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

Abstract

Background and aims: Thermal ablation is the standard-of-care treatment modality with curative intent for early-stage nonresectable HCC, but a durable response is limited, with up to 40% of patients with HCC eventually experiencing local recurrence on posttreatment surveillance. While thermal ablation has been established to cause immediate cell death in the center of the thermal ablation zone, its metabolic impact in the peri-ablational region remains unclear. We aimed to elucidate the metabolic mechanism by which Galectin-1 (Gal-1) promotes thermal-ablation-induced hyperthermia resistance in HCC and demonstrate the therapeutic potential of inhibiting Gal-1 in combination with thermal ablation in vivo .

Approach and results: Proteomic analysis was performed using an untargeted approach on pre-ablation formalin-fixed paraffin-embedded biopsy specimens of thermal ablation responders (n=32) and nonresponders (n=23). Gal-1 was found to be overexpressed in thermal ablation nonresponders compared with responders. Moreover, HCC with Gal-1 overexpression demonstrated reduced sensitivity to hyperthermia in vitro and increased utilization of glycolysis and the downstream tricarboxylic acid (TCA) cycle under hyperthermia-induced stress. Gal-1-overexpressing HCC enhanced its metabolic utilization through Gal-1-facilitated GM1-ganglioside breakdown, producing galactose to increase the metabolic influxes into glycolysis and consequently the downstream TCA cycle. In-vivo studies showed that inhibiting Gal-1 in combination with thermal ablation significantly reduced tumor size compared with either monotherapy thermal ablation or Gal-1 inhibition alone.

Conclusions: Gal-1 can mediate hyperthermia resistance in HCC and can potentially be modulated as a therapeutic target to reduce rapid progression after thermal ablation.

Keywords: GM1-ganglioside; leloir pathway; liver cancer recurrence; thermal-ablation; tricarboxylic acid cycle.

Figure 1:. Upregulation of Galectin-1 is correlated with nonresponsiveness to thermal-ablation therapy in HCC.

A) (Top) T1-weighted non-contrast MRI of an HCC patient showing a completely ablated HCC tumor (yellow arrow) and adequate margins (orange arrowhead) around the target HCC. (Bottom) Three-month follow-up post-contrast-arterial phase MRI demonstrating rapid new growth (green arrowhead) around the prior ablation zone. B) Kaplan-Meier curve analysis of tumor progression-free survival probability in HCC patients who had stable disease (responders) versus progressive disease (nonresponders) to thermal-ablation therapy. C)The plot displays protein identifications across 55 Formalin-Fixed Paraffin-Embedded (FFPE) liver tumor samples, stratified by treatment response (n=32 responders and 23 nonresponders). Data points represent individual samples, with responders shown in blue and nonresponders in red. The black line indicates the Loess smoothed trend with 95% confidence intervals (gray shading). Above the main plot, two annotation tracks display z-score-normalized values (white to red gradient) for tumor size and alpha-fetoprotein (AFP) levels, winsorized at ±2 standard deviations. D) Heatmap of differentially expressed proteins (nominal p < 0.05) in tumors classified as responders or nonresponders to ablation. Protein expression values are represented as winsorized Z-scores ranging from −2 to 2, with red indicating higher expression and blue indicating lower expression. Columns correspond to individual samples, clustered by hierarchical clustering, while rows correspond to proteins, sorted first by manually assigned functional category (left color bar) and then by t-value. Sample-associated annotations are displayed at the top, including tumor size (<3 cm vs. >3 cm), AFP levels (<200 ng/mL vs. >200 ng/mL), and response classification (responder vs. nonresponder). Functional categories include cytoskeleton, expression-related, immune-related, metabolism, mitochondria, and trafficking. E) Visualized as a barplot of −log₁₀(P-value) for enriched biological pathways. Pathways enriched in nonresponders (positive enrichment) are shown in red bars extending to the right, while pathways enriched in responders (negative enrichment) are shown in dark blue bars extending to the left. The x-axis represents the −log₁₀(P-value), indicating the statistical significance of enrichment, with higher absolute values reflecting greater significance. F) Upregulated protein levels of Galectin-1. The abundance levels were log₁₀-transformed. Box plots show median (central line), upper and lower quartiles (box limits), and 1.5 interquartile range (whiskers). G) Matched FFPE biopsy samples (n=17 responders and 11 nonresponders) and randomly-selected healthy liver samples (n=11) were stained for Galectin-1 using immunohistochemistry. H) Cells positive for the antibody were quantified using HALO software to calculate mean number of positive cells. The cell numbers were then normalized by total tissue areas. P-value was calculated using one-tailed and unpaired Student’s t-test with *p<0.05, **p<0.01, ***p<0.001.

