Panx1 deficiency exacerbates GAN diet-induced obesity by destabilizing β-catenin via GSK3β

Summary

Obesity is a global health challenge and a major risk factor for metabolic diseases. Here, we show that Panx1-deficient mice develop severe obesity when fed a Gubra-Amylin (GAN) diet, characterized by adipocyte hypertrophy and disrupted lipid metabolism, while no such phenotype is observed on a conventional high-fat diet. Mechanistically, Panx1 deficiency leads to reduced non-phosphorylated β-catenin levels and its downstream targets, indicating impaired Wnt/β-catenin signaling. We further demonstrate that Panx1 interacts with β-catenin and GSK-3β, stabilizing β-catenin and preventing its degradation. Loss of Panx1 reduces β-catenin activity, impairs adipocyte hyperplasia, and promotes hypertrophic adipose expansion, thereby exacerbating obesity. These findings establish Panx1 as a key regulator of adipose tissue remodeling and suggest its potential as a tissue-specific target for obesity and related metabolic disorders.

Graphical abstract

Highlights

• •Panx1 deficiency worsens obesity in GAN diet-fed mice
• •Panx1 plays distinct roles in adipose tissue and the liver
• •Panx1 stabilizes β-catenin by binding GSK-3β and β-catenin
• •Panx1 loss impairs adipocyte hyperplasia and promotes hypertrophy

Physiology; Molecular biology

Introduction

The global prevalence of obesity has risen steadily over the past five decades, emerging as a major worldwide public health challenge.^1,2 Obesity is a complex, multifactorial chronic disease that is fundamentally driven by a sustained energy imbalance, leading to the excessive accumulation of body fat.³ Genetic, physiological, and environmental factors all contribute to its pathogenesis.⁴ This condition is associated with an elevated risk of numerous comorbidities, including type II diabetes, hypertension, cardiovascular disease, metabolic dysfunction-associated steatotic liver disease (MASLD), stroke, joint injuries, dementia, osteoarthritis, and various cancers.⁵ A hallmark of obesity is the expansion of white adipose tissue (WAT), which is a key driver of metabolic syndrome.⁶ This expansion occurs primarily through two processes: hyperplasia (an increase in adipocyte number via preadipocyte proliferation) and hypertrophy (an increase in adipocyte size).⁶ Of these, adipocyte hypertrophy is particularly implicated in pathological adipose tissue remodeling, as it leads to cellular stress, inflammation, and insulin resistance.⁷

Pannexin 1 (Panx1) is a channel-forming glycoprotein belonging to the innexin family. It is widely expressed on the plasma membrane and facilitates the transmembrane exchange of ions and metabolites, most notably ATP.⁸ Through this mechanism, Panx1 plays a central role in numerous physiological and pathophysiological processes, including the regulation of inflammatory responses, immune activation, and programmed cell death.^9,10 Accumulating evidence has suggested a role for Panx1 in obesity and metabolic disorders. Clinically, Panx1 expression was significantly up-regulated in visceral adipose tissue of patients with obesity, as observed following bariatric surgery, suggesting a potential adaptive mechanism to counter severe insulin resistance.¹¹ Consistently, studies using tissue-specific knockout models have revealed distinct metabolic functions: whole-adipose Panx1 ablation attenuated brown adipose tissue (BAT) thermogenesis and suppressed the browning of epididymal white adipose tissue (eWAT) through the inhibition of β-adrenergic signaling.¹² In contrast, BAT-specific Panx1 knockout impaired adaptive thermogenesis without affecting white adipose browning, highlighting a remarkable degree of tissue-specificity in its action.¹² However, the specific mechanisms through which Panx1 regulates eWAT homeostasis are still not fully understood.

