Targeting metabolic activity in high-risk neuroblastoma through Monocarboxylate Transporter 1 (MCT1) inhibition

Abstract

Amplification of the MYCN oncogene occurs in approximately 25% of primary neuroblastomas and is the single most powerful biological marker of poor prognosis in this disease. MYCN transcriptionally regulates a range of biological processes important for cancer, including cell metabolism. The MYCN-regulated metabolic gene SLC16A1, encoding the lactate transporter monocarboxylate transporter 1 (MCT1), is a potential therapeutic target. Treatment of neuroblastoma cells with the MCT1 inhibitor SR13800 increased intracellular lactate levels, disrupted the nicotinamide adenine dinucleotide (NADH/NAD⁺) ratio and decreased intracellular glutathione levels. Metabolite tracing with 13C-glucose and 13C-glutamine following MCT1 inhibitor treatment revealed increased quantities of tricarboxylic acid (TCA) cycle intermediates and increased oxygen consumption rate. MCT1 inhibition was highly synergistic with vincristine under cell culture conditions, but this combination was ineffective against neuroblastoma xenografts in mice. Post-treatment xenograft tumors had increased expression of the MCT1 homolog MCT4/SLC16A, a known resistance factor to MCT1 inhibition. We found that MCT4 was negatively regulated by MYCN in luciferase reporter assays and its expression in neuroblastoma cells was increased under hypoxic conditions and following hypoxia-inducible factor (HIF1) induction, suggesting that MCT4 may contribute to resistance to MCT1 inhibitor treatment in hypoxic neuroblastoma tumors. Co-treatment of neuroblastoma cells with inhibitors of MCT1 and LDHA, the enzyme responsible for lactate production, resulted in a large increase in intracellular pyruvate and was highly synergistic in decreasing neuroblastoma cell viability. These results highlight the potential of targeting MCT1 in neuroblastoma in conjunction with strategies that involve disruption of pyruvate homeostasis and indicate possible resistance mechanisms.

Introduction

Neuroblastoma is a pediatric solid tumor that originates from persistent neural crest progenitors and is responsible for 15% of cancer-related deaths in children worldwide (1). The survival rate for patients classified as high-risk unfortunately remains at 50% (2, 3), creating a significant need for better therapeutic approaches to improve outcomes for these patients. The master transcription factor MYCN, a member of the MYC family of oncogenes, is an established driver of neuroblastoma progression, but is a challenging therapeutic target (4). Targeting downstream alterations arising from MYCN dysregulation has therefore been a strategic approach in developing novel treatments for high-risk neuroblastoma.

Dysregulated MYC family oncoproteins have been demonstrated to remodel cancer metabolic pathways via direct transcriptional upregulation of metabolic genes, including those involved in glycolysis and glutaminolysis (5–7). Accelerated glycolysis in cancer cells results in the mass conversion of glucose to pyruvate, followed by conversion of pyruvate to lactate by lactate dehydrogenase A (LDHA), in preference to conversion of pyruvate to acetyl-CoA for oxidative phosphorylation via the tricarboxylic acid (TCA) cycle in mitochondria (8). The preferential conversion of pyruvate to lactate is essential in part due to the subsequent production of oxidized nicotinamide adenine dinucleotide (NAD⁺) which is recycled for continuous glycolysis. Accumulated lactate is exported out of cells by members of the monocarboxylate transporter (MCT) family (9), particularly MCT1 and MCT4, and may contribute to tumorigenesis by serving as an alternative fuel source for aerobic cancer cells, promoting angiogenesis and tumor invasion, and inducing further metabolic reprogramming (10–13). The catabolism of glutamine to glutamate by glutaminases, or glutaminolysis, replenishes TCA cycle intermediates and supports the synthesis of lipids and amino acids (14). In neuroblastoma, MYCN has been demonstrated to promote glutaminolysis by directly activating glutaminase 2 transcription (GLS2) (15).

The therapeutic potential of blocking lactate transport by inhibition of the monocarboxylate transporter 1 (MCT1, encoded by the SLC16A1 gene) has been investigated in preclinical studies of adult cancers and shown to delay tumor growth in lung and breast cancer and Burkitt’s lymphoma xenografts (16–19), while one MCT1 inhibitor has entered early phase clinical trial (ClinicalTrials.gov Identifier NCT01791595). Since MCT1 is a transcriptional target of MYCN (20) with a vital role in lactate homeostasis in cancer cells, this study investigated the biological and metabolic effects of MCT1 inhibitor treatment in preclinical models of high-risk neuroblastoma. The results provide evidence for MCT1 as a potential therapeutic target for this childhood disease.

