Canagliflozin mediates tumor suppression alone and in combination with radiotherapy in non‐small cell lung cancer (NSCLC) through inhibition of HIF‐1α

Non‐small cell lung cancer (NSCLC) has a poor prognosis, and effective therapeutic strategies are lacking. The diabetes drug canagliflozin inhibits NSCLC cell proliferation and the mammalian target of rapamycin (mTOR) pathway, which mediates cell growth and survival, but it is unclear whether this drug can enhance response rates when combined with cytotoxic therapy. Here, we evaluated the effects of canagliflozin on human NSCLC response to cytotoxic therapy in tissue cultures and xenografts. Ribonucleic acid sequencing (RNA‐seq), real‐time quantitative PCR (RT‐qPCR), metabolic function, small interfering ribonucleic acid (siRNA) knockdown, and protein expression assays were used in mechanistic analyses. We found that canagliflozin inhibited proliferation and clonogenic survival of NSCLC cells and augmented the efficacy of radiotherapy to mediate these effects and inhibit NSCLC xenograft growth. Canagliflozin treatment alone moderately inhibited mitochondrial oxidative phosphorylation and exhibited greater antiproliferative capacity than specific mitochondrial complex‐I inhibitors. The treatment downregulated genes mediating hypoxia‐inducible factor (HIF)‐1α stability, metabolism and survival, activated adenosine monophosphate‐activated protein kinase (AMPK) and inhibited mTOR, a critical activator of hypoxia‐inducible factor‐1α (HIF‐1α) signaling. HIF‐1α knockdown and stabilization experiments suggested that canagliflozin mediates antiproliferative effects, in part, through suppression of HIF‐1α. Transcriptional regulatory network analysis pinpointed histone deacetylase 2 (HDAC2), a gene suppressed by canagliflozin, as a key mediator of canagliflozin's transcriptional reprogramming. HDAC2 knockdown eliminated HIF‐1α levels and enhanced the antiproliferative effects of canagliflozin. HDAC2‐regulated genes suppressed by canagliflozin are associated with poor prognosis in several clinical NSCLC datasets. In addition, we include evidence that canagliflozin also improves NSCLC response to chemotherapy. In summary, canagliflozin may be a promising therapy to develop in combination with cytotoxic therapy in NSCLC.


Introduction
Lung cancer is the leading cause of cancer-related mortality [1].At early stages, non-small cell lung cancer (NSCLC), which represents 85% of lung cancer cases, is reasonably well-managed with surgery or high-precision stereotactic radiotherapy (RT) [2].However, many NSCLC patients present with locally advanced (LA) or metastatic disease at initial diagnosis [3].Chemotherapy and RT are conventional treatment options for advanced disease, and concurrent chemo-RT is the therapy of choice for good performance status patients.Nevertheless, response to these therapies is limited as patients with LA-NSCLC exhibit 5-year survival rates of 24-32% after concurrent chemo-RT [4] and metastatic disease is incurable.
In recent years, studies with immune checkpoint inhibitor therapy in addition to chemo-RT in patients with unresectable LA-NSCLC improved standard therapy outcomes [5].However, this treatment also provided moderate improvement in 5-year survival (42.9% vs 33.4%) and a progression-free survival rate of 33.1% in patients who showed no progression after chemo-RT [5].This underscores an urgent need to explore alternative strategies to augment the efficacy of current NSCLC therapies.
Metabolic reprogramming is a hallmark of cancer, in which cancer cells exhibit distinct metabolic phenotypes to sustain high proliferation rates, adapt to changes in the tumor microenvironment and resist anticancer therapies [6].The evolutionarily conserved serine/threonine kinase, adenosine monophosphateactivated protein kinase (AMPK), is a heterotrimeric enzyme consisting of a-catalytic, and band cregulatory subunits, which senses metabolic stress and facilitates metabolic reprogramming by orchestrating anabolic and catabolic cellular functions [7].Elevated AMP/ADP levels bind to AMPK's c subunit inducing a conformational change and activation of its kinase activity [8].AMPK inhibits energy expenditure, in part, by suppressing de novo lipogenesis, protein synthesis, DNA replication, and the cell cycle; effects mediated through blockade of rate-limiting enzymes such as acetyl-CoA carboxylase (ACC), the mammalian target of rapamycin complex 1 (mTORC1) and induction of p53, respectively [7].
Moreover, aberrant activation of the phosphoinositide-3-kinase (PI3K)-protein kinase B (Akt)-mTORC1 signaling pathway is frequently observed in NSCLC secondary to activating mutations in the epidermal growth factor receptor, Kristen rat sarcoma viral oncogene homolog, or PI3K subunits [9,10].Hypoxia-inducible factor (HIF)-1a, a regulator of the hypoxic transcriptional response, is stabilized through mTORC1 signaling even under normoxic conditions [11], which contributes to a Warburg glycolytic phenotype [9], radioresistance, and poor prognosis [12,13].Hypoxia-inducible factor-1a (HIF-1a) is often overexpressed in NSCLC, and its silencing has been shown to suppress glycolysis and inhibit tumor growth [14][15][16].Thus, the unique ability of the AMPK-mTORC1 pathway to regulate cancer cell proliferation and survival has triggered interest in exploiting this pathway for cancer therapy.
Canagliflozin, a well-tolerated oral medication approved for treating type 2 diabetes, was developed to improve glycemia by blocking the renal sodium-glucose cotransporter 2 (SGLT2) [17].However, canagliflozin also suppresses mitochondrial complex-I-supported respiration, leading to AMPK activation and suppression of the mTORC1 pathway [18].Our group was the first to demonstrate canagliflozin's antitumor activity in NSCLC and prostate cancer models [19].Since then, several other groups have shown similar antitumor effects in liver, breast, pancreatic, and other cancers [20].While some groups have suggested that canagliflozin's anticancer properties are linked to the inhibition of glucose transport via SGLT2 [21][22][23], this transporter is functionally present only in certain cancers [23], and emerging evidence indicates that additional mechanisms are at play [24,25].
Although our earlier work suggested canagliflozin's promising antitumor activity in cultured NSCLC cells, the efficacy of this agent to improve NSCLC response to cytotoxic therapy remains unknown.In this study, we sought to investigate canagliflozin's regulation of the RT and chemotherapy response in cultured cells and human xenograft models and to identify potential mechanisms of action.We report that, within its therapeutic window, canagliflozin inhibits proliferation and tumorigenic potential and enhances cytotoxic therapy response in NSCLC.Furthermore, we show that canagliflozin exerts profound transcriptional reprogramming of key metabolic, growth, and survival pathways, and it exerts its anticancer effects, at least in part, via blockade of histone deacetylase 2 (HDAC2) and the mTORC1-HIF-1a signaling pathway.

