Impact of hypoxia on chemoresistance of mesothelioma mediated by the proton-coupled folate transporter, and preclinical activity of new anti-LDH-A compounds

Background Expression of proton-coupled folate transporter (PCFT) is associated with survival of mesothelioma patients treated with pemetrexed, and is reduced by hypoxia, prompting studies to elucidate their correlation. Methods Modulation of glycolytic gene expression was evaluated by PCR arrays in tumour cells and primary cultures growing under hypoxia, in spheroids and after PCFT silencing. Inhibitors of lactate dehydrogenase (LDH-A) were tested in vitro and in vivo. LDH-A expression was determined in tissue microarrays of radically resected malignant pleural mesothelioma (MPM, N = 33) and diffuse peritoneal mesothelioma (DMPM, N = 56) patients. Results Overexpression of hypoxia marker CAIX was associated with low PCFT expression and decreased MPM cell growth inhibition by pemetrexed. Through integration of PCR arrays in hypoxic cells and spheroids and following PCFT silencing, we identified the upregulation of LDH-A, which correlated with shorter survival of MPM and DMPM patients. Novel LDH-A inhibitors enhanced spheroid disintegration and displayed synergistic effects with pemetrexed in MPM and gemcitabine in DMPM cells. Studies with bioluminescent hypoxic orthotopic and subcutaneous DMPM athymic-mice models revealed the marked antitumour activity of the LDH-A inhibitor NHI-Glc-2, alone or combined with gemcitabine. Conclusions This study provides novel insights into hypoxia/PCFT-dependent chemoresistance, unravelling the potential prognostic value of LDH-A, and demonstrating the preclinical activity of LDH-A inhibitors.

CRS and HIPEC in the National Cancer Institute, Milan, Italy (from August 1995 to October 2013) were selected for pathological examination. TMAs were constructed by using a tissue-arraying instrument (Beecher Instruments, Silver Springs, MD, USA).
IHC staining of paraffin-embedded tissues from MPM patients for proton-coupled folate transporter (PCFT), carbonic anhydrase IX (CAIX) and lactate dehydrogenase (LDH-A) were performed as described previously. Before staining with specific polyclonal rabbit anti-human antibody for PCFT [Hou et al, 2012], or with the commercial anti-CAIX (dilution 1:500; monoclonal antibody ab15086, Abcam, Cambridge, MA) and anti-LDH-A LDH-A (dilution 1:100; ab9002, Abcam) antibodies, the TMA slides were deparaffinized using xylene, rehydrated in alcohol and microwaved at 400W (2 times, for 5 minutes). Immunostaining was performed by the avidin-biotin peroxidase complex technique. Negative controls were obtained by replacement of primary antibody with phosphate-buffered saline (PBS), while positive controls were obtained using sections of colorectal cancer. Immunoreactivity was enhanced by antigen retrieval for 30 minutes in 10 mM sodium citrate, pH 6.0. The sections were washed three times in PBS and blocked with Super Block (Skytek Laboratories, Logan, UT) and 3% hydrogen peroxide for 10 minutes prior to overnight incubation at 4°C with the primary antibody (dilution, 1:40). After overnight incubation with the primary antibodies, the sections were washed two times for 3 minutes in PBS and incubated with the appropriate kit containing the secondary antibody tagged with avidin-biotinylated horseradish peroxidase (Cell Marque revelation Kit, Sigma). Finally, the colorimetric reaction obtained with 3'-3' diaminobenzidine was counterstained with hematoxylin, and slides were permanently fixed with synthetic mounting. The sections were scored by two researchers blinded to clinical outcome, who also evaluated the extent of tissue loss, background staining and overall interpretability. Scoring for PCFT was described in our previous study [Giovannetti et al., 2017] Immunostaining intensity of CAIX was described in the supplemental table 1, while for LDH-A we used a previously proposed grading system with two LDH-A expression levels: strong cytoplasmic expression in >50% of cancer cells or nuclear expression in >10% of cancer cells was defined as high expression; otherwise, nuclear expression was considered low [Koukourakis et al., 2006].
Neoplastic cells were always uniformly stained and counting all the tumor cells in each spot made positivity assessment. To further implement the reproducibly of our technical procedures, as discussed previously for other biomarkers in MPM samples [Li Petri et al., 2018], all stained TMA sections were also digitally imaged at 40X, using a computerized high-resolution acquisition system (D-Sight, Menarini, Florence, Italy), equipped with the automated quantitative image analysis software algorithm DSight software 2.1.0. Multipreview of the images allowed editing of the area of interests.

