4 in A1 expression is still unclear, the data clearly indicate that c-myc significantly participates in driving A1 expression in multiple myeloma cells.

To investigate a role for A1 in vivo, we used a myc-driven model (17, 28) where plasmacytomas arise in mice doubly transgenic for IL-6 and myc within the B cell lineage at 4 to 5 months of age. RNA expression profiling in splenocytes, harvested prior to development of tumors, demonstrated a >4 x fold increase in A1 expression relative to litter mates singly transgenic for only the IL6 gene (Fig. 2A). An advantage of this model was the ability to harvest an individual plasma cell tumor, whose tumor cells remained viable during continuous culture, allowing genetic manipulation. This transgenic line, designated IL6Myc-1, expresses high levels of CD138, produces an IgA-kappa monoclonal protein, and is clonogenic in vitro and in vivo (17). Following subcutaneous challenge, it grows as a solid tumor. Following IV challenge, mice develop hind limb paralysis and osteolytic bone lesions (17).

We performed CRISPR-Cas9 deletion of the A1 gene in IL6Myc-1 cells by targeting two independent sequences in exon 1 (termed sequence 11 or 12). A non-targeting gRNA was used as a control. Single cell–derived clones were then generated. All clones from targeted sequences demonstrated deletion of the A1 gene and loss of A1 expression (Supplementary Fig. S2A). In addition, all clones had diminished expression of c-myc (Supplementary Fig. S2A). We selected 1 clone from each of the two targeted sequences to expand and study, designated #125 and #111. Both exhibited loss of A1 expression and decreased c-myc expression when compared with control cells from the non-targeted clone (Fig. 2B, left and right). The inhibitory effect on c-myc expression was specific as A1 deletion had no effect on YB1 expression. YB-1 was used as a specificity control as it is also a known myc-IRES ITAF (29).

We first tested if decreases in steady state myc RNA or myc protein half-life could explain diminished myc protein levels in A1-deleted cells. However, instead of a decrease, there were significant increases (Fig. 2C and D). These alterations may be compensatory mechanisms. To assess translational efficiency, we performed a polysome analysis. This assay is based upon the observation that well translated transcripts are associated with polysomes and poorly translated mRNAs are monosomal. Thus, polysomes were separated from monosomes and associated RNAs were quantified by qt-PCR. Figure 2E demonstrates a striking decrease in myc RNA associated with polysomes in the 111 and 125 A1-deleted cells, indicating inhibited myc translation. In contrast, translation of actin was not suppressed.

To assess effects on cap-dependent versus IRES-dependent translation, we utilized the dicistronic reporter assay where cells were transfected with reporter constructs shown in Fig. 2F. The myc 5′UTR, containing its IRES, was subcloned into the intra-cistronic space between the Renilla and Firefly luciferase ORFs in the pRF vector to yield the pRmF vector. The pRmF reporter's Firefly luciferase translation is driven by the myc 5′UTR and is a reflection of IRES-dependent translation whereas Renilla expression is due to cap-dependent translation. Results in Fig. 2G confirm the presence of the 5′UTR in pRmF reporter increases Firefly expression without effect on Renilla in control IL6/Myc-1 cells. It is also apparent that relative Firefly expression is decreased in A1-deleted 111 and 125 cell lines versus control cells. Thus, loss of A1 inhibits IRES-dependent translation of myc.

Although containing diminished c-myc levels, A1-deleted 111 and 125 cells were capable of in vitro expansion, doubling in viable cell numbers by 72 hrs (Fig. 3A). While in vitro growth was comparable with control cultures at 24 and 48 hours, there was a modest decrease at 72 hours, which was maintained at 96 hours but not more impressive (30%–35% decrease in viable tumor cell numbers in 111 and 125 cultures). Mice were next challenged SQ with control IL6Myc-1 cells [designated wild-type (WT) in Fig. 3B] or A1-deleted tumor cells. Individual mice received WT control inocula in one flank (9 inocula) or 111 cell challenges (n = 9) in the opposite flank (Group 1) or WT control inocula in one flank (n = 9) and 125 cell challenges in the opposite flank (n = 9) (Group 2). As shown in Fig. 3B and C, outgrowth of 111 and 125 tumors were inhibited compared with WT control tumors. This was especially true for 125 tumors, which were extremely small (Fig. 3D). Immunoblot of A1-deleted tumors demonstrated corresponding decreases in c-myc protein (Fig. 3E) and absence of hnRNP A1 (latter examples shown in Supplementary Fig. S2B).

