Liver-Secreted Hexosaminidase A Regulates Insulin-Like Growth Factor Signaling and Glucose Transport in Skeletal Muscle

Incidence of both type 2 diabetes (T2D) and nonalcoholic fatty liver disease (NAFLD) is increasing and is closely associated with the increasing prevalence of obesity (1). Insulin resistance and NAFLD have a bidirectional relationship: hepatic steatosis contributes to hepatic and systemic insulin resistance, and insulin resistance can drive NAFLD progression (2). NAFLD ranges from simple steatosis to more advanced liver disease, including nonalcoholic steatohepatitis (NASH) with accompanying hepatic fibrosis, which can further progress to cirrhosis and hepatocellular carcinoma, increasing the risk of liver-related death (1). Technological advances in mass spectrometry and other analytic tools have greatly expanded the understanding of the liver’s endocrine function, and it is now known that the liver is an active endocrine organ, secreting thousands of proteins, metabolites, and noncoding RNAs (3). Importantly, NAFLD affects the endocrine function of the liver, and NAFLD-induced hepatokines exert powerful effects on lipid and glucose metabolism in other tissues (3). Several NAFLD-regulated hepatokines affect glycemic control and insulin sensitivity (4–8), highlighting the likely impact of altered intertissue communication in the pathogenesis of T2D.

Using murine hepatocytes and precision-cut liver sections from patients with NAFLD, we focused this study on the identification of hexosaminidase A (HEXA) as a liver-secreted protein that is regulated in NASH and T2D. HEXA is part of the HexA heterodimer, which consists of subunit α (encoded by HEXA) and subunit β (encoded by HEXB). HexA is a lysosomal protein complex that degrades gangliosides, a family of glycated sphingolipids that are primarily localized to cell-surface lipid rafts (9). Specifically, HexA removes the terminal N-acetyl-d-galactosamine from GM2 gangliosides, forming GM3 gangliosides. Lipid rafts are ubiquitous subdomains of the plasma membrane and are enriched in cholesterol, glycated sphingolipids, and membrane proteins, including transporters [e.g., fatty acid transporters such as CD36 (10), or glucose transporters, including GLUT4 (11)] and receptors [e.g., the insulin receptor and β-adrenergic receptors (12)]. Thus, lipid rafts play an important role in cell-to-cell recognition, adhesion, signal transduction, and protein and lipid trafficking (9).

HexA deficiency is associated with the lysosomal storage disorders Tay-Sachs disease (mutations in HEXA) and Sandhoff disease (mutations in HEXB), which are characterized by GM2 ganglioside accumulation in neuronal lysosomes and lead to neurological defects (13,14). Tay-Sachs disease is further associated with muscle weakness and muscle wasting (15), highlighting a potential role for HEXA and gangliosides beyond neuronal metabolism. In addition, ganglioside composition is altered in liver and adipose tissue of obese and diabetic rodents (16,17), suggesting that gangliosides have a possible role in glycemic control. The presence of an N-terminal signal peptide in HEXA further suggests that HEXA is a classically secreted protein. Supporting this finding, HEXA is present in serum, and serum HEXA is positively associated with liver diseases, including cholestasis (18) and cirrhosis (19), and with cardiovascular disease (20) and type 1 diabetes (21). Despite these associations between circulating HEXA and various diseases caused by metabolic impairments, little is known about the endocrine function of HEXA. Herein, we identify HEXA as a NASH-induced hepatokine in both mice and humans and describe a role for HEXA in regulating skeletal muscle glucose uptake and whole-body glycemic control.

Human ethics of this study was approved by the University of Melbourne Human Ethics Committee (ethics identification [ID] 1851533), The Avenue Hospital Human Research Ethics Committee (ethics ID WD00006, HREC reference no. 249), and the Alfred Hospital Human Research Ethics Committee (ethics ID GO00005). Patient recruitment and the detailed procedure for all human liver-section experiments have been reported previously (6). For gene expression studies, obese patients undergoing bariatric surgery were prospectively enrolled in Melbourne (Australia) in 2015–2017, whereas for secretion studies, patients with obesity scheduled for bariatric surgery were screened to participate (2019–2021). Patient characteristics also have been published previously (6).