In this study, we used two complementary diet-induced obesity models to examine the role of Panx1 under different levels of metabolic stress. A conventional high-fat diet (HFD) induces obesity, insulin resistance and hepatic steatosis but usually causes only modest liver inflammation and fibrosis,^13,14 whereas the Gubra-Amylin (GAN) diet, which combines high fat with added fructose and cholesterol, imposes stronger metabolic stress, leading to a more pronounced obesogenic response and more advanced liver involvement, including NASH-like steatohepatitis and fibrotic changes.^15,16 Comparing Panx1-deficient and wild-type mice under these two regimens, we found that global Panx1 knockout produced a pronounced obesity phenotype only under GAN feeding, characterized by the marked expansion of epididymal white adipose tissue (eWAT) and adipocyte hypertrophy. Mechanistically, Panx1 deficiency disrupted Wnt/β-catenin signaling, reduced preadipocyte proliferation, and shifted adipose expansion from hyperplasia toward hypertrophy, thereby promoting obesity. Thus, our work refines the understanding of molecular mechanisms governing adipose tissue expansion and identifies Panx1 as a potential therapeutic target in obesity.

Results

Panx1 deficiency exacerbated Gubra-Amylin diet-induced obesity

To assess the role of Panx1 in the progression of metabolic disease, we generated Panx1 globally knockout mice (Panx1^−/−) and subsequently developed mouse models using HFD and GAN diet, respectively (Figure 1A; Figure S1A). The expression analysis revealed a significant decrease in the gene levels of Panx1 in the brain, liver, eWAT, brown adipose tissue (BAT), and colon, respectively (p < 0.05) (Figure 1B). A severe obesity phenotype was observed in mice following 20 weeks of GAN diet feeding, characterized by a substantial body weight gain (p < 0.001) (Figures 1C and 1D). The body composition analysis also showed a notable increase in fat content (p < 0.001) and a decrease in lean body mass (p < 0.001) (Figure 1E). However, body weight differences were not observed in HFD-fed mice (Figures S2A and S2B) or in mice maintained on standard chow for 24 weeks (Figure S3). The ratio of eWAT and subcutaneous adipose tissue (SAT) to body weight of the mice treated by GAN diet was remarkably higher in Panx1^−/− mice compared to WT (p < 0.001 and p < 0.05, respectively), while other tissues remained unchanged (Figure 1F). In contrast, no difference was observed in HFD-treated mice (Figures S2H and S2I). Fasting blood glucose levels, as well as glucose levels during the glucose tolerance test (GTT) and insulin tolerance test (ITT), were not significantly different between these two mouse models (Panx1^−/− vs. WT) under either GAN or HFD feeding (Figures 1H–1J; Figures S2C–S2G). These results suggest that global Panx1 deficiency promotes obesity primarily in the GAN diet-induced model.

Figure 1.

GAN diet induces obesity in Panx1^−/− mice

(A) Feeding protocol schematic. Eight-week-old Panx1^−/− and WT mice were fed a GAN diet for 20 weeks.

(B) Panx1 mRNA expression in different tissues (qRT-PCR, PPIA as control).

(C) Representative images of mice after 20 weeks of GAN diet.

(D) Body weight changes during GAN feeding.

(E) Body composition after 20 weeks.

(F) Organ-to-body weight ratios (liver, eWAT, SAT, and BAT).

(G) Representative images of eWAT.

(H) Fasting blood glucose.

(I) Baseline GTT and ITT.

(J) GTT and ITT one week before the endpoint. Data shown as mean ± SD, n = 6/group. Statistics in (B): ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, two-sided Student’s t test. Statistics in (D), (H), (I), and (J): ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, two-way ANOVA with repeated measures followed by Sidak’s multiple comparisons test. Statistics in (E) and (F): ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, two-sided Student’s t test.

A parallel cohort of female mice showed broadly similar patterns. Panx1^−/− and WT females displayed comparable body weight, adiposity and glucose homeostasis under HFD (Figures S4A–S4F), whereas Panx1^−/− females developed greater body-weight gain and increased eWAT and SAT mass than WT controls on the GAN diet, with the onset of excess weight gain occurring later than in males (Figures S4G–S4L).