We previously identified an inhibitor, C11, which prohibited myc IRES activity (10). C11 was biochemically modified to yield a more efficacious inhibitor, J007 (Fig. 4A; detailed synthesis in 22). IL6/Myc-1 cells exposed to J007 demonstrated a concentration-dependent decrease in myc expression (Fig. 4B). J007 was also effective in the RPMI8226 line and primary multiple myeloma specimens (Fig. 4B). Inhibited myc expression in IL6/Myc-1 cells was not associated with decreases in myc RNA abundance (Fig. 4C) nor protein stability (Supplementary Fig. S3). Polysome profiling (Fig. 4D) demonstrates an inhibitory effect of J007 (50 nmol/L) on translational efficiency of myc. In the dicistronic reporter assay, J007 significantly decreased Firefly expression while having no effect on Renilla expression (Fig. 4E). These results indicated that J007 could inhibit IRES-dependent translation in multiple myeloma cells. However, the ability of the inhibitor, used alone, to curtail overall protein expression was surprising since cap-dependent translation of myc should proceed unfettered. Thus, we performed additional assays to rule out IRES-independent adverse effects on c-myc. As A1 is also a splicing factor (21), we tested if a J007-A1 interaction regulates Max splicing. Max is spliced via A1 to generate Delta Max, a truncated version of the Max protein, which includes exon 5 (21). As shown in Fig. 4F, J007 exposure did not alter Max splicing in WT cells. As a positive control, the A1 deletion in 125 cells markedly reduced exon 5-containing Delta Max transcript levels. Furthermore, we tested effects of J007 on cap-dependent translation of myc by immunoprecipitating eIF-4E and assessing the relative amounts of associated c-myc mRNA. As shown in Fig. 4G, J007 did not inhibit c-myc RNA association with eIF-4E ruling out an effect on the mTOR/4EBP-1/eIF-4E cascade as an explanation for J007's inhibition of myc expression.

To test J007 in vivo, mice were challenged with SQ inocula of control IL6Myc-1 tumor cells in one flank and 111 tumor cells in the opposite flank. Half were treated with J007 and half with vehicle. The 111 multiple myeloma cells are moderately inhibited in tumor growth (Fig. 3B and C) but should not be further inhibited by J007 as the drug's target (A1) is absent. As shown in Fig. 5A, J007 prevented tumor outgrowth from control IL6Myc-1 inocula but had no effect on growth of 111 tumors, documenting the A1 specificity of the antitumor response. J007-treated mice demonstrated no toxicity or weight loss (Supplementary Fig. S4A). Immunoblot analyses confirmed inhibited myc expression in WT control tumors but no effect in 111 tumors (Fig. 5B). In a separate experiment, treatment of tumor-challenged mice with J007 resulted in prolonged survival (Fig. 5C).

We also tested if J007 was efficacious in mice challenged IV with luciferase-expressing IL6Myc-1 where tumor grows within the skeleton. Vehicle-treated control mice demonstrated progressive tumor growth, which was suppressed by both 20 and 40 mg/kg of J007 (Fig. 5D and E). In addition, while having no effect on weights (Supplementary Fig. S4B), both doses prolonged survival of IV-challenged mice (Fig. 5F) and markedly diminished serum M-protein (Fig. 5G). MicroCT demonstrated an inhibition of bone loss, which accompanied tumor growth in vehicle-control mice (Supplementary Fig. S5).