Proliferation
Cells (500 cells/well) were seeded in 96-well plates and left to adhere overnight.When untreated control wells reached 80% confluency, cells were fixed with 10% formalin and stained with 0.5% crystal violet.The stain was solubilized with 20% MeOH, and absorbance was measured at 570 nm on the SpectraMax iD5 Multi-Mode Microplate Reader (Molecular Devices: San Jose, CA).

Clonogenic survival
Cells were seeded in 12-well plates at different densities depending on the RT dose (500 cells/well for 0 Gy, 1 9 10 3 cells/well for 2 Gy and 2 9 10 3 cells/well for 4 Gy) and left to adhere overnight.After a 7-day incubation post-treatment administration or when viable colonies (defined as a cluster of ≥ 50 cells) formed in the untreated control wells, cells were fixed with 10% formalin, stained with crystal violet and colonies were counted.Data were normalized to seeding density.

Synergy analysis
The Highest Single Agent (HSA) model on the Syner-gyFinder web application (https://synergyfinder.org) was used to predict additivity, synergism, or antagonism of combined treatments [27].

Xenograft models
Six-to eight-week-old male athymic BALB/c nude mice (CAnN.Cg-Foxnnu/Crl; Charles River: Wilmington, MA) and male and female nonobese diabetic rag gamma (NRG) mice (NOD.Cg-Rag1(tm1Mom)Il2rg (tm1Wjl)/SzJ; bred in-house; Jackson Laboratory: Bar Harbor, ME) were housed in a pathogen-free facility with ad libitum access to food and water.NSCLC cells (1 9 10 6 cells) were injected subcutaneously into flanks of anesthetized mice.Tumors were measured with a caliper, and volume was calculated using V = ½ 9 (length9width 2 ).Animals were treated as described in Fig. 1

Functional metabolism assays
Cells were seeded at 2 9 10 4 cells/well, treated as indicated, and analyzed on the Seahorse Bioscience XFe96 Extracellular Flux Analyzer (Agilent Technologies: Santa Clara, CA).After a 48-h incubation with the treatments, cells were subjected to the Glycolytic Rate Assay (GRA) or the Mito Stress Test (MST) as per the manufacturer's protocol [28].Cells were assayed in Seahorse XF DMEM (25 mM D-glucose, 1 mM sodium pyruvate, and 4 mM Lglutamine) without drug treatments.GRA inhibitors: 0.5 lM rotenone/antimycin A and 50 mM 2-DG.MST inhibitors: 1.5 lM oligomycin, 1 lM FCCP, and 0.5 lM rotenone/antimycin A. After the assays, cells were stained with crystal violet as described in Section 2.2, and data were normalized to cell content.Wave Desktop software was used for data acquisition and analysis (version 2.6; Agilent Technologies).

Immunoblotting
Cells were lysed as described [19], and protein concentrations were determined using the BCA Protein Assay Kit (Thermo Fisher Scientific: Waltham, MA).Lysates were subjected to electrophoresis (SDS/PAGE), immunoblotting [19], and imaging on a Fusion FX system (Vilber Lourmat: Coll egien, France).See Table S1 for list of antibodies.

Immunohistochemistry (IHC)
Standard IHC protocols were followed on FFPE tumor sections as described [29].Tissues were incubated overnight with the indicated primary antibody at 4 °C, followed by biotinylated secondary antibody and HRP streptavidin.Slides were developed using the Vector NovaRED Substrate Kit (Vector Laboratories Inc: Burlingame, CA).Hematoxylin was used as a counterstain.See Table S1 for list of antibodies.

