Chemotherapy regimens induce inhibitory immune checkpoint protein expression on stem-like and senescent-like oesophageal adenocarcinoma cells

Highlights • OAC cells express several inhibitory immune checkpoint (IC) ligands and receptors.• Chemotherapy upregulates IC ligands and receptors on the surface of OAC cells.• ICs are enriched on stem-like and senescent OAC cells following chemotherapy.• PD-1 blockade induced apoptosis and enhanced chemotherapy toxicity in OAC cells.


Introduction
Oesophageal cancer (OC) is one of the deadliest cancers globally, ranked the third most common cancer of the gastrointestinal tract and sixth most commonly diagnosed cancer worldwide [1] . Oesophageal adenocarcinoma (OAC) affects the distal third of the oesophagus and is the predominant subtype of OC in Western countries and its incidence is rapidly increasing in this region [1] .
Treatment options for OAC patients include the FLOT or MAGIC combination chemotherapy regimens before surgery (neoadjuvant) and after surgery (adjuvant) [2] . The FLOT chemotherapy regimen includes 3 different classes of drugs with distinct mechanisms of action, the antimetabolite 5-fluorouracil (5-FU), a platinum-based DNA intercalator ox-OAC patients that achieve a complete pathological response after receiving neoadjuvant chemotherapy or chemoradiation have a significantly increased overall survival [7 , 8] . However, ~70% of OAC patients don't achieve a complete pathological response and are subjected to treatment associated toxicities without any apparent therapeutic benefit [7 , 8] . This highlights the need to develop better treatment strategies to enhance the efficacy of these regimens.
Interestingly, the central dogma that T cells and tumour cells exclusively express IC receptors and IC ligands respectively, no longer holds true [10] . Several studies have now shown that IC receptors are expressed on the surface of a subpopulation of cancer cells in a range of malignancies including OAC [11] . Emerging studies have shown that melanoma and breast cancer stem-like cells were enriched for the expression of PD-1 and PD-L1 [12] , respectively. Recent studies have demonstrated that activation of both inhibitory IC ligands and receptors including PD-L1, PD-L2, TIM-3 and PD-1 on the surface of cancer cells promoted various immune-independent hallmarks of cancer such as an altered metabolism, proliferation, invasion and metastasis, DNA repair and chemoresistance [13][14][15][16][17][18][19][20] , and immune checkpoint blockade (ICB) suppressed various hallmarks of cancer including invasion, chemoresistance, proliferation, glycolysis and DNA repair [13][14][15][16][17][18][19][20] .
This study profiles the expression of a wide range of inhibitory IC ligands and receptors on the surface of OAC cells in vitro and in ex vivo biopsies to identify potential therapeutic targets. The effect of clinically relevant combination chemotherapies (FLOT, CROSS chemotherapy (CROSS CT) and MAGIC) on the expression of ICs on OAC was assessed in vitro . Additionally, we also investigated the phenotype of OAC cells expressing ICs, specifically the stem-like and senescent-like properties of OAC cells expressing IC ligands and receptors. In light of the emerging studies highlighting novel immune-independent functions for ICs in cancer cells, we also investigated if blockade of PD-1 signalling in OAC cells reduces the viability of OAC cells and enhances chemotherapy toxicity in vitro .

Acquisition of human OAC tumour tissue biopsies
All patients involved in this study were enrolled from 2018-2020. Treatment-naïve tumour tissue biopsies were obtained from OAC patients undergoing endoscopy at St. James's Hospital ( n = 10) at time of diagnosis prior to initiation of chemotherapy or radiotherapy. Post-FLOT chemotherapy-treated ( n = 5) and post-CROSS chemoradiotherapy-treated OAC ( n = 5) tumour tissue biopsies were obtained at time of surgical tumour resection. The group consisted of 11 males and 5 females, with an average age of 63.6 years. The patient demographics are detailed in Table S1. The work was performed in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving human samples. Patients provided informed consent for sample and data acquisition, and the study received full ethical approval from the St. James's Hospital/AMNCH Ethical Review Board. Patient samples were pseudonymised to protect the privacy rights of the patients.

