Real-time metabolic profiling of oesophageal tumours reveals an altered metabolic phenotype to different oxygen tensions and to treatment with Pyrazinib

Oesophageal cancer is the 6th most common cause of cancer related death worldwide. The current standard of care for oesophageal adenocarcinoma (OAC) focuses on neoadjuvant therapy with chemoradiation or chemotherapy, however the 5-year survival rates remain at < 20%. To improve treatment outcomes it is critical to further investigate OAC tumour biology, metabolic phenotype and their metabolic adaptation to different oxygen tensions. In this study, by using human ex-vivo explants we demonstrated using real-time metabolic profiling that OAC tumour biopsies have a significantly higher oxygen consumption rate (OCR), a measure of oxidative phosphorylation compared to extracellular acidification rate (ECAR), a measure of glycolysis (p = 0.0004). Previously, we identified a small molecule compound, pyrazinib which enhanced radiosensitivity in OAC. Pyrazinib significantly inhibited OCR in OAC treatment-naïve biopsies (p = 0.0139). Furthermore, OAC biopsies can significantly adapt their metabolic rate in real-time to their environment. Under hypoxic conditions pyrazinib produced a significant reduction in both OCR (p = 0.0313) and ECAR in OAC treatment-naïve biopsies. The inflammatory secretome profile from OAC treatment-naïve biopsies is heterogeneous. OCR was positively correlated with three secreted factors in the tumour conditioned media: vascular endothelial factor A (VEGF-A), IL-1RA and thymic stromal lymphopoietin (TSLP). Pyrazinib significantly inhibited IL-1β secretion (p = 0.0377) and increased IL-3 (p = 0.0020) and IL-17B (p = 0.0181). Importantly, pyrazinib did not directly alter the expression of dendritic cell maturation markers or reduce T-cell viability or activation markers. We present a new method for profiling the metabolic rate of tumour biopsies in real-time and demonstrate the novel anti-metabolic and anti-inflammatory action of pyrazinib ex-vivo in OAC tumours, supporting previous findings in-vitro whereby pyrazinib significantly enhanced radiosensitivity in OAC.

www.nature.com/scientificreports/ time metabolic rate and clinical patient characteristics, OCR and ECAR were divided according to nodal status, tumour stage, stage of differentiation, body mass index, age at diagnosis and gender of the patients evaluated in this study. OCR and ECAR were shown to be independent of clinical patient characteristics, whereby there was no significant difference in the levels of OCR or ECAR when assessed according to patient characteristics (Supplemental Fig. 1, 2). All metabolic readouts were normalised to biopsy protein content following this assay; thus a limitation of this study was biopsy fragments directly adjacent to the fragment used in study had to be used to confirm the pathology of the tumour. In summary, OAC treatment-naïve biopsies have a significantly higher rate of OCR than ECAR and real-time metabolic rate is not significantly associated with clinical patient characteristics, which indicates that oxidative phosphorylation is an important metabolic pathway active across all n = 17 OAC tumours evaluated in this study, regardless of clinical patient characteristics. Pyrazinib (P3) significantly inhibited oxygen consumption rate in OAC treatment-naïve biopsies. Patient characteristics of the patient cohort evaluated in this study are outlined in Supplemental Table 1.