In a previous study (22), IRES inhibition had additive antitumor effects when administered with the mTOR inhibitor pp242. This was not surprising as both IRES-dependent myc translation would be prohibited as well as mTOR/cap-dependent translation (by pp242). We, thus, tested if J007 could interact with pp242 for heightened responses. Although mTORC1 was inhibited by pp242 in vitro (decreased phosphorylation of p70/S6/4EBP-1), there was no inhibition of c-myc expression (Supplementary Fig. S6A). Since mTOR inhibitors upregulate IRES activity (30, 31), the lack of pp242's effect on myc expression further supports the primacy of cap-independent myc translation in the multiple myeloma model. However, the addition of pp242 to a low concentration of J007 resulted in greater inhibition of c-myc expression as compared with J007 used alone (Supplementary Fig. S6B). To assess an interaction in vivo, tumor-challenged mice were treated with DMSO, pp242 (50 mg/kg/day), J007 (20 mg/kg/day), or the combination of drugs. As shown in Supplementary Fig. S6C, both pp242 or J007, used alone, induced a significant albeit modest slowing of tumor growth. The combination of drugs resulted in enhanced antitumor effect.

To confirm that diminished tumor growth in A1-deleted multiple myeloma cells was due to c-myc inhibition, we attempted to ectopically reexpress myc. However, as previous data suggested a dependence on A1-dependent IRES activity for myc translation, we were concerned that transfected myc would not be expressed in A1-deleted cells. Prior studies have identified sequences in the p27 mRNA leader that can direct cap-independent translation (32, 33) and the p27 IRES does not require A1 as an ITAF (18). Accordingly, we generated a construct in which we replaced the native 5′UTR of c-myc with the p27 leader containing its IRES. The transcription of c-myc-containing genes was driven by a CMV promoter. The A1-deleted 125 multiple myeloma line was transfected with an empty vector, the myc ORF construct (no 5′UTR) or the p27IRES-myc ORF (p27 IRES upstream of the myc ORF). After selection, c-myc expression was compared with control WT cells. As shown in Fig. 6A, the inhibited myc protein expression in EV-transfected 125 cells is again identified (versus control WT cells). Downregulated myc expression in 125 cells is not affected by transfection with CMV-myc ORF but the construct with upstream p27 IRES sequences allowed reexpression of myc protein. Fig. 6B demonstrates the myc protein reexpression data from three independent experiments and shows that myc RNAs were faithfully expressed. These cell lines were then used to challenge mice. As shown in Fig. 6C and D, tumor growth was rescued in A1-deleted 125 cells only when transfected with the p27IRES-myc ORF construct. Immunoblot assay (Fig. 6E) confirms reexpression of myc correlates with rescued tumor outgrowth. These data confirm that the decreased myc expression in A1-deleted cells is a key determinant of inhibited tumor growth and, furthermore, underscore the importance of the 5′UTR in A1's regulation of myc expression.

Although our data indicated that downregulation of myc translation mediates J007's anti–multiple myeloma effect, it was possible that other alterations could play a role. We, thus, performed an unbiased evaluation of altered gene expression in J007-treated multiple myeloma cells. Genome wide RNA-seq data was obtained from WT cells treated ± J007 (1 μmol/L for 72 hours). The viability of both cell preparations was >97% at the time of RNA harvesting. We subsequently identified 257 genes that were upregulated greater than 2× fold and 230 genes that were downregulated at least 50% by J007. Gene set enrichment from the RNA-seq assay is shown in Supplementary Fig. S7A. As anticipated, the most significantly altered gene set was that of ‘myc targets’ (enrichment plot in Supplementary Fig. S7B. Of the myc target activation gene set, several significantly downregulated genes were of particular interest for the myeloma model, such as heat shock protein 1, exportin 1, PCNA, and several proteasomal subunits. However, it is interesting that the ‘angiogenesis’, ‘mTORC1 signaling’, ‘protein secretion’, ‘G2M checkpoint’, ‘unfolded protein response’, and ‘fatty acid metabolism’ gene sets were also decreased by J007 treatment. Some of these additional alterations may contribute to the anti–multiple myeloma effect of J007.

