Galectin-1 modulates glycolysis through a GM1-galactose-dependent pathway to promote hyperthermia resistance in HCC.

INTRODUCTION
HCC is the most common primary liver cancer and the third leading cause of cancer-related death in the world.1 Thermal ablation has historically been utilized with curative intent for early-stage, nonsurgical patients with HCC.2 While the goal of thermal ablation is to treat the targeted HCC with adequate margins to achieve complete pathologic necrosis, there are select cases where there is rapid HCC progression after ablation, even in cases with technical success with adequate margins.3,4

The rationale for rapid HCC progression lies in the fact that sublethal hyperthermia is created at the edges of the thermal-ablation zone.5 The center zone has been well characterized in thermal-dose models and is usually associated with immediate cell death.6 However, this sublethal-hyperthermic environment is insufficient to eliminate peripheral tumor cells and can create a protumorigenic milieu, exacerbating metabolic derangements associated with more aggressive HCC.5 These metabolic interactions can be amplified in imaging-occult-HCC types, such as in poorly differentiated or macrotrabecular subtypes, when these are exposed to hyperthermic regions of the peri-ablational zone.7

Metabolic alterations within the tumor microenvironment are known to play a key role in postablation HCC cell survival and progression,8 as rapid growth requires increased energy production, macromolecular biosynthesis, and maintenance of redox balance. However, in-depth analysis of early-stage HCC tumor microenvironment and associated metabolic reprogramming, particularly under hyperthermic conditions, has not been well evaluated. Part of the reason for this knowledge gap is that studying the molecular aberrations of HCC requires clinical samples, which is currently not within the standard guidelines for HCC diagnosis.2 HCC management guidelines for diagnosing HCC primarily rely on imaging alone, restricting the use of needle biopsy in only select equivocal cases, rather than routine diagnosis.2 Consequently, the underlying mechanisms behind early-stage HCC postablation progression and associated metabolic pathways remain understudied.
A potential target that has been implicated in HCC progression is Galectin-1 (Gal-1), an evolutionarily conserved, glycan-binding protein associated with cancer invasion and treatment resistance.9–11 Gal-1 has been found to be overexpressed and modulate metabolism in other tumor types such as prostate, pancreas, and brain.11–13 Specifically, in glioma, Gal-1 functions as an intermediate-signaling activator for aerobic glycolysis13—a key ATP-producing mechanism conserved across many cancers, also known as the Warburg effect.14 However, the direct metabolic role of Gal-1 has not yet been linked to posttreatment cancer metabolism, especially after ablation in HCC. This study aimed to elucidate the mechanism by which Gal-1 mediated glycolysis and consequently the downstream tricarboxylic acid (TCA) cycle to enable HCC cells to resist peri-ablational hyperthermic stress. Specifically, we characterized the role of Gal-1 in promoting the hydrolysis reaction between GM1 and catalytic enzyme β-galactosidase (β-gal), ultimately creating galactose and subsequently metabolic influxes for glycolysis and the TCA cycle.
To elucidate the mechanism and metabolic role of Gal-1, we first employed liquid chromatography-mass spectrometry to analyze differential Gal-1 expression in a unique retrospective set of pre-ablation needle-biopsy samples from early-stage HCCs of patients subsequently identified as thermal-ablation responders and nonresponders. Subsequently, we demonstrated that Gal-1 modulation directly correlated with hyperthermia sensitivity and mediated the enhanced ability of HCC cells to utilize glycolysis and the downstream TCA cycle in vitro, particularly under hyperthermic stress. Finally, these results were validated in an orthotopic murine model, where the combined approach of ablation and Gal-1 inhibition resulted in a significant tumor reduction compared with monotherapy ablation or Gal-1 inhibition alone.

METHODS

Retrospective clinical study
This retrospective study was performed under written informed consent and approval from the UCLA Institutional Review Board (IRB#23-000131). A total of 58 patients who had been diagnosed with moderate-differentiated (n=54) or poorly differentiated (n=4) HCC were included in this study. These patient cohorts were matched for demographic, clinical baseline, and other tumor characteristics as shown in Supplemental Table S1, http://links.lww.com/HEP/J823.

Mouse studies
All mouse studies were performed in accordance with UCLA institutional ethical guidelines and are detailed in Supplemental Methods, http://links.lww.com/HEP/J822. Briefly, 5-week-old nude mice (homozygous for Foxn1<nu>, male, cat#002019, Jackson Laboratories) were orthotopically implanted with SNU449-cell–derived tumors. The study animals were then subjected to thermal ablation and/or 5 mg/kg OTX008 administration with the timeline (Figure 8A).