Having demonstrated in real-time that OAC treatment-naïve biopsies are metabolically active we sought to investigate the effect of our anti-metabolic agent, pyrazinib (P3), on real-time metabolic rates in OAC treatment-naïve biopsies. Pyrazinib (P3) treatment significantly inhibited OCR in OAC treatment-naïve biopsies (p = 0.0039) ( Fig. 2A). Pyrazinib (P3) induced a 35% reduction in OCR compared to the baseline OCR reading. Oligomycin, an ATP synthase inhibitor, was used as a positive control and resulted in a significant reduction in OCR (p = 0.0098) ( Fig. 2A). No significant change in metabolic rate was seen following treatment with the 0.1% dimethyl sulfoxide (DMSO) control from baseline OCR ( Fig. 2A). To determine if the reduction in OCR following treatment with pyrazinib (P3) was dependant on certain clinical patient characteristics, the percentage reduction in OCR was assessed according to the following clinical patient characteristics: nodal status, tumour stage, stage of differentiation and body mass index (Supplemental Fig. 3A,C,E,G). Importantly, the percentage reduction in OCR was independent of clinical patient characteristics suggesting that pyrazinib (P3) could function across our patient cohort. Regarding ECAR, treatment with pyrazinib (P3) did not significantly alter ECAR in OAC treatment-naïve biopsies (Fig. 2B). Percentage change in ECAR following treatment with pyrazinib (P3) was independent of patient characteristics: nodal status, tumour stage, stage of differentiation and body mass index (Supplemental Fig. 3B,D,F,H). Therefore, pyrazinib (P3) produced a significant reduction of oxygen consumption rate in OAC treatment-naïve biopsies and its function was independent of clinical patient characteristics.
Pyrazinib (P3) significantly inhibited real-time metabolic rate in OAC treatment-naïve biopsies cultured under hypoxia (0.5% O 2 ). Clinical patient characteristics of the patient cohort used in this study are outlined in Supplemental Table 2. Hypoxic tumours are inherently resistant to treatment thus we investigated both real-time metabolic rate and the action of pyrazinib (P3) under hypoxic conditions of 0.5% O 2 in OAC treatment-naïve patient biopsies from 7 male patients. To determine if OAC biopsies adapt their metabolic rate to changes in oxygen levels, we evaluated real-time basal metabolic rate under normoxia and realtime metabolic rate of the same biopsies again following culture in 0.5% O 2 for 6 h. Basal metabolic rate demonstrates OCR is significantly higher than ECAR under normoxic conditions (p = 0.0156) (Fig. 3A). In contrast to OAC biopsies cultured under normoxic conditions, following culture of OAC treatment-naïve biopsies under 0.5% O 2 , no significant differences were observed between OCR and ECAR (Fig. 3A). The ratio of OCR:ECAR was significantly higher in OAC treatment-naïve biopsies at baseline under normoxic conditions compared to the same biopsies cultured under hypoxia for 6 h (p = 0.0469) (Fig. 3B). Interestingly, the shift in metabolic Figure 1. Oxygen consumption rate is significantly higher than extracellular acidification rate in OAC treatment-naïve biopsies. OCR, a measure of oxidative phosphorylation and ECAR, a measure of glycolysis were assessed in real-time in OAC treatment-naïve biopsies using the Seahorse Biosciences XFe24 Analyser. (A) Basal OCR and ECAR rates in OAC pre-treatment biopsies, (n = 17). (B) OCR is significantly elevated in OAC treatment-naïve biopsies, (n = 17), Wilcoxon signed rank test, ***p < 0.001. (C) Relative metabolic ratio OCR:ECAR compared to ECAR:OCR in OAC treatment naïve biopsies (n = 17), paired t-test, **p < 0.01. Data expressed as + SEM. OAC treatment-naïve biopsies display a heterogeneous inflammatory secretome. The clinical patient characteristics for the patient cohort used in this study are outlined in Supplemental Table 3. In addition to an altered metabolic phenotype, inflammation has been reported to play a significant role in the progression and treatment response of OAC tumours, whereby elevated levels of pro-inflammatory mediators such as LIF, C3a, C4a and IL-1β have been associated with poor treatment response 8,9,11 . To profile the inflammatory secretions of OAC treatment-naïve tumours, OAC biopsies were cultured for 24 h and the secreted levels of 54 proteins in the tumour conditioned media (TCM) were evaluated by multiplex ELISA. The secreted levels of    Fig. 5E. Notably, there is a high level of variability of the secreted levels of the proteins detected in this screen, highlighting the high level of heterogeneity between OAC treatmentnaïve patient tumours. Furthermore, no significant correlations were seen when secreted inflammatory factors were divided according to clinical patient characteristics including tumour stage, nodal status, body mass index, stage of differentiation and age at diagnosis (data not shown).