OAC tumour tissue digestion
Biopsies were enzymatically digested to perform OAC cell phenotyping. Briefly, tissue was minced using a scalpel and digested in collagenase solution (2 mg/ml of collagenase type IV (Sigma) in Hanks Balanced Salt Solution (GE healthcare) supplemented with 4% (v/v) foetal bovine serum) at 37 °C and 1500 rpm on an orbital shaker. Tissue was filtered and washed with FACs buffer (PBS containing 1% foetal bovine serum and 0.01% sodium azide). Cells were then stained for flow cytometry.

Cell culture of OAC cell lines
Human Het1A cells, OE33 cells and SK-GT-4 cells were purchased from European Collection of Cell Cultures. Het1A cells were grown in flasks precoated with a mixture of fibronectin (0.01 mg/ml, Merck, Germany), collagen type I (0.03 mg/ml, Corning, USA) and bovine serum albumin (0.01 mg/ml, Sigma, USA) dissolved in PBS and cultured in serum free Bronchia/Trachea Epithelial Cell Growth Medium (511-500, Merck, Germany). Het1A cells were detached from flask using accutase (Sigma, USA) diluted 1:15 in PBS. OE33 and SK-GT-4 cells were grown in RPMI 1640 medium with 2 mM L-glutamine (Gibco) and supplemented with 1% (v/v) penicillin-streptomycin (50 U/ml penicillin 100 g/ml streptomycin) and 10% (v/v) foetal bovine serum (Gibco). All cell lines were maintained in a humidified chamber at 37 °C 5% CO 2 and were tested regularly to ensure mycoplasma negativity.

Cell viability CCK-8 assay
A CCK-8 assay (Sigma, USA) was used to assess the effect of single agent 5-FU, oxaliplatin, docetaxel, epirubicin, cisplatin, carboplatin and paclitaxel on the viability of OE33 and SK-GT-4 cells and combination chemotherapy regimens FLOT, CROSS CT and MAGIC. Additionally, the effect of nivolumab, atezolizumab or dual nivolumab-atezolizumab in the absence or presence of FLOT and CROSS CT regimens on the viability of OE33 and SK-GT-4 cells was investigated. 5 × 10 3 OAC cells were adhered in a 96 well plate at 37 °C, 5% CO2 overnight. Cells were treated with increasing doses of either single agent chemotherapy or a combination of chemotherapies including FLOT (5-FU, oxaliplatin, docetaxel), CROSS CT (carboplatin and paclitaxel) or MAGIC (epirubicin, cisplatin and 5-FU) at a range of inhibitory concentrations (IC 10 , IC 25 , IC 50 ) (Table S2 and Table S3) for 48 h. The doubling time of OE33 and SK-GT-4 cells is 33 and 36 h, respectively, therefore, a 48 h time point allowed sufficient time for the chemotherapies to be incorporated into the cells and induce cytotoxicity. Additionally, cells were treated with and without FLOT or CROSS CT regimen in the absence or presence of single agent atezolizumab (10 ug/ml), nivolumab (10 ug/ml) or dual atezolizumab (10 ug/ml) and nivolumab (10 ug/ml). An IC 25 dose of each chemotherapy in the FLOT and CROSS CT regimen was used to treat OE33 cells. An IC 50 and IC 25 dose of each chemotherapy in the FLOT and CROSS CT regimens respectively, were used to treat the SK-GT-4 cells as shown in Table S3. 5 l of CCK-8 solution was added to each well, followed by a 1.5 h incubation in the dark at 37 °C, 5% CO2. The optical density at 450 nm and 650 nm (reference wavelength) was measured using the Versa Max microplate reader (Molecular Devices, Sunnyvale, CA, USA) to determine a viable cell number. All of the data were analysed from three independent experiments.

Aldehyde dehydrogenase (ALDH) assay
Aldehyde dehydrogenase (ALDH) enzyme activity was assessed using the Aldefluor® assay (Stem Cell Technologies), according to the manufacturer's instructions. Briefly, cells were trypsinised and resuspended at a density of 1 × 10 6 cells/mL in Aldefluor® assay buffer containing ALDH substrate (bodipy-aminoacetaldehyde) (5 L/mL). Immediately following this, half of the resuspended cells were added to a tube containing the ALDH inhibitor diethylaminobenzaldehyde (DEAB) to provide a negative control. Cells were acquired using BD FACs CANTO II (BD Biosciences) using Diva software and analysed using FlowJo v10 software (TreeStar Inc.).