Mass spectrometry data analyses

Proteomic analysis
MSFragger software with a specific protein sequence database was used for identification. Raw MS1 was calibrated for m/z (mass-to-charge) based on all identified ions. Sample intensities were normalized using sample preparation and injection volumes.

Metabolomic analysis
Following data acquisition as detailed in Supplemental Methods and Materials, http://links.lww.com/HEP/J822, data were analyzed using Maven (version 8.1.27.11). Metabolites were identified based on accurate mass (±5 ppm) and previously established retention times of pure standards. Data analysis was performed using in-house R scripts.

Statistical analysis
The log-rank test was utilized for tumor progression-free survival rates. Multiple linear regression, Benjamin-Hochberg, and WGCNA were used in proteomics statistical analysis. Chi-square and unpaired 1-tailed Student t test were used for statistical analysis of clinical characteristics and in vitro and in vivo experiments, respectively. p values <0.05 were considered statistically significant.
Additional details about all the experiments in this study can be found in the Supporting Methods and Materials, http://links.lww.com/HEP/J822.

RESULTS

Thermal-ablation nonresponse correlates with Gal-1 overexpression in formalin-fixed paraffin-embedded needle-biopsy samples and HCC cells in vitro
Thermal ablation has become the standard of care for early-stage and nonresectable HCCs within BCLC 0-A criteria.2 Most patients experience an initial complete response after ablation, but up to 40% of patients will eventually progress (Figure 1A), despite having similar risk profiles.3,4 While the center of the ablation zone is associated with immediate cell death,6 the peripheries are associated with sublethal-hyperthermic microenvironment (Supplemental Figure S1A, http://links.lww.com/HEP/J824) which can promote HCC-aggressive phenotypes.5 A retrospective study of patients with matched-propensity scores (Supplemental Table S1, http://links.lww.com/HEP/J823) was conducted using a unique cohort of pre-ablation needle biopsies of early-stage HCCs. Patients were categorized as thermal-ablation responders, defined as those who had a complete response by localized mRECIST criteria for up to 2 years, compared with thermal-ablation nonresponders, defined as those who had progressive disease by localized mRECIST criteria within the 2 years, a time frame used in prior work.15 Ablation responders (n=34) had significantly longer tumor progression-free survival compared with nonresponders (n=24) (57.0±1.6 [median not reached] vs. 8.3±0.5 months [median: 13.6 mo], p <0.001), respectively (Figure 1B).
Proteomic analysis was performed using an untargeted-proteomic approach and identified 1712 proteins from 32 responsive and 23 nonresponsive formalin-fixed paraffin-embedded (FFPE) biopsy samples (Figure 1C). There were 59 proteins that were differentially expressed (nominal p< 0.05) (Figure 1D) and 19 pathways that were found enriched with log10-transformation of significant p values as shown (Figure 1E). Among the identified proteins, Gal-1 was selected for further study due to its previously established association with cancer aggressiveness and treatment resistance,9–11 and, importantly, because its direct role in HCC metabolism remains largely unexplored. Therefore, understanding the role of Gal-1 in metabolic reprogramming could potentially identify breakthrough and drug-targetable points for improving HCC clinical outcomes. In addition, Gal-1 levels were analyzed using immunohistochemistry staining on respective samples to further confirm its differential expression levels found on FFPE proteomic analysis. This analysis revealed that Gal-1 was significantly more expressed in nonresponders compared with responders (Figures 1G, H). Normal liver tissues were also stained for Gal-1, which showed significantly less Gal-1 expression compared with nonresponders but similar to responders (Figures 1G, H).
Next, an in vitro model was created to simulate the clinical thermal-ablation model to further investigate Gal-1 functions. HCC-derived cell lines (SNU423, SNU449, and HepG2/C3a) were exposed to either normothermic (37°C) or sublethal-hyperthermic (47°C) temperatures in a water bath for 25 minutes, with the initial 15 minutes for equilibration (Supplemental Figures S1B, C, http://links.lww.com/HEP/J824). Consistent with prior studies simulating the peri-ablational zone,16,17 sublethal temperature defined as 47°C was used to induce partial cell death without necrosis to allow for in-depth metabolic investigations. Cell-death percentage under hyperthermia was then assessed at 24, 48, and 72 hours. SNU449 was found to be significantly more resistant against hyperthermia, with a markedly lower cell-death percentage at 47°C compared with SNU423 (Figures 2A, B). This resistance pattern was verified with bright-field microscopy (Supplemental Figure S1D, http://links.lww.com/HEP/J824), confirming SNU423 and SNU449 as hyperthermia-responsive and resistant cells, respectively. The relation of Gal-1 to hyperthermia resistance, previously noted in FFPE samples (Figure 1F), was then evaluated. Immunoblotting confirmed Gal-1 upregulation in hyperthermia-resistant SNU449 compared with hyperthermia-responsive SNU423 cells (Figure 2C). These findings suggest a link between Gal-1 and thermal-ablation responsiveness in patients with HCC and HCC cell lines.

