Silibinin down-regulates PD-L1 expression in nasopharyngeal carcinoma by interfering with tumor cell glycolytic metabolism

Abstract

The upregulation of checkpoint inhibitor PD-L1 expression has recently been associated with nasopharyngeal carcinoma (NPC) resistance to therapy. The mechanism of induction of PD-L1 has also been linked to enhanced aerobic glycolysis promoted by HIF1-α dysregulation and LDH-A activity in cancer. Here, we investigated the effect of the anti-tumoral compound Silibinin on HIF-1α/LDH-A mediated cancer cell metabolism and PD-L1 expression in NPC. Our results demonstrate that exposure to Silibinin potently inhibits tumor growth and promotes a shift from aerobic glycolysis toward oxidative phosphorylation. The EBV+ NPC cell line C666–1 and glycolytic human tumor explants treated with Silibinin displayed a reduction in LDH-A activity which consistently associated with a reduction in lactate levels. This effect was accompanied by an increase in intracellular citrate levels in C666–1 cells. Accordingly, expression of HIF-1α, a critical regulator of glycolysis, was down-regulated after treatment. This event associated with a down-regulation in PD-L1. Altogether, our results provide evidence that silibinin can alter PD-L1 expression by interfering with HIF-1α/LDH-A mediated cell metabolism in NPC. These results provide a new perspective for Silibinin use to overcome PD-L1 mediated NPC resistance to therapy.

Introduction

Treatment of nasopharyngeal carcinoma (NPC) poses a particular challenge in terms of disease recurrence management and availability of effective accessible therapies. It is estimated that approximately 15–30% of NPC patients treated with traditional radiotherapy would develop local to regional recurrence ^1,2, and that patients diagnosed with locally advanced tumors would experience recurrence in 30–80% cases ^3–5. Considering that in such cases most treatment options are limited to palliative approaches, it has become necessary to identify new molecules which may help overcoming the barrier of resistance to therapy ^2,6–9

In NPC, as for most epithelial-originated cancers, immune checkpoint blockade PD-L1 overexpression has been associated with poor prognosis ^10,11. Importantly, in recent years, targeting the PD-1/PD-L1 interaction has risen to the forefront of the possible options for head and neck cancer squamous cell carcinoma management ^12–14; thereby making the disruption of the pathway upstream of PD-L1 an exciting therapeutic alternative for patients with relapsing or metastatic tumors (RM-NPC).

Metabolic reprogramming is regarded as a hallmark of cancer and contributes to cancer treatment resistance ¹⁵. The hypoxia inducible transcription factor HIF-1α, is known to induce metabolic rewiring through its α subunit ^16,17, and, as for a number of different malignancies, HIF-1α upregulation in NPC patients has been associated with poor prognosis ^18–22. HIF-1α regulates the expression of lactate dehydrogenase (LDH-A), a key glycolytic enzyme catalyzing the conversion of pyruvate to lactate, whose overexpression is sufficient to enhance the glycolytic flux ²³. Accumulation of lactate in the tumor microenvironment promotes tumor growth, invasiveness, immune escape and is positively correlated with metastasis development ²⁴. Interestingly, HIF-1α upregulation positively regulates the expression of PD-L1, which in turn delivers immune inhibitory signals to T cells via PD-1 to favor tumor immune escape and recurrence ²⁵. Recent clinical trials targeting the PD-1/PD-L1 pathway have shown promising antitumor activity in patients with RM head and neck cancers ^26,27.

In Epstein-Barr virus (EBV) induced NPC, the viral oncoprotein LMP-1 is constitutively expressed on the tumor cell surface and is involved in mediating an oncogenic signal which can upregulate PD-L1 expression either via inflammatory signals involving IFN-γ or via mitogenic stimulation ²⁸. Under normoxic conditions, LMP-1 signaling directly upregulates HIF-1α transcription and promotes the expression of multiple glycolytic genes including GLUT1 and LDH-A ^11,29,30.