Figure 2:. Galectin-1 overexpression induces sublethal hyperthermia resistance in HCC cells.

A-B) SNU449 cell survival percents (n=3 each condition) (A) post thermal exposure at 37°C and 47°C at 24, 48, and 72 hours. Corresponding cell growth curve (B) at the same time intervals. C) Western blot showing the expression of Gal-1 in SNU449, SNU423, and HepG2/C3a (positive control). β-Tubulin expression was used as loading controls. D-E) SNU449 cell-growth curve (n=3 each condition) (D) post-thermal exposure at 37°C and 47°C with Gal-1 selective inhibitor OTX (50μM) or DMSO at 24, 48, and 72 hours. Corresponding cell-survival percentage (E) at the same time intervals. F-G) shControl and shGal-1-SNU449 cell growth curve (n=3 each condition) (F) post-thermal exposure at 37°C and 47°C at 24, 48, and 72 hours. Corresponding cell-survival percentage (G) at the same time intervals. H-I) shGal-1-SNU449 (Gal-1 knockdown, control group) and pGal-1-SNU449 (Gal-1 overexpression using pLentivirus-ORF particles, experimental group) cell growth curve (n=3 each condition) (H) post-thermal exposure at 37°C and 47°C at 24, 48, and 72 hours. Corresponding cell-survival percentage (I) at the same time intervals. P-values were calculated using one-tailed, unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001.

Figure 3:. Galectin-1 overexpression increases the complex formation between Gal-1 and N-terminal of P-glycoprotein via an O-GlcNAcylation-dependent pathway under hyperthermia (47°C) in HCC cells.

A) Western blot showing levels of Galectin-1 (Gal-1) in shGal-1-SNU449 (SNU449 cells with Gal-1 silencing via lentiviral particles) and respective shControl-SNU449 cells after thermal exposure at 37°C or 47°C. β-Tubulin expression was used as loading controls. N=2 duplicates per group. B-C) Western blot showing levels of GM1 binding to the complex formed by N-terminal-half of P-glycoprotein and Gal-1 (N-P-gp/Gal-1) in shGal-1-SNU449 and respective shControl-SNU449 cells after thermal exposure at 37°C or 47°C. GM1 levels were measured using western blot to stain for cholera toxin B (CTBx) subunits which have high affinity for GM1. Therefore, after the thermal treatments, cells were added 10μM of CTBx to stain for GM1. The cells were then harvested and stained using polyclonal CTBx antibody for GM1 staining (B) or Gal-1 for Gal-1 staining (C). N=2 duplicates per group. Dotted lines indicate discontinuous wells on the same membrane. D-E) Western blot showing levels of GM1 binding to the complex formed by N-terminal-half of P-glycoprotein and Gal-1 (N-P-gp/Gal-1), in SNU449 WT treated with DMSO or Wortmannin (Wort, 200μM) to reduce P-gp, and subsequently N-terminal of P-gp expression. Immunoblotting was performed using polyclonal CTBx antibody for GM1 staining (D) or Gal-1 for Gal-1 staining (E). N=2 duplicates per group. Dotted lines indicate discontinuous wells on the same membrane. F) Western blot showing O-GlcNAcylation (O-GlcNAc) levels post-thermal exposure. PUGNAc (100μM) was used to treat to preserve the full production of O-GlcNAc in shControl-SNU449 cells 24 hours prior to thermal exposure. Immunoblotting was then performed using O-GlcNAc antibody. G) shControl-SNU449 cells were treated with DON (L-6-Diazo-5-oxonorleucine) (400μM) to inhibit O-GlcNAc for 3 days before thermal exposure. Cytoplasmic fractions were extracted using NuCLEAR extraction kit. Immunoblotting was performed using Gal-1 antibody. α-tubulin was used as loading controls. H) Illustration showing the mechanistic association between thermal stress and formation of GM1/Gal-1/N-P-gp complex. I) After DON and thermal treatments as shown in G), shControl-SNU449 cells were added 10μM of CTBx to stain for GM1. Immunoblotting was then performed using polyclonal CTBx antibody for GM1 staining. J) After DON and thermal treatments as shown in G), shControl-SNU449 and shGal-1-SNU449 cells were added 10μM of CTBx to stain for GM1. Immunoblotting was then performed using polyclonal CTBx antibody for GM1 staining. Dotted lines indicate discontinuous wells on the same membrane. K) Illustration showing the correlation between Gal-1 expression and formation of GM1/Gal-1/N-P-gp complex under hyperthermia.

