NAD+ depletion enhances reovirus-induced oncolysis in multiple myeloma.

Cancer cell energy metabolism plays an important role in dictating the efficacy of oncolysis by oncolytic viruses. To understand the role of multiple myeloma metabolism in reovirus oncolysis, we performed semi-targeted mass spectrometry-based metabolomics on 12 multiple myeloma cell lines and revealed a negative correlation between NAD+ levels and susceptibility to oncolysis. Likewise, a negative correlation was observed between the activity of the rate-limiting NAD+ synthesis enzyme NAMPT and oncolysis. Indeed, depletion of NAD+ levels by pharmacological inhibition of NAMPT using FK866 sensitized several myeloma cell lines to reovirus-induced killing. The myelomas that were most sensitive to this combination therapy expressed a functional p53 and had a metabolic and transcriptomic profile favoring mitochondrial metabolism over glycolysis, with the highest synergistic effect in KMS12 cells. Mechanistically, U-13C-labeled glucose flux, extracellular flux analysis, multiplex proteomics, and cell death assays revealed that the reovirus + FK866 combination caused mitochondrial dysfunction and energy depletion, leading to enhanced autophagic cell death in KMS12 cells. Finally, the combination of reovirus and NAD+ depletion achieved greater antitumor effects in KMS12 tumors in vivo and patient-derived CD138+ multiple myeloma cells. These findings identify NAD+ depletion as a potential combinatorial strategy to enhance the efficacy of oncolytic virus-based therapies in multiple myeloma.

Introduction

Multiple myeloma (MM) is a plasma cell neoplasm that accounts for approximately 10% of all hematologic malignancies. MM has a 5-year survival rate of under 50% and is invariably fatal, with current therapies failing to overcome refractory disease or prevent recurrence. Thus, there is a desperate need for new and efficacious MM therapeutic options. Oncolytic viruses (OVs) have shown promising results in a wide range of cancers including MM and represent a novel anticancer therapeutic avenue., OVs act by preferentially targeting and killing cancer cells through both direct lysis and induction of potent antitumor immune responses while leaving normal cells unharmed. One OV under investigation as an anti-MM therapy is reovirus, a naturally occurring dsRNA virus that is benign in humans. Several studies have demonstrated that reovirus has the potential to kill MM cells through apoptosis or autophagy., However, in a Phase 1 clinical trial performed on MM patients (NCT01533194), the anticipated oncolysis was less than optimal. Thus, approaches enhancing the oncolytic capacity of reovirus, for instance by combination therapies,, are much needed.

Metabolic reprogramming can increase the susceptibility of cancer cells to OV-driven oncolysis. Viruses hijack host cell metabolism to support their own growth. Exploiting this strategy, metabolic reprogramming has been used to enhance viral replication and oncolysis., Recently, we found that an infection-induced block of pyruvate flux into mitochondria via inhibitory phosphorylation of pyruvate dehydrogenase (PDH), protected breast cancer cells from reovirus-induced oncolysis. Consequently, pharmacologic or genetic inhibition of PDH kinases enhanced reovirus-induced oncolysis in breast cancer cells. Whether MM harbors such therapeutically exploitable metabolic vulnerabilities in the context of OV therapies is unknown.

Here, we observed that reovirus-induced oncolysis was negatively correlated with both nicotinamide adenine dinucleotide (NAD+) levels and activity of nicotinamide phosphoribosyl transferase (NAMPT; the rate-limiting enzyme within the NAD+ salvage pathway). Pharmacological inhibition of NAMPT with FK866 enhanced reovirus-induced oncolysis in several MM lines that especially expressed functional p53. Mechanistically, the combination of reovirus + FK866 lowered glucose flux into the tricarboxylic acid (TCA) cycle, thus leading to enhanced autophagy and oncolysis. Most importantly, following reovirus + FK866 combination therapy, patient-derived CD138+ MM cells underwent enhanced oncolysis, and mice xenografted with MM tumors showed prolonged survival compared with monotherapies. These results identify reovirus coupled with NAD+ synthesis inhibition as a novel combinatorial therapeutic strategy against MM.