IHC quantification and necrosis evaluation
For evaluation of p-ACC Ser79 , CC3 Asp175 , and p-H3 Ser10 , 10 random high-power fields (409) of each slide were quantified.The cytoplasmic intensity for p-ACC Ser79 and the percentage of nuclear positivity for CC3 Asp175 and p-H3 Ser10 were quantified using IMAGEJ software (version 1.53a; National Institutes of Health: Bethesda, MD).For necrosis evaluation, the ratio of necrotic area to whole tumoral section was quantified using IMAGEJ [30].

Cell cycle analysis
Following treatment, cells were fixed in 70% ethanol and stained with FxCycle Propidium Iodide (PI)/ RNase Staining Solution (Thermo Fisher Scientific).At least 1 9 10 5 events/sample were analyzed on the CytoFLEX LX Flow Cytometer (Beckman Coulter: Brea, CA; 488 nm excitation and 610 nm emission).FLOWJO software (version 10.8.0; FlowJo LLC: Ashland, OR) was used for analysis.

Real-time quantitative PCR (RT-qPCR)
The RNeasy Mini Kit (QIAGEN: Hilden, Germany) was used for total RNA purification.Samples were reverse transcribed to cDNA using SuperScript IV reverse transcriptase (Invitrogen: Waltham, MA).Amplification and detection were performed in a qPCR thermocycler (Corbett Rotor Gene 6000: Montreal Biotech Inc: Dorval, Canada) using TaqMan Assay fluorogenic 5 0 nuclease chemistry (Invitrogen).Relative gene expression was calculated using the Livak method [31] normalizing to GAPDH.See Table S1 for list of probes.

RNA sequencing (RNA-seq)
Purified RNA samples were subjected to RNA-seq at the McMaster Genomics Facility.RNA quality was confirmed using the 2100 Bioanalyzer System (Agilent Technologies).mRNA was enriched from 1 lg of total RNA using the NEBNext Poly(A) mRNA Magnetic Isolation Module, and Illumina libraries were prepared using the NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs: Ipswich, MA).Each run was done with the Illumina HiSeq 1500 using the HiSeq Rapid v2 chemistry with onboard cluster generation and a 1 9 50 bp configuration (Illumina: San Diego, CA), aiming for approximately 25 M clusters per sample.

Gene set and regulatory network analysis
Gene ontology analysis was performed using Gene Set Enrichment Analysis (version 4.1.0)[36].Cytoscape's iRegulon plugin (version 3.9.1)was used to predict the regulators responsible for modulating gene expression.

Prognostic analysis
The Kaplan-Meier plotter (https://kmplot.com/analysis/ ) was used to analyze gene expression and survival data of 1925 lung cancer cases from 12 cohorts [37].

Statistical analyses
Ordinary one-way or two-way ANOVA followed by Tukey's post hoc test was used to evaluate the significance between treatments unless otherwise indicated.Proliferation and clonogenic survival IC 50 values were calculated using a nonlinear regression model with a normalized slope.Statistical analyses were performed using GRAPHPAD PRISM (version 9.1.0;GraphPad: San Diego, CA).Statistical significance was considered at P-value < 0.05.

Proliferation
We examined the antiproliferative effects of clinically relevant doses of canagliflozin (5-30 lM) and RT; 2 or 4 Gy in human adenocarcinoma (A549, H1299 and H1975), squamous cell (SK-MES-1), and large cell (H460) NSCLC cells.Canagliflozin and RT dosedependently inhibited the proliferation of all cell lines, with A549 cells showing the highest sensitivity to canagliflozin and H460 the least (Fig. 1A,B; see figure legend for IC 50 values).All NSCLC cell lines treated with RT exhibited additional reduction of proliferation with canagliflozin (Fig. 1C).Synergy analysis revealed that the addition of canagliflozin to RT provided additive antiproliferative effects in A549, H1299, and H1975 cells and synergistic effects in SK-MES-1 and H460 cells (Fig. 1D).

Clonogenic survival
In clonogenic survival assays, canagliflozin (2.5-30 lM) suppressed the clonogenic potential of A549, H1299, and H1975 cells (IC 50 values: 4.4, 5.7, and 7.2 lM, respectively) alone and in combination with RT (2 or 4 Gy; Fig. 1E).Suppression of survival by canagliflozin and RT was additive in A549 and H1299 cells and synergistic in H1975 cells (Fig. 1F).These data demonstrate that clinically achievable concentrations of canagliflozin effectively inhibit tumorigenic potential and enhance the efficacy of RT in NSCLC cells.

Tumor growth
We investigated the effects of oral canagliflozin (diet designed to deliver 60 mgÁkg À1 Áday À1 ; see Section 2), RT (5 Gy) and combination therapy in A549, H1299, and H1975 xenograft models (Fig. 2A).Consistent with previous studies in rodents and humans [38], the polyuria induced by canagliflozin led to greater food and water intake and resulted in lower body mass compared with mice fed the control diet (Fig. S1A-E).The estimated daily dose of canagliflozin in mice harboring A549, H1299, and H1975 tumors was 81, 55, and 78 mgÁkg À1 Áday À1 , respectively.
Canagliflozin and RT monotherapies reduced tumor growth by 18-26% and 32-44% compared with control mice; however, this did not reach statistical significance at endpoint (Fig. 2B).In contrast, the combination therapy significantly suppressed tumor growth in all three cohorts (51-65% tumor suppression compared with controls).