Mouse experiments were approved by the Monash University Animal Ethics Committee (no. MARP/2016/073) and the University of Melbourne Anatomy & Neuroscience, Pathology, Pharmacology, and Physiology Animal Ethics Committee (ethics ID 1814403) and conformed to the National Health and Medical Research Council of Australia guidelines regarding the care and use of experimental animals. Male C57BL/6 mice aged 8–10 weeks were obtained from the Monash University Animal Research Platform breeding facility or Animal Resources Centre (Canning Vale, Australia); male db/db mice (BKS.Cg-Dock7m^+/+ Leprdb/J) were bred in-house. All mice were maintained in a temperature-controlled room (22°C ± 1°C) with a 12-h light/dark cycle and ad libitum access to food and water.

To induce NASH, male C57BL/6 mice were fed a diet enriched in fat (40% of calories from lard), fructose (30% of calories), and cholesterol (20 g/kg) (SF16–033; Specialty Feeds) for 40 weeks; the respective control animals received standard rodent chow (5% of energy from fat; Specialty Feeds).

For experiments using recombinant protein and adeno-associated virus (AAV) vectors, mice were fed either a standard rodent chow or a diet enriched in fat and sucrose (43% of energy from fat) (SF04–001; Specialty Feeds) starting at 8 weeks of age for a total of 8–12 weeks.

For hepatocyte-specific HEXA overexpression, an AAV serotype 8 (AAV8) vector driven by a thyroxine-binding globulin (TBG) promoter and containing a cDNA sequence encoding for murine HEXA (AAV8-TBG-m-HEXA) was injected via the tail vein at 1 × 10¹² genome copies per mouse, whereas control mice received AAV8-TBG-Null. Metabolic phenotyping was carried out 8 weeks after AAV injection. For muscle-specific overexpression of HEXA, male, 8-week-old C57BL/6 mice received an intramuscular injection of AAV6-CMV-m-HEXA or a AAV6-CMV-Null, as previously described (22).

Murine HEXA recombinant protein was generated by the Monash University Protein Production Unit (Melbourne, Australia), and HEXA(R178H) protein was generated by Gene Universal. Protein was injected intraperitoneally (i.p.) at 1 mg/kg body weight either 60 min before a glucose or insulin tolerance test, or once daily for 28 days, and tissues were collected 48 h after the last dose. Plasma HEXA was assessed by ELISA (Aviva Systems Biology).

Primary murine hepatocytes were isolated by collagenase perfusion (23), and secreted factors were collected for 16 h in EX-CELL 325 PF CHO Serum-Free medium (Sigma-Aldrich).

Fat and lean mass were measured using the EchoMRI-900 (EchoMRI Corporation). Whole-body energy expenditure, the respiratory-exchange ratio, food intake, and locomotor activity were assessed using the Promethion System (Sable Systems International).

Mice were fasted for 4 h, then gavaged with glucose (1 or 2 g/kg body weight for db/db and C57BL/6 mice, respectively) or injected i.p. with insulin (0.3 and 1 units/kg for C57BL/6 mice or 2 units/kg for db/db mice). Blood glucose levels were monitored (Accu-Chek II; Roche Diagnostics), and plasma insulin determined (Ultra-Sensitive Mouse Insulin ELISA; Crystal Chem). For insulin-stimulated glucose disposal, mice were fasted for 4 h and then injected i.p. with 1 unit/kg body weight insulin and 10 µCi 2-[1-¹⁴C]-deoxy-d-glucose (2DG)/mouse (Perkin Elmer). Tissue-specific 2DG uptake was determined as described previously (24).

RNA was extracted using TRI-Reagent (Sigma-Aldrich), DNAse treated (Ambion DNA-free Kit; Thermo Fisher), and reverse transcribed using iSCRIPT Reverse Transcriptase (Invitrogen), and qPCR was performed (Quantinova SYBR Green PCR Kit; QIAGEN) using the CFX Connect Real-Time PCR Detection System (Bio-Rad). Primer sequences are listed in Supplementary Table 1. For immunoblot analysis, proteins were resolved by SDS-PAGE and analysis conducted was as described elsewhere (25). Antibodies are listed in Supplementary Table 2.