Results

Sensitivity of MM cells to OV oncolysis negatively correlates with NAD+ levels

Recently, we showed that metabolic reprogramming of cancer cells can enhance the oncolytic ability of OV. To expand our understanding of the role of metabolism in reovirus-driven oncolysis of MM, we performed semi-targeted mass spectrometry-based metabolomics on a panel of 12 MM cell lines prior to infection and at a pre-lethal time point of 24 h post-infection (h.p.i.). This elucidated both basal (relative to the average of all cell lines) and reovirus-induced MM metabolic signatures of all cell lines (Figures S1A and S1B). Next, we determined the sensitivity to reovirus of the 12 MM cell lines as measured by cytotoxicity 72 h.p.i (Figure 1A). To identify metabolites that might contribute to reovirus-induced oncolysis, we correlated oncolysis with either reovirus-induced metabolic changes or basal metabolite levels in all cell lines (Figures 1B and S1C). One of the most negatively correlated metabolites with oncolysis was NAD+ (both at basal levels and after reovirus infection; Figures 1C and 1D). Supporting a role for NAD+, basal levels of the NAD+ -associated metabolites, glutathione (both reduced and oxidized) and glutamate negatively correlated with reovirus-induced cytotoxicity (Figures S1D–S1F)., Furthermore, basal expression levels of several NAD+-consuming enzymes (e.g., SIRT6), dehydrogenases (e.g., NDUFS7), or enzymes essential for maintaining NAD+ homeostasis (e.g., GOT1) positively correlated with oncolysis, although other NAD+ -related enzymes, including NAD+ kinase (NADK), negatively correlated with oncolysis (Figures 1E–1G, S1G, and S1H). Importantly, all but one cell line were highly infected (>45%, exception MOLP8) at 24 h.p.i (Figure S1I), and the percentage of infected cells did not correlate with either oncolysis or NAD+ levels (Figures S1J–S1L), ruling out differential infection rates as an explanation for varying sensitivities. Together, these results suggest that reovirus oncolysis is determined by factors downstream of infection and at least in part are driven by NAD+ -related pathways in MM.

Figure 1

Basal and reovirus-induced NAD+ levels negatively correlate with reovirus oncolysis

(A) Percentage of cell death in MM cells caused by reovirus (MOI 10), 72 h.p.i. Data are means ± standard deviation from five independent experiments each performed in triplicates. (B) The correlation factor and –log10 probability of reovirus-induced changes in metabolites 24 h.p.i. versus percent oncolysis 72 h.p.i. in all MM cells. (C and D) Comparison of the respective susceptibility to reovirus oncolysis with the basal (C; log 2-fold compared with average of all cells) and reovirus-induced (D; log 2-fold compared with non-infected) NAD+ levels in MM cells. (E) The correlation factor and –log10 probability of gene expression (data from CCLE) and percent cell death by reovirus in MM cells. Blue circles represent proteins that significantly correlate with oncolysis; red dots represent proteins that are involved in NAD+ metabolism. (F and G) Examples of the genes that correlate with oncolysis.

Inhibition of NAMPT enhances reovirus oncolysis in MM cell lines that express functional p53 and favor mitochondrial metabolism

In most cancers including MM, NAD+ is mainly synthesized through the NAD+ salvage pathway. In line with a prior report we found that, of 1,019 cell lines across 36 different cancers in the cancer cell line encyclopedia (CCLE), MM cell lines had the second highest average mRNA expression of NAMPT, the rate-limiting enzyme of the NAD+ salvage pathway (Figure S2A). Complementing these data, MM cells have a much higher dependency on NAMPT for survival than other cancers according to the CERES CRISPR-Cas9 essentiality screen (Figure S2B). To investigate the role of the NAD+ salvage pathway in reovirus-induced cytotoxicity, we first measured NAMPT protein levels in the context of reovirus infection. In accord with expression data, basal or reovirus-induced protein levels of NAMPT did not correlate with oncolysis (Figures S2C–S2F). However, as protein levels may not correlate with enzyme activity, we measured the absolute log 2-fold change of NAD+ after treatment with a specific NAMPT inhibitor, FK866, to determine NAMPT activity. In keeping with a contribution of NAD+ to oncolysis in MM, NAMPT activity negatively correlated with reovirus-induced cytotoxicity (Figures 2A, S2G, and S2H).