Global Pannexin 1 knockout did not exacerbate liver injury in mouse models

To assess the impact of Panx1 deficiency on liver injury, hepatic lipid accumulation, inflammation, and collagen deposition were evaluated. In both GAN diet and HFD-induced mice, liver size was comparable between Panx1^−/− and WT mice (Figure 2A; Figure S5A). Histological analysis of liver sections revealed a slight increase in lipid droplet accumulation in Panx1-deficient mice compared with WT controls under the GAN diet, which was accompanied by significantly elevated hepatic triglyceride (TG; p < 0.05) and total cholesterol (TC; p < 0.05) levels (Figures 2B and 2C). Despite this hepatic lipid accumulation, serum TG and TC concentrations remained comparable between WT and Panx1^−/− mice, indicating that Panx1 deficiency does not cause overt systemic dyslipidemia in this model (Figure 3C). At the transcriptional level, no significant differences were observed in the expression of key lipogenic genes, including acetyl-CoA carboxylase alpha (ACACA), ATP citrate lyase (ACLY), sterol regulatory element-binding protein 2 (SREBP2), fatty acid synthase (FASN), and diacylglycerol acyltransferases (DGAT1 and DGAT2), or in fatty acid oxidation-related genes such as acyl-CoA oxidase 1 (ACOX1), carnitine palmitoyltransferase 1A (CPT1A), and peroxisome proliferator-activated receptor α (PPARα) (Figure 2D). In contrast, genes involved in hepatic fatty-acid uptake and intracellular trafficking showed a more selective pattern. Under GAN feeding, hepatic CD36 and FABP1 mRNA levels, both involved in fatty acid uptake and intracellular trafficking, were significantly increased in Panx1^−/− mice compared with WT controls (CD36, p < 0.01; FABP1, p < 0.05), whereas the expression of microsomal triglyceride transfer protein (MTTP), a key factor in very-low-density lipoprotein (VLDL) assembly and secretion, was unchanged (Figure 2D).

Figure 2.

Detection of hepatic lipid metabolism, fibrosis, and inflammation of the Panx1^−/− mouse

(A) Gross liver morphology after 20 weeks of GAN diet.

(B) H&E staining and lipid droplet quantification (scale bars, 100 μm).

(C) Hepatic TC and TG levels.

(D) qRT-PCR analysis of genes involved in hepatic lipid uptake, intracellular transport and export, as well as fatty acid synthesis, oxidation, triglyceride synthesis, fibrosis, and inflammation (PPIA as control).

(E) Sirius Red and MPO staining with the quantification of fibrotic area and MPO-positive cells (scale bars, 100 μm and 50 μm).

(F) Immunofluorescence of F4/80+ macrophages with quantification (scale bars, 50 μm).

(G) Hepatic cytokine and chemokine levels were measured by Luminex assay and normalized to total protein concentration. Data are mean ± SD, n = 4–6 per group. Statistics in (C)–(E) and (G): ∗p < 0.05, ∗∗p < 0.01, two-sided Student’s t test.

Figure 3.

Knockout of Panx1 decreased inflammation without affecting the lipid metabolism, angiogenesis or browning level of eWAT in obese mice

(A) Schematic of WAT remodeling (The figure was generated using Figdraw (www.figdraw.com)).

(B) H&E staining and adipocyte size quantification of eWAT and SAT (scale bars, 50 μm).

(C) TC and TG levels in eWAT and serum, normalized to total protein concentration.

(D) qRT-PCR analysis of genes related to lipolysis, fatty acid synthesis, and oxidation (PPIA as control).

(E) Western blot of lipolysis-related proteins (β-actin as loading control).

(F) qRT-PCR of angiogenesis-related genes (PPIA as control).

(G) Western blot of thermogenic proteins (β-actin as loading control).

(H) Immunohistochemistry of F4/80+ cells in eWAT (scale bars, 50 μm).