L6-GLUT4myc myoblasts were cultured as described previously (26). HEXA (100 ng/mL) was added to the medium at day 2 of differentiation for 3 days. For membrane fractionations, myotubes were incubated in absence or presence of HEXA, 2 mmol/L methyl-β-cyclodextrin (β-MCD) for 30 min, or 10 nmol/L insulin for 10 min. Membrane fractions were obtained by differential ultracentrifugation (23). AAV6 vectors expressing shRNA to knock down the insulin receptor and the IGF-1 receptor (IGF1R) were generated in-house (22) and added to the medium at 1 × 10⁶ viral genomes per cell at day 2 postdifferentiation for 3 days.

For assessment of glucose oxidation, extensor digitorum longus (EDL) muscle and precision-cut liver sections were incubated for 2 h in high-glucose DMEM containing 0.2% BSA and 1 μCi/mL U-¹⁴C-glucose (Perkin Elmer), the medium was acidified in 1 mol/L HClO₄, CO₂ was captured in 1 mol/L NaOH, and radioactivity was counted (TRI-CARB 4910TR, Perkin Elmer), as described elsewhere (4). Liver glycogen was assessed as described previously (27). For assessment of glucose output, hepatocytes were incubated in absence or presence of HEXA in glucose-free DMEM containing 25 mmol/L pyruvate for 24 h, followed by detection of glucose in the culture medium [Infinity Glucose (Ox); Fisher Scientific].

For determination of glucose uptake, cells and tissues were incubated in glucose-free DMEM containing 25 mmol/L 2-deoxy-glucose and 0.1% BSA for 60 min, in the absence or presence of 2 mmol/L β-MCD or dextrin or 2 µg/mL filipin (Sigma Aldrich) for 30 min, and 10 nmol/L insulin/IGF-1 for 10 min, followed by glucose-free DMEM containing 25 mmol/L 2-deoxy-glucose, 0.1% BSA, and 1 µCi/mL 2DG (PerkinElmer) for 10 min. Cells and tissues were washed and lysed in PBS, and radioactivity was counted.

Lipid rafts were isolated by sucrose gradient ultracentrifugation, as described previously (6,28).

Lipids were extracted in 100 μL 1:1 butanol to methanol, containing 10 μL SPLASH II LIPIDOMIX Standard (catalog 330709 W; Avanti Polar Lipids), 10 μL Ceramide LIPIDOMIX Standard (catalog 330712×; Avanti Polar Lipids), and 1 µg C18:0 GM3 Ganglioside-d5 (catalog 860073W; Avanti Polar Lipids), as described previously (6,28). Samples were analyzed by ultra-high-performance liquid chromatography (UHPLC) coupled to tandem mass spectrometry (MS/MS) using a Vanquish UHPLC linked to an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific), with separate runs in positive and negative ion polarities, as described previously (6,28).

Proteomics sample processing has been published (6,28). Peptide samples were analyzed by liquid chromatography–MS/MS using an Orbitrap mass spectrometer (Thermo Fisher Scientific) coupled online to an RSLC nano HPLC (Ultimate 3000 UHPLC; Thermo Fisher Scientific) (6,28).

Data are presented as mean ± SEM. Data were assessed for normal distribution using the D’Agostino and Pearson test, with parametric tests used in cases of normal distribution. These included a two-tailed unpaired Student t test and one- or two-way ANOVA followed by Bonferroni multiple comparisons tests. Statistical significance was determined at P < 0.05.

All data analyzed during this study are included in this article (and its online supplementary files), including all proteomics and lipidomics data. No applicable resources were generated or analyzed during this study.

Hepatic Hexa expression is induced by lipid availability and increased in murine NASH (28). To examine the human relevance of this finding, we assessed hepatic HEXA expression in individuals with no liver pathology (No-NAFLD), NAFLD, or NASH, with varying degrees of liver fibrosis. Patient characteristics are described elsewhere (6). Briefly, the male to female ratio was 30%:70%, with the following characteristics (± SD): age, 47 ± 11 years; BMI, 43 ± 7 kg/m²; fasting plasma ALT, 43.9 ± 39.7 IU/L; triglyceride, 1.5 ± 0.7 mmol/L; cholesterol, 4.1 ± 1.0 mmol/L; blood glucose, 5.9 ± 2.0 mmol/L; and plasma insulin, 10.3 ± 10.5 mU/L. Hepatic HEXA expression tended to increase in individuals with NASH compared with individuals with NAFLD (P = 0.0523) but was not different when compared with No-NAFLD (Fig. 1A). HEXA expression was significantly increased in individuals with hepatocyte ballooning (Supplementary Fig. 1A) but was not associated with steatosis, inflammation, or fibrosis (Supplementary Fig. 1B–D).