Results

High MCT1/SLC16A1 expression is associated with poor clinical outcome in primary neuroblastoma and is a feature of MYCN-amplified tumor cells

To determine whether lactate transporter expression is associated with clinical outcome in neuroblastoma, we analysed mRNA expression of SLC16A1 (encoding for MCT1) and its homolog SLC16A3 (encoding for MCT4) in an expression array dataset from a prospectively accrued primary neuroblastoma cohort of 649 patients (21). SLC16A1 expression, but not SLC16A3 expression, was strongly associated with both poor event-free survival (EFS) and overall survival (OS) when mRNA expression was dichotomized at the median (Figure 1A–D). Multivariate Cox regression analyses with SLC16A1 expression and established prognostic indicators for neuroblastoma (age at diagnosis, INSS stage and MYCN amplification status) as variables, revealed that SLC16A1 is an independent prognostic marker for EFS and OS in neuroblastoma (Table 1). SLC16A1 gene expression levels were significantly higher in MYCN-amplified tumors than in MYCN non-amplified tumors (Figure 1E), while SLC16A3 levels were lower in MYCN-amplified tumors (Figure 1F). High expression of lactate dehydrogenase A (LDHA), which generates the MCT1/4 substrate lactate from pyruvate, was also strongly associated with both worse EFS and OS and was higher in MYCN-amplified tumors (Supplementary Figure S1). Multivariate Cox regression analyses revealed that LDHA is also strongly prognostic of EFS and OS, independent of established prognostic indicators for neuroblastoma (Supplementary Table S1). Consistent with tumor mRNA expression data, MYCN-amplified neuroblastoma cell lines typically expressed MCT1 protein at a substantially higher level than MYCN non-amplified cell lines, while MCT4 was expressed at very low levels in MYCN-amplified lines (Supplementary Figure S2). Similarly, LDHA protein levels were substantially higher in MYCN-amplified cell lines, with the exception of SK-N-BE(2) and its sub-clone BE(2)-C, where LDHA protein was absent. These associations suggest the potential importance of lactate homeostasis genes in the biology of MYCN-amplified, high-risk neuroblastoma.

Figure 1. SLC16A1 (MCT1) expression is prognostic for poor outcome in primary neuroblastoma.

(A–D) Kaplan-Meier curves for event-free survival (EFS) and overall survival (OS) based on SLC16A1 or SLC16A3 mRNA expression in an expression array dataset from a prospectively accrued primary neuroblastoma cohort of 649 patients (21). Gene expression was dichotomized at the median. SLC16A1 EFS: HR = 3.33, 95% CI = 2.48–4.47, P < 0.001. SLC16A1 OS, HR = 6.23, 95% CI = 4.04–9.60, P < 0.001. SLC16A3 EFS: HR = 0.98, 95% CI = 0.75 SLC16A3–1.28, P < 0.001. SLC16A3 OS: HR = 1.04, 95% CI = 0.744–1.44, P < 0.001. (E–F) SLC16A1 mRNA expression was higher in MYCN-amplified compared to MYCN non-amplified tumors, while SLC16A3 expression was lower. ****P<0.0001.

Table 1.

Multivariate Cox regression analysis of SLC16A1 expression and outcome in neuroblastoma.

Variable	Event-free survrival (n=628)		Overall survival (n=649)
	HR (95% CI)b	P value	HR (95% CI)	P value
MYCN statusa	1.39 (0.99–1.94)	0.055	2.43 (1.67–3.54)	<0.001***
INSS tumor stage	1.62 (1.18–2.24)	0.003**	2.30 (1.48–3.59)	<0.001***
Age at diagnosis	2.22 (1.62–3.03)	<0.001***	4.96 (3.16–7.79)	<0.001***
SLC16A1 expression	2.14 (1.52–3.00)	<0.001***	2.35 (1.45–3.81)	0.001**

^aVariables adjusted for in the multivariable analysis included MYCN status (amplified vs non-amplified), INSS tumor stage (1, 2, and 4S vs 3, 4), age at diagnosis (<18 months vs ≥ 18 months) and SLC16A1 expression (dichotomised at the median, low vs high).

^bHR, hazard ratio; CI, confidence interval.

^**P<0.01,

^***P<0.001

The MCT1 inhibitor SR13800 disrupts lactate homeostasis and NAD⁺/NADH ratio in neuroblastoma cells and decreases cell growth

Next, we directly assessed the importance of MCT1 in the MYCN-amplified cell lines Kelly and LA-N-1, both of which express high levels of MCT1 protein and low levels of MCT4 protein (Supplementary Figure S2). Treatment with the MCT1 inhibitor SR13800 resulted in a large increase in intracellular lactate levels in both cell lines (Figure 2A–B; P<0.05) with a corresponding increase in NADH/NAD⁺ ratio (Figure 2C–D; P<0.05) and depletion of intracellular GSH (Figure 2E–F). To determine the effects on cell growth, we treated a panel of the neuroblastoma cell lines, as well as the human breast cancer line MCF7 and the human skin fibroblast cell line WI-38, with SR13800 in short-term dose-response assays. The IC₅₀ of SR13800 in the neuroblastoma and MCF7 lines ranged from 14.6μM to 31.3μM (P=0.1379, Table 2). The IC₅₀ for SR13800 in the WI-38 fibroblast cell line was substantially higher at 167.4μM (Table 2). Taken together, these results indicate that MCT1 inhibition effectively disrupts lactate homeostasis and reduces growth in neuroblastoma cells.