Gal-1 overexpression induces sublethal-hyperthermia resistance in HCC cells in vitro
Given the observed Gal-1 overexpression in hyperthermia-resistant SNU449 (Figure 2C), selective modulation of Gal-1 was performed to isolate its role in promoting hyperthermia resistance. SNU449 was treated with a predetermined dose of 50 µM Gal-1 inhibitor (OTX008) (Supplemental Figures S2A, B, http://links.lww.com/HEP/J824) before thermal exposure.11 SNU449 growth in the OTX-treated group under hyperthermia was markedly diminished compared with that exposed only to hyperthermia at 47°C (Figure 2D). Cell-death percentage demonstrated that Gal-1 inhibition also significantly increased hyperthermic sensitivity (Figure 2E). Similar increases in hyperthermic sensitivity were noted in SNU423 with Gal-1 inhibition (Supplemental Figures S2C, D, http://links.lww.com/HEP/J824).
Gal-1 was then silenced using lentiviral-shRNA particles in SNU449 (shGal-1-SNU449) (Supplemental Figure S3A, http://links.lww.com/HEP/J824). Growth-rate analysis showed shGal-1-SNU449 was significantly more susceptible to hyperthermia than shControl-SNU449 (Figure 2F). Cell survival studies corroborated the cell growth results (Figure 2G), further suggesting that silencing Gal-1 enhanced hyperthermic sensitivity. To investigate the rescuing effect of increased Gal-1 expression, Gal-1-knockdown SNU449 (shGal-1-SNU449) was then overexpressed with Gal-1 using pLentivirus-ORF particles (pGal-1) (Supplemental Figure S3B, http://links.lww.com/HEP/J824). Cell growth and survival analyses under hyperthermia showed that pGal-1 was significantly more resistant to hyperthermia than its respective control shGal-1 (Figures 2H, I). Altogether, these findings underscore the critical role of Gal-1 in promoting hyperthermia resistance in HCC.