Although HEXA contains an N-terminal signal peptide, suggesting that it is a classically secreted protein, little is known about its endocrine function. Therefore, we next assessed HEXA secretion from precision-cut liver sections in a separate cohort of individuals. HEXA secretion was significantly increased in individuals with NASH when compared with No-NAFLD (P = 0.005), with a trend for increased secretion in NASH compared with NAFLD (P = 0.08) (Fig. 1B). These effects were related to steatosis (Fig. 1C) but not inflammation, hepatocyte ballooning, or hepatic fibrosis (Supplementary Fig. 1E–G). Similar responses were observed in mice, with a trend for increased HEXA secretion from primary hepatocytes isolated from mice with NASH compared with those with no pathology (P = 0.087) (Supplementary Fig. 1H).

Subanalysis in individuals identified a significant positive correlation between fasting blood glucose value and HEXA secretion (P = 0.006) (Supplementary Fig. 1I), whereas no correlations were observed between HEXA secretion and all other clinical and biochemical variables (data not shown). Furthermore, plasma HEXA was increased sixfold in obese insulin-resistant (IR) mice and 20-fold in db/db mice with T2D, when compared with lean insulin-sensitive mice (Fig. 1D), highlighting a potential role for circulating HEXA in glycemic control.

To investigate the role of circulating HEXA in glycemic control, recombinant murine HEXA protein was administered to mice. Injection of recombinant HEXA at 1 mg/kg body weight (the lowest effective dose on glucose clearance) (Supplementary Fig. 1J) increased plasma HEXA 2.5-fold by 2 h after administration, with plasma levels returning to baseline after 4 h (Fig. 2A). Next, HEXA was injected into lean mice and an oral glucose tolerance test was performed 1 h later. HEXA improved blood glucose control (Fig. 2B) despite marked reductions in levels of fasting and glucose-stimulated plasma insulin (Fig. 2C and D). Acute HEXA administration also improved insulin tolerance in lean mice (Fig. 2E). HEXA’s effects on glucose clearance and insulin levels were recapitulated in obese IR mice (Fig. 2F and G), with similar trends also observed in db/db mice (Fig. 2H–J). In contrast, acute administration of enzymatically inactive HEXA(R178H) did not affect glucose or insulin tolerance in lean mice (Fig. 2K and L). These results support a likely endocrine role for HEXA in blood glucose control and insulin secretion.

To assess whether a chronic increase in circulating HEXA improves glycemic control, HEXA protein was administered (1 mg/kg body weight) once daily for 4 weeks to obese IR mice. Metabolic assessments were conducted 48 h after the last protein injection to differentiate acute and chronic effects of HEXA (Fig. 2). Chronic HEXA administration increased plasma HEXA levels by 12-fold (Fig. 3A), improved blood glucose control (Fig. 3B), and tended to reduce plasma insulin level (Fig. 3C). Chronic HEXA did not affect insulin tolerance (Supplementary Fig. 2A), potentially because of rapid clearance of plasma HEXA (Fig. 2A). Furthermore, chronic HEXA did not affect body weight, adiposity, energy expenditure, the respiratory-exchange ratio, food intake, or locomotor activity when compared with that of untreated mice (Supplementary Fig. 2B–G). Overall, these data highlight that both acute and chronic HEXA administration improve blood glucose control in mice.

Because HEXA is an endogenously produced hepatokine, we overexpressed HEXA in the livers of obese IR mice using AAV (HEXA-AAV); control mice received AAV containing scrambled cDNA (Control-AAV). Liver Hexa was increased 13-fold with no change in other tissues examined (Fig. 3D), whereas plasma HEXA was increased approximately threefold in HEXA-AAV mice compared with Control-AAV mice (Fig. 3E). We recently reported that hepatic HEXA overexpression does not alter body weight, adiposity, energy expenditure, food intake, or locomotor activity (28). Although HEXA-AAV mice did not exhibit any change in glucose tolerance (Fig. 3F), glucose-stimulated plasma insulin level was reduced (Fig. 3G), and insulin sensitivity was improved compared with that of Control-AAV mice (Fig. 3H). Overall, these data highlight that the liver is an important contributor to circulating HEXA levels and that increasing hepatic and plasma HEXA improves aspects of glycemic control.