Real-time metabolic profiles were significantly correlated with inflammatory secretions in OAC tumour biopsies. To investigate the relationship between real-time metabolic rate (OCR and ECAR) and the inflammatory protein secretions in OAC treatment-naïve biopsies, we correlated basal metabolic rate with inflammatory secretions in 10 matched patients, as per patient characteristics outlined in Supplemental Table 1. OCR was significantly positively correlated with ECAR, (r = 0.8505, p < 0.0001). In addition, OCR was significantly correlated with the secreted levels of 3 of 54 proteins in the TCM including Vascular endothelial growth factor A (VEGF-A) (r = 0.7091, p = 0.0268), interleukin-1 receptor antagonist (IL-1RA) (r = 0.7939, p = 0.0088) and TSLP (r = 0.6727, p = 0.0390) shown in Table 1 Pyrazinib (P3) significantly altered the secretion of IL-1β, IL-3 and IL-17B from OAC treatment-naïve biopsies. Inflammation drives development of OAC, however not all types of inflammation are detrimental to the host, e.g. T helper 1 (T H 1) profiles are associated with a good response to immunotherapeutic drugs 14 , whereas myeloid cell abundance in tumours is associated with worse survival 15 . To investigate if our anti-metabolic pyrazine compound pyrazinib (P3) affects altered inflammatory secretions ex vivo, we cultured OAC treatment-naïve biopsies with 10 µM pyrazinib (P3) or 0.1% DMSO for 24 h and compared the inflammatory secretions from the OAC treatment-naïve biopsies from the same patient. Of the 54 factors screened for in the multiplex ELISA, pyrazinib (P3) treatment significantly alters the secretions of 3 proteins: IL-1β, IL-3 and IL-17B. Treatment with pyrazinib (P3) significantly reduced the secretion of IL-1β from OAC www.nature.com/scientificreports/ treatment-naïve tumour biopsies (p = 0.0377) (Fig. 6A). In contrast, pyrazinib (P3) significantly increased the secretions of IL-3 (p = 0.0020) and IL-17B (p = 0.0181) in ex vivo OAC treatment-naïve biopsies (Fig. 6B,C).

Pyrazinib (P3) does not directly alter the expression of maturation markers on CD11c + dendritic cells.
In addition to the anti-metabolic and anti-inflammatory activity of pyrazinib (P3), it was critical to investigate its effect on immune cells, including dendritic cells which play a critical role in anti-tumoural immunity. Thus, we investigated whether pyrazinib (P3) altered the expression of dendritic cell maturation markers. Dendritic cells are professional antigen presenting cells which play a key role in orchestrating antitumour immune responses, via T cell polarisation and activation. In the development of a new compound, it is crucial to determine if such treatment will affect the function of important anti-tumour immune cells such as  www.nature.com/scientificreports/ dendritic cells, as this could hinder clinical potential. Patient information is outlined in Supplemental Table 3.   www.nature.com/scientificreports/ early apoptotic (Fig. 8B), late apoptotic (Fig. 8C) and necrotic cells (Fig. 8D). The percentage of live cells treated with the varying concentrations of pyrazinib (P3) at 24 h and 48 h incubation periods did not differ, with a maintained range of 94-98% Jurkat cell viability. There was some variance seen in the proportion of cells that underwent early apoptosis (Annexin V + only), but at very low levels (≤ 3%), therefore no significant difference in the percentage of cells that underwent early apoptosis. The percentage of cells that underwent necrosis (PI + only) remained at very low levels (≤ 3%) with no significant difference observed between samples. Late apoptosis (Annexin V + , PI + ) was detected in ≤ 2% of cells, and there was no significant difference between different treatment conditions. In summary, pyrazinib (P3) is not toxic to Jurkat T cells, at 0-10 μM concentrations after 24 h and 48 h treatment.