Regulation of gene expression: Induction of oxidative phosphorylation (OxPhos) and downregulation of hypoxia pathways
To identify potential mechanisms contributing to these effects of canagliflozin and RT, we conducted RNA-seq analysis and found that canagliflozin (10 lM) significantly upregulated 102 and downregulated 81 genes (Fig. 3A), whereas RT (5 Gy) upregulated 2562 and downregulated 3226 genes (Fig. S2A).The volcano plot in Fig. 3B illustrates the most significantly downregulated genes by canagliflozin (IGFBP1, SLC2A3, NDRG1, PGK1, EGLN3, and P4HA1; involved in growth factor activity, glucose uptake and glycolysis, cell growth and differentiation, and response to hypoxia) and upregulated genes (ASNS, PHGDH, PSAT1, SLC1A5, CYP1B1, and RNR1; and involved in amino acid uptake, metabolism, and ribosomal function).To understand the biological processes behind the DEGs, DEGs were  subjected to Gene Set Enrichment Analysis.The majority of significantly downregulated pathways were associated with cellular response to hypoxia, whereas the most significantly upregulated pathways include amino acid transport, unfolded protein response, and mitochondrial OxPhos (Fig. 3C).Interestingly, a large proportion of genes modulated by canagliflozin (146 of 183 DEGs) were also influenced by RT (Fig. S2B).

Mitochondrial respiration
Given the observed modulation of OxPhos gene expression, we examined cellular respiration.In nonirradiated cells, canagliflozin (5-30 lM) caused small but significant dose-dependent suppression of the oxygen consumption rate (OCR) and enhancement of the extracellular acidification rate (ECAR; Fig. 4A-C, Fig. S3A,B).RT (2 and 8 Gy) did not alter OCR consistently but significantly repressed ECAR (Fig. 4A,B).The addition of canagliflozin to RT suppressed OCR and reversed the RT-induced inhibition of ECAR (Fig. 4A,B).The ECAR/OCR ratio, an indicator of anaerobic metabolism (metabolic stress), was reduced with RT but was effectively enhanced by canagliflozin in both nonirradiated and irradiated cells (Fig. 4C).Overall, canagliflozin mediated modest metabolic stress in NSCLC cells as indicated by changes of 54% or lower in OCR or ECAR.
To better understand canagliflozin's impact on mitochondrial respiration, we compared its effects to those of the potent and specific OxPhos complex-I inhibitors, BAY-87-2243 and IACS-010759.Seahorse MST and Fig. 3. Bioinformatics analysis of DEGs in A549 NSCLC cells treated with canagliflozin.RNA-seq was employed to assess the transcriptome profile of A549 cells treated with 10 lM canagliflozin for 24 h (n = 3).(A) The number of significantly upregulated and downregulated DEGs in A549 cells following canagliflozin treatment with a FDR q-value < 0.05.(B) A volcano plot depicting the total upregulated (positive log 2 (Fold Change)) and downregulated (negative log 2 (Fold Change)) genes by canagliflozin.DEGs with a Àlog 10 (P-value) < 2 are separated with a dotted line and have an FDR q-value < 0.05.(C) The most significantly regulated gene sets from Gene Set Enrichment Analysis of canagliflozin-treated A549 cells.Data were derived from a single RNA-seq experiment.

Regulation of metabolic stress and growth factor pathways
To understand the time course of metabolic stress signals induced by canagliflozin, we analyzed the effects of canagliflozin (30 lM) in A549 cells at different time points (Fig. 5A).The activating AMPK Thr172 phosphorylation (p-AMPKa Thr172 ) was detected within 1 h of drug incubation but continued to increase up to 48 h.The inhibitory phosphorylation of Ser79 on ACC (p-ACC Ser79 ) and Ser792 on Raptor (p-Raptor Ser792 ), direct AMPK targets [39,40], was detected by 8 h and maintained for up to 48 h.Consistent with these effects, suppression of Ser2448 phosphorylation on mTOR (p-mTOR Ser2448 ) continued to 48 h (Fig. 5A).Therefore, subsequent analyses were performed after 48 h of canagliflozin treatment.
To examine the metabolic effects of canagliflozin in combination with RT, cells were pretreated for 5 h with 10 or 30 lM canagliflozin and analyzed 48 h after 2 or 8 Gy RT exposure.Canagliflozin alone induced p-AMPKa Thr172 and p-ACC Ser79 in nonirradiated and irradiated cells indicating activation of metabolic stress signaling (Fig. 5B,C).The mTOR pathway was suppressed by canagliflozin, and RT provided further inhibitory effects (Fig. 5D).This is evident by the inhibition of p-mTOR Ser2448 and its downstream targets: p-p70S6K Thr389 , p-S6 Ser240/244 , and p-4EBP1 Ser65 (Fig. 5E).Similar results were obtained in H1299 cells (Fig. S4A,B).
Compared with untreated controls, canagliflozin, RT, and the combined treatment reduced the proportion of p-H3 Ser10 -positive cells in A549 and H1299 tumors, with the combination treatment showing more apparent additive effects in A549 tumors (Fig. 5J).Furthermore, IHC for CC3 Asp175 detected low but increasing numbers of apoptotic cells in tumors treated with canagliflozin, RT, and combination therapy in H1299 tumors (P = 0.0401; Fig. 5J).However, we could not detect consistent changes in levels of tumor necrosis in response to any of the treatments (Fig. S5).
Transcript levels rose transiently with canagliflozin treatment at 2 h and returned to lower levels by 24 h (Fig. S6).Consistent with modulation of the cell cycle, canagliflozin alone caused a slight nonsignificant increase in p21 cip1 and p27 kip1 protein levels after a 48h treatment (Fig. 6C,D).While RT led to significant induction of p21 cip1 protein levels, we did not detect further significant induction of p21 cip1 or p27 kip1 protein beyond that caused by RT (Fig.