Gal-1 overexpression increases the complex formation between Gal-1 and the N-terminal of P-glycoprotein through an O-GlcNAcylation–dependent pathway under hyperthermia in HCC cells
To further understand the role of Gal-1 in hyperthermia resistance, Gal-1 expression was assessed under hyperthermia and normothermia in shControl-SNU449 and shGal-1-SNU449. This analysis revealed no significant changes in Gal-1 (15 kDa) levels in both cell lines (Figure 3A). Because Gal-1 is colocalizing with GM1 (monosialotetrahexosyl-ganglioside),18 a GM1-immunoblotting study was performed by using cholera-toxin subunit B and cholera-toxin subunit B antibody.19 This analysis showed a significant upregulation of a protein around 105 kDa after hyperthermia exposure in both cell lines (Figure 3B). This protein was investigated further and found to be composed of Gal-1 because the same 105 kDa band appeared when Gal-1 was probed (Figure 3C). In addition to Gal-1 and GM1, the protein makeup was suspected to require additional components because Gal-1 and GM1 together still did not fully account for the total weight. Prior work has suggested a strong affinity between P-glycoprotein (P-gp), an ATP-binding transporter, and Gal-1.18,20 Specifically, Gal-1 binds to N-terminal of P-gp (N-P-gp) in the cytosol.18,21 Therefore, it was hypothesized that reducing P-gp, and consequently N-P-gp, would decrease the 105 kDa protein expression.
To reduce P-gp, a selective inhibitor (Wortmannin/Wort) of the PI3K pathway regulating P-gp expression,10 was used under normothermia at a predetermined concentration from a titration study (Supplemental Figure S4A, http://links.lww.com/HEP/J824). This investigation was performed in SNU449-WT (wild type) as these cells were overexpressed with P-gp (Supplemental Figure S4B, http://links.lww.com/HEP/J824). GM1 (Figure 3D) and Gal-1 (Figure 3E) were then probed, and both demonstrated a decrease in the 105 kDa protein expression upon P-gp reduction. These findings collectively indicate that the 105 kDa protein was formed by GM1 binding to Gal-1, which then binds to the N-P-gp.
The mechanism of hyperthermia-induced upregulation of GM1/Gal-1/N-P-gp complex (Figures 3B, C) was then evaluated. P-gp binds to Gal-1 through N-acetylglucosamine residues on P-gp.18,20,21 These residues are subject to O-GlcNAcylation (O-GlcNAc), a stress-responsive posttranslational modification that mediates the binding interactions between O-linked N-acetylglucosamine with surrounding proteins.22 Furthermore, O-GlcNAc has been found to be modified by hyperthermia exposure in HCC.16,22 Therefore, we hypothesized that O-GlcNAc may impact the formation of GM1/Gal-1/N-P-gp complex under hyperthermia. shControl-SNU449 was used for this study because its Gal-1 overexpression would allow for the opportunity to analyze the formation of GM1/Gal-1/N-P-gp. First, O-GlcNAc was assessed and found to be significantly decreased in shControl-SNU449 under hyperthermia compared with normothermia (Figure 3F). PUGNAc (100 µM), an O-GlcNAc-breakdown inhibitor,16 was used to preserve the full O-GlcNAc production, allowing for a complete evaluation of the hyperthermic impact. The relationship between Gal-1 and O-GlcNAc reduction under hyperthermia was then investigated. Specifically, cytosolic Gal-1 levels were assessed because O-GlcNAc occurs in the cytosol.22 When O-GlcNAc was maximally suppressed by combining hyperthermia with O-GlcNAc inhibition (l-6-Diazo-5-oxonorleucine [DON]),23 cytosolic Gal-1 was found to be significantly reduced compared with either hyperthermia or DON-treated alone groups (Figure 3G). This finding indicates that decreased O-GlcNAc led to increased Gal-1 secretion and consequently reduced Gal-1 remaining in the cytosol, consistent with prior work.23 As the number of Gal-1 molecules translocating through the membrane increased upon O-GlcNAc reduction, the interactions between Gal-1 and a membrane protein like P-gp would increase.24 Thus, the formation of GM1/Gal-1/N-P-gp would hypothetically increase upon decreased O-GlcNAc levels under hyperthermia as illustrated (Figure 3H). Expectedly, when O-GlcNAc level was maximally suppressed in the combined hyperthermia/DON-treated group, GM1/Gal-1/N-P-gp complex exhibited the largest increase in expression compared with either hyperthermia or DON-treated alone (Figure 3I). In addition to being mediated by O-GlcNAc, GM1/Gal-1/N-P-gp complex formation was demonstrated to be dependent on Gal-1 expression. When O-GlcNAc was suppressed maximally in the combined hyperthermia/DON-treated group, GM1/Gal-1/N-P-gp was significantly more expressed in Gal-1-overexpressing-shControl-SNU449 than in Gal-1-underexpressing-shGal-1-SNU449 (Figures 3J, K). Overall, these findings demonstrate that the formation of GM1/Gal-1/N-P-gp complex under hyperthermia depends on both O-GlcNAc and Gal-1 expression.