Figure 2. Cell biology and metabolic effects of SR13800 on neuroblastoma cells.

Response of neuroblastoma cell lines Kelly and LA-N-1 to SR13800 treatment. In lactate assays, treatment with the MCT1 inhibitor SR13800 for 4h increased intracellular lactate in Kelly (A) and LA-N-1 (B). (C–D) NADH/NAD⁺ ratio in Kelly (C) and LA-N-1 (D) cells treated with 0.1μM SR13800 for 6h. (E–F) In GSH assays, Kelly and LA-N-1 were treated with vehicle, 5μM BSO control or SR13800 for 24h and intracellular levels of GSH were determined. Data is presented as means ± SD (n=3). *P < 0.05, ***P<0.001, ****P<0.0001 compared to vehicle control.

Table 2.

SR13800 cytotoxicity assays on a panel of neuroblastoma cell lines

Cell line	SR13800 IC50 (μM)
LA-N-1	31.3 ± 3.0a
Kelly	31.4 ± 2.1
IMR-32	14.6 ± 3.1
BE(2)-C	18.4 ± 1.8
CHP-134	28.9 ± 2.6
SH-SY5Y	22.3 ± 2.6
MCF7	22.3 ± 2.6
WI-38	167.4 ± 11

^aMean ± SD of three independent experiments (n= 3)

SR13800 treatment in Kelly neuroblastoma cells causes an increase in tricarboxylic acid (TCA) cycle activity derived from glucose

The metabolic effects of SR13800 on MYCN-amplified Kelly cells were studied using GC-MS profiling of polar metabolites derived from heavy isotope labeling with 13C-glucose and 13C-glutamine. Examination of overall metabolic quantities (labeled and unlabeled) confirmed a large increase in intracellular lactate upon SR13800 treatment (Figure 3A). Examination of TCA metabolites revealed increases in the amounts of α-ketoglutarate, malate and citrate (Figure 3A), consistent with increased TCA cycle activity (Supplementary Figure S3). Overall metabolic quantification also revealed a large decrease in intracellular aspartate, a metabolite derived from the TCA cycle (Figure 3A). These observations corresponded with decreased lactate secretion and glucose uptake in media samples (Figure 3B–C).

Figure 3. MCT1 inhibition causes increased TCA activity in Kelly neuroblastoma cells.

(A) GC-MS metabolic profiling of intracellular polar metabolite quantities in Kelly treated with 0.1μM SR13800 and vehicle treated controls for lactate, TCA intermediates α-ketoglutarate, malate and citrate, and aspartate. YSI measurements on media samples from SR13800-treated Kelly demonstrated decreased lactate secretion (B, 6h) and decreased glucose uptake (C, 24h). (D) Fractional labelling derived from 13C-glucose in TCA metabolites, succinate, fumarate, malate, α-ketoglutarate and citrate fragments in Kelly treated with 0.1μM SR13800. OCR (E) and ECAR (F) 2h post treatment with 1μM SR13800 compared to vehicle treated Kelly (n=3). (G) Relative intracellular ATP levels were determined in Kelly treated with SR13800 (n=3). *P < 0.05, **P<0.01 compared to vehicle control.

Examination of 13C fractional labelling revealed that SR13800 treatment resulted in an increase in the proportion of the TCA intermediates succinate, fumarate, malate, α-ketoglutarate and citrate that were derived from 13C glucose (Figure 3D). While MYCN-amplified neuroblastoma cells can utilise glutamine as an alternative fuel source to glucose (15, 22), examination of the 13C fractional label from 13C glutamine revealed no significant changes in the fraction of any metabolites in Kelly cells in response to SR13800 treatment (Supplementary Table S2). In support of this, cell viability assays conducted on Kelly and LA-N-1 cells treated with SR13800 in the presence or absence of glutamine revealed little change in IC₅₀ values, whether under hypoxic or normoxic conditions (Table 3).

Table 3.

SR13800 cytotoxicity assays in the presence or absence of glutamine.

	SR13800 IC50 (μM)
Cell line	Glutamine	Glutamine-free	P value
Kelly	25.4 ± 2.48a	21.9 ± 1.49	0.300
LA-N-1	24.8 ± 0.684	23.3 ± 1.57	0.360

^aMean ± SD of three independent experiments (n= 3)

These observations suggest that MYCN-amplified neuroblastoma cells redirect pyruvate into the TCA cycle in response to SR13800 treatment, resulting in increased oxidative phosphorylation levels. To further test this possibility, real-time metabolic analyses were performed on live Kelly cells following SR13800 treatment. An increase in oxygen consumption rate (OCR) (Figure 3E, P<0.05 at 2h; Supplementary Figure S4A), a decrease in extracellular acidification rate (ECAR) levels (Figure 3F, P=0.11 at 2h; Supplementary Figure S4B) and an increase in intracellular ATP levels (Figure 3G; P<0.01) were all observed, consistent with increased oxidative phosphorylation.