Figure 4:. Galectin-1 overexpression is associated with glycolytic resistance via a GM1-dependent pathway.

A) Illustration showing glycolysis pathway and associated metabolites and catalytic enzymes. B-D) Upregulation of glycolytic proteins, Glucose-6-Phosphase Isomerase (GPI) (B), Glyceraldehyde 3-Phosphate Dehydrogenase (GADPH) (C), Phosphoglycerate Kinase (PGK) (D) in HCC nonresponders (n=23) compared to responders (n=32). The abundance levels were log₁₀-transformed. Box plots show median (central line), upper and lower quartiles (box limits), and 1.5 interquartile range (whiskers). E) Glycolytic activities in SNU449 and SNU423 (n=5 each condition) were measured based on the basal extracellular acidification rates (ECAR) via a Seahorse XF96 analyzer. F) ATP production rates via glycolysis in SNU449 and SNU423 were measured using a Seahorse XF96 analyzer. G) Glycolytic activities in SNU449 with Gal-1 inhibitor OTX (150μM) or DMSO (vehicle control) (n=10 each condition) were measured. H) Corresponding ATP production rates under the same conditions. I) Schematic diagram showing the interactions between Galectin-1 and β-Galactosidase with P-gp as well as between β-Gal and GM1 ganglioside. J) Western blot showing levels of β-Gal binding to N-P-gp proteins after SNU449 WT cell were treated with Wort (200μM) to reduce N-P-gp expression. β-Tubulin expression was used as loading controls. N=2 duplicates per group. Dotted lines indicate discontinuous wells on the same membrane. K) Western blot showing levels of β-Gal binding to N-P-gp proteins in SNU449 WT treated with PBS or Caffeine (β-Gal inhibitor) (15mM). N=2 duplicates per group. Dotted lines indicate discontinuous wells on the same membrane. L) Western blot showing levels of GM1 binding to the complex formed by N-P-gp and Gal-1 in the corresponding conditions as shown in (K). N=2 duplicates per group. Dotted lines indicate discontinuous wells on the same membrane. P-value was calculated using one-tailed and unpaired Student’s t-test with **p<0.01, ***p<0.001.

Figure 5:. Hydrolysis of Galectin-1-bound GM1 mediates glycolysis through galactose production.

A) Schematic diagram showing the mode of action of Gal-1 binding to GM1 and P-gp to bridge GM1 and its catalytic enzyme, β-Gal. These interactions ultimately lead to GM1 breakdown to produce galactose and subsequent glucose-6-phosphate for glycolysis. B) Galactose levels were assessed in shControl and shGal-1 (n=3 each condition) using an Abcam galactose assay kit. RFU = relative fluorescence units. C) β-Galactosidase (β-Gal) activity was assessed in shControl and shGal-1-SNU449 (n=3 each condition). D) Glycolytic activities in shControl and shGal-1 (n=5 each condition) were measured based on the basal extracellular acidification rates (ECAR) via a Seahorse XF96 analyzer. E) ATP production rates via glycolysis in shControl and shGal-1 were measured using a Seahorse XF96 analyzer. F) Schematic diagram showing the metabolism of galactose via Leloir pathway and subsequently glycolysis. This diagram also shows the isotopologue tracing when U-¹³C₆ galactose is used. G) Percent enrichment of galactose-1 phosphate (Gal-1P) product from the metabolism of U-¹³C₆ galactose in shControl and shGal-1-SNU449 cells (n=3 each condition). H) Percent enrichment of remaining U-¹³C₆ galactose after its metabolism in shControl and shGal-1-SNU449 cells. I-K) Percent enrichment of glucose-6 phosphate (G6P) (M+6) (I), 3/2 phosphoglycerate (M+3) (J), lactate (M+3) (K) products from the metabolism of U-¹³C₆ galactose in shControl and shGal-1-SNU449 cells (n=3 each condition). P-values were calculated using one-tailed, unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001.

Figure 6:. Galectin-1 silencing reduces metabolites and activities of glycolysis and mitochondrial TCA cycle under hyperthermia.