Gal-1 complexes with the N-P-gp to modulate glycolysis through GM1-hydrolyzed galactose production
The signaling functions of Gal-1 have been extensively studied,10,11 but its direct role in HCC metabolism remains unclear. Glycolysis, a key metabolic pathway in HCC and many other cancers, is crucial for tumor progression.14 The FFPE-biopsy-proteomic analysis revealed enrichment of glycolysis-related pathways (Figure 1E), including β-catenin–mediated glycolysis through pyruvate dehydrogenase kinase isozyme 1 expression and NOTCH-mediated glycolysis through p53 or PI3K/AKT signaling.25 Furthermore, levels of glycolysis-related proteins (Figure 4A)—glucose 6-phosphate isomerase, glyceraldehyde-3-phosphate dehydrogenase, and phosphoglycerate kinase—from the differential expression protein analysis (Figure 1D) were also found to be upregulated in nonresponders compared with responders (Figures 4B–D). These findings led us to specifically characterize the mechanistic underpinnings between increased glycolysis and Gal-1 overexpression. This investigation started with SNU423 as hyperthermia-responsive and SNU449 as hyperthermia-resistant cells, respectively (Figures 2A, B). Glycolysis was noted to be significantly upregulated in SNU449 compared with SNU423, resulting in increased glycolytic ATP production (Figures 4E, F). When Gal-1 was inhibited with OTX in SNU449 WT, glycolysis was expectedly decreased along with a concomitant decrease in glycolytic ATP production (Figures 4G, H), establishing the direct link between Gal-1 and glycolysis.
The specific mechanism of Gal-1–mediated glycolysis was then evaluated. An increase in the complex formation of GM1/Gal-1/N-P-gp was identified under hyperthermia (Figures 3B, C), providing the basis for our rationale. β-gal then emerged as a key molecule of interest due to its role in catalyzing GM1 hydrolysis to produce galactose,26 a precursor for the glycolytic substrate glucose 6-phosphate. β-gal has a high affinity for the N-P-gp, which separately binds to Gal-1 as illustrated (Figure 4I).21,26,27 Therefore, P-gp expression reduction, and consequently N-P-gp expression, would reduce N-P-gp/β-gal complex. To test this, P-gp expression was reduced using Wort, leading to a decrease in N-P-gp/β-gal at ~193 kDa (Figure 4J), consistent with prior work.27 To confirm the association of N-P-gp and β-gal, a selective β-gal inhibitor (Supplemental Figure S4C, http://links.lww.com/HEP/J824), was used. Expectedly, N-P-gp/β-gal complex levels were increased upon β-gal inhibition (Figure 4K). This trend suggests that reduced β-gal catalytic activity decreased its dissociation from N-P-gp. Consequently, β-gal inhibition decreased GM1 hydrolysis, resulting in increased intact GM1 to bind Gal-1, which binds to N-P-gp (Figure 4L). Collectively, these findings further support the mechanism of Gal-1–mediated glycolysis through an N-P-gp/β-gal-GM1–dependent pathway as illustrated (Figure 4I).
The proposed mechanism (Figure 5A) indicates that Gal-1 overexpression would allow cells to hydrolyze more GM1 to produce more galactose. To further study this, galactose levels were then measured in Gal-1-overexpressing-shControl and Gal-1-underexpressing-shGal-1 using an Abcam-Galactose assay. This analysis showed a significant decrease in galactose in shGal-1 compared with shControl (Figure 5B). Moreover, the catalytic activity of β-gal in GM1 hydrolysis to produce galactose was reduced in shGal-1 compared with shControl (Figure 5C). To confirm if decreased galactose would reduce glycolysis, ECAR (extracellular acidification rates, an indicator of glycolysis) was assessed. Compared with shControl, shGal-1 showed a marked reduction in glycolysis and glycolytic ATP production (Figures 5D, E). These findings suggest that galactose metabolism, and thus glycolytic influx, is less upregulated in shGal-1.
To validate this implication, U-13C6 galactose, a carbon-13–labeled form of d-galactose, was utilized to trace galactose metabolism within the Leloir pathway, the primary metabolic pathway for galactose utilization,28 and glycolysis (Figure 5F). U-13C6 galactose-1-phosphate (M+6) was increased in shGal-1 (Figure 5G), suggesting that shGal-1 cells were not as heavily consuming galactose-1-phosphate to promote increased glycolysis as shControl cells. In addition, a significantly higher concentration of unmetabolized U-13C6 galactose was found in shGal-1 (Figure 5H), further confirming the downregulation of the Leloir pathway in shGal-1. The products of U-13C6 galactose metabolism were then traced for their contribution to downstream glycolysis. Consistent with decreased glycolysis in Gal-1-underexpressing-shGal-1 cells (Figure 5D), glycolytic metabolite U-13C6 glucose-6 phosphate (M+6) was decreased in shGal-1 compared with shControl (Figure 5I). Other downstream glycolytic metabolites, 3/2-phosphoglycerate (3/2-PG, M+3) (Figure 5J) and lactate (M+3) (Figure 5K), were also markedly reduced in shGal-1 compared with shControl. In summary, these findings confirm the role of Gal-1–mediated glycolysis through galactose production to support HCC proliferation.