Silibinin (SBN), an FDA and EMA approved drug for the treatment of liver diseases, extracted from Silybum marianum (L.) Gaertn., fructus, has been reported to have in vitro and in vivo anti-tumor effects in a variety of tumor models, including skin, bladder, oral, lung, breast, prostate and kidney carcinomas ³¹. The drug has been shown to possibly exert its inhibitory effects by targeting either HIF-1α, LDH or PD-L1 expression ^32–34. However, to the best of our knowledge, the effect of this molecule against NPC has not been yet explored. In the present study, we aimed to investigate whether NPC tumors treatment with silibinin could overcome PD-L1 expression by interfering with NPC glycolytic metabolism.

Materials and methods

Study population characteristics

Primary NPC biopsies (n=20) were collected, before treatment, from NPC untreated patients (mean age: 36.3± 20.20 years, Stage II/IV) from Otorhinolaryngology departments of Mustapha Pacha and Bachir Mentouri hospitals in Algiers upon informed consent. This study was approved by the ethics committee of the National Agency for Research Development in Health (ATRSS).

NPC biopsies culture

NPC samples were processed immediately after endoscopic biopsy. The biopsies were rinsed with PBS plus antibiotics (streptomycin/penicillin) prior to culture. Blood clots and the underlying connective tissue were discarded.

The biopsy specimens were cut into equal size pieces and cultured in 96 flat-bottom well plates with complete Dulbecco’s Modified Eagle’s medium (DMEM) (Sigma, USA) and incubated in a humidified chamber at 5% CO₂ and 37°C with increasing concentrations of SBN (Sigma, USA)(range= 0–200µM) for 24hours. The supernatants were stored at −20°C until use and the biopsies were formalin-fixed paraffin-embedded (FFPE) for hematoxylin and eosin staining and immunohistochemistry (IHC) analysis.

Protein extraction

Proteins extraction was performed by adding 20 l RIPA lysis buffer (Thermo Scientific™ Pierce™, USA) to each well of a 96 well culture plate. Cells were incubated on an agitator at 37°C for 15min. The quantification of the total protein extract per well was assessed using Pierce™ BCA Protein Assay Kit (Thermo Scientific™, USA) following manufacturer’s instructions. Briefly, 20 l of cell lysate were transferred in a normal 96 well plate to which 200 l of BCA reagent were added. After 30 min incubation at 37 °C, the ODs were read at 562nm on a SpectraMax i3 plate reader.

Immunostaining

FFPE sections (4 m) were prepared in silanized slides before immunostaining (IS). Sections were deparaffinized in xylene and rehydrated in decreased concentrations of ethanol.

Antigen retrieval was performed in a water bath for 40min using TRIS-EDTA buffer pH 9.0 (Novocastra, Leica, UK) for PD-L1 staining and using Trilogy buffer (Cell marquee, USA) for HIF-1α staining. Endogenous peroxidase activity was quenched using peroxide block (Cell marque, USA). Sections were stained using the primary antibodies HIF-1α (AbDSerotec, USA, clone ESEE122, dilution 1:100) and rabbit anti-human PD-L1 (Zeta corporation, USA, clone ZR3, dilution 1:100) at room temperature for 90min and 60min respectively. TBS was used as washing buffer. HiDef Detection^™ HRP Polymer kit (Cell marque, USA) was used as secondary detection system. Sections were incubated with Diaminobenzidine tetrachloride (DAB) (DAKO, USA) for 5min to visualize the antibody-antigen complex.

IS was scored according to the H-scoring method ³⁵. The number of positive cells and the intensity of staining were taken in consideration. Five non overlapping randomly selected, high-power fields were analyzed for quantification. The tissue sections were read independently by two investigators.

C666–1 cell line culture and proliferation assay

The C666–1 (EBV+ LMP-1+) undifferentiated human NPC cell line was kindly provided by Dr. Bill Sugden (University of Wisconsin, USA). For the proliferation assay, C666–1 cells were seeded at a density of 10,000 cells/well in complete RPMI 1640 medium (MSKCC media preparation core facility, USA) in 96 flat-bottom well plate, and incubated with increasing concentrations of SBN (range= 0–200µM) in a humidified chamber at 5% CO ₂ and 37°C. After addition of propidium iodide (PI, 2µg/ml, Nexcelom Bioscience, USA), live and dead cells were imaged and counted at 4 time points after treatment (0, 24, 48, 72h) with the Celigo imaging cytometer (Nexcelom Bioscience, USA) to monitor proliferation and cell death kinetics.