Figure 2

Inhibition of NAMPT enhances reovirus oncolysis in MM cell lines

(A) Correlation between NAMPT activity and reovirus oncolysis. (B) Percent cell death 72 h after reovirus infection, FK866 treatment, and/or NMN addition. Cell lines are shown in the order of their sensitivity to reovirus. Data are means ± standard deviation from five independent experiments in triplicates. ∗p < 0.05 versus NT, #p < 0.05 versus Reo, τp < 0.05 versus FK866. (C) Heatmap summarizing cell death and mutation status of TP53 and KRAS. Cell lines are shown in the order of their sensitivity to combination therapy (combo). MUT, mutation; WT, wild type; LOE, loss of expression; ∗mutation (C)1010G > T maintains partial function; N/A, data unavailable; see Table S1 for more details. (D) The correlation factor and –log10 probability of gene expression (data from CCLE) and percent cell death by reovirus + FK866 in MM cells. Blue circles represent proteins that significantly correlate with combination-induced oncolysis; red dots represent genes of interest. (E–G) Examples of the genes that correlate with combination-induced oncolysis. (H and I) The correlation factor and –log10 probability of basal metabolic levels (H) compared with percent cell death by combination treatment in all MM cells, with thiamine used as a representative metabolite (I).

To test whether inhibition of NAMPT affects the sensitivity of MM cells to reovirus, we investigated oncolysis following the combination of FK866 reovirus. First, we measured FK866-induced cytotoxicity and found a range of sensitivities, consistent with published literature, that positively correlated with reovirus-induced oncolysis (Figures S2I and S2J). To confirm the specificity of FK866, nicotinamide mononucleotide (NMN, the product of NAMPT enzymatic reaction and precursor to NAD+) was used to rescue cells from the effects of NAMPT inhibition. NMN supplementation recapitulated NAD+ levels in FK866-treated cells to those of untreated counterparts in all cell lines (Figures S2G–S2I). Next, by combining FK866 + reovirus, we were able to determine the role of NAD+ in reovirus-induced oncolysis. The responses to FK866 and reovirus were cell dependent: KMS11 and RPMI8226 were relatively resistant to both treatments, whereas 5TGM1, MOLP8, JJN3, KMS27, and MM1R were more sensitive to FK866 and showed no synergistic effect when combined with reovirus. KMM1 and U226 were more sensitive to reovirus and showed no synergistic effect when combined with FK866 (Figures 2B, 2C, and S3A). Finally, reovirus-mediated oncolysis was significantly enhanced by FK866 treatment in three cell lines: IM9, KMS12, and MM1S (Figures 2B, 2C, and S3A). In all cases, FK866-induced cell death was rescued by addition of NMN.

Previous studies of other cancers have reported that optimal sensitivity to FK866 and reovirus monotherapies require a specific gene expression profile.25, 26, 27, 28 To better understand why certain cell lines were sensitive to combination therapy, we investigated global basal gene expression and mutational status of two genes known to play a role in monotherapy efficacy, the tumor suppressor TP53 and the proto-oncogene KRAS.25, 26, 27, 28 In line with the literature for monotherapies, cell lines that are most sensitive to combination therapy (KMS27, MOLP8, IM9, KMS12, and MM1S) expressed wild-type or partially functional TP53, and in most cases, a wild-type KRAS (IM9 and KMS12, exception MM1S).25, 26, 27, 28, 29, 30, 31 Conversely, cell lines that were most resistant contained loss-of-function or gain-of-(oncogenic)-function mutations in TP53 (KMS11, RPMI8226, KMM1, U226, and JJN3) and gain-of-function mutations in KRAS (RPMI8226 and KMM1) (Figure 2C and Table S1).25, 26, 27, 28, 29, In support of mutation analysis, basal expression levels of TP53 and p53 stabilizer MDM4 positively correlated with sensitivity to combination therapy, whereas levels of KRAS negatively correlated with the combination-induced cytotoxicity (Figures 2D, 2E, S3B, and S3C). Functional p53 and mutated KRAS are known to have counteractive effects on several pathways including metabolism; p53 favors oxidative phosphorylation, and mutated KRAS favors aerobic glycolysis., Indeed, p53-repressed glycolytic enzymes PFKP and PFKM negatively correlated with sensitivity to combination therapy, whereas the electron transport chain subunit succinate dehydrogenase cytochrome b560, positively correlated with sensitivity to combination therapy (Figures 2F, 2G, and S3D)., These data suggest FK866 + reovirus combination therapy requires MM to express functional p53 and wild-type KRAS and to have a metabolic profile that favors oxidative phosphorylation over glycolysis for optimal efficacy.