Gal-1 silencing and inhibition further reduce metabolites and activities of glycolysis and mitochondrial TCA cycle under hyperthermia
Thus far, the present study has demonstrated the mechanism by which Gal-1 mediates glycolysis in HCC under normothermia (Figure 5A). We then sought to determine the effects of Gal-1 modulation by genetic silencing or pharmacological inhibition under hyperthermia by examining the metabolites and activities of glycolysis and the downstream TCA cycle. This study employed U-13C6 glucose (M+6), a carbon-13–labeled form of d-glucose, to track glucose metabolism in Gal-1-overexpressing-shControl and Gal-1-underexpressing-shGal-1 (Figure 6A). Lactate, a glycolysis end product whose concentration reflects glycolytic activity,14 (M+3)-isotopolog levels were significantly reduced in shGal-1 compared with shControl under hyperthermia (Figure 6B). Moreover, glycolytic metabolite phosphoglycerate (3/2-PG) (M+3) was reduced in shGal-1 compared with shControl under hyperthermia (Figure 6C). These findings suggest that low Gal-1 expression further reduces glycolysis under hyperthermia. In addition, metabolites of the TCA cycle were further decreased in shGal-1 compared with shControl under hyperthermia. Specifically, citrate (M+0 to M+6) isotopologs were markedly reduced in shGal-1 compared with shControl under hyperthermia (Figure 6D). Similarly, malate (M+0 to M+4) levels were significantly decreased in shGal-1 compared with shControl under hyperthermia (Figure 6E). Total levels of lactate (Figure 6F), 3/2-PG (Figure 6G), citrate (Figure 6H), and malate (Figure 6I) were also measured, which consistently showed significant decreases in shGal-1 compared with shControl under hyperthermia. These metabolic reductions could link decreased Gal-1 expression with significant reductions in glycolysis and TCA cycle activities. To confirm this implication, the glycolytic and mitochondrial TCA cycle activities were measured under the hyperthermic conditions by using a Seahorse analyzer to quantify ECAR (an indicator of aerobic glycolysis) and oxygen consumption rate (an indicator of the TCA cycle activity through oxidative phosphorylation).29 This analysis showed marked decreases in glycolysis and TCA cycle activities (Figures 6J, K) along with reduced ATP production rates (Supplemental Figures S5A, B, http://links.lww.com/HEP/J824) in shGal-1 compared with shControl under hyperthermia.
To further investigate the effects of Gal-1 modulation on glycolysis and the TCA cycle, the isotopolog tracings of U-13C6 glucose metabolism were performed in SNU449 WT with Gal-1 inhibition (OTX) (Figure 7A). Compared with control, the Gal-1–inhibited group exhibited reductions in various glycolytic metabolites, lactate (M+3), fructose 1,6-bisphosphate (M+6), 3/2-PG (M+3), and phosphoenolpyruvate (M+3), under hyperthermia (Figures 7B–E). Also, the Gal-1–inhibited group exhibited significant decreases in citrate (M+0 to M+6) and malate (M+0 to M+4) within the TCA cycle compared with the control under hyperthermia (Figures 7F, G). Total levels of these metabolites were then measured and showed significant reductions in the Gal-1–inhibited group compared with the control under hyperthermia (Figures 7H–M). To confirm whether the metabolic decline would lead to a reduction in the activities of glycolysis and the downstream TCA cycle, ECAR and oxygen consumption rate, respectively, were quantified. Indeed, this analysis showed significant decreases in the activities of glycolysis and TCA cycle (Figures 7N, O) as well as associated ATP production rates (Supplemental Figures S5C, D, http://links.lww.com/HEP/J824) in the Gal-1–inhibited group compared with control under hyperthermia. Collectively, these findings underscore the importance of Gal-1 in facilitating energy production through glycolysis and the TCA cycle for HCC resistance against hyperthermia-induced stress.