To assess the tissue(s) responsible for the improvements in systemic insulin sensitivity with HEXA-AAV, we repeated the insulin tolerance test in the presence of radiolabeled 2-deoxy-glucose (Supplementary Fig. 3A). HEXA-AAV increased insulin-mediated glucose disposal into skeletal muscle (Fig. 4A) but not in heart, kidney, or white and brown adipose tissues (Supplementary Fig. 3B). The increase in skeletal-muscle insulin action in HEXA-AAV mice was confirmed at the signaling level by increased Akt S473 phosphorylation (Fig. 4B). These findings were accompanied by a twofold increase in glucose oxidation in EDL muscle in HEXA-AAV mice compared with Control-AAV mice (Fig. 4C) and after chronic HEXA protein administration (Fig. 4D).

Further experiments demonstrated that HEXA-AAV did not affect liver glucose metabolism or insulin sensitivity, as assessed by measures of glucose oxidation (Supplementary Fig. 3C), gene expression of the gluconeogenic regulators phosphoenolpyruvate carboxykinase 1 (Pck1) and glucose-6-phosphatase (G6pc) (Supplementary Fig. 3D and E), glycogen content (Supplementary Fig. 3F), and hepatocyte glucose output after chronic HEXA administration (Supplementary Fig. 3G). Taken together, these data highlight that the HEXA-mediated improvements in glycemic control are likely due to improved skeletal muscle insulin action.

To gain a better understanding of how HEXA improves muscle insulin action, we focused on HEXA’s primary function, conversion of GM2 to GM3 gangliosides, which occurs primarily in lipid rafts localized on the cell surface (9) (Fig. 4E). Total muscle GM3 content (Fig. 4F) and various GM3 lipid species (Supplementary Fig. 4A) were reduced by ∼30% in obese IR mice, with almost complete depletion in obese db/db mice. In contrast, GD1, GD3, and GT1 ganglioside content was increased in muscle of IR mice (Fig. 4G) but was not detected in db/db muscle (Fig. 4F). Muscle GM3 content was increased in HEXA-AAV mice (Fig. 4H, Supplementary Fig. 4B) in the absence of changes in other ganglioside classes or species (Fig. 4H) or other sphingolipids, including sphingomyelin and ceramide (Supplementary Fig. 4C–F), diacylglycerols (Supplementary Fig. 4G and H), and phospholipid or lysophospholipid classes (Supplementary Fig. 4I and J). In contrast, there were no changes in muscle GM3 or other ganglioside classes after chronic recombinant HEXA treatment (Supplementary Fig. 4K–M), which is likely related to muscle being collected 48 h after the last HEXA injection.

Using L6 myotubes stably overexpressing GLUT4 (L6-GLUT4), a validated model for the examination of lipid raft physiology (29), we confirmed that chronic HEXA administration increased insulin-stimulated glucose uptake (Fig. 4I). When lipid rafts were disrupted using either filipin or β-MCD (Fig. 4I), but not the structurally related carbohydrate dextrin (Supplementary Fig. 5A), HEXA no longer enhanced insulin-mediated glucose uptake, demonstrating that intact lipid rafts are required for HEXA’s effect on insulin-stimulated glucose uptake. Importantly, enzymatically inactive HEXA(R178H) did not affect glucose uptake in L6-GLUT4 myotubes (Fig. 4J), supporting the notion that HEXA’s actions toward GM2 are required for its impact on glucose uptake. HEXA also affects muscle glucose uptake through a lipid raft–dependent mechanism in vivo. HEXA-AAV was injected into the tibialis anterior and EDL muscles of mice (Fig. 4K), and glucose uptake assessed in EDL muscle ex vivo. Similar to the findings in myotubes, HEXA overexpression increased insulin-stimulated glucose uptake and these effects were dependent on intact lipid rafts (Fig. 4L).