(I) Luminex assay of cytokines and chemokines in eWAT (normalized to total protein concentration) and in serum. Data are mean ± SD, n = 4–6/group. Statistics in (B)–(D), (F), (H), and (I): ∗p < 0.05, ∗∗p < 0.01, two-sided Student’s t test.

Histological and molecular markers of liver fibrosis, including Sirius Red staining and the expression of COL3A1 and ACTA2, were not significantly different between GAN diet-fed Panx1^−/− and WT mice (Figures 2D and 2E). In contrast, no significant differences were detected in inflammatory markers, including neutrophil presence (MPO staining), macrophage infiltration (F4/80 staining), or inflammatory cytokine levels (Figures 2E–2G). In addition, Panx1^−/− mice showed no differences from WT controls in any parameters assessed in HFD-induced models (Figures S5B–S5E). Taken together, these data suggest that Panx1 deficiency under GAN feeding enhances CD36⁻and FABP1-dependent hepatic fatty-acid uptake and trafficking and thereby contributes to the selective increase in hepatic TG and TC, but this occurs without the detectable aggravation of liver damage, inflammation or fibrosis in either the GAN or HFD models.

Pannexin 1 deficiency promoted adipocyte hypertrophy

To elucidate how Panx1 deficiency promoted adipocyte dysfunction, we investigated processes including lipid metabolism, browning, angiogenesis, and inflammation in eWAT, all of which had been implicated in the development of obesity (Figure 3A).¹⁷ H&E staining revealed a significant enlargement of adipocytes in both eWAT and SAT of Panx1-deficient mice compared to WT controls fed a GAN diet (p < 0.05) (Figure 3B). However, this difference was not observed in the HFD group (Figure S4J). TG and TC levels were significantly elevated in the eWAT of Panx1^−/− mice (p < 0.05 and p < 0.01, respectively) (Figure 3C). Additionally, the expression of genes involved in lipid metabolism, such as perilipin 2 (Plin2; p < 0.05), comparative gene identification-58 (CGI-58; p < 0.05), and ACC1 (p < 0.05), was elevated in eWAT tissue (Figure 3D). A marked decrease in the expression of PPARα, stearoyl-CoA desaturase 1 (SCD1), ACOX1, and CPT1A gene was also observed in the eWAT of Panx1^−/− mice (p < 0.05) (Figure 3D), indicating impaired lipid oxidation. In contrast, the protein levels of the lipolysis-related proteins CGI-58 and ATGL showed no significant difference (Figure 3E).

No significant differences were observed between WT and Panx1^−/− mice in the expression of angiogenesis-related genes, including angiopoietin-1 (ANG1), angiopoietin-2 (ANG2), platelet-derived growth factor A (PDGFA), vascular endothelial growth factor A (VEGFA), and vascular endothelial growth factor B (VEGFB), or in the expression of browning-associated proteins, such as uncoupling protein 1 (UCP1) and PR domain-containing 16 (PRDM16) (Figures 3F and 3G). With respect to inflammation, macrophage infiltration in eWAT was markedly reduced in Panx1^−/− mice compared to WT controls (p < 0.05) (Figure 3H). Furthermore, the levels of inflammatory cytokines, including IL-1α and IL-6, were significantly decreased in eWAT (p < 0.05), whereas IL-1β, IL-10, VEGF, and TNFα were not significantly different between Panx1^−/− and WT mice (Figure 3I). In contrast, no significant changes were detected in the serum levels of these cytokines (Figure 3I). This divergence between eWAT and serum cytokine profiles indicates that inflammatory changes in Panx1^−/− mice remain largely confined to the local adipose tissue microenvironment and do not translate into overt systemic low-grade inflammation.

The above data suggested that adipocyte hypertrophy resulting from Panx1 knockout was mediated by a disorder in lipid metabolism and occurred independently of inflammation, browning, and angiogenesis.