Discussion
This study highlights a novel method for measuring the real-time metabolic profiles of OAC treatment-naïve tumour biopsies which could be applied to multiple cancer types. Ex vivo, real-time metabolic profiling demonstrated that oxidative phosphorylation was significantly higher in OAC treatment-naïve biopsies compared to glycolysis. This supports previous findings by our group which have reported the importance of oxidative phosphorylation in OAC, and its previous association with radiation resistance 10 . The metabolic rate of OAC biopsies was shown to be independent of clinical patient characteristics. Whilst Warburg initially found cancer cells to be reliant on aerobic glycolysis, numerous studies also support the functional role of oxidative phosphorylation in tumorigenesis 1 . Real-time metabolic profiling of OAC biopsies supports previous findings at the in vitro level, which demonstrates that mitochondrial respiration is a predominant metabolic pathway used by OAC cancer cells 10 . Numerous studies have evaluated the mitochondrial function of tumour cells and reported that tumour cells predominantly have functional mitochondria which have retained the ability to carry out oxidative phosphorylation 16 .
Our novel small molecule compound pyrazinib (P3) significantly inhibited OCR in OAC treatment-naïve biopsies and no significant upregulation of ECAR occurred following treatment with pyrazinib (P3). Importantly, the activity of pyrazinib (P3) was independent of clinical patient characteristics, indicating that pyrazinib (P3) can maintain its anti-metabolic activity irrespective of patient's clinical characteristics such as tumour stage, nodal status, tumour differentiation or patient BMI. The significant effect of pyrazinib (P3) on OAC tumour oxidative phosphorylation rate is not only likely to have an anti-cancer effect given the prominent utility of oxidative phosphorylation in OAC tumours, it may also enhance the radioresponse of these tissues, as seen previously in vitro in an isogenic model of OAC radio-resistance 7 . Targeting oxidative phosphorylation has been reported in a number of studies as a novel mechanism to enhance radiosensitivity and reduce tumour growth 5,17 . A study by Benej et al. demonstrated that targeting mitochondrial respiration with the ergot alkaloid papaverine significantly inhibited mitochondrial respiration and enhanced tumour oxygenation and subsequently enhanced radiosensitivity in pulmonary adenocarcinoma cells 17 . An interesting study carried out by Bol et al. also demonstrated reprogramming of tumour mitochondria improves responses to radiation, a mitochondrial dysfunctional cell line which were exclusively glycolytic were found to be more radiosensitive than wild type oxidative phosphorylation proficient cells 18 . A novel in vivo model of oncogenic ablation-resistant pancreatic cancer cells which were responsible for tumour relapse were reported to depend on mitochondrial function for survival 5 . Targeting oxidative phosphorylation in this subpopulation of surviving pancreatic cells was shown to significantly decrease tumour spheroid growth 5 . Furthermore, oxidative phosphorylation is significantly upregulated in breast cancers deficient in RB1, a protein lost in 20-30% of basal-like breast cancers 19,20 . Tigecycline, a mitochondrial translational inhibitor attenuated growth of RB1-deficient breast tumours in vivo 20 . Taken together with the findings in this study, targeting oxidative phosphorylation in the neo-adjuvant setting could enhance radiosensitivity in OAC, but also oxidative phosphorylation could be targeted in the adjuvant setting to specifically target any remaining surviving cancer cells. Given the predominance of the oxidative phosphorylation   22 . Interestingly, this study found the levels of both malic acid and citric acid were significantly lower in more advanced SCC tumours when compared to early stage tumours which may be associated with a downregulation of the tricarboxylic cycle in late stage tumours 22 . Furthermore, a metabolomics study which utilised urinary samples from SCC patients and healthy controls demonstrated oesophageal SCC was associated with alterations in fatty acid β-oxidation and the