To explore the role of metabolism in combination therapy, we completed semi-targeted metabolomics analysis on all cell lines pre- and post-treatment with reovirus + FK866. Basal levels of thiamine (a vitamin essential for mitochondrial NAD+ -dependent dehydrogenases [pyruvate dehydrogenase and alpha-ketoglutarate dehydrogenase]) and the mitochondrial substrate glutamine positively correlated with combination-induced cytotoxicity (Figures 2H, 2I, and S3E), again suggesting that sensitivity to FK866 + reovirus therapy is linked to mitochondrial metabolism. Also supporting a role for energy metabolism, combination-induced production of ADP, an indicator of low-energy status, positively correlated with sensitivity to combination therapy (Figures S3F and S3G). Together, these correlation data support a hypothesis wherein FK866 + reovirus therapy is most effective in MM cells that inherently rely on oxidative phosphorylation over aerobic glycolysis in a p53/KRAS-dependent manner., To elucidate the underlying mechanism, we focused on KMS12 cells, as the FK866 + reovirus combination achieved the greatest synergism in this line (Figure S3A).

Reovirus + FK866 combination enhances oncolysis in KMS12 cells independently of reovirus replication

To aid our search for the mechanism facilitating the enhanced sensitivity of KMS12 cells to combination treatment, we comparatively evaluated global proteomes of KMS12 cells untreated or following 24 h of reovirus/NMN/FK866 treatments. We quantified the levels of 5,826 proteins in KMS12 cells (Figure 3A). Principal component analysis (PCA) revealed that reovirus + FK866 induced a protein signature different from monotherapies and controls (Figure 3B). However, the proteins that were altered by combination and reovirus treatments were similar, and the differences between the two groups were subtle (no protein levels were different by more than 2-fold, and enrichment analysis did not reveal any significant differences). In both cases, most proteins that changed were regulated by interferon and were involved in “defense against virus” pathways (Figures S4A–S4D). Examples of proteins increased by reovirus and reovirus + FK866 include OASL, LEG9, IFTI1, IFIT3, ISG15, and OAS2 (Figures S4E–S4J). Furthermore, we did not observe any differences in levels of virus proteins or viral replication at early (12 h.p.i.), mid (24 h.p.i.), or late (48 h.p.i.) replication stages (Figures 3C and 3D). Also, cell surface expression of JAM-A (reovirus receptor), which has previously been reported to correlate with reovirus-induced oncolysis in MM, was not altered by reovirus + FK866 treatment (Figure S4K). Our data thus suggest that increased sensitivity of KMS12 cells to reovirus + FK866 treatment was not caused by early changes in host protein levels or by differences in viral replication.

Figure 3

NAMPT inhibition does not enhance reovirus infection or replication in KMS12 cells

(A) Heatmap representing log 2-fold change of 5,826 proteins compared with non-treated (NT) controls by FK866 (20 nM, FK), NMN (100 μM), FK866 + NMN (F + N), reovirus (10 MOI, Reo), reovirus + FK866 (R + F), reovirus and NMN (R + N), or reovirus + FK866 + NMN (R + F + N) 24 h post-treatment. Values represent data from one experiment. (B) PCA of proteomics dataset from (A). (C) Levels of four reovirus proteins extracted from the proteomics data. (D) Percent KMS12 cells infected with reovirus. Data are means ± standard deviation from five independent experiments, each performed in triplicate.

Reovirus + FK866 combination decreases glucose uptake, glucose flux into TCA cycle, and oxygen consumption in MM cells