Flow cytometry analysis

PD-L1 surface staining was performed by incubating C666–1 cells with PE-Cy7-labeled anti-human CD247 (Clone: MIH1, BD Biosciences, USA, dilution 1:100) for 30min after blocking Fcγ receptors (FcγR) with the FcR Blocking Reagent (Miltenyi Biotec, USA) for 10min.

For intracellular staining of HIF-1α, C666–1 cells were fixed and permeabilized (Foxp3/Transcription Factor Fixation/Permeabilization Buffer, eBioscience, USA) after 30min incubation with the FcR Blocking Reagent and an eFluor506 fixable viability dye (Invitrogen, USA). Fixed and permeabilized cells were then incubated with a PE-labeled anti-human HIF-1α antibody (clone 241812, R&D Systems, dilution 1:20). Isotype-matched control antibodies were used to detect non-specific staining. Samples were acquired on a Fortessa flow cytometer (BD Biosciences, USA) using BD FACSDiva software (BD Biosciences, USA) and data were analyzed with FlowJo10.2 software (Tree Star Inc.).

Glycolysis analysis

Glycolytic activity of C666–1 cells was measured using a Seahorse XF96 Extracellular Flux Analyzer (Agilent Seahorse XF Technology, USA). Cells were seeded at 40,000 cells/well in Seahorse XF96 well plates and allowed to attach overnight in a complete RPMI medium at 37°C and 5% CO ₂ and then treated with SBN (0–100µM) for 3h. The medium was changed in each well and replaced with assay medium (DME medium without red phenol, supplemented with 2mM glutamine, 10mM glucose, 1mM pyruvate and 5mM HEPES pH7.4). Cell incubation was extended for an hour at 37°C in a non-CO ₂ incubator to allow media temperature and pH to reach equilibrium before measurement. The Proton Efflux Rate (PER) and extracellular acidification rate (ECAR) were measured under basal conditions and after sequential addition of 18ul of Rotenone/Antimycin A (ROT/AA, 5uM final concentration) into each well followed by another injection of 20µl of 2-Deoxy-D-Glucose (2-DG, 50mM final concentration, Sigma, USA). Total PER and ECAR were measured by plotting them as a function of time (pmol/min) and (mpH/min) respectively in XF96 Glycolysis report generator. Data were normalized to the total protein amount present in each individual well using Wave software.

LDH activity

LDH colorimetric activity was measured in C666–1 cell lysates and in NPC culture biopsies’ supernatants after 24h incubation with SBN (range = 0–100µM) using LDH colorimetric assay kit (Biovision, USA) following the manufacturer’s instructions. The ODs were read at 450nm on a SpectraMax i3 plate reader, USA.

Lactate measurement

Lactate production was measured in NPC biopsies and C666–1 cell line culture supernatants after 24h incubation with SBN (0–100µM) using the lactate plus Meter Test strips (Nova Biomedical, USA) following the manufacturer’s instructions.

Citrate measurement

Citrate rates variations were measured in C666–1 cell lysates upon 24h incubation with SBN (0–100µM) using Biovision Citrate colorimetric assay kit. The ODs were read at 570 nm on a SpectraMax i3x plate reader, USA.

Statistical analysis

All statistical analyses were performed using GraphPad Prism 6.0.1 software. The data were presented as mean ± SEM. Mean comparison between groups was performed using t test and One-way analysis of variance (ANOVA one-way). A p value ≤0.05 was considered statistically significant.

Heat-mapping

Heat-mapping was performed using centered and normalized data with Genesis 1.8.1.