To gain mechanistic insights into the metabolic pathways that contribute to HEXA’s effects on glucose uptake, we assessed the impact of HEXA on the myotube proteome. A total of 2,542 proteins were identified in control and HEXA-treated myotubes, with principal component analysis showing distinct proteome clustering (Supplementary Fig. 5B). Reactome pathway analysis highlighted increased GLUT4 translocation to the plasma membrane as a metabolic pathway affected by HEXA, with altered abundance of 15 proteins involved in GLUT4 exocytosis (Fig. 4M) (30,31). To assess the impact of HEXA on GLUT4 localization to the cell surface, cell fractions were isolated from L6 myotubes after exposure to HEXA, the lipid raft disruptor β-MCD, and/or insulin, and plasma membrane GLUT4 abundance was measured (Fig. 4N and O). HEXA treatment increased membrane GLUT4 by 70%, and this effect was ablated with disruption of lipid rafts (Fig. 4N and O). Together, these data demonstrate that HEXA directly acts on myotubes to increase GLUT4 localization to the cell surface and promote glucose uptake.

To elucidate the mechanism underpinning HEXA’s effects on GLUT4 localization to the cell surface, we incubated L6 myotubes chronically with HEXA, isolated myotube lipid rafts, and performed label-free proteomics. Principal component analysis revealed protein clustering of control and HEXA-treated myotubes (Supplementary Fig. 5C), with 198 lipid raft proteins regulated by HEXA treatment (n = 95 increased and 103 decreased) (Fig. 5A). The proteins increased in lipid rafts included IGFR1 and IGFR2, which are transmembrane receptors that are activated both by IGFs and insulin. HEXA treatment also increased lipid raft localization of sortilin 1 (Sort1) (Fig. 5A), a GLUT4-associated glycoprotein that rescues GLUT4 from degradation and plays a role in protein trafficking to the cell surface (32).

Because the activation of IGFR signaling in muscle alters gene expression, protein synthesis, and/or glucose metabolism (33), we refocused our attention to the observed changes in the myotube proteome after HEXA treatment. Accompanying the substantial changes in lipid raft localized proteins (Fig. 5A), volcano plot analysis indicated substantial regulation of protein expression by HEXA (Fig. 5B). From 2,542 identified proteins, 850 proteins were regulated by HEXA (Fig. 5B), of which 112 were IGFR target genes (Fig. 5B). An IGF signaling signature was confirmed by Reactome pathway analysis (Fig. 5C).

The IGF1R and insulin receptor are highly homologous and show pronounced overlap in ligand binding and signaling capacities implicated in metabolism and growth (34), suggesting that both receptors affect insulin action in muscle. Of note, we did not detect the insulin receptor within lipid rafts (Fig. 5A), and insulin receptor protein expression was not affected by HEXA (Fig. 5B), suggesting that HEXA might promote glucose metabolism via activation of IGFR signaling. Confirming the proteomics data, HEXA dramatically increased IGF1R localization to the cell surface, both in the basal and insulin-stimulated state (Fig. 5D and E). Disruption of lipid rafts completely abolished the effects of both insulin and HEXA on IGFR localization to the cell surface (Fig. 5D and E). Although HEXA increased IGF1R at the cell surface, cell-membrane insulin-receptor abundance was unexpectedly reduced after chronic HEXA exposure (Fig. 5D and E), further suggesting that the effects of HEXA are related to increased IGF-1 signaling. This was confirmed by independent experiments showing that although HEXA increased insulin-stimulated glucose uptake in myotubes, this effect was substantially magnified by IGF-1 (Fig. 5F).

To ascertain whether HEXA improves muscle glucose uptake through IGF-1 signaling, we silenced either the IGF1R (by 51%) or insulin receptor (by 65%) in L6 myotubes (Fig. 5G and H) and incubated myotubes in the absence or presence of HEXA, β-MCD, and/or insulin, followed by assessment of glucose uptake. Confirming findings of previous experiments (Fig. 4I), HEXA increased insulin-stimulated glucose uptake in the presence of intact lipid rafts. However, knockdown of both IGF1R and the insulin receptor completely abolished the stimulation of glucose uptake by either HEXA and/or insulin (Fig. 5I), suggesting that the HEXA-mediated improvements in glucose uptake are likely dependent on both intact IGF1R and insulin receptor signaling.