metabolism of purines, amino acids, and pyrimidines 23 . These studies amongst others highlight the importance of metabolic alterations in oesophageal cancer compared to healthy controls and highlight the potential for biomarker development for disease diagnosis and progression 24,25 . It would be of interest to employ a similar approach in OAC treatmentnaïve tumour tissue across the various stages of disease progression to further elucidate the role of altered energy metabolism in OAC and compare the findings to real-time metabolic analysis. Hypoxia promotes the transformation of tumour cell metabolism from oxidative metabolism to anaerobic glycolysis, which protects tumour cells, promotes tumour growth and the development of treatment resistance in tumour stem cells 26 . We sought to investigate if OAC treatment-naïve biopsies adapt their metabolic profile to conditions of hypoxia and if pyrazinib (P3) could inhibit mitochondrial respiration under hypoxic conditions of 0.5% O 2 . Under normoxic conditions, OAC tumours had a significantly higher rate of OCR, but adapted their  www.nature.com/scientificreports/ metabolic rate to cope with hypoxic conditions, therefore there was no significant difference in OCR compared to ECAR. The OCR:ECAR ratio was significantly higher in OAC tumour under normoxia versus hypoxia and the ECAR:OCR ratio was significantly higher in hypoxic versus normoxic biopsies, highlighting the metabolic adaptation of OAC tumours to their environment. Pyrazinib (P3) significantly inhibited both oxidative phosphorylation and glycolysis. The ability of pyrazinib (P3) to inhibit both oxidative phosphorylation and glycolysis is a critical finding, which shows even in OAC tumours which can adapt their metabolic profiles to hypoxic conditions, pyrazinib (P3) can still inhibit both OCR and ECAR. In a study by Wang et al., in genetically modified macrophages overexpressing HIF-1α the OCR:ECAR ratio was dramatically decreased compared to non-HIF-1α overexpressing macrophages demonstrating a shift to glycolysis metabolism compared to mitochondrial oxidation in HIF-1α overexpressing macrophages 27 . Oxygen is a potent radiosensitiser and solid tumours with areas of hypoxia are the most aggressive and difficult tumours to treat 26 . A number of strategies which have attempted to increase oxygen delivery to the tumour have failed in the clinic largely due to the heterogenic nature of tumour vasculature 28 . In a pancreatic xenograft, the selective HIF-1α inhibitor PX-478 was found to potentiate the effect of fractioned chemoradiation therapy 29 . Targeting intra-tumoural oxygen consumption with compounds targeting oxidative phosphorylation may present a novel means to overcome tumour hypoxia and enhance anti-cancer activity in tumours where oxidative phosphorylation is upregulated, but also to improve treatment response rates in hypoxic treatment resistant tumours 17 . Notably, elevation of oxygen by as little as 2% is sufficient to produce oxygen enhancement 16 . Targeting oxidative phosphorylation in mammary tumours with papaverine was found to enhance hypoxic tumour oxygenation, sensitise tumours to radiation therapy and significantly reduce tumour growth 17 . Taken together with our current and our previous findings in vitro which demonstrated the anti-metabolic and radiosensitising activity of pyrazinib (P3) suggests pyrazinib (P3) has the potential to function as an anti-cancer agent in vivo 7 . Of note, the fresh patient samples used in our hypoxia metabolism study were all male, whilst OAC is a male dominant disease, previous studies have suggested a gender bias may exist in oesophageal cancer patients in relation to treatment response, thus it would be of importance to address the influence of gender on hypoxia metabolism in a much larger prospective study across multiple sites 30 . Tumour metabolism is tightly linked with both the local and systemic inflammatory response. OAC is an inflammatory driven upper gastrointestinal cancer 11 , thus we sought to characterise the inflammatory profile of the tumour conditioned media from OAC treatment-naïve biopsies. A multiplex ELISA demonstrated the heterogeneity of secreted factors from OAC biopsies including inflammatory, angiogenic