SR13800 augments the potency of vincristine in neuroblastoma cells

To determine whether SR13800 potentiates the activity of chemotherapeutic agents used in the treatment of high-risk neuroblastoma, LA-N-1 and Kelly cells were treated with SR13800 in combination with vincristine and doxorubicin and evaluated in cell viability assays. Both vincristine and doxorubicin were found to be synergistic with SR13800 in LA-N-1 cells (Figure 4A–B). In Kelly cells, doxorubicin and SR13800 were not synergistic, however vincristine and SR13800 were strongly synergistic with 100% loss of cell viability at the lowest dose combination (Figure 4C–D). This striking synergy was also observed in clonogenic assays (Figure 4E). As vincristine functions primarily by inhibition of mitosis at metaphase, cell cycle analysis was performed following treatment of Kelly cells with SR13800 in combination with vincristine. Low dose treatment with SR13800 or vincristine alone did not significantly affect the percentage of cells in G2/M state, however the combination of SR13800 and vincristine significantly increased G2/M arrest in Kelly cells (Figure 4F), consistent with potentiation of vincristine function.

Figure 4. Vincristine augments the potency of SR13800 in neuroblastoma cells.

(A–D) In cell viability assays, treatment with SR13800 was synergistic with vincristine (VCR) in both LA-N-1 (A) and Kelly cells (C). SR13800 treatment was also synergistic with doxorubicin (Dox) in LA-N-1 (B) but not Kelly (D) in cell viability assays. CI value of 0.3–0.7 denotes synergism. (E) In colony assays, the combination of vincristine and SR13800 in Kelly was highly synergistic (CI of 0.316). (F) Cell cycle analysis in propidium-iodide stained Kelly cells showed a significant increase of percentage of cells in G2 arrest with the SR13800 (18μM) and vincristine (0.021μM) combination compared to single drug treatment. Analysis at 18h post-treatment. (G–H) ROS analysis in neuroblastoma cells. Kelly (G) were treated with vehicle, BSO control (5μM), SR13800 (17μM) and/or vincristine (2.1×10⁻³μM). LA-N-1 (H) were treated with vehicle, control BSO (5μM), SR13800 (18μM) and/or vincristine (1μM) for 18h. All data is presented as means ± SD (n=3). *P<0.05, ***P<0.001, ****P<0.0001 compared to vehicle control unless indicated.

Both SR13800 and vincristine are reported to be associated with increased intracellular ROS in cancer cells (17, 23). To investigate whether increased ROS could account for the synergistic decrease in cell viability, LA-N-1 and Kelly cells were treated with SR13800 and vincristine and assayed for ROS levels after 18h treatment. While vincristine treatment alone did not produce a significant increase in intracellular ROS in either cell line at the concentration at which maximal synergy was observed, SR13800 increased ROS to a similar extent as the BSO positive controls (Figure 4G–H). The combination of the two drugs further increased intracellular ROS levels, suggesting that the synergistic effect observed in the viability assays may in part be due to a large increase in intracellular ROS, without further depleting GSH levels (Supplementary Figure S4C).

MCT4 expression is inversely correlated with MYCN in neuroblastoma cells

MYCN has been shown to positively regulate SLC16A1 expression at the transcriptional level (20). Using two MYCN inducible systems, SHEP-TET21/N (Tet-off) and SHEP-MYCN3 (Tet-on), where MYCN is repressed or induced respectively by the addition of doxycycline, we confirmed that MCT1 protein expression paralleled MYCN expression (Figure 5A–B). We also assessed expression of MCT4, which is a resistance factor for MCT1 inhibition in other cancers (16, 24). MCT4 expression was found to be inversely related to MYCN expression in both systems (Figure 5A–B). To confirm these findings in an independent cell line, MYCN was suppressed in Kelly cells using two independent siRNA duplexes (dx1, dx2). MCT4 protein expression was upregulated following MYCN knockdown (Figure 5C), while MCT1 expression was reduced. To exclude MCT4 upregulation as a compensatory response to MCT1 downregulation rather than a direct effect lower MYCN levels, siRNA-mediated MCT1 knockdown was performed on Kelly cells. No significant alteration of MCT4 expression was detected upon MCT1 knockdown (Figure 5D).

Figure 5. MCT4 expression is suppressed by MYCN.

MCT1 expression is associated with MYCN in neuroblastoma cells (A–D). In both SH-EP-Tet21/N cells (MYCN Tet-off system) (A) and SH-EP-MYCN3 (MYCN Tet-On system) (B) neuroblastoma cells at 24h and 48h post-treatment with doxycycline, MCT1 expression was positively associated with MYCN expression while MCT4 was negatively associated with MYCN expression. (C) MCT4 expression was upregulated in Kelly cells when endogenous MYCN expression was suppressed with two independent MYCN siRNAs. However, MCT4 expression is unaffected by MCT1 knockdown using two different siRNAs in Kelly neuroblastoma cells (D). Actin or tubulin served as a loading control. Representative blots from three independent experiments. (E) In luciferase reporter assays, increased luciferase activity was detected in SH-EP-TET21/N cells transfected with two SLC16A3/MCT4 promoter/luciferase gene reporter constructs when MYCN expression was suppressed with doxycycline (Dox). Data presented as means ± SD (n=3). ***P<0.001 compared to no doxycycline treatment. **P<0.01, ***P<0.001 compared to vehicle control.