Results

SBN treatment decreases tumor cells viability by affecting NPC glycolytic activity

To assess the anti-proliferative efficacy of SBN on LMP-1+ NPC tumor cells (C666–1), cultures were exposed to a range of concentrations of the drug varying between 0–200µM and cell viability was monitored every 24h over a period of 72h. We observed that exposure to the drug resulted in a significant inhibition of cell growth, in a dose and time dependent manner. Compared with control (vehicle DMSO), SBN treatment at 100µM resulted in a partial suppression of tumor cell viability to reach a 50% inhibition at 72h post-treatment. Exposure to a concentration of 200µM induced a rapid and important cytotoxic effect on C666–1 cells which translated into 95 % cell death at 24h post-treatment (Fig.1A). Drug concentrations under 50µM did not show any significant cytotoxic activity.

Fig.1. SBN treatment affects NPC tumor cells viability by decreasing tumor cell glycolytic activity.

A) Cells were plated at 10,000 cells/well in a 96 well plate and treated SBN (25, 50, 100 and 200 M) diluted in DMSO or DMSO as vehicle control. Propidium Iodide (PI) was added to monitor kinetic viability at 0, 24, 48, 72h incubation time points using the Celigo imaging cytometer and imaging with the bright field and 531/629 ex/em channel to count total and PI-positive dead cells. B) SBN reduces Bcl-2 expression in NPC biopsies. Biopsies (n=6) were treated with SBN (0–100 M) for 24h. Bcl-2 expression was tested by IHC (B.1). Scoring evolution by patient. The scoring was based on the percentage of positive cells (B.2). Differences between groups were analyzed using t test (** p=0.006). C) GlycoPER (proton efflux rate from glycolysis) and D) Metabolic flux ECAR (Extracellular acidification rate) and in C666–1 cells were measured using Seahorse metabolic analyzer. Differences between controls and SBN treated cells were tested using ANOVA (**** p˂0.0001). One representative of 3 experiments is shown.

To confirm the anti-proliferative/pro-apoptotic activity of SBN on primary NPC tissues, expression of the anti-apoptotic marker Bcl-2 was assessed on human biopsies (n=6) exposed to 100µM SBN for 24h. A comparison of the percentage of cells expressing the pro-survival biomarker in treated versus non treated biopsies showed with interest, that exposure to the drug reduced expression of Bcl-2 by 30% on average (p=0.006) (Fig.1B.1-2). A morphological analysis of the tested tissues did not show presence of necrotic features 24h post-treatment.

Considering that cancer cells rely on glucose metabolism to obtain the sufficient energy to proliferate³⁶ and the putative inhibitory effects of SBN on glycolysis ^32,34,37, we hypothesized that SBN decreased tumor cell growth and viability by slowing down tumor cell metabolic activity. Therefore, we next tested if the observed anti-proliferative effect of SBN occurred in association with a possible anti-glycolytic activity. To this end, we incubated C666–1 cells with SBN at concentrations that did not show direct cytotoxic activity as early as 3h after treatment (0–100 M) and monitored changes in the glycolytic rate assessed by either proton efflux selectively from the glycolytic pathway into the extracellular media (GlycoPER, which occurs subsequently to the breakdown of glucose to lactate during glycolysis) and extracellular acidification (ECAR) using the Seahorse XF96 Extracellular Flux Analyzer. This approach provides accurate measurements of glycolytic rates for basal conditions and compensatory glycolysis following mitochondrial inhibition using Rotenone (ROT) and Antimycin A (AA). Reductions in proton release and milieu acidification should reflect the capacity of SBN to block glycolysis.

ROT/AA addition to C666–1 cell cultures triggered instant ECAR and glycoPER increases in the control wells treated with DMSO (vehicle). This effect was subsequently abolished by addition of 2-DG, a glycolysis inhibitor. Interestingly, addition of SBN to the NPC cell cultures significantly reduced ECAR and GlycoPER in a dose dependent manner compared to the control. An optimal inhibitory effect, reflected by a net reduction of glycolysis and extracellular acidosis, was observed in presence of 100µM SBN (p ˂0.0001; Fig.1C-D.1-2).