A) Schematic diagram showing the isotopologue tracing of U-¹³C₆ glucose metabolism via glycolysis and TCA cycle. B-E) shControl and shGal-1-SNU449 cells (n=3) were cultured with U-¹³C₆ glucose (10mM) in glucose-free RPMI and then exposed to 37°C or 47°C.The metabolic extraction was performed 3 hours post thermal exposures. The metabolic abundances were then assessed using mass spectrometry. These levels were then compared to the respective controls to evaluate the relative changes in the enrichments among the experimental conditions. The isotopologues (lactate (M+3) (B), 3- or 2-phosphoglycerate (3/2-PG) (M+3) (C),citrate (M+2 to M+6) (D), malate (M+0 to M+4) (E)) presented here were those that directly originated from U-¹³C₆ glucose or its direct intermediates. F-I) the total abundances of lactate (E), 3/2-PG (G), citrate (H), and malate (I) were also quantified. These levels were then compared to their respective controls to evaluate metabolic changes among the experimental groups. J-K) Glycolytic (J) and TCA cycle (K) activities in shControl and shGal-1 under hyperthermia (47°C) and normothermia (37°C) (n=10 each condition) were measured based on the basal extracellular acidification rates (ECAR) and oxygen consumption rates (OCR) via a Seahorse XF96 analyzer. P-values were calculated using one-tailed, unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001.

Figure 7:. Galectin-1 inhibition reduces metabolites and activities of glycolysis and mitochondrial TCA cycle under hyperthermia.

A) Schematic diagram showing the isotopologue tracing of U-¹³C₆ glucose metabolism with Gal-1 inhibition (OTX) via glycolysis and TCA cycle. B-E) SNU449 WT cells (n=3) were cultured with U-¹³C₆ glucose (10mM) in glucose-free RPM and treated with OTX (150μM) or DMSO (vehicle control) for 24 hours before exposing cells to 37°C or 47°C.The metabolic extraction was performed 3 hours post thermal exposures. The metabolic abundances were then assessed using mass spectrometry. These levels were then compared to the respective controls to evaluate the relative changes in the enrichments among the experimental conditions. The isotopologues, lactate (M+3) (B), fructose 1,6bisphosphate (F1,6BP, M+6) (C), 3- or 2-phosphoglycerate (3/2-PG) (M+3) (D), phosphoenolpyruvate (PEP, M+3) (E), citrate (M+0 to M+6) (F), malate (M+0 to M+4) (G)) presented here were those that directly originated from U-¹³C₆ glucose or its direct intermediates. H-M) the total abundances of lactate (H), F1,6BP (I), 3/2-PG (J), PEP (K), citrate (L), and malate (M) were also quantified. These levels were then compared to their respective controls to evaluate metabolic changes among the experimental groups. N-O) Glycolytic (N) and TCA cycle (O) activities in SNU449 WT with Gal-1 inhibition (150μM OTX) under hyperthermia (47°C) and normothermia (37°C) (n=10 each condition) were measured based on the basal extracellular acidification rates (ECAR) and oxygen consumption rates (OCR) via a Seahorse XF96 analyzer. P-values were calculated using one-tailed, unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001.

Figure 8:. Targeting Galectin-1 by selective inhibitor OTX008 enhances thermal-ablation efficacy in hyperthermia-resistant-HCC tumors.

A) Schematic diagram showing the timeline for tumor implantation and for drug OTX and thermal ablation treatments in vivo. B) Images of tumors from the 4 study groups: S-ablation DMSO (S=sham) (n=6), ablation DMSO (n=6), S-ablation OTX (n=6), ablation OTX (n=5, O = complete disappearance). C) Tumor weights from the 4 study groups (n=6 for S-ablation DMSO, Ablation DMSO, and S-Ablation OTX, n=4 for Ablation OTX). D) Galactose levels were assessed in the tumors from the four study groups, by using an Abcam galactose assay kit. These levels were then compared to the control (S-ablation DMSO) to evaluate for galactose changes among the study groups. E) Schematic diagram showing galactose metabolism (Leloir pathway) and glycolysis. F) 50mg of the harvested tumors from all groups were subject to metabolic extraction and submitted to mass spectrometry to assess levels of glucose-1 phosphate (Gal-1P). G) Western blot showing expression levels of Gal-1 binding to N-terminal-half of P-glycoprotein (N-P-gp/Gal-1) and free Gal-1 in the tumors from the four study groups. β-Tubulin expression was used to as loading controls. N=2 duplicates per group. H-K) Metabolic extracts from the tumors were also assessed for fructose-1,6-bisphosphate (F1,6BP) (H), phosphoenolpyruvate (PEP) (I), citrate (Cit) (J), and malate (Mal) (K). These levels were then compared to the control to evaluate for metabolic changes among the study groups. P-values were calculated using one-tailed, unpaired Student’s t-test, *p<0.05, **p<0.01, ***p<0.001.