Determination of Mitochondrial Function and Glycolysis in Mesothelioma cells
Oxygen Consumption Rate (OCR) and ECAR (Extracellular Acidification Rate) were measured in the mesothelioma cell lines MSTO and H2452 by using the Seahorse XFp Metabolic Flux Analyzer (Agilent Technologies, Inc, Santa Clara, Ca, USA), as described previously [Schipper et al, 2017]. One day before the assay, cells were seeded at a density of 40,000 per well in a final volume of 80 µl in a Seahorse plate. The mitochondrial stress test was performed according to manufacturer's instructions.
Before the analysis, medium was changed with Seahorse XFp RPMI medium, enriched with 1 mM pyruvate, 2 mM glutamine and 10 mM glucose and then incubated for 45 minutes. After three baseline measurements, Oligomycin (inhibitor of V Complex), Carbonyl cyanide-4 (trifluoromethoxy) phenylhydrazone (FCCP, mitochondrial uncoupler) and a mix of Rotenon and Antimycin A (inhibitors of Complexes I and III) were sequentially injected into each well to final concentrations of 1.5 µM, 0.5 µM 0.5 µM. The obtained data allowed for calculations of ATP-linked Respiration, Maximal Respiration, Spare Capacity, Proton Leak and Non-Mitochondrial Oxygen Consumption. The Glycolysis Stress Test was conducted with similar seeding conditions. At the day of analysis, the Seahorse XFp RPMI medium was enriched with 2 mM glutamine and adjusted to pH 7.4. Glucose (substrate for glycolysis), Oligomycin (inhibitor of mitochondrial ATP production) and 2-deoxyglucose (inhibitor of hexokinase) were consecutively added into wells to final concentrations of 10 mM, 5 µM and 100 mM, respectively. The Glycolytic capacity, Glycolytic reserve, Glycolysis and Non-Glycolytic Acidification were estimated and normalized to mg of protein.
Quantitative-RT-PCR (qRT-PCR). PCR reactions were performed in triplicate with 5 µL of cDNA, 12.5 µL of TaqMan Universal PCR Master Mix, and 5 µL of probe, and forward and reverse primers in a final volume of 25 µL. Samples were amplified by the following thermal profile: an initial incubation at 50°C for 5 minutes to prevent the reamplification of carry-over PCR products by AmpErase uracil-N-glycosylase, followed by incubation at 95°C for 10 minutes to supress AmpErase UNG activity and denature the DNA, 50 cycles of denaturation at 95°C for 15 seconds, followed by annealing and extension at 60°C for 1 minute. Specific forward and reverse primers and probes were obtained from Applied Biosystems Assay-on-Demand products. All the samples were amplified in triplicate with appropriate non-template controls, and the coefficient of variation was less than 2%. Spheroids. MPM spheroids were established by seeding 1000 MPM cells per ml in DMEM/F12 GlutaMAX-I (1:1, Invitrogen), in ultra-low attachment plates (Corning Incorporated, Corning, NY). These spheroids were generated for 7-10 days, and then harvested for RNA isolation and analysis of PCFT expression, as described above.
We also performed an exploratory analysis using a sequential trypsin digestion of spheroids of H2452 that had reached a diameter of approximately 500 μm. These In parallel experiments we also evaluated whether pemetrexed was able to affect spheroid formation by counting the number of spheroids created in cells exposed immediately after seeding to 0.1, 1, 10 and 20 µM pemetrexed for 72 hours, compared to untreated cells.
MESO II and STO spheroids with a diameter of approximately 300 µm were created in 96-well flat bottom plates coated with 1.5% agarose. DMEM/F12 medium was replaced with drug-free medium or medium containing gemcitabine or PI-FLY161 (6 wells per condition). Images of spheroids were taken with an automated phasecontrast microscope (Universal Grab software, Digital Cell Imaging Labs). To detect the amount of light passing through the spheroids, pixel intensities of 8-bit black/whiteconverted images were calculated using ImageJ Software (U.S. National Institutes of