Targeting Gal-1 by selective inhibitor OTX008 enhances thermal-ablation efficacy in hyperthermia-resistant HCC tumors
Hyperthermia-resistant SNU449 cells were used to evaluate the efficacy of Gal-1 inhibition in improving postablation tumor control in an orthotopic model. This cell line was used because success in controlling its tumor progression would further emphasize the critical role of Gal-1 in mediating glycolysis and consequently the TCA cycle in promoting metabolic plasticity. After subcutaneous SNU449-cell–derived tumors were orthotopically implanted into the livers of 5-week-old male nude mice, each subject was administered with OTX before ablation as shown (Figure 8A). In accordance with institutional ethical guidelines, mice were sacrificed 1 week after ablation, and tumors were harvested. Tumor growth was significantly reduced in the combined ablation and OTX-treated group compared with ablation or OTX alone (Figures 8B, C). As expected, there was no tumor reduction observed in the monotherapy ablation group compared with control, given that SNU449 cells were previously shown to be resistant to hyperthermia (Figures 2A, B, Supplemental Figures S1C, D, http://links.lww.com/HEP/J824).
The tumor-metabolic profile was then characterized using an Abcam-Galactose assay. Galactose levels were substantially reduced in the combined-treatment group compared with thermal ablation or OTX alone (Figure 8D). Levels of galactose-metabolism product, galactose-1-phosphate (Figure 8E), were also measured using mass spectrometry, which showed a marked decrease in the combined treatment compared with ablation or OTX alone (Figure 8F). These findings suggest reduced galactose metabolism in the combined-treatment group, potentially due to decreased Gal-1 expression. This aligns with previous in vitro studies linking decreased galactose metabolism to reduced Gal-1 (Figure 5B). To further confirm this, tumor Gal-1 levels were assessed, which showed the largest Gal-1 decrease in the combined-treatment group (Figure 8G). While unbound-Gal-1 expression was decreased, the levels of Gal-1 binding to N-P-gp were increased in the combined-treatment or thermal-ablation groups compared with control (Figure 8G). This finding aligns with our previous observation of increased Gal-1/N-P-gp complex under hyperthermia (Figure 3C). Glycolytic metabolites, fructose 1,6-bisphosphate and phosphoenolpyruvate (Figure 8E), were assessed and found to be markedly decreased in the combined treatment compared with ablation or OTX alone groups (Figures 8H, I), further confirming decreased glycolysis. Citrate and malate, TCA cycle metabolites (Figures 8J, K), were also markedly reduced in the combined group compared with ablation or OTX alone, corroborating previous in vitro findings of decreased TCA cycle metabolites upon Gal-1 inhibition (Figures 7L, M). Collectively, the findings in vitro and in vivo highlight the critical role of Gal-1 in mediating glycolysis and consequently the downstream TCA cycle to allow cells to meet their increased energy demand under hyperthermia.

DISCUSSION
This study investigated the role of Gal-1 in mediating glycolysis and consequently the downstream TCA cycle through a GM1-galactose–dependent pathway to promote hyperthermia resistance and postablation progression in early-stage and nonresectable HCC. Gal-1 upregulation was retrospectively identified from the pre-ablation FFPE biopsy samples of thermal-ablation nonresponders. Gal-1 overexpression was then linked to HCC tumor cells’ enhanced ability to utilize glycolysis and the downstream TCA cycle under sublethal hyperthermia–induced stress. This process was facilitated by upregulating the formation of GM1/Gal-1/N-P-gp complex through an O-GlcNAc–dependent pathway, to bridge GM1 to the catalytic enzyme β-gal. In addition, Gal-1 inhibition (OTX008) or knockdown sensitized HCC cells to sublethal hyperthermia by diminishing the metabolic fluxes and activities of glycolysis and the downstream TCA cycle. Importantly, in vivo studies using an orthotopic murine model showed that the combination of Gal-1 inhibition and ablation led to significant tumor size reduction compared with ablation alone. These results suggest that Gal-1 mediation of glycolysis and consequently the TCA cycle may contribute to thermal-ablation resistance and postablation progression in early-stage HCC.
Thermal ablation causes immediate cell death at the center of the ablation zone.2,6 However, the peri-ablational zones associated with hyperthermia have been correlated with rapid tumor progression in more aggressive subtypes of HCC.5,16,17 Hyperthermia-induced glycolysis has been proposed to be a key factor contributing to HCC tumor progression.16 The present study extends beyond glycolysis to reveal the critical role of the TCA cycle in HCC postablation progression. These mechanisms of action are found to be specifically orchestrated by Gal-1, a glycan-binding protein whose overexpression correlates with aggressive clinicopathological features such as tumor invasion, angiogenesis cancers, and drug resistance.9–11 Unlike these prior studies focusing on the genetic-regulatory functions of Gal-1 in cancer aggressiveness, this study demonstrates the direct metabolic regulation of Gal-1 in HCC progression, particularly after ablation.
Using an in vitro peri-ablational hyperthermia model in the present study, Gal-1 was shown to directly modulate aerobic glycolysis through GM1 hydrolysis to produce galactose. The galactose metabolite is then metabolized to produce glucose 6-phosphate for glycolysis. This glycolytic modulation consequently led to corresponding changes in the metabolic levels and activity of the downstream TCA cycle. Galactose metabolism (Leloir pathway) combined with glycolysis has been hypothesized to yield zero-net ATP compared with glycolysis alone, due to the energy cost of producing UDP-glucose.30,31 Nevertheless, the produced UDP-glucose can participate in multiple catalytic cycles to generate UDP-galactose, thereby offsetting the initial energy expenditure.32 Ultimately, the combined pathways will still yield 2 net ATP molecules and result in increases in the glycolytic influx and consequently the metabolic influx and activity of the TCA cycle for further ATP production.28 Ultimately, this mechanism allows Gal-1–overexpressing HCCs to meet increased energy demand under hyperthermia. Gal-1 can thus serve as a promising therapeutic target, particularly for patients with HCC with an elevated risk of recurrence following ablation.7