Statistical analysis
Data were analysed using GraphPad Prism (GraphPad Prism, San Diego, CA, USA) software and was expressed as mean ± SEM. Statistical differences between two treatments in a particular cell line were analysed using a paired parametric Student's t -test. To compare the statistical differences between two different cell lines an unpaired parametric t -test was conducted. To compare differences between patients and different treatments an unpaired non-parametric t -test was conducted. Statistical significance was determined as p ≤ 0.05. To determine if IC expression on OAC cells in the treatment-naïve setting and post-treatment setting correlated with patient clinical features, Spearman correlations were performed to analyse correlation data.

A subpopulation of normal oesophageal epithelial cells and OAC cells express inhibitory IC ligands and inhibitory IC receptors in vitro
Emerging studies have shown that cancer cells not only express inhibitory IC ligands but also inhibitory IC receptors including PD-1, TIGIT and TIM-3 in OAC [11] , melanoma [22] , pancreatic [15] , gastric [18 , 20] cervical [23] , lung [16 , 24] , ovarian [25] , endometrial [25] and colorectal cancer [26] . In addition, IC receptor TIM-3 has also been identified on normal gastric epithelial cells as well as normal cervical epithelial cells. Therefore, we screened normal oesophageal epithelial (Het1A cells) and OAC cells for the expression of both inhibitory IC ligands and receptors. OE33 and SK-GT-4 OAC cell lines used in this study were established from tumours that had progressed from the pre-malignant Barrett's oesophagus condition to OAC. While the OE33 cells are poorly differentiated, the SK-GT-4 cells are well differentiated, which helps to encapsulate the heterogeneous nature of OAC.

Chemotherapy promotes a more immune-resistant phenotype through upregulation of inhibitory IC ligands on OAC cells in vitro
Previous studies have shown that single agent cisplatin and 5-FU increase PD-L1 on the surface of lung cancer [27] and OAC cells [28] in vitro respectively, which may drive immune evasion and resistance to chemotherapy. Therefore, we sought to investigate the effect of clinically-relevant single agent chemotherapies on the expression of a range of inhibitory IC ligands (PD-L1, PD-L2 and CD160) on the surface of OE33 cells by flow cytometry. Importantly, we also assessed the effect of the combination chemotherapy regimens FLOT, CROSS CT and MAGIC on the expression of a range of inhibitory IC ligands on the surface of OE33 and SK-GT-4 cells by flow cytometry. For single agent chemotherapies an IC 50 dose was used. To obtain IC 50 doses for combination FLOT, CROSS CT and MAGIC regimens; OE33 and SK-GT-4 cells were treated with increasing doses of single agent chemotherapies that comprise the FLOT, CROSS CT and MAGIC regimens to identify IC 10 , IC 25   S2.). Subsequently, OE33 and SK-GT-4 cells were then treated with the combination regimens for 48 h using an IC 10 , IC 25 or IC 50 dose of each drug to obtain an inhibitory concentration of combination doses which killed 50% of the cells (Fig. S3.). These combination doses that reduced the viability of OE33 and SK-GT-4 cells by 50% were used for all the experiments in this study. Single agent 5-FU, oxaliplatin and docetaxel significantly increased the percentage of OE33 cells expressing PD-L1 (untreated 1.31 ± 0.5%, 5-FU 26.35 ± 6.4% (p = 0.01), oxaliplatin: 13.59 ± 2.1% (p = 0.0007) and docetaxel: 3.40 ± 2.1% (p = 0.02)), ( Fig. 2 A.). The combination FLOT regimen also significantly increased PD-L1 expression on OE33 cells (1.16 ± 0.2% vs. 14.08 ± 3.5%, (p = 0.01)) and SK-GT-4 cells (1.62 ± 0.3% vs. 43.0 ± 3.6%, ( p = 0.01)), ( Fig. 2 B.). Similarly, single agent 5-FU, oxaliplatin and docetaxel as well as the combined FLOT regimen significantly increased the percentage of OE33 and SK-GT-4 cells expressing PD-L2 ( Fig. 2 A. and Fig. 2 B.). Single agent carboplatin (but not paclitaxel or CROSS CT) significantly increased PD-L1 and PD-L2 expression on the surface of OE33 cells. However, CROSS CT significantly increased PD-L1 and PD-L2 on the surface of SK-GT-4 cells ( Fig. 2 A.). Only single agent carboplatin significantly increased CD160 on the surface of OE33 cells, as did FLOT and CROSS CT regimens. There was no significant changes observed for CD160 expression on the SK-GT-4 cell line ( Fig. 2 A.). Single agent cisplatin and 5-FU but not epirubicin significantly increased PD-L1 expression on OE33 cells (untreated 1.44 ± 0.6% vs. cisplatin 12.65 ± 4.6% (p = 0.04) and 5-FU 26.35 ± 6.4% ( p = 0.01)), ( Fig. 2 A.). Similar results were observed for PD-L2 on OE33 cells ( Fig. 2 A.). The combination MAGIC regimen had no effect on IC ligand expression on OE33 cells, however PD-L1 was significantly increased post-MAGIC treatment on the surface of SK-GT-4 cells (untreated 1.62 ± 0.3% vs. MAGIC 21.13 ± 3.3%, ( p = 0.03)), ( Fig. 2 B.).
Overall, 5-FU and the combination FLOT regimen had the greatest effect at upregulating inhibitory IC ligands, in particular PD-L1, on the surface of OAC cells in vitro .