To investigate the role of metabolism in enhanced cytotoxicity by reovirus + FK866, we performed global metabolomics analysis in KMS12 cells 24 h post-treatments (Figure 4A). Combination treatment resulted in several unique changes to metabolite levels; for example, reovirus + FK866 increased ADP levels compared with either monotherapy, suggesting decreased energy production (Figure S5A). Also, two TCA cycle intermediates, citrate and malate, were decreased by reovirus + FK866 compared with either monotherapy (Figure 4A), suggesting decreased metabolic flux into mitochondria. As a first approach to determine the mechanism by which combination therapy altered metabolism, we measured glucose uptake 24 h post-reovirus/FK866/NMN treatment. Like the findings from other cancers, reovirus infection alone increased glucose uptake compared with non-infected cells,,40, 41, 42, 43, 44, 45 whereas reovirus + FK866 reduced glucose uptake compared with monotherapies and non-treated controls (Figure 4B). To explore the consequences of altered glucose uptake, we conducted untargeted [U-13C]glucose flux analysis beginning 24 h post-treatments. Within 6 h of [U-13C]glucose supplementation, we identified 13C labeling in 35 different metabolites representing a range of pathways including nucleotide, amino acid, and central energy metabolism (Figure S5B). Like other viruses, reovirus infection resulted in a shift of glucose flux toward the pentose phosphate pathway, as was evident by increased 13C labeling of adenine and uracil compared with non-treated controls (Figure S5B).46, 47, 48 One of the largest discrepancies between reovirus + FK866 and monotherapies was flux into the TCA cycle (Figure S5B). Although reovirus alone decreased the flux of [U-13C]glucose into the TCA cycle (as shown by lesser amount of [2-13C]citrate and [2-13C]malate in cells at 6-h post-[U-13C]glucose supplementation), reovirus + FK866 treatment caused a significantly larger drop in 2-13C citrate and malate labeling compared with monotherapies and non-treated control (Figures 4C and S5C). Both reovirus alone and reovirus + FK866 decreased production of [3-13C]lactate from [U-13C]glucose 6 h post-supplementation compared with non-treated controls (Figure S5D). These data suggested that reovirus infection shifted glycolytic flux toward the pentose phosphate pathway, whereas reovirus + FK866 decreased glucose uptake and minimized metabolic flux into the TCA cycle. Supporting our flux analysis, reovirus and combination treatments both decreased extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) compared with non-treated controls, whereas reovirus + FK866 decreased OCR compared with either monotherapy (Figure 4D). Consistent with a reduced glycolytic flux and reduced mitochondrial function, reovirus + FK866 caused a reduction in overall redox capacity compared with monotherapies (Figure 4E). Similar to findings from both monotherapies in other cancer models,, our data suggest that the reovirus + FK866 combination results in an early synergistic negative effect on central energy metabolism with a drop in mitochondrial function in KMS12 cells (Figure 4F).

Figure 4

Combination of FK866 + reovirus caused mitochondrial dysfunction in KMS12 cells

(A) Treatment-induced metabolite levels relative to non-treated (NT). (B) Median fluorescent intensity (MFI) of 2-NBDG uptake in 30 min by KMS12 24 h post-treatment. Data are means ± standard deviation from three independent experiments in triplicates. (C) 2-13C-labeled citrate levels 2 and 6 h post [U-13C]glucose supplementation (cells were infected with reovirus and treated with FK866/NMN 24 h before [U-13C]glucose supplementation). Values determined by average peak area of three independent replicates. Indicated statistical significance is for the 2-13C-labeled citrate levels. (D) Oxygen consumption rates (OCR) and extracellular acidification rates (ECAR) measured 24 h post-treatment. Data are means ± standard deviation from three independent wells. Minimal cell death was observed during this time frame. (E) Alamar blue levels of KMS12 cells 24 h post-treatment. Data are means ± standard deviation from three independent experiments in triplicate. (F) Metabolic flux model. Thicker lines represent increased flux. PPP, pentose phosphate pathway; pyr, pyruvate; AcCoA, acetyl-CoA; Lac, lactic acid; mito, mitochondria. (G) Percentage of dead cells 72 h post-treatment. Chloroquine (50 μM) was added at the same time as NMN/FK866/reovirus. Data are means ± standard deviation from three independent experiments in triplicates. ∗p < 0.05.

Reovirus + FK866 combination induces autophagy-dependent cell death in KMS12 cells

One potential consequence of decreased mitochondrial function is the upregulation of autophagy. Of note, increased autophagic flux has already been linked to both FK866 treatment and reovirus infection in MM cells., We found that several autophagy proteins, e.g., Beclin 1, were changed following the treatment with reovirus + FK866 in our proteomics dataset (Figures S6A and S6B). To precisely understand the role of autophagy during reovirus + FK866 combination, we measured autophagic flux by immunoblot analysis of LC3B and STQM1/p62 proteins 24 h after inhibition with chloroquine in KMS12 cells (Figure S6C). As shown (Figures S6D and S6E), the LC3BI-to-LC3BII conversion as well as the levels of p62 were significantly higher following reovirus + FK866 combination compared with monotherapies. These data suggested that reovirus + FK866 treatment promotes autophagic flux in MM cells. To test whether this increase in autophagy contributed to KMS12 cell death following reovirus + FK866 treatment, we measured cell viability in the presence of chloroquine. Supporting the role of autophagy in our model, chloroquine prevented the synergistic effect of combination therapy without affecting the sensitivity to monotherapies (Figure 4G). These data suggest that reovirus + FK866 therapy leads to autophagy-dependent cell death in KMS12 cells.