Transcriptional regulation
To help understand the processes underlying canagliflozin's transcriptional reprogramming, we implemented gene regulatory network analysis with Cytoscape's iRegulon tool.Consistent with the observed suppression of the HDAC2 gene itself (Fig. 8A), iRegulon analysis determined HDAC2 as the transcriptional regulator responsible for canagliflozin's reprogramming [false discovery rate (FDR) q-value = 0.022].Fig. 8A illustrates the wide variety of HDAC2-regulated biological processes we found to be downregulated by a clinically achievable concentration of canagliflozin in NSCLC cells.

Canagliflozin target genes are associated with poor prognosis in NSCLC
The Kaplan-Meier Plotter database was employed to evaluate the prognostic value of DEGs targeted by canagliflozin in NSCLC datasets [37].The database is generated from the Cancer Biomedical Informatics Grid, the Gene Expression Omnibus, and The Cancer Genome Atlas cancer datasets and includes NSCLC patients at different stages with various histologies and treatment regimens (see Fig. 9 legend for trial datasets included) [37].Interestingly, results revealed that higher expression of several genes downregulated by canagliflozin is associated with poor prognosis of NSCLC.This included genes involved in transcription (HDAC2), hypoxia response and extracellular matrix synthesis (P4HA1, PLOD2), glucose metabolism (ALDOA, LDHA, and PGK1), growth factor response, proliferation, and prevention of apoptosis (NDRG1, NEK2, and BIRC5; Fig. 9).