and vascular injury, chemokine, cytokine and T H 17 related proteins. To investigate a potential relationship between metabolic rate and the OAC inflammatory secretion profile, we correlated protein secretions with baseline real-time metabolic rate in matched patient samples. OCR was significantly correlated with ECAR in all patients at baseline. Both OCR and ECAR were significantly positively correlated with the secretion of VEGF-A, IL-1RA and TSLP in OAC treatment-naïve biopsies. In addition, ECAR was positively correlated with IL-13, MIP-3α and TNF-α. Oxidative phosphorylation and glycolysis are known to be influenced by systemic inflammation thus it is not surprising that metabolic rate correlated with a number of inflammatory mediators 31 . The significant correlation between VEGF-A secretion and OCR and ECAR highlights the tight links which exist between the two biological processes of angiogenesis and metabolism, whereby there are elevated levels of the angiogenic mediators VEGF-A and TSLP in tumours with higher levels of oxidative phosphorylation 32,33 . TNF-α was only significantly correlated with ECAR and not OCR in OAC treatment-naïve biopsies. TNF-α was previously shown to induce aerobic metabolism in prostate epithelial cells and glycolytic reliance in mammary carcinoma cells 34,35 . In addition, in a previous study, MIP-3α was shown to be significantly correlated with the levels of HIF-1α, a mediator of glycolytic induction, in Barrett's oesophagus tissue 36 .
Treatment of OAC treatment-naïve biopsies with pyrazinib (P3) significantly inhibited IL-1β secretion and increased IL-3 and IL-17B secretion. The significant reduction of IL-1β secretion following pyrazinib (P3) treatment is a critical finding which may contribute to pyrazinib's (P3) anti-cancer effect. Previous work by our department found elevated levels of IL-1β in tumour samples compared to squamous epithelium from the same patients, and IL-1β levels were significantly decreased in the TCM generated from post treatment biopsies, compared to the TCM generated from pre-treatment biopsies in matched patients who achieved a complete pathological response to neoCRT 9 . In addition, IL-1β was previously shown to be significantly correlated with clinical outcome in oesophageal SCC, whereby patients with IL-1β positive tumours had a poor response to treatment compared to patients with IL-1β negative tumours, the suggested underlying mechanism of this difference in tumour response was increased epithelial-mesenchymal transition aggressive tumour growth in IL-1β-positive tumours 37 . Inhibition of IL-1β was shown to attenuate tumour growth and invasion and ameliorate treatment resistance 37 . Furthermore, in a study by Deans et al., tumoural IL-1β expression levels were significantly correlated with systemic inflammation as measured by C-reactive proteins levels, which is a marker of reduced survival in oesophagogastric cancer patients 38 . In an in vivo melanoma model, IL-1β inhibition was shown to stably reduce tumour growth by limiting inflammation and inducing the maturation of immature myeloid cells into M1 macrophages. Furthermore, in an in vitro model of pancreatic chemoresistance, administration of an IL-1 receptor blocking antibody, as a means of targeting IL-1β signalling, reduced NF-κB activation and the acquisition of chemoresistance in these cells 39 . Reports from the literature suggest the significant inhibition of IL-1β in response to pyrazinib (P3) is a positive effect which may contribute to the anti-cancer activity of this drug in addition to its effects on oxidative phosphorylation in vitro and ex vivo and radiosensitivity in vitro. IL-3 has been reported to exert paradoxical effects in cancer including pro-tumourigenic as well as anti-tumourigenic cellular responses 40,41 . Importantly, IL-3 has been reported to play an important role in anti-tumoural immunity. IL-3 is able to enhance antigen presentation by dendritic cells and activate macrophages to increase the expression of class II MHC molecules and IL-1 42 . Elevated gene expression of IL-3 in fibrosarcoma xenografts (FSA-JmIL-3 tumours) was associated with enhanced response to radiation compared to parental tumours, where FSA-JmIL-3 