To examine for possible transcriptional repression of SLC16A3 by MYCN, two overlapping promoter regions of SLC16A3 were cloned into the luciferase reporter plasmid, pGL3-Basic, which was subsequently transfected into SHEP-Tet21/N cells and tested for luciferase activity. Luciferase activity of both SLC16A3 promoter/pGL3-Basic constructs was increased by over two-fold following doxycycline-induced MYCN downregulation (Figure 5E; P<0.01), suggesting that MYCN negatively regulates SLC16A3 transcription.

In vivo resistance to MCT1 inhibitor treatment is associated with increased levels of MCT4 protein

To determine whether the combination of SR13800 and vincristine is effective in vivo, Balb/c nude mice were xenografted with Kelly cells and established tumors treated with SR13800 alone or in combination with vincristine (Figure 6A). While vincristine modestly delayed tumor growth compared to the vehicle-treated group (P=0.01), SR13800 alone or combined with vincristine were both ineffective at delaying tumor growth (P=0.77 for SR13800, P=0.11 for SR13800/VCR). Tumors harvested at endpoint from mice treated with SR13800 or SR13800 and vincristine in combination demonstrated increased MCT4 protein expression (Figure 6B), potentially contributing to MCT1 inhibitor resistance.

Figure 6. MCT4 expression is elevated in neuroblastoma under conditions of hypoxia and HIF-1α induction.

(A) Response of Kelly xenografts in BALB/c nude mice to vehicle, SR13800 (30 mg/kg), vincristine (VCR, 0.2 mg/kg) and VCR/SR13800. (B) Tumor protein levels of MCT4 harvested at end point. (C–D) MYCN-amplified Kelly (C) and MYCN non-amplified SH- SY5Y (D) cells were incubated in normoxic or hypoxic conditions for 24, 48 and 72h and samples were analysed by Western blot for MCT1, MCT4 and LDHA. Kelly (E) and SH-SY5Y (F) were treated with HIF inducers desferrioxamine (DFO, 20mM) or deferasirox (DFX, 100mM) over a timecourse (6h–72h) and levels of HIF1α, MCT1, MCT4 and LDHA were analysed by Western blot. Tubulin served as a loading control. Representative blots from three independent experiments.

MCT4 is upregulated by hypoxia

As MCT4 is regulated by hypoxia inducible factor HIF-1α (25), we investigated the impact of hypoxia and HIF-1α induction on MCT4 protein expression in neuroblastoma cells. Kelly, LA-N-1, and MYCN non-amplified SH-SY5Y cells were maintained under both normoxic and hypoxic (1% oxygen) conditions over a period of 24h–72h and analyzed by Western blot. Under hypoxic conditions, MCT4 expression was upregulated in each cell line, particularly at the later time points, while no consistent changes were observed with MCT1 expression between normoxic and hypoxic conditions (Figure 6C–D, Supplementary Figure S5A). Cells were also treated with three different HIF-1α inducers, including the iron chelators desferrioxamine (DFO) and deferasirox (DFX) (26) and the prolyl hydroxylase inhibitor dimethyloxalylglycine (DMOG) (27), each of which increased MCT4 expression (Figure 6E–F, Supplementary Figure S5B, C). We also observed that 5μM SR13800 treatment of Kelly cells in vitro caused small increases in MCT4 protein expression after 72h treatment (Supplementary Figure S5D), suggesting that SR13800 treatment may partially contribute to the increased tumor expression of MCT4.

SR13800 modulates pyruvate levels in neuroblastoma cells and is highly synergistic with the LDHA inhibitor FX11 in reducing cell viability

Although MCT1 is widely recognized to regulate lactate transport, the inhibition of MCT1-mediated pyruvate export may also lead to cancer cell apoptosis (18). Accordingly, we investigated the effect of SR13800 treatment on pyruvate levels in neuroblastoma cells. SR13800 treated Kelly cells demonstrated a decrease in extracellular pyruvate as detected by GC-MS (Supplementary Figure S6A). To further modulate pyruvate levels, we treated with the LDHA inhibitor FX11, which inhibits the conversion of pyruvate to lactate. Although FX11 decreased intracellular lactate and ATP in Kelly and LA-N-1 cells (Supplementary Figure S6B–E), FX11 treatment had markedly few effects on the polar metabolic profile of Kelly cells (Supplementary Table S3). One notable exception was that FX11 treatment caused a significant increase in levels of extracellular pyruvate in Kelly (Supplementary Figure S6F). To follow up on this observation, we performed extracellular and intracellular pyruvate assays with Kelly and LA-N-1 treated with either SR13800 or FX11, or both drugs in combination. SR13800 treatment decreased extracellular pyruvate in both cell lines (Figure 7A–B) and increased intracellular pyruvate in LA-N-1, however in Kelly intracellular pyruvate levels remained unchanged (Figure 7C–D). While LA-N-1 showed a significant increase in intracellular pyruvate upon FX11 treatment, which would be expected with LDHA inhibition, Kelly instead showed a significant increase in extracellular pyruvate (Figure 7A–D). This suggested that excess accumulation of pyruvate caused by FX11 treatment is effluxed from Kelly cells, and this efflux is likely to be mediated by MCT1.