Discussion
Resistance of tumor cells to cytotoxic therapy presents a significant obstacle to successful treatment.Mounting evidence suggests that metabolic reprogramming through activation of the PI3K-Akt-mTOR pathway is a principal mechanism of resistance [42].Previously, we showed that canagliflozin blocks mitochondrial complex-I-supported respiration leading to AMPK activation and suppression of the mTOR pathway in cancer cells [19].Although SGLT2 is expressed in NSCLC cells (see Fig. S8) and canagliflozin inhibits glucose uptake in cancer cells, in that study, we found that the drug's anticancer action was independent of the extracellular glucose concentration [19].Moreover, pyruvate or acetate supplementation in cell culture media did not reverse canagliflozin's antiproliferative activity (see Supplemental Data in Villani et al. [19]).Papadopoli et al. [43] also found that canagliflozin's antitumor activity is through SGLT2-independent mechanisms, as canagliflozin's effects occurred in the absence of glucose and with knockdown of SGLT2 in breast cancer cells.Furthermore, we showed that canagliflozin's action was partially inhibited by overexpression of the biguanide-resistant Saccharomyces cerevisiae complex-I, NADH dehydrogenase 1 (ND1) [19].These off-target findings have since been corroborated by others [44].Collectively, these data indicate that canagliflozin's antitumor activity does not rely upon reduced carbohydrate availability to cancer cells but relies, at least in part, on mitochondrial OxPhos complex-I inhibition.8. Regulatory network analysis and identification of HDAC2 as the modulator of canagliflozin's gene expression profile.(A) All downregulated DEGs; FDR q-value < 0.05 from the RNA-seq data were analyzed using Cytoscape's iRegulon plugin to predict the master regulators that mediate canagliflozin's transcriptional program.HDAC2, which was suppressed by canagliflozin, was identified as the key mediator associated with the drug's gene expression profile.Gene regulatory network analysis illustrates the HDAC2-regulated genes that were significantly downregulated by canagliflozin in the RNA-seq dataset (Normalized Enrichment Score = 3.66; FDR q-value = 0.022).(B) Representative protein immunoblot and (C) quantification of HDAC2 in A549 cells pretreated with canagliflozin (10 or 30 lM) for 5 h and subsequently treated with RT (single fraction of 2 or 8 Gy) and analyzed 48 h later.(D) Representative protein immunoblots and of HDAC2, HIF-1a, and p21 cip1 in A549 cells treated with negative control siRNA (5 nM) or HDAC2 siRNA (5 nM) for 72 h (quantification of in (E)).All western blot data were normalized to b-actin.(F) Proliferation assay of A549 cells treated with canagliflozin (5-30 lM) and either negative control siRNA (5 nM) or HDAC2 siRNA (5 nM) for 5 days.Results were derived from three independent experiments and presented as mean AE SEM.Ordinary one-way ANOVA (E) or two-way ANOVA (C, F) followed by Tukey's multiple comparisons test was used to determine statistical significance.*P < 0.05, **P < 0.01, ****P < 0.0001.
The primary objectives of the present study were to (1) investigate whether canagliflozin can further enhance the efficacy of cytotoxic therapy in preclinical models of NSCLC and (2) gain additional insights into the molecular mechanisms involved.
Here, we report that canagliflozin effectively suppressed NSCLC cell proliferation and survival and potentiated the efficacy of RT in vitro.Additive and synergistic effects were detected with clinically achievable canagliflozin and RT doses, an effect observed with even lower drug doses in clonogenic assays (Fig. 1).Importantly, the latter results were reproduced in three xenograft models (Fig. 2).Given that the purpose of this study was to focus on the efficacy of clinically achievable doses of canagliflozin and to investigate the combined therapy, we refrained from using high doses of cytotoxic therapy.Therefore, it is not unexpected that neither canagliflozin nor RT generated statistically significant antitumor efficacy as single agents, but combined therapy mediated this effect.
We observed that a clinically relevant dose of canagliflozin (10 lM) rapidly (24 h) mediated reprogramming of NSCLC transcriptional activity leading to a gene expression profile consistent with suppression of growth factor signaling, cell cycle progression, and response to hypoxia, the latter of which induces glycolytic gene expression, survival, and radioresistance (Fig. 3).Therapeutically relevant canagliflozin doses suppressed mitochondrial respiration in NSCLC cells (Fig. 4), which triggered an expected induction of electron transport chain genes (Fig. 3C).
Compared with the selective mitochondrial complex-I inhibitors, BAY-87-2243 and IACS-010759, canagliflozin mediated moderate metabolic stress, reduced rates of extracellular acidification and increased antiproliferative activity (Fig. 4D-H).This was consistent with suppression of glycolytic genes such as aldolase A (ALDOA), phosphoglycerate kinase 1 (PGK1), 6-phosphofructo-2kinase/fructose-2,6-biphosphatase 3 (PFKFB3), lactate dehydrogenase A (LDHA; Fig. 7A), as well as SLC16A3, a gene encoding for monocarboxylic acid transporter 3, which transports pyruvate and lactic acid across membranes (Figs 3B and 8A; Fig. S7).The sustained metabolic stress after treatment with the OxPhos inhibitors (BAY-87-2243 and IACS-010759) suggests that they irreversibly blocked mitochondrial complex-I.A possible explanation is that the incomplete electron transfer in the mitochondria due to complex-I blockade generated free radicals, which damaged mitochondrial components, including inactivating complex-I activity [49].The cells metabolically switched toward a glycolytic phenotype to maintain energy levels as indicated by elevated ECAR (Fig. 4D-H).These findings highlight an antitumor activity of canagliflozin with a reduced risk for extracellular acidification and lactic acidosis.It indicates a significant advantage for canagliflozin over selective and potent mitochondrial OxPhos inhibitors, which have failed to show benefit in clinical trials [50,51].
Blockade of mitochondrial metabolism is intimately linked with AMPK activation and suppression of de novo lipogenesis, protein synthesis, and cell proliferation [7].We found that canagliflozin mediated activation of AMPK, inhibition of ACC and the mTOR-HIF-1a pathway, and an associated suppression of the mitosis marker, p-H3 Ser10 , in NSCLC cells.Similar results were obtained in other cancer models [22,25].In irradiated cells, canagliflozin amplified these effects, providing increased suppression of mTOR and its effectors.Notably, monotherapy with RT (5Gy) and oral canagliflozin induced sustained activation of the AMPK-ACC pathway and reduced p-H3 Ser10 in tumors, and these effects were enhanced with the combined treatment (Fig. 5I,J).The latter was coupled with the inhibition of activating ERK1/2 phosphorylation (Thr202/204) in nonirradiated and irradiated NSCLC cells (Fig. 5H).These data suggest that: (1) continuous oral delivery of canagliflozin mediates activation of AMPK, suppression of growth factor signaling and metabolic stress response in tumors, (2) RT also induces sustained elevation of baseline AMPK activity, and (3) the combined treatment further enhances these events in surviving tumor cells.