In vivo experiments using orthotopic and subcutaneous mouse models and live imaging
In vivo experiments were performed in nu/nu athymic female mice 4 weeks old with average 23 g of weight (range, 22-24 g) at the arrival, while the weight during the experiment is reported in the Supplemental Figure S8). The animals were purchased purchased from Harlan (Horst, The Netherlands). In the study design and report we followed the ARRIVE reporting guideline and provide a completed ARRIVE checklist as a supplemental file specifying where in the manuscript each item is reported.
In particular, we undertook the following steps to minimise the effects of subjective bias when allocating animals to treatment: randomisation procedure (matching for tumor volume and animal weight) and blinding assessment of the results by the pathologists. The animals were hosted in cages (3 animals/cage), under pathogen free [SPF] condition, with standard light/dark cycle, and temperature conditions, and free access to food and water, environmental enrichment). Welfare-related assessments and interventions were carried out prior to, and during the experiment.
Orthotopic primary DMPM models (n=5 tumors per treatment group) were generated by injection of 3×10 6 Fm/GC primary cells into the peritoneal cavity of the mice.  High CAIX High CAIX High CAIX "High CAIX" include the scores 4,5 and 6 "Low CAIX" includes the scores 2 and 3 Supplemental table 3. List of genes significantly up-regulated in cells growing as monolayers or spheroids, after exposure to hypoxic conditions as well as after PCFT silencing (showing a PCFT reduction of 80% at RT-PCR) and, using the Hypoxia RT2 ProfilerTM array. The genes are reported in alphabetic order (NC, negative control, genes in blue are in common between the "Spheroids vs. monolayers" and the "Hypoxia vs. Normoxia" groups, genes in green are in common between the "Spheroids vs. monolayers" and the "PCFT siRNA vs. NC siRNA" groups, genes in red are in common between the "Hypoxia vs. Normoxia" and the "PCFT siRNA vs. NC siRNA" groups, LDH-A is the only gene in common among all these three groups). P values <0.01; Fold change= 3 Spheroids vs. monolayers N=25 Hypoxia vs. Normoxia N=49 PCFT siRNA vs. NC siRNA N=14 EGR1  PFKFB4  EPO  PGK1  GBE1  PGK1  GBE1  PIM1  HIF3A  PKM2  HNF4A  SLC2A1  IER3  SLC2A3  IGFBP3  JMJD6  LDH-A  LOX  MIF  MXI1  NDRG1  NOS3  ODC1  P4HA1  P4HB  PKM2  PLAU  RBPJ  RUVBL2  SERPINE1  SLC2A1  SLC2A3  TFRC  TPI1  TXNIP  USF2  VDAC1  VEGFA   Supplemental Table 4. Outcome of the patients in the MPM patients according to clinicopathological characteristics.
Abbreviations: EORTC, European Organization for Research and Treatment of Cancer; PFS, progression-free survival; OS, overall survival (Survival data were available for all patients and the minimum follow-up at the time of analysis was 27 months).
Notes: The potential association of LDH-A expression with clinicopathological characteristics was compared between groups using the chi-square test, with continuity correction when cells had expected count less than 5, while the correlation with PFS and OS was evaluated using the log-rank test  Table 5. Outcome of the patients in the DMPM patients according to clinicopathological characteristics.
Notes: The potential association of LDH-A expression with clinicopathological characteristics was compared between groups using the chi-square test, with continuity correction when cells had expected count less than 5, while the correlation with PFS and OS was evaluated using the log-rank test        Representative BLI images of mice harbouring orthotopic DMPM tumors. At the start of the experiment, mice were stratified into groups with comparable BLI signal, and then treated with NHI-GLC-2 administered i.p. for 2 weeks. Lower panel: Tumor growth as detected by BLI analysis. Columns, mean values obtained from the measurements in three mice; bars, SD.

Figure S9. Body weight of animals inoculated subcutaneously with DMPM cells.
Median weight loss was always <10%, without toxic deaths. These results support the conclusion that the gemcitabine/NHI-Glc-2 combination displays a safe antitumor activity against in vivo models of DMPM.

Figure S10. LDH-A immunohistochemical analyses of MPM tissue microarrays (TMAs).
Representative TMA cores in the cohorts of MPM patients, illustrating cases with high (upper panels) and low (lower panels) LDH-A expression (at 4X, 10X and 40X original magnification). Of note, at 40X the LDH-A staining seems membrane-bound. However, this is an artefact. As clearly detected at the other magnification the staining is indeed cytoplasmatic. As clearly detected at the other magnifications the staining is indeed cytoplasmatic. Since these cells have very big nuclei, these nuclei are taking most of the space in the cytoplasm and the staining seems located near to the membrane at 40X. However, this staining is not a "membrane staining", as clearly demonstrated also by the comparison to our previous staining of EGFR in MPM tissues