The in vivo aspect of this study provides key insights into the therapeutic potential of targeting Gal-1 in combination with ablation. This combined strategy led to significantly greater reductions in tumor size and metabolic levels of glycolysis and downstream TCA cycle compared with either ablation or OTX-treatment alone. These reductions also correlated with Gal-1 downregulation in the combined group. Gal-1 downregulation was likely due to increased activity of the 26S proteasome, which degraded the complex formed by Gal-1 and Gal-1 inhibitor OTX at a faster rate under hyperthermia.11,33 These results indicate that systematically administering Gal-1 inhibitor (OTX008) beforehand to form a Gal-1-OTX complex that is then rapidly degraded under thermal ablation may provide significant clinical insights. This approach would lead to a reduction in glycolysis and consequently TCA cycle activity, both of which correlate with poor prognosis in patients with HCC.34,35 The safety of Gal-1 inhibitor OTX008 has previously been demonstrated in a first-in-man trial in 2013, where patients with treatment-refractory metastatic colorectal cancer were administered with OTX008 as a daily subcutaneous injection.36 The dose escalation study showed that a daily subcutaneous injection at 65 mg was safe without dose-limiting toxicity, although plasma concentration was dependent on body weight. Overall, the strategy of combining thermal ablation with a Gal-1 inhibitor holds considerable clinical potential for improving patient outcomes.
In addition, targeting Gal-1 offers broader clinical implications for improving HCC patient outcomes. As HCC biopsy becomes more widely accepted, biomarkers associated with upregulated aerobic glycolysis and poor survival in HCC (eg, pyruvate kinase M2 or hexokinase-2)35 can identify patients who would benefit from neoadjuvant Gal-1 inhibition therapy. Improving local control through Gal-1 modulation can also potentially keep an early-stage patient with HCC within the Milan criteria and reduce the risk for transplant drop-off.2 For late-stage HCC, Gal-1 has been established as a potential mediator for angiogenesis and promoter of immune evasion in late-stage HCC9,37—where checkpoint inhibition and antiangiogenic drugs have taken center stage.38 Gal-1 overexpression, known to upregulate PD-1/PDL-1 ligands and induce immunotherapy resistance in other tumors,37 suggests that Gal-1 inhibition could enhance checkpoint inhibitor efficacy in HCC. Given the implicated roles of Gal-1 in both early- and late-stage HCC, its inhibition offers a potential adjuvant or neoadjuvant approach to extend response duration across all stages.
There are several limitations in this study. The biopsy samples from early-stage HCCs may be subject to intratumoral heterogeneity. However, we focused on evolutionarily conserved biomarkers across multiple cell types to mitigate the heterogeneity risks. Another limitation of the in vivo studies is that they were performed using a single HCC cell line (SNU449). Nevertheless, this thermally resistant model was effective in highlighting the potential of Gal-1 inhibition in overcoming ablation resistance. Future studies will be useful in validating the results in patient-derived HCC tumors. Furthermore, the focus on the roles of Gal-1 in mediating glycolysis and consequently the TCA cycle, while important for tumor metabolism, may have also inadvertently overlooked other pathways, namely Gal-1–mediated lipid metabolism.39 Finally, longer follow-up in vivo is required for a comprehensive assessment of the survival benefits of Gal-1 inhibition.
Rapid postablation progression for early-stage, nonresectable HCC has been an active area of research, particularly because ablation is often applied with curative intent or as a bridge to transplant.2 This present study provides compelling evidence that Gal-1 plays a critical role in regulating metabolic plasticity in HCC after ablation. By facilitating glycolysis through a GM1-galactose–dependent pathway and consequently the TCA cycle, Gal-1 enables HCC cells to persist under hyperthermic conditions. Inhibiting Gal-1 can potentially disrupt this metabolic adaptation, which was demonstrated in vitro and in vivo. Thus, combining Gal-1 inhibitor with ablation can potentially improve patient outcomes after ablation. While additional research is needed to fully understand the broader metabolic impacts of Gal-1 inhibition, this study represents an important step forward in the development of more effective treatments for early-stage, nonresectable HCC.

Supplementary Material