Figure 7. SR13800 and FX11 treatment altered pyruvate levels in neuroblastoma cells and are synergistic with each other in cell viability assays.

Pyruvate assays on Kelly (A, C) and LA-N-1 (B, D) treated for 6h (intracellular) or 24h (extracellular) with SR13800 (0.1μM), FX11 (20μM) or both drugs in combination. (E–F) In cell viability assays, the combination of SR13800 and FX11 was highly synergistic in reducing cell viability in both (E) Kelly (CI = 0.176) and (F) LA-N-1 (CI = 0.598). Data presented as means ± SD (n=3). *P<0.05, ** P<0.01, *** P<0.001, **** P<0.0001 compared to vehicle control.

Taken together, these observations led to the hypothesis that the combination of SR13800 and FX11 leads to an accumulation of intracellular pyruvate in both cell lines, regardless of the differential response between the two cell lines to SR13800 and FX11 as individual agents. Indeed, combined treatment of SR13800 and FX11 on Kelly resulted in a large increase in intracellular pyruvate (Figure 7C). A small increase in intracellular pyruvate was also detected in LA-N-1 for the combination versus single treatments (Figure 7D). To investigate if accumulation of intracellular pyruvate by the combined treatment of SR13800 and FX11 is associated with inhibition of cell growth to neuroblastoma cells, cell viability assays combining SR13800 and FX11 were performed on Kelly and LA-N-1 cells. Strikingly, this combination was strongly synergistic in both Kelly and LA-N-1 cells, resulting in almost 100% loss of cell viability (Figure 7E–F).

Discussion

The clinical application of MCT1 inhibitors to block export of excess intracellular lactate, a critical step in maintenance of high glycolytic rates in cancer cells, is under investigation for adult cancer therapy. This study investigated whether MCT1 inhibition has antitumor activity in neuroblastoma, a pediatric tumor characterized by MYCN amplification. Our finding that high expression of SLC16A1/MCT1 is strongly and independently prognostic for poor survival in patients and is associated with MYCN amplification, supports MCT1 as a potential therapeutic target in this tumor. MCT1 expression was increased with high levels of MYCN, and, interestingly, we showed for the first time that high MYCN levels repress MCT4 expression. Strong synergy was observed between the MCT1 inhibitor SR13800 and vincristine in vitro. Despite this finding, this combination was ineffective in vivo in Kelly neuroblastoma xenografts. Our observations suggest that resistance mechanisms could include an increase in MCT4 levels in the hypoxic neuroblastoma tumor environment, and SR13800 treatment itself leading to an increase in MCT4 levels as well as increased OCR in Kelly cells. SR13800 was also highly synergistic with the LDHA inhibitor FX11 in vitro, and this synergy was associated with an accumulation of intracellular pyruvate. Because MCT4 has a very low affinity for pyruvate transport (9), combining an MCT1 inhibitor with an LDHA inhibitor could circumvent induced MCT4 resistance. Therefore, the therapeutic potential of this combination should be tested when suitable LDHA inhibitors become available.

The regulation of metabolites is a complex homeostatic process involving large numbers of regulators, enzymes, transporters, signalling pathways and metabolites, and an accumulation of metabolites can cause cascading effects on the metabolic network (28). Cancer cells have considerable capacity to undergo continuous global metabolic rewiring to adapt to changing environment conditions (29) and are flexibly respond to fluctuations in metabolites and activate or repress pathways via the action of master transcription factors such as MYCN and HIF-1α (6). Relatively little is understood about the metabolic phenotype of neuroblastoma, its metabolic preferences, or its capacity to switch between different metabolic pathways in response to various stimuli, nutrients or pathway inhibition (6, 30).