FK866 potentiates the antitumor efficacy of reovirus-based cancer therapy in vivo and reovirus-based oncolysis in patient-derived CD138+ MM cells

To determine the clinical applicability of our findings, we investigated whether the in vitro synergistic effect of reovirus + FK866 combination was translatable to in vivo contexts. For this purpose, KMS12 tumor-bearing NOD-SCID immunodeficient mice were injected intratumorally three times with PBS, FK866, and/or reovirus regimens, as shown in Figure 5A. Paralleling our in vitro results, tumor-bearing animals treated with the reovirus + FK866 combination showed better survival compared with the tumor-bearing mice treated with either FK866 or reovirus monotherapy or with PBS alone (Figure 5B). Also, tumor volume was generally smaller in KMS12 tumor-bearing mice injected with the combination therapy compared with mice injected with PBS, FK866, or reovirus alone (Figure S7A). These results demonstrated that FK866 enhances the oncolytic efficacy of reovirus in preclinical settings.

Figure 5

FK866 enhanced antitumor efficacy of reovirus-based cancer therapy in vivo and in patient-derived MM samples

(A) Schematic representing the timeline and experimental procedures used during animal experiments. (B) Survival analysis of KMS12 tumor-bearing NOD-SCID mice after three injections of reovirus (1 × 107 PFU), FK866 (10 mg/kg), or a combination of both (Combo). Total mice per treatment: PBS, six mice; FK866, seven mice; reovirus, seven mice; combination, eight mice. (C) Percent cell death as measured by percent annexin V- and/or 7AAD-positive cells 72 h post-treatment in CD138+ patient-derived bone marrow samples. (D) Cell proliferation 120 h post-treatment of CD138+ patient-derived bone marrow samples. Data are means ± standard deviation from three independent experiments, each performed in triplicate. ∗p < 0.05.

Finally, we tested the susceptibility of patient-derived clinical samples to reovirus and/or FK866 (Figures 5C, 5D, and S7B). For this, we propagated cancerous CD138+ and non-cancerous CD138–bone marrow cells from eight MM patients.53, 54, 55 The samples were acquired from patients that ranged from responders to non-responders to standard chemotherapies (Table S2). Only one sample, a non-responder, MM106, contained a loss-of-function mutation in TP53; however, other non-responders and partial responders (MM21, MM43, MM57) had a high mutational burden in a wide range of oncogenes, most of which are known to drive aerobic glycolysis (including a frameshift mutation in the p53 activator ATM and a missense mutation in EGR1, a transcription cofactor that normally facilitates p53-mediated repression of glycolytic genes in MM21; loss-of-function mutations in FAM46C, which is normally a repressor of the known aerobic glycolysis activator AKT, in MM43; V600E mutation in BRAF, which causes p53 inactivation, in MM57; gain-of-function mutation in KRAS in both MM43 and MM57), whereas very low mutations were detected in the responders (MM44, MM55, and MM67) (Table S3). Of the three responders, only MM55 had two mutations. One, c.5558A > T in ATM, has been reported as having no clinical significance,, while the other, c.1890C > A, is synonymous. Another sample, MM12, only partially responded to chemotherapies but had no detectable mutations. All together, the mutations in the high-mutational-burden samples align with a well-known link between resistance to chemotherapies and high glycolysis/low oxidative phosphorylation in myelomas.62, 63, 64, 65, 66, 67, 68 In accord with a cancer-specific effect of reovirus and FK866, monotherapies and combination therapy had no effect on CD138– cells (Figure S7B). Most importantly, the combination of reovirus + FK866 led to an increase in cytotoxicity and/or a decrease in cell proliferation in CD138+ cells from the low-mutational-burden patient samples (MM12, MM44, MM55, MM67) compared with either of the monotherapies (Figures 5C and 5D). The enhanced cytotoxic effect observed with the reovirus + FK866 combination was abolished upon the addition of NMN (Figures 5C and 5D). The reason that CD138+ cells from the non-responders/high-mutational-burden samples (MM21, MM43, MM57, and MM106) did not respond to combination therapy remains unknown. Nonetheless, it may be linked to an underlying mutational landscape driving aerobic glycolysis (Figures 2C–2I, S3B–S3G; Table S1). Altogether, these data suggest the possibility of using reovirus + FK866 combination against patient-derived MM cells.