Glycolysis inhibition associates with loss of LDH-A mediated lactate release in SBN treated tumors

Cancer cells, which conduct high levels of glycolysis in aerobic conditions (Warburg effect), produce large amounts of lactate, through the catalytic reduction of pyruvate by lactate dehydrogenase (LDH-A), and extrude it in the extracellular environment, provoking acidosis^38–40. Recent studies have demonstrated that NPC overexpressing LDH-A, may release the glycolytic enzyme in the environment via exosomes to mediate immune evasion and metastasis ^41–43. To test if reductions in glycolysis and extracellular acidification rates upon SBN treatment were related to inhibition of LDH-A and lactate production, we next sought to investigate the influence of SBN (100µM) on LDH-A enzymatic activity- and lactate release in NPC cell cultures. Interestingly, we observed that whereas LDH-A activity was reduced by 36% in SBN-treated compared to control C666–1 cultures 24h post-treatment (p=0.002), extracellular lactate concentrations were reduced by 40% (p=0.001) (Fig.2A-B).

Fig.2. SBN reduces LDH activity and lactate levels in C666–1 cells.

Cell cultures were treated with SBN (0–100µM) for 24h and LDH activity and/or lactate production were assessed. LDH activity (A) and lactate levels (B). Measurements were performed in triplicates. Shown data are representative of one out of 3 experiments. Differences between groups were analyzed using t test. ** p=0.002 (A), *** p=0.001 (B).

To verify if the observed effects of SBN on the C666–1 cell line could be extended to patients’ tumors, we next treated explants isolated from 20 NPC patients with SBN (100µM). A heatmap analysis of LDH-A and lactate synthesis showed to our surprise that only 12 out of 20 untreated tumors displayed increased LDH-A activity and/or elevated lactate levels. This result indicated that the glycolytic phenotype was heterogeneous among patients’ tumors. Analysis of the effect of the drug on the most glycolytic tumors, identified by LDH-A^High-Lactate^High and LDH-A^High –Lactate ^Low profiles (n=12), showed presence of an inhibitory pattern imprinted by SBN on most tumors. Interestingly C666–1 cells showed a similar pattern (Fig.3A, Fig.4B). Further analysis indicated that SBN exerted significant inhibitory effects on the expression and/or release of studied molecules (LDH-A: 53.11%, p=0.0005; Lactate: 15.41%, p=0.059) (Fig.3B-C). To visualize the consequences of SBN treatment on LDH-A/Lactate association, a Pearson analysis was performed. We observed a loss of association linking LDH-A synthesis to lactate when comparing treated tumor explants and untreated controls (r = −0.18, p = 0.57 vs. r = 0.51, p = 0.08) (Fig. 3D.1-2).

Fig.3. SBN reduces LDH activity and lactate levels in NPC biopsies.

A) Heatmap representation of LDH activity and lactate levels in untreated and SBN treated NPC biopsies (SBN (0–100µM) for 24h). B-C) Analysis of the effect of the drug on most glycolytic tumors identified by LDH-A^High-Lactate^Low and LDH-A^High-Lactate^High profiles. Measurements were performed in triplicates. Differences between groups were analyzed using t test. *** p=0.0005. D.1–2) Correlation analysis between LDH-A and lactate levels in NPC supernatants after treatment (0–100 M) (n=12). The correlation was performed using Pearson r test.

Fig.4. SBN increases citrate rate in C666–1 cells at the expense of aerobic glycolysis.

A) C666–1 cells were treated with SBN (0–100µM) for 24h. Citrate rate was evaluated using Citrate colorimetric assay. Measurements were performed in triplicates. The experiment was repeated 3 times. Differences between groups were analyzed using t test. *** p=0.0008. B) Heatmap shows relative levels of glycolytic and TCA cycle metabolites from C666–1 cells treated with SBN (0–100 µM). Data are representative of 3 independent experiments.

Together these results demonstrated that SBN inhibited aerobic glycolysis in NPC tumor cells, likely, by targeting LDH-A activity and lactate dependent acidosis.