Our results suggest that canagliflozin treatment alone triggered G 1 /S phase arrest (Fig. 6A), which is in agreement with previous reports [52,53].Canagliflozin rapidly upregulated transcripts of the cell cycle regulation genes, TP53, CDKN1A, and CDKN1B at 2 h (Fig. 6B).Importantly, in irradiated cells, canagliflozin treatment led to enhancement of the RT-induced G 2 / M checkpoint and reduction of cells in G 1 and S phases.The G 2 /M phase is regarded as the most radiosensitive stage of the cell cycle [54].Therefore, this effect can be biologically important in the setting of fractionated RT where RT fractions are delivered daily or every 2-3 days.Our analysis of tumor death mechanisms suggested a detectable increase in apoptosis with canagliflozin, RT, and the combined treatment but no clear evidence of increased necrosis (Fig. 5; Fig. S5).The lack of extensive evidence of cell death in tumors weeks after cytotoxic therapy is not unexpected.The sustained improved suppression of growth in tumors receiving combined treatments may result from initial enhanced cell death with combined drug and RT treatment and continued cytostatic effects of canagliflozin in the long term.
HIF-1a is a key driver of glucose metabolism that promotes cell survival, migration, and angiogenesis, all of which contribute to tumor resistance to cytotoxic therapy [55].The work presented here shows suppression of HIF-1a activity by canagliflozin.Our RNA-seq data underlined the importance of canagliflozinmediated downregulation of the HIF-1a pathway.While we did not find HIF1A to be among the downregulated DEGs, its regulation by canagliflozin may be post-translational.The results of HIF-1a knockdown or pharmacological stabilization validated the notion of HIF-1a suppression as a critical mechanism for canagliflozin's activity (Fig. 7).It appears that loss of HIF-1a activity generates a vulnerability to metabolic stress induced by canagliflozin suggesting generation of a positive feedback loop whereby canagliflozin mediates sensitivity to its own effects and those of RT.A study by Faubert et al. [56] using A549 cells suggested that loss of the tumor suppressor, liver kinase B1 (LKB1), leads to metabolic reprogramming mediated by HIF-1a that involves increased expression of genes like ALDOA, PDK1, and LDHA and increased glucose and glutamine metabolism through the tricarboxylic acid (TCA) cycle.They postulated that HIF-1a targeting would generate an opportunity for synthetic lethality in LKB1-deficient tumors.Nevertheless, in this study, we observed effective suppression of HIF-1a, proliferation, and tumor growth in both LKB1deficient (A549) and LKB1-proficient (H1299) NSCLC cells and tumors (Fig. 7B,C; Fig. S4B).These findings support a strong potential for canagliflozin to reprogram tumor metabolism, inhibit NSCLC survival, and enhance the effects of cytotoxic therapy.
Our transcriptional regulatory network analysis identified HDAC2 as the putative gene mediating canagliflozin's transcriptional program (Fig. 8A).Histone deacetylases remove acetyl groups and are markers of epigenetic repression.They play global roles in regulating gene transcription, proliferation, and survival but are also notoriously overexpressed in solid malignancies [57].HDAC2 is shown to inhibit cell cycle checkpoints through suppression of genes like CDKN1A (p21 cip1 ) [58] and mediate cell migration and invasion through HIF-1a stabilization [59].Here, we found that canagliflozin and radiation suppress HDAC2 transcript (Fig. S7A) and the cellular HDAC2 protein levels while combined treatment does so more effectively (Fig. 8B,  C).The results of our HDAC2 knockdown experiments (Fig. 8) demonstrated that HDAC2 controls cellular HIF-1a and p21 cip1 levels and confers resistance to canagliflozin's antiproliferative activity consistent with the regulation of HIF-1a (Fig. 7).Recently, another study suggested that canagliflozin directly targets and suppresses HDAC6 in gastrointestinal cancer cells [60], indicating that HDAC targeting may be a common pathway for the activity of this drug in tumors.Many of the genes that were significantly upregulated by RT (FGFR1, ALDOA, RNF145, PGK1, P4HA1, PLOD2, IGFBP3, NDRG1, PFKFB3, and BNIP3L) are associated with glycolysis, cell cycle progression, prevention of apoptosis, angiogenesis, cholesterol homeostasis, and survival, which contribute to resistance to cytotoxic therapy.In contrast, canagliflozin downregulated these genes, highlighting additional pathways for this drug to curtail radioresistance induced by RT (see Fig. S7A,B for a complete list of canagliflozin-regulated DEGs and comparative effects of RT).
Studies with pancreatic and liver cancer xenograft models suggested that canagliflozin can improve chemotherapy responses [23,61].While the focus of this study was to investigate the combination of canagliflozin with RT for the treatment of NSCLC, we also found that canagliflozin improved the efficacy of the key chemotherapeutic agents of NSCLC, cisplatin, and etoposide, as well as concurrent chemo-RT in vitro (Fig. S9).These findings lay out arguments for future clinical investigations of canagliflozin in advanced NSCLC where combined cytotoxic treatment with chemo-RT is the standard of care.
Our findings that canagliflozin suppressed several genes involved in tumor survival, which hold poor prognostic value in clinical NSCLC datasets (Fig. 9), further strengthen the rationale for clinical investigation of this agent in NSCLC.However, beyond strengths shared with other agents with tumorsuppressive and RTÀ/chemo-sensitizing activities, canagliflozin holds a unique quality that may lead to an improved therapeutic ratio if combined with standardof-care therapies for LA-NSCLC.Clinical studies in the past 5 years show that unintentional delivery of RT to the heart during chest irradiation for LA-NSCLC, which often cannot be avoided due to the anatomical presentation of tumors, leads to poor overall survival [62,63].Notably, based on phase III randomized trial evidence, canagliflozin is now also approved for the prevention and management of heart failure [64], indicating a potential for cardio-protection.This notion and the lack of evidence associating canagliflozin with any toxicity in patients receiving cancer therapy provide additional rationale for canagliflozin to be investigated in LA-NSCLC.