tumours were associated with increased lymphocyte infiltration and elicited immune responses 41  www.nature.com/scientificreports/ that the enhanced secretion of IL-3 following pyrazinib (P3) treatment is a positive effect of this small molecule compound. Dendritic cells are professional antigen presenting cells which are responsible for induction of antigen specific T cell responses, thus it is critical that function of dendritic cells remains intact even in the presence of our small molecule compound pyrazinib (P3). In this study, we investigated both the effect of the secretions from the TCM and pyrazinib (P3) on the expression of dendritic cell maturation markers. Increased expression of several cell surface markers including CD54, PD-L1, CD40, CD83, and HLA-DR on dendritic cells is associated with dendritic cell maturation and T cell activating ability 43 . Direct treatment with pyrazinib (P3) showed no effect on CD83, CD54, PD-L1, CD40 and HLA-DR expression in response to LPS, whereas both control and pyrazinib (P3) treated TCM significantly reduced the expression of CD83, suggesting that mediators secreted from the tumour microenvironment specifically exert an inhibitory effect on dendritic cell maturation. Pyrazinib (P3) treatment does not negatively affect the expression of dendritic cell maturation markers, indicating the function of these cells remains intact even in presence of this compound. This is a critical finding, as a previous study by our group demonstrated that both control and bevacizumab treated colorectal conditioned media significantly inhibited LPS-induced maturation and function of dendritic cells 43 . The adaptive immune system is associated with tumour control and elimination, particularly the T H 1 phenotype 14,44,45 . Pyrazinib (P3) did not significantly alter the viability of a Jurkat T cell line. This is an important finding, which highlights the potential to use pyrazinib (P3) within the clinical setting because it does not deplete or kill T cells. We also examined the effect of pyrazinib on Jurkat T cell activation status, using pre-activated Jurkats. Pyrazinib (P3) has been previously shown to enhance radiosensitivity in vitro and a study by Voos et al. demonstrated that radiation doses of ≥ 2 Gy activate Jurkat T cells and stimulates pro-inflammatory immune responses, through upregulation of IL-2, IFN-γ and CD25 surface expression 46 . Importantly, pyrazinib (P3) did not affect expression of T cell activation markers by activated or unactivated Jurkat cells. CD8 + T cells are more susceptible to becoming exhausted upon constitutive activation than CD4 + T cells 47 and one of the limitations to this study is the use of Jurkat T cells in this in vitro setting. Further research is required in relation to this study, both in patient-derived PBMCs and in the in vivo setting at multiple timepoints to gain a better understanding of the effect pyrazinib (P3) may have on other immune cells within the tumour microenvironment.
In summary, we report a new method for profiling the metabolic rate of human OAC tumour biopsies in real time, highlighting the importance of the oxidative phosphorylation pathway in OAC tumours, and that these tumours can adapt their metabolic profiles in line with changes in oxygen tension. We have demonstrated the novel anti-metabolic and anti-inflammatory action of pyrazinib (P3) in ex vivo OAC treatment-naïve biopsies, in addition to its radiosensitising properties. It will be critical to further evaluate the anti-cancer potential of pyrazinib (P3) now in a murine model of OAC. written informed consent, diagnostic biopsy specimens were taken from OAC patients being treated with curative intent, by a qualified endoscopist, prior to neo-adjuvant therapy. Histologic confirmation of tumour tissue in biopsies was performed by a pathologist using routine haematoxylin and eosin staining. All patient tumour tissue used in this study was taken prior to the initiation of neo-adjuvant treatment (treatment-naïve tissues). All experimental protocols were approved by the joint St James's Hospital/AMNCH ethical review board and carried out in accordance with the relevant guidelines of the joint St James's Hospital/AMNCH ethical review board.