Pannexin 1 knockout impairs Wnt/β-catenin signaling in epididymal white adipose tissue

To explore how Panx1 deficiency contributes to obesity at the molecular level, we carried out transcriptomic profiling of eWAT. Principal component analysis (PCA) of gene expression (FPKM) values demonstrated a distinct separation between Panx1^−/− and WT mice (Figure 4A). Gene set enrichment analysis (GSEA) of differentially expressed genes (DEGs) further suggested that loss of Panx1 suppresses the Wnt/β-catenin signaling pathway (Figure 4B), a pathway known to regulate embryonic development, cell differentiation, and proliferation.¹⁸ To refine this observation, we compared DEGs with Wnt/β-catenin pathway genes curated in the GSEA database. Fifty-two genes were found to overlap between the two sets (Figure 4C). Notably, several components of the β-catenin destruction complex—such as Axin1, CK1α, CK1ε, β-TrCP, and CUL1, which together facilitate the GSK-3β-mediated phosphorylation and degradation of β-catenin—were significantly reduced in Panx1^−/− eWAT relative to WT controls (Figure 4D). Conversely, many proteasome-related genes implicated in β-catenin degradation, including Psma1, Psmc3, Psma4, Psmb5, Psmb7, Psmd8, Psmd12, Psmd13, and Psmd14, were increased in Panx1-deficient mice (Figure 4D). Collectively, these data suggest that Panx1 plays an essential role in modulating Wnt/β-catenin signaling, particularly by influencing β-catenin degradation.

Figure 4.

Panx1 knockout downregulated the Wnt/β-Catenin pathway in eWAT tissue

(A) PCA of gene expression profiles (n = 6/group).

(B) GSEA enrichment of the Wnt/β-catenin pathway in eWAT.

(C) Venn diagram showing overlap of DEGs with Wnt/β-catenin pathway genes.

(D) Heatmap of Wnt-related DEGs in WT and Panx1^−/− mice.

(E) Western blot of active β-catenin and downstream proteins (β-actin as loading control). Data are mean ± SD, statistics in (E): ∗p < 0.05, ∗∗p < 0.01, two-sided Student’s t test.

Analysis of eWAT protein expression showed that total β-catenin and the non-phosphorylated (active) forms at Ser45 and Ser33/37/Thr41 were all significantly reduced in Panx1^−/− mice compared with WT controls (p < 0.05; Figure 4E). Among β-catenin targets, c-Jun and PPARγ were also significantly lower in Panx1^−/− eWAT (p < 0.05), whereas c-Myc, CCND1, and C/EBPα did not differ significantly between genotypes (Figure 4E). Together with the transcriptomic data, these protein-level changes support an overall attenuation of Wnt/β-catenin signaling in Panx1^−/− eWAT.

Pannexin 1 enhances β-catenin activity and nuclear translocation

To investigate the role of Panx1 in the Wnt/β-Catenin pathway, we established stable cell lines with overexpression (Panx1-OE) and knockout (Panx1^−/−) of Panx1 using 3T3-L1 preadipocytes (Figure 5A). The Panx1 protein is known to exist in three glycosylation states: a non-glycosylated form (Gly0, lower band), a high-mannose glycoprotein (Gly1, intermediate band), and a fully mature complex glycoprotein (Gly2, upper band), with the Gly2 species representing the glycoform predominantly localized to the plasma membrane. These post-translational modifications are critical for its intracellular trafficking, membrane localization, and biological activity.¹⁹ To validate the glycosylated form of Panx1 in 3T3-L1 overexpressing cell lines, we treated cell lysates with two distinct deglycosylation enzymes²⁰: PNGase F, which removes nearly all N-linked oligosaccharides, including Gly1 and Gly2 of Panx1, and Endo H, which removes high-mannose and hybrid-type N-glycans, such as Gly1 of Panx1. The results showed the intermediate band (Gly1) disappeared after treatment with Endo H, while both the upper (Gly2) and intermediate (Gly1) bands vanished after treatment with PNGase F (Figure 5B). Meanwhile, the lower band showed a noticeable increase (Figure 5B). This result suggests that the exogenous Panx1 protein (FLAG tag) in the overexpressing cell line exhibits normal glycosylation, making it suitable for further related research.