Recent studies have uncovered potential glycolytic and glutaminolytic targets for neuroblastoma, but few reports to date have studied the metabolic profile of neuroblastoma cells. We showed that SR13800 treatment caused an increase in TCA activity in Kelly cells, suggesting that pyruvate is redirected to the TCA cycle, which also corresponded to an increase in intracellular ATP levels. However, this contrasts with SR13800 administered to Raji lymphoma cells which did not alter levels of TCA intermediates, but instead showed decreased ECAR without changing basal OCR, and decreased intracellular ATP levels, most likely caused by reduction in glycolytic rates (17). The ability of neuroblastoma cells to redirect excess pyruvate into the TCA cycle and generate additional ATP via oxidative phosphorylation is in concordance with a recent report showing that human lung NSCLC tumors can take up lactate and pyruvate to supplement the TCA cycle (10). The decreased aspartate levels in response to SR13800 treatment in Kelly cells may indicate reduced utilisation of the TCA cycle to produce aspartate from oxaloacetate. This may be due to limited NAD⁺ availability leading to inhibition of malate dehydrogenase activity, or alternatively, aspartate may be rapidly used as a precursor for other metabolites as a result of MCT1 inhibition. Our data also indicate that while these neuroblastoma cells can compensate for excess pyruvate caused by MCT1 inhibition by redirecting pyruvate to the TCA cycle, the addition of an LDHA inhibitor abolishes pyruvate homeostasis. Thus, an in-depth understanding of neuroblastoma metabolic pathways will be helpful in developing and refining relevant therapies for the disease.

An important resistance mechanism for MCT1 inhibitors is compensatory upregulation of the related lactate transporter MCT4 (16, 25). While MYCN is known to directly regulate MCT1 transcription (20), our study made the novel finding that MYCN normally suppresses SLC16A3/MCT4 transcription in neuroblastoma cells. MYCN, a well characterized transcriptional activator, also acts as a gene repressor via binding to Sp1 (31, 32), and this function may contribute to neuroblastoma progression by inhibiting genes that antagonise tumor growth such as angiogenic inhibitors and pro-apoptotic genes (32, 33). Although MYCN is active in remodeling metabolic pathways (6, 34), there is evidence for MYCN acting as a repressor of metabolic genes. The ability of MYCN to activate SLC16A1/MCT1 transcription and simultaneously repress SLC16A3/MCT4 is reflected in the pattern of endogenous expression in the cell lines. However, MCT1 inhibitor treatment was ineffective in vivo, which may suggest that the regulation of MCT4 by HIF1α in neuroblastoma tumors is likely to overcome the repressive ability of MYCN, resulting in MCT4-mediated resistance to MCT1 inhibition. In support of this, a recent study also indicated that some solid tumor cell lines, including neuroblastoma and lung cancer lines, demonstrated only partial sensitivity to MCT1 inhibition and generally had higher levels SLC16A3 mRNA expression compared to hematological tumors (35).

The combinatorial use of metabolic inhibitors with conventional cancer agents has been investigated as a strategy to overcome resistance, generally targeting the importance of metabolite balance for cellular or chemoresistance responses or cell damage repair (36–38). In our in vitro studies, we found that the effect of SR13800 was greatly enhanced in combination with vincristine, resulting in increased intracellular ROS and G2/M arrest. However, this combination administered in vivo was ineffective partially due to MCT4 upregulation as a result of functional inhibition of MCT1 via SR13800. Another possible strategy that may avoid MCT4 upregulation may be to exploit cancer cells with high MCT1 expression by developing a MCT1 targeting substrate agent like 3-BrPA (20, 39). While 3-BrPA is not suitable for clinical use, 3-BrOP, a derivative of 3-BrPA, has been reported to have activity against neuroblastoma cells (40) and it would be interesting to investigate if existing analogs can be further developed as MCT1 substrate targets.

SR13800 has inhibitory activity not only against MCT1 but also the related transporter MCT2, which can also export pyruvate (41) and is expressed by various cancers including neuroblastoma (20, 42). It is possible that the effectiveness of combining MCT1 and LDHA inhibition in dramatically reducing neuroblastoma cell viability is partially attributable to the action of SR13800 on MCT2. Regardless, the very high synergy demonstrated by this combination suggests that neuroblastoma cells are particularly sensitive to large increases in intracellular pyruvate and are unable to adequately redirect it to other pathways. Pyruvate is also required by histone deacetylases (HDACs), and pro-survival proteins which induce apoptosis upon association with pyruvate (43). Alternatively, accumulation of pyruvate may exert added pressure on the TCA cycle, leading to increases in ROS activity, although we could find no evidence for this in our analyses. Unfortunately, FX11 is unlikely to be suitable for clinical use particularly due to poor solubility and insufficient potency (44, 45). Although improved LDHA inhibitors have been developed, these have not been reported as suitable for in vivo use, but once such development occurs these may warrant testing in combination with MCT1 inhibitors.

In conclusion, this study has investigated the potential application of MCT1 inhibition as a therapeutic strategy for MYCN-amplified high risk neuroblastoma. In particular, the combination of MCT1 and LDHA inhibition was found to be highly synergistic in decreasing cell viability of neuroblastoma cells. An interesting novel discovery is the role of MYCN as a repressor of MCT4 while positively regulating MCT1. Although in vivo Kelly xenografts were resistant to MCT1 inhibitor treatment likely via various resistance mechanisms, importantly, our results suggest that there may be unique opportunities to develop MCT1 inhibition strategies for neuroblastoma therapy in combination with strategies that disrupt pyruvate homeostasis.