Methods
Real-time metabolic profiling of OAC tumour biopsies. Three biopsies per patient were collected at endoscopy, immediately placed on saline-soaked gauze and were transported to the laboratory within 10 min. Each biopsy was placed into a separate well of a 24 well XF24 Islet Capture Microplate (Agilent Technologies, Santa Clara, CA, USA) containing 1 mL M199 medium (Gibco) supplemented with 10% FBS (Gibco), 1 μg/mL insulin (Sigma) and 1% penicillin/streptomycin (Gibco). 1 mL of complete M199 was placed in four background control wells. A XF24 capture screen was placed over each biopsy to prevent the biopsy touching utility plate probes during assay. The XF24 microplate was placed at 37 °C for 30 min to allow biopsies to equilibrate. Three baseline measurements of OCR and ECAR were taken over 24 min consisting of three repeats of mix (3 min), wait (2 min), measurement (3 min) to establish basal respiration, using a Seahorse Biosciences XFe24 analyser (Agilent Technologies, Santa Clara, CA, USA). Basal respiration of each patient was established by taking the average OCR and ECAR readout from the three individual biopsies obtained from same patient. Following basal metabolic profiling of biopsies, capture screens were removed and biopsies and corresponding media were transferred to a new XF24 islet capture microplate and treated with one of the following; 0.1% DMSO (control), 6 µM of oligomycin (positive control) or 10 µM of pyrazinib (P3). Following treatment biopsies were cultured for 24 h at 37 °C in 5% CO 2 /95% air. Following 24 h culture, a capture screen was placed on each biopsy and three basal measurements of OCR and ECAR were taken over 24 min consisting of three repeats of mix (3 min), wait (2 min), measurement (3 min) to establish the effect of drug treatment with our novel small molecule pyrazinib (P3) on OCR and ECAR. The effect of treatment was determined as the percentage change in metabolic rate readout from the baseline reading of each individual biopsy to the reading following treatment of that individual biopsy. The metabolic rate of each biopsy was normalised to tumour protein content using the BCA assay Scientific RepoRtS | (2020) 10:12105 | https://doi.org/10.1038/s41598-020-68777-7 www.nature.com/scientificreports/ (Pierce) and tumour biopsies were snap frozen and stored at − 80 °C. Tumour conditioned media (TCM) was collected and stored at − 80 °C.

Real-time metabolic profiling of OAC tumour biopsies cultured under hypoxic conditions.
Basal metabolic rate was determined as described as stated above. Following evaluation of basal OCR and ECAR using the Seahorse Biosciences XFe24 analyser; biopsies and corresponding media were transferred to a new XF24 islet capture microplate and cultured in the Whitley H35 hypoxystation (Don Whitley Scientific) at 0.5% O 2 at 37 °C in 5% CO 2 for 6 h. Following 6 h culture, capture screens were placed over each biopsy in each well and the plate was transferred to Whitley i2 workstation containing the XFe24 Seahorse analyser maintained at 0.5% O 2. Real-time OCR and ECAR were assessed under 0.5% O 2 , three measurements of OCR and ECAR were taken over 24 min consisting of three repeats of mix (3 min), wait (2 min) and measurement (3 min), to establish the effect of a 6 h hypoxia culture on real-time metabolic rate in OAC patient biopsies. Following metabolic profiling of biopsies, capture screens were removed and biopsies and corresponding media were transferred to a new XF24 islet capture microplate and treated with one of the following; 0.1% DMSO (control), 6 µM of oligomycin (positive control) or 10 µM of pyrazinib (P3) for 14 h under 0.5% O 2 at 37 °C in 5% CO 2. Following 14 h treatment, real-time OCR and ECAR measurements were taken over 24 min consisting of three repeats of mix (3 min), wait (2 min), measurement (3 min) whilst the Seahorse Biosciences XFe24 analyser was maintained under 0.5% O 2 to establish the effect of treatment on metabolic rate of OAC biopsies maintained under 0.5% O 2. Following metabolic profiling, capture screens were removed, biopsies were snap frozen and both biopsies and TCM were stored at − 80 °C. The metabolic rate of each biopsy was normalised to tumour protein content using the BCA assay (Pierce).