Chemotherapy promotes a more immune-resistant phenotype through upregulation of inhibitory IC receptors on OAC cells in vitro
As we demonstrated that single agent chemotherapy and combination chemotherapy regimens significantly altered the expression profile of inhibitory IC ligands on the surface of OAC cells, we also assessed the effect of these chemotherapies on the expression profile of a panel of inhibitory IC receptors on OAC cells. Treatment with single agent chemotherapies or combination chemotherapy regimens did not significantly affect the percentage of OE33 cells expressing PD-1 or TIGIT ( Fig. 3 A.). However, FLOT and CROSS CT regimens significantly increased PD-1 expression on SK-GT-4 cells (untreated: 7.88 ± 2.7% vs. FLOT: 10.85 ± 3.2%, (p = 0.002) and CROSS CT: 10.30 ± 3.2%, (p = 0.01)), and MAGIC significantly increased the expression of TIGIT on the surface of SK-GT-4 cells (untreated: 6.77 ± 2.1% vs. MAGIC: 12.11 ± 1.9%, (p = 0.02)), ( Fig. 3 A.). Single agent 5-FU (but not oxaliplatin or docetaxel) and the combined FLOT regimen significantly increased the percentage of OE33 cells expressing TIM-3 (untreated: 2.6 ± 0.6% vs. 5-FU: 11.81 ± 2.9%, (p = 0.04)), and untreated: 0.80 ± 0.2% vs. FLOT: 8.10 ± 0.9% ( p = 0.0003)) ( Fig. 3 A. and Fig. 3 B.). Similarly, FLOT significantly increased the percentage of SK-GT-4 cells expressing TIM-3 ( Fig. 3 A. and 3 B). Single agent carboplatin (but not paclitaxel or CROSS CT) significantly increased the percentage of OE33 cells expressing TIM-3 ( Fig. 3 A. and 3 B). Similarly, CROSS CT significantly increased the percentage of SK-GT-4 cells expressing TIM-3 ( Fig. 3 A. and 3 B). The MAGIC regimen did not significantly affect the percentage of OE33 or SK-GT-4 cells expressing TIM-3, however, single agent cisplatin and 5-FU (but not epirubicin) significantly increased the percentage of OE33 cells expressing TIM-3 ( Fig. 3 A. and 3 B). The percentage of OE33 cells expressing LAG-3 was significantly increased following single agent 5-FU and docetaxel treatment (but not oxaliplatin) as well as combination FLOT treatment ( Fig. 3 A. and Fig. 3 B.). Furthermore, carboplatin (but not paclitaxel) significantly increased LAG-3 expression in OE33 cells, as did the CROSS CT regimen. Additionally, FLOT and CROSS significantly increased LAG-3 expression in the SK-GT-4 cells ( Fig. 3 A. and 3 B). MAGIC had no effect on the percentage of OE33 and SK-GT-4 cells expressing LAG-3, however, single agent cisplatin and 5-FU (but not epirubicin) significantly increased LAG-3 expression in OE33 cells ( Fig. 3 A. and 3 B). 5-FU and docetaxel (but not oxaliplatin) significantly increased the percentage of OE33 cells expressing A2aR, as did the combined FLOT regimen. Furthermore, single agent carboplatin (but not paclitaxel) significantly increased the percentage of OE33 cells expressing A2aR, as the CROSS CT regimen ( Fig. 3 A. and 3 B). While epirubicin significantly decreased the percentage of OE33 cells expressing A2aR and 5-FU significantly increased the expression of A2aR on OE33 cells, (single agent cisplatin or the combination MAGIC regimen had no significant effect on A2aR expression). Only the FLOT regimen significantly increased the percentage of SK-GT-4 cells expressing A2aR ( Fig. 3 A. and 3 B).
Similarly, in addition to single agent 5-FU and combination FLOT chemotherapy regimen having the greatest effect in upregulating IC ligand expression on OAC cells, 5-FU and FLOT also had the greatest effect