Conclusions
Our results show that the widely used SGLT2 inhibitor, canagliflozin, exerts tumor-suppressive activity and enhances NSCLC response to standard cytotoxic therapy.Canagliflozin's antineoplastic effects are attributed, in part, to mitochondrial respiration blockade, inhibition of the mTOR-HIF-1a pathway, and substantial transcriptional reprogramming mediated likely through suppression of HDAC2.Canagliflozin is a well-tolerated medication that is approved for the treatment of type 2 diabetes and heart failure.It is well tolerated by nondiabetics and there is no known increase in toxicity when combined with standard cytotoxic therapy, making it a favorable agent for clinical investigation in NSCLC.

Fig. 4 .
Fig. 4. Effects of canagliflozin and RT on mitochondrial respiration in A549 NSCLC cells.A549 cells were pretreated for 5 h with the indicated canagliflozin (5-30 lM) doses prior to RT (single fraction of 2 or 8 Gy), and OCR and ECAR were measured with the Seahorse MST 48 h after RT.Effects of the treatments on (A) basal OCR, (B) basal ECAR, and (C) basal ECAR/OCR ratio were plotted relative to the untreated control.See Fig. S3 for time-dependent data from the MST.A549 cells were treated with canagliflozin (30 lM) and the mitochondrial complex-I inhibitors, BAY-87-2243 (10 nM) or IACS-010759 (10 nM), for 48 h and subjected to the Seahorse MST to measure (D) OCR and (E) ECAR.(F) Basal OCR and ECAR values from 'D' and 'E' were plotted relative to the untreated control.(G) A549 cells were treated for 48 h with canagliflozin (30 lM), BAY-87-2243 (10 nM), or IACS-010759 (10 nM) and subjected to the Seahorse GRA to measure the basal PER.All Seahorse data were normalized to cell content.(H) A proliferation assay of A549 cells treated with BAY-87-2243 (10 nM), IACS-010759 (10 nM), and/or canagliflozin (30 lM) for 5 days.Drug treatments were administered to the cells concurrently.All experiments were obtained from three independent experiments and presented as mean AE SEM.Data were analyzed using two-way ANOVA (A-C, F, G) or ordinary one-way ANOVA (H) followed by Tukey's multiple comparisons test.ns: nonsignificant, */ § P < 0.05, **/ § § P < 0.01, ***P < 0.001, ****/ § § § § P < 0.0001.

Fig. 6 .
Fig. 6.Regulation of the cell cycle by canagliflozin and RT in A549 NSCLC cells.(A) Cell cycle analysis using flow cytometry of A549 cells pretreated for 5 h with canagliflozin (10 or 30 lM) before RT (single fraction of 8 Gy) and analyzed 48 h later.(B) Relative mRNA expression, determined with RT-qPCR, of cell cycle regulation genes in A549 cells treated with canagliflozin (10 or 30 lM) for 2 h.See Fig. S6 for the canagliflozin time course up to 24 h.(C) Representative protein immunoblots of cell cycle regulation markers, p21 cip1 and p27 kip1 , in A549 cells pretreated for 5 h with canagliflozin (10 or 30 lM) followed by RT (single fraction of 2 or 8 Gy) and analyzed 48 h later (quantification in (D) normalized to b-actin).Data were obtained from three independent experiments and presented as mean AE SEM.Two-way ANOVA (D) or ordinary one-way ANOVA (B) followed by Tukey's multiple comparisons test was used to detect statistical significance.ns: nonsignificant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig.
Fig.8.Regulatory network analysis and identification of HDAC2 as the modulator of canagliflozin's gene expression profile.(A) All downregulated DEGs; FDR q-value < 0.05 from the RNA-seq data were analyzed using Cytoscape's iRegulon plugin to predict the master regulators that mediate canagliflozin's transcriptional program.HDAC2, which was suppressed by canagliflozin, was identified as the key mediator associated with the drug's gene expression profile.Gene regulatory network analysis illustrates the HDAC2-regulated genes that were significantly downregulated by canagliflozin in the RNA-seq dataset (Normalized Enrichment Score = 3.66; FDR q-value = 0.022).(B) Representative protein immunoblot and (C) quantification of HDAC2 in A549 cells pretreated with canagliflozin (10 or 30 lM) for 5 h and subsequently treated with RT (single fraction of 2 or 8 Gy) and analyzed 48 h later.(D) Representative protein immunoblots and of HDAC2, HIF-1a, and p21 cip1 in A549 cells treated with negative control siRNA (5 nM) or HDAC2 siRNA (5 nM) for 72 h (quantification of in (E)).All western blot data were normalized to b-actin.(F) Proliferation assay of A549 cells treated with canagliflozin (5-30 lM) and either negative control siRNA (5 nM) or HDAC2 siRNA (5 nM) for 5 days.Results were derived from three independent experiments and presented as mean AE SEM.Ordinary one-way ANOVA (E) or two-way ANOVA (C, F) followed by Tukey's multiple comparisons test was used to determine statistical significance.*P < 0.05, **P < 0.01, ****P < 0.0001.
legend.Tumors were formalin-fixed and paraffin-embedded (FFPE) by the McMaster Histology Core Facility.The McMaster University Animal Ethics Research Board approved all animal procedures (animal utilization protocol 20-12-47).