To determine whether IGF1R and/or insulin receptor localization to muscle lipid rafts are regulated by HEXA in vivo, we assessed the protein content of both receptors in lipid raft and nonraft fractions (the latter representing receptor both on the cell surface but not contained to lipid rafts and intracellular levels) from mice treated chronically with HEXA. HEXA reduced the proportion of insulin receptor in lipid rafts, similar to the findings in myotubes, but did not affect the proportion of IGF1R (Fig. 5J and K, Supplementary Fig. 5D and E). Although plasma insulin level was reduced after acute or chronic HEXA administration, and in HEXA-AAV mice (Figs. 2 and 3), there were no changes in plasma IGF-1 levels with either acute or chronic increases in HEXA (Supplementary Fig. 5F).

The liver secretes ∼1,500 proteins (6), with evidence of substantial regulation of hepatokine secretion in the presence of hepatic steatosis (5) and NASH (6). Despite recent advances in understanding the endocrine function of several NAFLD-regulated proteins (3), including FGF21 (35), fetuin B (5), SMOC1 (4), and ARSA (6), hepatokine regulation of systemic energy metabolism remains incompletely defined. Here, we show that HEXA secretion is increased with NASH in mice and humans. HEXA remodels lipid raft protein composition in skeletal muscle and promotes skeletal muscle glucose uptake, leading to systemic improvements in glycemic control and insulin action. Mechanistically, HEXA increases IGF1R content in lipid rafts, augments IGF-1 signaling, and increases GLUT4 localization to the plasma membrane (Fig. 6). Disrupting lipid rafts reverses the impact of HEXA on IGF-1 signaling, membrane GLUT4, and muscle glucose uptake.

Previous studies have shown that HEXA is present in plasma (36) and that plasma HEXA is increased in patients with gestational diabetes (37), type 1 diabetes, and early signs of diabetic nephropathy (21), cholestasis (18), and liver cirrhosis (19). In addition, plasma HEXA is positively correlated with serum triglycerides and liver enzymes (20), blood glucose levels (20,21), and HbA_1c (21), suggesting a possible role for HEXA in NAFLD pathogenesis and glycemic control. Our finding that plasma HEXA is increased with insulin resistance and T2D corroborates these previous observations. The present study advances beyond associations and shows that HEXA is a bona fide hepatokine, that liver HEXA secretion is increased in patients with NASH, and that this is associated with hepatic steatosis. The link with steatosis is consistent with our previous finding that HEXA gene expression is induced by increased lipid availability in murine hepatocytes (28). In this study, we also report a positive correlation between liver HEXA secretion and fasting blood glucose levels in patients with NAFLD, further supporting a potential role for HEXA in regulating glycemic control. The beneficial effects of HEXA on glycemic control might therefore be surprising and, in some respect, paradoxical, given that plasma HEXA is increased in NASH. Changes in liver-secreted factors were previously shown to promote insulin resistance in NASH (6), and in this pathological context, the increase in plasma HEXA is insufficient to improve glycemic control; however, increasing HEXA to supra-physiological levels can induce therapeutic benefits. This is consistent with previous studies showing that NAFLD-induced hepatokines, including FGF21 (38), SMOC1 (4), and activin E (8), exert beneficial effects on glucose and/or lipid metabolism at concentrations higher than those attained in the progression from no pathology to NAFLD.

Indeed, HEXA improves glycemic control whether administered by recombinant protein or AAV8-mediated delivery to the liver, and this improvement occurs in obese mice with insulin resistance or T2D. Mechanistically, these improvements are mediated by increased muscle glucose uptake, which is driven by HEXA’s actions on increasing muscle lipid raft GM3 ganglioside content, increased localization of IGF1R to lipid rafts, enhanced IGF-mediated signaling, and increased GLUT4 at the plasma membrane. We find that HEXA does not affect GLUT4 protein content but increases its localization to the cell surface, which is analogous to insulin action on GLUT4 (39). Insulin-stimulated GLUT4 translocation requires the lipid raft–localized, small GTP-binding protein TC10 and exocyst protein complex (40), and disruption of lipid rafts also inhibits insulin stimulation of GLUT4 translocation (11). Similarly, we find that disrupting lipid rafts abolishes the effects of both HEXA and insulin on GLUT4 localization to the cell surface, on one hand confirming the importance of lipid rafts in muscle glucose homeostasis but also highlighting, on the other hand, the importance of lipid rafts in HEXA’s actions on muscle glucose uptake.