The FLOT chemotherapy regimen enhances a stem-like phenotype and preferentailly upregulates PD-L1 and TIM-3 on the surface of stem-like OAC cells in vitro
Our data suggests that chemotherapy regimens enhance a more immune resistant phenotype in OAC cells, demonstrated by the upregulation of inhibitory IC ligands and receptors on OAC cells following FLOT and CROSS CT treatment. Therefore, we sought to investigate if chemotherapy regimens could enhance a more stem-like phentoype in OAC cells, which are thought to be responsible for tumour recurrence and treatment resistance [29 , 30] . OE33 and SK-GT-4 cells were treated with FLOT, CROSS CT and MAGIC regimens for 48 h and the level of ALDH activity was measured, which is an established marker of stemness in OAC [31] ( Fig. 4 A.). Following FLOT treatment, ALDH activity in OE33 cells was significantly increased (untreated: 15.59 ± 3.4% vs. FLOT: 33.97 ± 4.6%, (p = 0.01) ( Fig. 4 A.). However, following CROSS CT the level of ALDH activity was significantly decreased in OE33 cells (untreated: 15.59 ± 3.4% vs. 7.42 ± 1.8%, (p = 0.02)) ( Fig. 4 A.). MAGIC did not significantly alter ALDH activity in OE33 cells ( Fig. 4 A.). Following FLOT treatment of SK-GT-4 cells, ALDH activity was significantly increased (untreated: 29.40 ± 9.9% vs. FLOT: 55.17 ± 4.9%, ( p = 0.02)) ( Fig. 4 A.). However, CROSS CT and MAGIC did not significantly alter ALDH activity in SK-GT-4 cells ( Fig. 4 A.).
Overall, the FLOT regimen enhanced a stem-like phenotype in both OE33 and SK-GT-4 cells demonstrated by an increase in ALDH activity. Inhibitory ICs PD-L1 and TIM-3 were enriched in the stem-like compartment of OAC cells following chemotherapy treatment. However, PD-1 and TIGIT were preferentialy expressed on non-stem-like OAC cells basally and following chemotherapy treatment.

The FLOT chemotherapy regimen enhances a senescent-like phenotype and upregulates A2aR on the surface of senescent-like OAC cells in vitro
Senescent cancer cells play an important role in conferring treatment resistance and in orchestrating a tumour-promoting milieu of immuosuppressive cells within the tumour microenvirnoment via the secretion of a range of pro-inflammatory markers known as the senescenceassociated secretory phentoype (SASP). Characterisation of senescent cells can be performed by assessing multiple markers such as an enlarged morphology, activation of p53-p21 and/or p16-Rb tumour suppressor pathways, induction of p16 INK4a , the presence of persistent DNA damage response, an increase in -galactosidae ( -gal + ) expression, and the appearance of senescent-associated distension of satellites and telomereassociated DNA damage foci [32 , 33] . We aimed to determine the effect of OAC chemotherapy regimens on the induction of a senescent-like phenotype in OAC cells using the well-established marker -gal + as a marker of senescent-like cells.
Given that FLOT, CROSS CT and MAGIC regimens induced the expression of the senescene marker -galactosidase in OAC cells in vitro we sought to investigate if ICs are expressed on these OAC cells and whether chemotherapy upregulates them further ( Fig. 5 B.). As ICs play an important role in immune evasion the expression of ICs on senescent OAC cells may represent a drug targetable mechanism of immune evasion. Following FLOT treatment the percentage of senescent OE33 cells expressing TIM-3 was significantly increased (untreated: 0.54 ± 0.1% vs. FLOT: 14.40 ± 1.9% ( p = 0.02)) ( Fig. 5 C.). Similarly, following FLOT treatment the percentage of OE33 cells expressing A2aR was significantly increased (untreated: 0.85 ± 0.1% vs. FLOT: 22.10 ± 3.0% ( p = 0.02)) ( Fig. 5 C.). There was no significant differences observed in the SK-GT-4 cell line ( Fig. 5 C.). We also found that TIGIT was expressed at significnatly higher levels on the surface of -gal + SK-GT-4 compared with -gal − SK-GT-4 cells, basally and following FLOT, CROSS CT and MAGIC treatment (but not OE33 cells). Overall, FLOT had the greatest effect at inducing the senescence marker -galactosidase in OAC cells and enriched -gal + OE33 cells for TIM-3 and A2aR expression.