Materials and methods

General reagents

DMEM and RPMI media, FCS and Lipofectamine 2000 and RNAiMAX transfection reagents were obtained from Life Technologies Australia (Mulgrave, VIC). MCT1 inhibitor SR13800 and LDHA inhibitor FX11 were obtained from Merck Australia (Bayswater, VIC). Vincristine was obtained from Selleck (Houston, Texas). Deferasirox was obtained from Ontario Chemicals (Canada). All other chemicals were obtained from Sigma-Aldrich Australia (Castle Hill, NSW).

Reactive oxygen species (ROS) assay and cell cycle analysis

For ROS assays, after drug treatment for 18h, cells were incubated with 2′,7′-Dichlorofluorescein diacetate (DCF) (Sigma-Aldrich) for 1h at 37°C, harvested and analysed on a FACSCalibur (BD Biosciences) using FlowJo Software (Ashland, Oregon, USA).

Cell cycle analysis was performed as described (46) on cells treated with vehicle, SR13800 and/or vincristine for 18h, harvested and stained with propidium iodide (25μg/mL, Sigma-Aldrich) containing RNAse (2μg/mL)(Roche Diagnostics, Castle Hill, NSW), then fixed with 70% ethanol before flow analysis.

13C metabolic labeling, metabolomic and real-time metabolic analyses

13C-metabolic labeling was performed following published methods (47). Kelly were plated at 5 × 10⁵/well in 6-well plates in RPMI/10% FBS overnight before medium was changed to MEM (Sigma-Aldrich #51416C) containing 10% FBS, NEAA, 2g/L glucose and 2mM 50% 13C-uniformly labeled glutamine. For 13C glucose labeling, the same medium was used except that the glucose component was 50% 13C-uniformly labeled glucose (Sigma-Aldrich) and glutamine was unlabelled (2mM). Inhibitor or vehicle was added to cells for 6h or 24h before harvesting for media and polar metabolite fractions (47). GC-MS metabolomic analyses were performed as published (48) on polar metabolite or media fractions. Media was also analysed for glucose and lactate using a YSI model 7100 enzyme analyzer.

For real-time metabolic analyses, 80,000 Kelly cells were plated in wells of a XF24 plate (Seahorse Bioscience), and next day media was replaced with base media containing 2g/L glucose and 200 mM glutamine and analysed on a Seahorse XF24-3 Bioanalyser. After determination of baseline ECAR and OCR, 1μM of SR13800 or vehicle was injected onto cells. Results were normalized by cell count in five fields of methanol-fixed cells stained with May-Grunwald (Sigma-Aldrich) and Giemsa (Sigma-Aldrich).

Generation of MCT4 promoter constructs and luciferase reporter assay

Two overlapping promoter regions of the human SLC16A3 (MCT4) gene were PCR-amplified and cloned into luciferase reporter vector pGL3- Basic (Promega) using standard procedures. Promoter 1 (−1600 to +500 from the transcriptional start site (TSS) of NM_001206950.1) and promoter 2 (−1000 to +600 from the TSS of NM_004207.3) (25) were PCR-amplified using primers (Supplementary Table S6) on genomic DNA from Kelly cells. MCT4 constructs or empty pGL3-Basic luciferase vector were transiently co-transfected with Renilla construct (Promega) into SHEP-TET/21N and after 4h, media was replaced with or without doxycycline (2μg/mL). At 48h post-transfection, cells were assayed using a Dual Luciferase Kit (Promega).

Animal studies

All animal studies were approved by the University of New South Wales Animal Care and Ethics Committee and conducted according to the Animal Research Act, 1985 (NSW, Australia) and the Australian Code of Practice for Care and Use of Animals for Scientific Purposes (2013). The toxicity of single doses of SR13800 (in 1:1:8 DMSO-Tween80-water) administered i.p. was evaluated in Balb/c mice, and the maximum tolerable dose of 30 mg/kg was evaluated for toxicity in combination with vincristine (0.2 mg/kg). 10⁷ Kelly cells were subcutaneously engrafted into one flank of Balb/c nude mice with Matrigel Matrix (BD Biosciences). When tumor size reached 100 mg as measured by vernier calipers, mice were administered i.p. vehicle, SR13800 (30 mg/kg) for five days per week for the duration of the experiment and/or vincristine (0.2 mg/kg) for five consecutive days total. Mice were sacrificed when tumor size reached 1000 mg.

Statistical Analysis

Data are presented as means ± standard deviation (SD) of three independent experiments unless otherwise stated. All statistical analyses were performed using GraphPad Prism (San Diego, CA, USA). P<0.05 was considered statistically significant. Unless indicated, two-sided Student’s t-test was used to evaluate statistical differences. Combination index (CI) was calculated using the Calcusyn program (Biosoft, Cambridge, UK), where a CI<1 indicates synergy, CI = 1 indicates additive effect and CI>1 indicates antagonistic effect (49). Survival analyses were computed by the Kaplan-Meier method and compared between groups using the log-rank test. Univariate and multivariate analyses for event-free survival (EFS) and overall survival (OS) were performed using SPSS version 24 (IBM, Mainz, Germany) as described (50).