Discussion

Here, we demonstrate how targeting NAD+ synthesis can be used to enhance the efficacy of reovirus-based therapies in MM. Our unbiased metabolomics analysis found that NAD+ levels negatively correlated with oncolytic effects of reovirus in MM (Figures 1C and 1D). We then hypothesized that lowering NAD+ would promote greater sensitivity to reovirus in MM cells. Indeed, we found that the reovirus + FK866 combination promotes significantly higher cytotoxicity in selected MM cell lines, promotes greater survival in MM tumor-bearing animals, and selectively targets patient-derived CD138+ MM cells. Mechanistically, we found that these enhanced cytotoxic effects occur through combination therapy-induced mitochondrial dysfunction promoting autophagy.

There are several documented links between NAD+ and host-pathogen interactions., For example, HIV-infected lymphocytes display decreased NAD+ levels, whereas plasmodium-infected erythrocytes and Leishmania infantum-infected macrophages show increased NAD+ levels compared with non-infected cells., Most of these infection-driven changes in NAD+ levels were mediated through altered NAMPT activity.,, Comparable to our findings, group A streptococcus-mediated lysis of human keratinocytes requires a NADase-dependent drop in NAD+ levels and increased autophagic flux (Figures 2B, 2C, 4G, and S6C–S6E)., The precise connection between NAD+ and pathogen defense in mammalian cells remains unclear. However, in plants, NAD+ protects the host against pathogens through regulation of mitochondrial-generated reactive oxygen species (ROS). Our study suggests a mechanism whereby high NAD+ levels maintain mitochondrial activity in MM cells with functional p53, thus preventing the energy depletion that is required for microbe-driven lysis of the host cell.

NAD+ has been identified as a therapeutic target in MM cells by several studies.,,79, 80, 81, 82 Accordingly, we showed that the activity of NAMPT was negatively correlated with the susceptibility of MM cells to reovirus-induced oncolysis (Figure 2A). These data informed our decision to inhibit NAMPT to enhance oncolysis in MM cells. Indeed, FK866 enhanced reovirus oncolysis compared with monotherapies in three of twelve MM cell lines (Figures 2B, 2C, and S3A). We showed that sensitivity to combination therapy positively correlates with expression levels and functionality of p53, which is a known promoter of mitochondrial metabolism and repressor of glycolysis and has previously been shown to be required for both reovirus- and FK866-mediated cytotoxicity (Figures 2C–2E and S3B; Table S1).25, 26, 27, 28 Also, levels of mitochondria-related metabolites and expression of mitochondrial genes all positively correlated with sensitivity to combination therapy, suggesting a role for mitochondria in combination-induced cytotoxicity (Figures 2F–2I and S3D–S3G). In line with our findings, both FK866 and reovirus monotherapies are known to have better responses in cancers with high mitochondrial function.,,,, All together, these data suggest that optimal response to reovirus + FK866 combination therapy in MM requires an inherent dependence on mitochondrial metabolism.

We considered various possible mechanisms by which FK866-mediated depletion of NAD+ levels could have enhanced reovirus oncolysis. Although we found no evidence that reovirus infection or replication was altered by combination therapy (Figures 3C and 3D), we did find several indicators of combination therapy-induced mitochondrial dysfunction (Figures 4A–4F), suggesting that the combination of FK866 + reovirus resulted in a synergistic negative effect on mitochondrial function in MM cells. This proposed two-pronged attack on mitochondrial function would potentially overcome the well-documented limitations of FK866 and reovirus as monotherapies., As mitochondrial dysfunction promotes autophagy, and it is known that both FK866 and reovirus alone can perturb the autophagic homeostasis in MM cells,,, we examined the effect of FK866 + reovirus combination on autophagy. We observed that MM cells treated with FK866 + reovirus displayed higher autophagic flux than that induced by either monotherapy (Figures S6C–S6E). Commonly known as a pro-survival process, autophagy can alternatively lead to cell death under certain circumstances,, this autophagy-induced cytotoxicity can be prevented by chloroquine.87, 88, 89 Corresponding to autophagic cytotoxicity during combination treatment, we observed that chloroquine prevented cytotoxicity induced by FK866 + reovirus (Figure 4G). Our findings suggest a model wherein the reduction of NAD+ levels through NAMPT inhibition enhances reovirus oncolysis through mitochondrial dysfunction and energy depletion, eventually leading to autophagic cell death.