A subpopulation of OAC cells expressing inhibitory IC ligands and inhibitory IC receptors were identified in OAC tumour tissue biopsies ex vivo
Following our in vitro studies demonstrating that OAC cells express a range of inhibitory IC ligands and receptors, which are upregulated following chemotherapy treatment on stem-like and senescent-like OAC cells, we screened treatment-naïve ( n = 10), post-FLOT chemotherapy ( n = 5) and post-CROSS chemoradiotherapy ( n = 5) OAC tumour biopsies for the expression of a range of inhibitory IC ligands and receptors ex vivo . OAC cells were characterised as CD45 − CD31 − as previously reported [21] .
Following identification of a subpopulation of OAC cells expressing inhibitory IC ligands and receptors in vitro and ex vivo, we investigated the correlative relationship between IC expression on patient OAC tumour cells and clinical patient characteristics and tumour features ( Table 1 ). We observed a strong correlation between IC expression on OAC cells with the expression of additional ICs in the treatmentnaïve and post-treatment setting of OAC patients. There was a strong and significant positive correlation ( r = 0.81) between the percentage of OAC tumour cells expressing LAG-3 and treatment response ( p = 0.03) in the treatment-naïve setting ( Table 1

Blockade of PD-1 axis intrinsic-signalling in OAC cells reduces the viability of OAC cells inducing apoptosis and enhancing chemotherapy toxicity in vitro
Nivolumab and atezolizumab are among the few ICIs approved for clinical use and have demonstrated some success in the clinic, therefore, we investigated whether blocking the PD-1 axis could reduce the viability of OAC cells and induce direct cytotoxicity in vitro . In this study we have demonstrated that chemotherapy regimens upregulate PD-1 and PD-L1 on OAC cells. Therefore, we investigated if combining nivolumab, atezolizumab or dual nivolumab-atezolizumab treatment combined with FLOT or CROSS CT regimens could enhance cytotoxicity in OE33 cells and SK-GT-4 cells following 48 h treatment.
Overall, we found that blockade of the PD-1 axis significantly reduced OAC cell viability and enhanced apoptosis induction, particularly enhancing the toxicity of the FLOT chemotherapy regimen in vitro.
The key findings from our study are summarised in Fig 8 .