These results support roles for A1 and myc IRES in multiple myeloma. They also indict myc as an inducer of A1 expression in multiple myeloma. A1 RNA expression was considerably higher in IL6/Myc transgenic B cells versus single IL6 transgenic mice and overexpression/knockdown experiments in human multiple myeloma lines also support a role for myc. These results point to a positive feedback circuit where myc stimulates A1 RNA expression and A1 enhances myc translation.

It was remarkable that IRES targeting, by itself, was so effective because mTOR/cap-dependent translation of myc should have proceeded without restriction. In fact, short term exposure to the mTOR inhibitor, pp242, had no effect on myc expression (Supplementary Fig. S6). These results suggest that overall myc translation and expression in multiple myeloma is more dependent on IRES activity than on mTOR activity. It is certainly possible that A1 may also enhance cap-dependent translation in addition to its role as a myc ITAF (34). Nevertheless, the rescue of myc expression by transfection which required a functional IRES upstream of the myc ORF, underscores the importance of A1-dependent IRES activity. In addition, although the mTORC1 pathway was a gene set significantly impacted by J007 in RNA-seq analysis (Supplementary Fig. S7), J007 had no effect on binding of myc RNA to eIF-4E, indicating mTORC1/cap-dependent translation of myc was not constrained by J007.

A particular ER stress in myc-driven multiple myeloma may explain heightened sensitivity to IRES disruption as depicted in the model in Supplementary Fig. S8. ER stress inhibits mTORC1 (10) and promotes a shift to IRES-dependent translation via the unfolded protein response. This renders multiple myeloma cells completely dependent on IRES activity for myc translation. In similar fashion, A1 mediates activation of IRES-dependent SREBP-1a translation in response to ER stress (35). Previous work with additional myc ITAFs (29, 36, 37) also supports the critical role of the myc IRES in multiple myeloma cells.

An unbiased transcriptome analysis suggested additional J007-induced alterations that could participate in an anti–multiple myeloma response. Some of these (for example, angiogenesis) may simply be sequelae of inhibited myc target activation. However, direct effects on G2M cell cycle transit, the unfolded protein response and fatty acid metabolism may be contributory and deserve to be tested.

Targeting of the myc oncogene has been a long-sought goal for many years (38, 39). The molecular structure of the myc gene and its RNA in multiple myeloma may render it sensitive to IRES targeting. Although lymphoma cells with 8;14 translocations contain myc breakpoints which theoretically disrupt or decapitate the IRES, only 15% of multiple myeloma specimens contain a myc translocation (40) and even multiple myeloma cell lines with myc translocations contain an intact exon 1 (41). Thus, a strategy of targeting myc with a drug like J007 in multiple myeloma is promising.

B. Dhakal reports personal fees from Bristol-Myers Squibb, Karyopharm, Janssen, Arcellx, and Natera; and personal fees from Sanofi outside the submitted work. J.F. Gera reports grants from NIH/NCI; and grants from US Department of Veterans Affairs during the conduct of the study; in addition, J.F. Gera has a patent for “Inhibitors of IRES-mediated protein synthesis,” 16/098784 issued. S. Janz reports grants from NCI/NIH during the conduct of the study. A. Lichtenstein reports grants from NIH/NCI during the conduct of the study; in addition, A. Lichtenstein has a patent for “Inhibitors of IRES-mediated protein synthesis”, 16/098784 issued. No disclosures were reported by the other authors.

Y. Shi: Conceptualization, formal analysis, supervision, funding acquisition, investigation, methodology, writing–original draft. F. Sun: Investigation, methodology. Y. Cheng: Investigation, methodology. B. Holmes: Investigation, methodology. B. Dhakal: Methodology. J.F. Gera: Methodology. S. Janz: Conceptualization, funding acquisition, investigation, methodology. A. Lichtenstein: Conceptualization, formal analysis, supervision, funding acquisition, investigation, writing–original draft.

This work was supported by NIH grants RO1CA111448, RO1CA214246, RO1CA217820, and RO1CA151354, the VA, the MMRF, and The William G Schuett Jr Multiple Myeloma Research Endowment.

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.