Exposure to SBN restores citrate synthesis at the expense of aerobic glycolysis

In cancer cells, pyruvate is shunted away from the mitochondrial matrix and converted into lactate thus depauperating the source of citrate for the tricarboxylic acid cycle (TCA). Citrate is an intermediate metabolite in cellular oxidative metabolism which exerts a negative feedback control on glycolysis; it is a key indicator of glycolysis decoupling from oxidative phosphorylation (OXPHOS), and it mirrors the reduction of pyruvate carbon entry in the TCA cycle ³⁶. To determine if down regulation of LDH-A under SBN treatment would restore TCA cycle, we measured variations in intracellular citrate levels in C666–1 cells after SBN treatment (100µM). We observed that in comparison to control, citrate synthesis increased by 419.18% in cells 24h post treatment (p=0.0008) (Fig.4A). A heatmap analysis confirmed the consistency of the pattern imprinted by SBN on citrate synthesis at the expense of LDH-A and lactate synthesis (Fig.4B). Our results show the capacity of SBN to inhibit LDH-A and lactate release and to stimulate citrate synthesis in human NPC.

These results suggest that SBN treatment may promote activation of the TCA cycle in NPC at the expense of aerobic glycolysis.

SBN decreases HIF-1α expression in nasopharyngeal tumors

Considering the pivotal transcriptional function of HIF-1α in glycolytic metabolism, its enrichment in RM-NPC, and the recent demonstration of the capacity of SBN to inhibit its expression in prostate cancer ^{22,32,44–47}, we hypothesized that SBN-mediated glycolytic inhibition would extend to HIF-1α expression in NPC cells. To verify our hypothesis, we took advantage of the constitutive induction of HIF-1α expression in EBV-immortalized C666–1 cells to measure the impact of SBN treatment on its expression ^30,48. Flow cytometry analysis 24h post treatment showed slight but significant reductions in HIF-1α expression in C666–1 cells exposed to SBN compared to controls (p=0.006) (Fig.5A). In order to confirm these results in primary tumor biopsies, we analyzed changes in HIF-1α expression in treated patients’ explants by IHC (n=6; 100µM). An H-score evaluation determined that SBN induced in average a 42.89% reduction in HIF-1α expression in treated compared to control tumor biopsies (p=0.07) (Fig. 5B). We therefore concluded that SBN would interfere with NPC cell survival by inhibiting HIF-1α expression and LDH-A dependent glycolytic activity.

Fig.5. SBN decreases HIF-1α expression in NPC tumors.

C666–1 cells and NPC biopsies (n=6) were treated with SBN (0–100 M) for 24h. A) HIF-1α expression in C666–1 cells by flow cytometry in one representative of 2 independent experiments performed in triplicate. MFI, median fluorescence intensity. B) Representative IHC staining of HIF-1α (x20 and x40 magnification) after SBN treatment (0–100 M) and quantification using H score. Differences between groups were analyzed using t test. ** p=0.006.

SBN down-regulated expression of the immune checkpoint blockade PD-L1 concomitantly with HIF-1α and lactate synthesis

A growing body of evidence showed over the past decade that HIF-1α would act in a variety of immune and tumor cells, as a transcription factor of PD-L1, to mediate immune escape in lactate enriched microenvironments ^49,50. Considering the prognostic value of PD-L1 in RM-NPC patients ^25,40,51, we next thought to monitor the effect of loss of HIF1-α expression on lactate synthesis and PD-L1 expression in SBN treated C666–1 cells and primary tumors (n=6). To this aim, flow cytometry and IHC analyses were performed to monitor PD-L1 expression on the above cited systems. Of note, we observed that SBN significantly reduced PD-L1 expression (mean fluorescence intensity) in the tested cell line (p=0.02) (Fig.6A). Likewise, PD-L1 expression was significantly down-regulated in NPC treated tumor tissues compared to controls (p=0.03) (Fig.6B). Despite the limited number of analyzed tumors which did not allow us to analyze the correlation linking the expression of the analyzed molecules, we observed using a heatmap analysis, a consistency in the inhibitory activity of SBN on HIF-1α, PD-L1 and lactate synthesis in the analyzed systems (Fig.6C).

Fig.6. SBN decreases HIF-1α and lactate associated PD-L1 expression in NPC tumors.