Discussion
We demonstrated that the FLOT chemotherapy regimen enhanced a more stem-like and senescent-like phenotype in OAC cells and perhaps may be selecting for a more therapeutically resistant phenotype. Interestingly, the CROSS CT regimen significantly reduced ALDH activity in OE33 cells. This may suggest that the chemotherapies com- Fig. 8. Graphical summary of the key findings from the study. Each section of the figure corresponds to the key findings of this study. Section 3.1 displays the relative expression levels of ICs on Het1A, OE33 and SK-GT-4 cells. Section 3.2 summarises the relative effect of single agent chemotherapies and combination chemotherapy regimens on IC expression. A table showcasing the breakdown of the relative effect of single agent chemotherapies and their combinations on IC expression is also included. Section 3.3 highlights the effect of chemotherapy regimens on a stem-like phenotype in OAC cells and highlights that FLOT preferentially upregulates PD-L1 and TIM-3 on the surface of stem-like OAC cells in vitro . Section 3.4 demonstrates that combination chemotherapy enhance a senescent-like phenotype in OE33 cells and that FLOT preferentially upregulates A2aR on the surface of OAC cells in vitro . Section 3.5 highlights the expression levels of ICs on OAC cells in tumour tissue and demonstrates that post-FLOT TIM-3, LAG-3 and A2aR IC receptors are significantly decreased on OAC cells (characterised as CD45 − CD31 − ) ex vivo . Section 3.6 shows that treatment with nivolumab, atezolizumab or dual nivolumab-atezolizumab enhances the toxicity of the FLOT regimen in vitro .
prising the CROSS CT regimen may be attenuating a stem-like phentoype in certain OACs. Our study demonstrates that FLOT, CROSS CT and MAGIC chemotherapy regimens enhance a senescent-like phenotype in OAC cells in vitro, whereby again we saw that FLOT substantially enhanced a senescent-like phenotype. Chemotherapy regimens are cytotoxic to tumour cells and often promote immunogenic cell death, this study highlights the double-edged sword of chemotherapy, where we observed that the FLOT regimen in particular may also elicit tumour promoting effects through selecting for a more aggressive cancer cell phenotype [10] .
The percentage of OAC cells in tumour tissue expressing LAG-3, A2aR and TIM-3 were significantly lower post-FLOT treatment compared with the treatment-naïve setting. However, treatment of OE33 and SK-GT-4 cells with FLOT for 48 h in vitro significantly increased IC expression on the surface of OAC cells including LAG-3, A2aR and TIM-3. The in vitro experiments recapitulate the direct effects of chemotherapy on IC expression on OAC cells using an in vitro culture system. In contrast, the analysis of IC expression in tumour biopsy tissue encapsulates the effect of the entire tumour microenvironment on IC expression in the treatment-naïve setting as well as the combined direct and indirect effects of FLOT and CROSS CRT ~6 weeks post-treatment. It is also likely that changes in IC expression dynamically and longitudinally occur over this time. Interestingly, we observed that single agent chemotherapies upregulate IC ligands and receptors on the surface of OAC cells, whereby 5-FU had the most substantial effect. Additionally, the FLOT regimen consistently upregulated both inhibitory IC ligands and receptors on OAC cells in vitro to a greater extent than the CROSS CT and MAGIC chemotherapy regimens. ICs play an integral role in maintaining immunotolerance and are key players in mediating tumour immune evasion, therefore this suggests that FLOT may induce immunogenic cell death in OE33 and SK-GT-4 cells in vitro . Studies have shown that oxaliplatin and 5-FU, which comprise the FLOT regimen have immunostimulatory properties and induce immunogenic cell death in lung and colon cancer cells [35 , 36] . Complementary studies have aslo shown that 5-FU increases PD-L1 on the surface of OE33 and HCT-116 cells [28] and similarly, cisplatin increases PD-L1 on the surface of lung cancer cells [27] .
FLOT and CROSS CT treatments had the greatest effect in upregulating PD-L1, PD-L2, CD160, TIM-3, LAG-3 and A2aR ICs on OAC cells in vitro. However, PD-1 and TIGIT were only minimally increased. Further studies are required to further determine how biologically significant this is as PD-1 and TIGIT are expressed on only a sub-population of OAC cells in vitro and in vivo and therefore, they may be expressed on aggressive cancer cell clones that often exist in low frequencies and PD-1 and TIGIT signalling pathways may contribute to their survival or treatment resistance.
The upregulation of ICs on the surface of OAC cells suggests these ICs may offer some level of protection against chemotherapy or perhaps they may be upregulated in an attempt to repair the chemotherapyinduced damage to the cell. Emerging studies have shown that IC receptors and ligands directly enhance glycolysis [38] , proliferation [13] , invasion, migration [23 , 39] and DNA repair [40] via cancer cell-intrinsic signalling. Several studies have demonstrated that DNA damage sig-nalling upregulates PD-L1 expression on the surface of cancer cells [41][42][43] and PD-L1 cancer cell-intrinsic signalling mediates DNA repair [42 , 43] . Blockade of PD-L1 on the surface of OAC cells could potentially prevent repair of the chemotherapy-induced DNA damage thereby, enhancing chemotherapy toxicity. It is unclear whether the viable OAC cells that survived chemotherapy treatment have upregulated inhibitory ICs on their cell surface or if the chemotherapy treatment selectively kills OAC cells that lack inhibitory IC expression enriching for OAC cells that express inhibitory ICs. The former may suggest that OAC cells upregulate inhibitory ICs perhaps as a survival advantage to facilitate immune evasion or perhaps IC-intrinsic signalling is promoting immuneindependent mechanisms of resistance to chemotherapy-induced cell death. The latter suggests that ICs are expressed on the surface of OAC cells that are more resistant to chemotherapy-induced cell death and persist following treatment. Our results demonstrate that chemotherapy preferentially upregulates PD-L1 and TIM-3 ICs on a more stem-like OAC cell phenotype in vitro . Additionally, FLOT chemotherapy significantly upregulated TIM-3 and A2aR on a subpopulation of senescent-like OAC cells. Other studies have shown that PD-L1 is enriched on cancer stem cells and provides a mechanism of immune escape [12] . TIM-3 has also been identified on cervical and gastric cancer cells and induced invasion and migration of HeLa cells in vitro [17] . High levels of TIM-3 expression on gastric cancer cells correlated with metastasis in gastric cancer patients [18] . Shi et al., demonstrated that A2aR signalling via PI3K-AKT-mTOR upregulated stemness-associated and EMT-like proteins in gastric cancer cells in vitro and that A2aR knockout murine models resulted in a decrease in the number and size of micrometastatic lesions in the lungs of mice [20] . Ultimately, the expression of inhibitory IC ligands and receptors on OAC cells may function in tandem to offer OAC cells a survival advantage via promoting a range of cancer hallmarks.
Our study demonstrates for the first time that single agent and combination ICIs reduce OAC cell viability and induce apoptosis in OAC cells directly and independent of the immune system, a novel finding in the context of OAC. Further studies are required to determine if blockade of PD-1 or PD-L1 could reduce tumour cell growth and induce tumour cell death via immune-independent mechanisms using in vivo pre-clinical models which will help further elucidate the biological role and clinical potential for targeting these pathways to reduce tumour growth in OAC.
Furthermore, we investigated the potential for combining ICIs with chemotherapy to enhance chemotherapy-induced OAC cell death. Although we did not observe a substantial amount of synergism between ICI-chemotherapy combinations we demonstrated that combining ICIs with chemotherapy had a limited but significant effect in enhancing chemotherapy-induced OAC cell death. Similarly, Liu. et al., demonstrated that PD-1 blockade enhanced chemosensitivity to 5-FU in a 5-FU resistant gastric cancer cell line [37] . These important clinically relevant findings do question whether ICIs might synergise with combination chemotherapy regimens in patients to enhance chemotherapyinduced OAC cell death in an immune-independent manner and highlights the need for further studies to answer this question. In addition, these findings may be reflective of the limited benefit observed from clinical trials testing combination ICI-chemotherapy regimens, but this may translate to a measurable improvement in clinical outcomes. We also demonstrated that PD-L1 is upregulated on stem-like OAC cells in vitro and therefore, blockade of the PD-1 signalling axis may be reducing the survival of stem-like OAC cells and subsequently enhancing the efficacy of chemotherapy which could translate to a clinically meaningful improvement in outcomes for patients. This may account for the limited effect observed between the addition of ICIs to chemotherapy as stemlike cells only comprise a subpopulation of a tumour cell population.