Our work reveals that the therapeutic inhibition of NAD+ synthesis can be combined with OV to establish an efficacious cancer therapy. It should be noted that, in addition to oncolysis, multiple factors are required for optimal efficacy of OV therapy, including an OV-induced immune response. Although our study shows that NAD+ depletion enhances the oncolytic ability of reovirus, we did not investigate the effect of global NAD+ depletion on other components of the anti-MM response, such as immune cells. In fact, FK866 is known to negatively influence immune cells. Thus, potential therapies involving the combination of NAD+ depletion with OV therapy should further consider possible negative impacts of this combination on OV-induced anti-MM immunity. Nevertheless, considering the already reported therapeutic implications for both NAD+ and reovirus in MM,,,,, we strongly believe that a combinatorial strategy inhibiting NAD+ synthesis during reovirus therapy promises enhanced anti-MM effects and should be tested in clinical trials.

Materials and methods

Cell lines and reovirus

Reovirus (serotype 3, Dearing strain) was propagated as previously described.,

Human KMS11, IM9, MM1S, U226, JJN3 (from ATCC), KMM1 (from JCBR), KMS12PE (from Dr. Linda Pilarski [University of Alberta]), and MOLP8, RPMI8226, KMS27, MM1R (from Dr. Jonathan Keats [Translational Genomics Research Institute (TGen)]), and mouse 5TGM1 (from Dr. Babatunde Oyajobi [UT San Antonio]) cell lines were grown and maintained in RPMI 1640 medium supplemented with 10% heat-inactivated FBS, GlutaMAX, sodium pyruvate, non-essential amino acids, and antibiotic-antimycotic.

Animal study

The experimental procedures were governed by the approval of the Ethics Committee at Dalhousie University. Human KMS12 cells were used to generate tumors in 8-week-old female NOD SCID mice (NOD.CB17-Prkdcscid/NCrCrl; Charles River). Visible tumors were injected with a series of three injections (PBS, FK866 [10 mg/kg], and/or reovirus [1 × 107 PFU]), each 2 days apart.

Patient-derived primary multiple myeloma cells

Experimentation on de-identified clinical samples was approved (REB #1024942) by the Nova Scotia Health Authority (NSHA). CD138–and CD138+ cells were cultured in a method adopted from Stockbauer et al. DNA was extracted from purified CD138+ cells positively selected from bone marrow as described previously. DNA mutation analysis was performed and analyzed using a bioinformatic pipeline described previously (additional detail in supplemental methods).

Flow cytometry

Cell death and glucose uptake were determined with 7AAD/annexinV [eBioscience #00-6993-50/#88-8005-74] and 2-NBDG [Thermo #N13195], respectively, as previously described. For reovirus infectivity detection, cancer cells were permeabilized and incubated with a rat anti-reovirus antibody (generated in-house) as previously described. Data acquisition and analysis were completed with BD-FACSCantoII and FCS Express (De Novo), respectively.

Mass spectrometry-based metabolomics

Semi-targeted metabolomics using a data-dependent analysis approach was performed on MM cells. For untargeted metabolic flux analysis, isotopically labeled compounds were identified by comparing labeled and unlabeled samples, using a workflow provided in Compound Discoverer 3.0 (Thermo Scientific). For more information refer to supplemental methods.

Proteomics

Samples were prepared and analyzed (data available, Mendeley Data: https://doi.org/10.17632/7cty2zbz4c.1) with an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific), using an MS3 method as described previously.,95, 96, 97 Protein identification was performed as previously described.,

Metabolic assays

The alamar blue assay was performed according to the manufacturer’s instructions (Sigma)., Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) by 8 × 104 KMS12 cells were measured with an XF24 Analyzer according to the manufacturer's protocol for non-adherent cells (Agilent).

Bioinformatics

All mRNA levels were acquired through the CCLE database (data were available from 9 of 12 cell lines; no data for 5TGM1, IM9, or MM1R). TP53 and KRAS mutation status were acquired from the IARC database (http://p53.iarc.fr/).

Statistics

Statistical analysis was done using ANOVA or Kaplan-Meier survival analysis coupled with log rank test (both with 95% confidence interval); p values of <0.05 were considered to be statistically significant. Correlation factors were calculated from datasets that were sampled from Gaussian distribution using Pearson correlation coefficients measuring two-tailed probability. Panther analysis of biological pathways was described previously. For original data, please contact shashi.gujar@dal.ca.