C666–1 cells and NPC biopsies (n=6) were treated with SBN (0–100 M) for 24h. A) PD-L1 expression in C666–1 cells by flow cytometry in one representative of 2 independent experiments performed in triplicate. MFI, median fluorescence intensity. B) Representative IHC staining of PD-L1 (x20 and x40 magnification) after SBN treatment (0–100 M) and quantification using H score. C) Heatmap shows SBN effects on the relative expression levels of HIF-1α, lactate and PD-L1 in C666–1 cells treated with SBN 100 µM. Data are representative of 3 independent experiments. Differences between groups were analyzed using t test. * p=0.02 (A), * p=0.03 (B).

Discussion

Immune checkpoint blockade therapy constitutes to date one of the most recent innovative therapeutic approaches which allows restoration a potent anti-tumor immune response able to overcome tumor resistance to therapy. A growing body of evidence suggests that therapeutic failure associating with PD-1/PD-L1 pathway signaling in cancer patients, would mostly rely on metabolic rewiring in the context of a lactate-rich tumor microenvironment⁵⁰. Recent independent demonstrations showed that SBN can negatively modulate expression of several metabolic mediators like lactate, LDH-A and HIF-1α which support PD-L1 expression^32,49,50,52. However, these processes have remained unexplored in NPC.

In this study, we tested the efficacy of SBN to interfere with tumor glycolysis and PD-L1 overexpression and the possible consequences on tumor cell survival in human NPC.

Our observation that SBN exerts a constraint on NPC survival is in agreement with a number of studies conducted on a variety of cancers (skin, bladder, lung, breast, prostate and kidney carcinomas) and thereby further extends the spectrum of application of the compound to head and neck tumors ³¹.

To date, the anti-tumoral activity of SBN has been associated with alteration of several cellular pathways including those related to survival, angiogenesis and immune escape³¹. More recent studies have shown that the drug could potently interfere with key tumor cells activities to mediate metabolic reprogramming³². Our observation, that SBN may directly target oncogenic EBV driven glucose uptake and induce intracellular metabolic reprogramming is in line with that understanding. The precise mechanism behind SBN induced glycolysis inhibition remains to be explored. For example, it would be important to clarify if the observed effect results from direct targeting of GLUT-1 expression or viral oncogene LMP-1, which regulates GLUT-1 expression^29,52,53.

The fact that reductions in extracellular acidification accompany the inhibition of tumor cell proliferation and survival, prompted us to examine the impact of the in-vitro treatment on the enzymatic processes involved in glycolysis and lactate synthesis. The net reduction of the extracellular acidification rate was shown to be tightly linked to significant reductions in the LDH-A/lactate pathway. This result is in agreement with Shukla et al. earlier report showing that lactate synthesis reduction by SBN in pancreatic cells was linked to LDH-A expression inhibition⁵². Considering that tumors extrude the intracellular H+ and lactate, via monocarboxylate transporters 1 and 4 (MCT1/4) to avoid intracellular pH (pHi) increase and cell death, and the recent demonstration showing that SBN potently inhibits MCT1 expression^54,55, we hypothesize here that a prolonged exposure (over 48h) to the drug would also mobilize a similar process in NPC cells and lead to cell death as a consequence of intracellular acidosis.

The deregulation of glycolysis and LDH-A activity and lactate synthesis has been described to occur in several tumors. In NPC, these effects have been attributed to the signaling activity of the EBV oncogene LMP-1 ⁴⁸. Considering that LDH-A upregulation in EBV+ nasopharyngeal tumors has been associated with poor prognosis and that its signaling showed to be pivotal to tumor increased invasiveness in-vitro and recurrence in patients ^40,56, we tested the capacity of SBN to inhibit LDH-A function using the EBV+ NPC cell line C666–1 as a model system and primary NPC biopsies. Our observations that SBN exerts a significant inhibitory effect on LDH-A activity in NPC cells and tissue biopsies, is in line with the literature and further extend the spectrum of efficacy of the molecule to head and neck cancers ³³. It is likely, that this inhibition would result from an inhibition of LDH-A transcription, as previously demonstrated by Shukla S.K. et al.⁵²; this implies that SBN, can be envisioned as a candidate molecule which can interfere with LDH-A/lactate activity mediating head and neck tumor recurrence^24,56,57. To verify that possibility, further investigations are necessary to confirm the inhibition of LDH-A at a transcriptional level and to assess the impact on lactate suppression on the angiogenic and metastatic processes which support tumor recurrence.