Future perspective
A therapeutic rationale is highlighted for combining ICIs with the FLOT, CROSS and MAGIC regimens in OAC to prevent immune evasion mediated through chemotherapy-induced upregulation of ICs on OAC cells. The expression of inhibitory IC ligands or receptors on the surface of stem-like and senescent-like OAC cells may represent a mechanism of immune escape for these OAC cell clones. IC blockade may therefore facilitate immune-mediated clearance of stem-like and senescentlike OAC cells in patients and enhance clinical outcomes. Further studies are required to determine if the stem-like and senescent-like OAC cells are truly cancer stem cells and senescent OAC cells, respectively. Importantly, future studies are warranted to determine the biological and clinical significance of IC-intrinsic signalling in either promoting the survival or function of both tumour cells and the inherently treatment resistant cancer stem cells.

Specific author contributions
All authors have contributed to this manuscript. Maria Davern collected and processed samples, carried out experiments, performed data analysis, interpreted the results and contributed to conception and design of the study. Noel E. Donlon and Andrew Sheppard collected patient samples and carried out experiments. Fiona O' Connell performed data analysis. John V. Reynolds, Narayanasamy Ravi, Dermot O'Toole and Anshul Bhardwaj were involved in collecting and processing clinical samples and data. Stephen G. Maher and Niamh Lynam-Lennon contributed to design of the project. Joanne Lysaght made substantial contribution to the conception and design of the project, interpretation of the data, securing funding and overall supervision of the project. All authors were involved in the drafting and critical appraisal of the manuscript.

Funding
This work was funded by an Irish Research Council scholarship ( GOIPG/2017/1659 ) and the CROSS oesophageal cancer research charity.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.