Lipid rafts are enriched in glycated sphingolipids, including the substrates and products of HEXA’s enzymatic activity (i.e., GM2 and GM3 gangliosides, respectively) (41). We show that skeletal muscle GM3 content is reduced in obese IR mice and almost completely depleted in muscle of mice with T2D, associations that indicate the likelihood of reduced GM3 contributing to diabetes progression. We further demonstrate that increasing skeletal muscle GM3 content via HEXA overexpression enhances muscle glucose uptake and improves glycemic control, which is primarily mediated by IGF1R signaling, rather than signaling via the insulin receptor. These results align with those of a recent study showing that enrichment of GM3 in membranes leads to the dissociation of the insulin receptor–caveolin 1 interaction, which is considered essential for insulin receptor–dependent insulin signaling (42). Indeed, mice lacking GM3 synthase have less GM3 in muscle and show enhanced insulin receptor phosphorylation (43). Although our data support the idea that the HEXA-mediated remodeling of lipid raft protein composition is associated with increased localization of both IGF1R and IGF2R to lipid rafts and enhanced IGF-mediated signaling, whether these effects are directly mediated by increased lipid raft GM3 requires more investigation, particularly given the conflicting findings that increasing GM3 ganglioside content suppresses IGF1R signaling (44) and GM3 depletion promotes IGF-1 signaling in keratinocytes (44–46). Furthermore, given the substantial remodeling of protein composition in lipid rafts after HEXA exposure, HEXA may have a broader role beyond IGF-1–specific signaling, and a more comprehensive assessment of HEXA-specific signaling events is required.

Taken together, we identified HEXA as a hepatokine that improves skeletal-muscle glucose uptake and systemic glycemic control, and this study’s findings provide impetus to further study HEXA biology and muscle lipid raft remodeling in the context of glucose metabolism. Skeletal muscle is a logical target tissue for therapeutic enhancement of glycemic control because it accounts for ∼80% of insulin-stimulated glucose disposal (47) and is the primary site of insulin resistance in individuals with T2D (48). Despite this, skeletal muscle is an under-represented target tissue for T2D pharmacotherapies, with current therapies primarily targeted to adipose (thiazolidinediones), gut (α-glucosidase inhibitors), liver (metformin), kidney (gliflozins), and pancreas (sulphonylureas, GLP1R agonists, DPP4 inhibitors). Increasing our understanding of the regulation of muscle glucose homeostasis, including the importance of lipid rafts, will possibly open new avenues for therapeutic advancement. Tempering this enthusiasm is the recent finding that increasing liver HEXA enhances VLDL secretion, which would be predicted to contribute to dyslipidemia and ectopic lipid deposition in peripheral tissues (28). However, because AAV-mediated gene therapy aimed at increasing or reinstating HEXA expression in patients with Tay-Sachs disease has been considered a safe and viable treatment option (49), it will be of interest to assess whether glycemic control is improved in these patients and if HEXA therapy is a potential strategy for patients with T2D.

Acknowledgments. This work was supported by infrastructure and technical assistance from the Melbourne Mouse Metabolic Phenotyping Platform at the University of Melbourne.

Funding. These studies were supported by funding from the National Health and Medical Research Foundation of Australia (grants APP1162511 and APP2011540) and Diabetes Australia (grant Y21G-MONM). M.K.M. was supported by a research fellowship from the National Health and Medical Research Council (APP1143224).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.K.M. is responsible for study design, designed and conducted experiments, analyzed data, and wrote the manuscript. A.Z.T., J.B., K.I.W., W.d.N., and S.N.K. conducted experiments and edited the manuscript. S.N., P.G., and N.A.W. analyzed the data and edited the manuscript. G.J.O., W.A.B., and P.R.B. provided human samples and edited the manuscript. M.J.W. provided study oversight and wrote the manuscript. M.K.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Australasian Diabetes Congress, 8–10 August 2022, Brisbane, Australia; at ComBIO2022, 3–5 September 2022, Melbourne, Australia; and at the International Congress on Obesity, 18–22 October, Melbourne, Australia.