The consistent ability of the treatment to increase citrate synthesis in NPC C666–1 cells can be interpreted as the consequence of pyruvate rerouting towards mitochondria and the TCA cycle. Considering the body of knowledge on the role of citrate in tumor development, indicating that citrate accumulation inhibits tumor growth through glucose metabolism and TCA cycle inhibition⁵⁸, further analyses should not only assess citrate accumulation in human tumors, but also determine the consequences of intracellular citrate increase, induced upon SBN treatment, on OXPHOS, mitochondrial functions and apoptosis ^36,59,60. Our observation that the key anti-apoptotic molecule Bcl-2 is downregulated in SBN treated biopsies supports the hypothesis that these effects may associate with to mitochondrial dysfunction.

In an effort to understand whether SBN activity on LDH-A was coupled with a potential effect on HIF1-α, we next verified the impact of SBN treatment on HIF-1α expression using the NPC cell line C666.1 and patients’ NPC biopsies where the hypoxia factor is constitutively induced by LMP-1 due to EBV latency program ^30,48. In line with Deep G. et al. study on prostate cancer³², we report here that SBN targets HIF1-α expression in the context of EBV induced NPC. Considering that this effect overlaps with a reduction in lactate synthesis, we propose that SBN would exert its suppressive activity by inhibiting the constitutive signaling pathway linking HIF1-α to LDH-A. Considering that such inhibitions may negatively regulate a number of critical angiogenic factors among which VEGF (vascular endothelial growth factor), next investigations should determine the implications of such an event on tumor vascularization and metastasis, and more importantly to tumors sensitization to therapy as HIF −1α inhibition may improve response to radiotherapy in NPC⁶¹.

With regards to the fact that HIF1-α has been incriminated in regulating PD-L1 expression – a major mediator a tumor escape from immune system destruction ⁴⁹ – we hypothesized that SBN targeting of HIF1-α would also lead to a down-regulation of PD-L1 in NPC tumors. To our surprise, not only expression of the immune checkpoint inhibitor was negatively affected by SBN, but we also found that, that event would most likely concur with an inhibition of HIF1-α and lactate synthesis. Our demonstration on the effect of SBN on PD-L1, is in line with the findings reported by Cuyàs et al. in lung carcinoma ³⁴. Despite the fact that further analyzes are necessary to understand the significance of this event, we speculate that LDH-A/lactate signaling inhibition would impair PD-L1 signaling and restore a functional anti-tumoral immune response, including those related to T and NK cell function ^59,62.

Overall, in this study we show that the antitumoral properties of SBN can be extended to the head and neck carcinoma through inhibition of HIF-1α and tumor glucose metabolism. In addition, we find that these effects are associated with reductions in tumor extracellular acidification and expression of the immune checkpoint inhibitor PD-L1, indicating the SBN may be of benefit to treat reccurent tumors relying on the se mechanisms. Considering the recent success met in targeting the PD-1/PD-L1 pathway in NPC, SBN utilization may constitute an accessible therapeutic agent for countries with low income to treat EBV associated head and neck carcinoma. Alternatively, this appoach may be attractive to reduce the costs associated with the development and use of an antibody. Despite the numerous questions which remain to be addressed, including the potential epigenetic modifications imprinted by the drug⁶³, the possible synergy of SBN with traditional chemotherapy and its impact on radiosensitization^64,65, we provide here important arguments in favor of SBN use as a therapeutic alternative for patients with relapsing or metastatic NPC tumor.

Highlights.

• SBN mediated NPC cell growth inhibition associates with a reduction of glycolysis and extracellular acidosis

• SBN treatment disrupts correlation between LDH-A and lactate in patients’ tumors

• SBN restores citrate synthesis at the expense of aerobic glycolysis in NCP cell line

• SBN interferes with NPC cell survival by interfering with HIF-1α expression and LDH-A signaling

• PD-L1 expression is significantly decreased in SBN treated NPC tumor tissues