Figure 5.

Panx1 increases β-Catenin activity and promotes its nuclear translocation

(A) Western blot shows Panx1 overexpression and knockout efficiency (GAPDH as loading control).

(B) Glycosylation patterns of exogenous Panx1.

(C and D) Total and non-phosphorylated β-catenin levels in whole cell and nuclear fractions (Lamin B1 as loading control).

(E) Immunofluorescence of β-catenin nuclear translocation (red; nuclei, blue). Scale bars, 25 μm.

(F) qRT-PCR of Wnt/β-catenin downstream gene expression normalized to PPIA.

(G) Western blot of Wnt/β-catenin downstream proteins (GAPDH as loading control). Data are mean ± SD, statistics in (C), (D), (F), and (G): ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001, two-sided Student’s t test.

Both total and non-phosphorylated β-catenin levels were significantly increased in Panx1-OE cells (total β-Catenin, p < 0.01; Ser45, p < 0.05; Ser33/37/Thr41, p < 0.05) compared to EV control in the absence of Wnt3a stimulation, whereas Panx1 knockout resulted in a significant decrease (p < 0.05) (Figure 5C). Consistent with these findings, nuclear protein extracts from the stable cell lines showed a comparable increase (total β-catenin, p < 0.05; Ser33/37/Thr41, p < 0.01 in Panx1-OE cells), which was further supported by enhanced nuclear β-catenin signals detected via immunofluorescence staining in Panx1-OE cells (Figures 5D and 5E). In contrast to the WT, the transcript expression levels of β-Catenin (p < 0.001) and its downstream genes, such as C/EBPα (p < 0.001), CCND1, c-Jun (p < 0.001), and PPARγ (p < 0.001), increased in Panx1-OE cells (Figure 5F). Protein expression analysis revealed an increase in the levels of CCND1, C/EBPα, c-Myc and PPARγ in Panx1-OE cells (p < 0.05), while these levels were reduced in Panx1^−/− cells (p < 0.05 for C/EBPα, c-Myc, and PPARγ) (Figure 5G).

Because Wnt3a is a prototypical canonical Wnt ligand,¹⁸ we used short-term Wnt3a stimulation (40 ng/mL for 6 h) as a positive control to acutely activate Wnt/β-catenin signaling and to test whether Panx1-dependent effects on β-catenin are maintained upon ligand stimulation. As shown in Figure S6, Wnt3a treatment increased total and non-phosphorylated β-catenin levels in all groups, yet the relative differences between Panx1-OE and Panx1^−/− cells remained similar to those observed under basal conditions. Together, these data indicate that Panx1 regulates β-catenin stability, nuclear accumulation, and downstream transcriptional activity at baseline, and that this regulatory effect is mediated predominantly through an intracellular, ligand-independent mechanism.

We next induced the differentiation of Panx1-OE and Panx1^−/− 3T3-L1 preadipocytes into mature adipocyte-like cells (monitoring by BODIPY staining, Figures S7A and S7B), respectively, and then examined the impact of Panx1 on the Wnt/β-Catenin signaling pathway. The results indicated that there were no significant changes in gene expression levels of β-Catenin, c/EBPα, CCND1, c-Myc, and PPARγ (Figure S6C), as well as in the protein levels of β-Catenin, non-phosphorylated β-Catenin, CCND1, and c-Myc (Figures S7D and S7E). Taken together with the findings in proliferating preadipocytes, these results suggest that the modulatory effects of Panx1 on Wnt/β-catenin signaling are most pronounced in the preadipocyte state and become largely attenuated once cells have undergone adipogenic differentiation.

We also conducted a preliminary assessment of whether Panx1 directly affected adipocyte differentiation in mature adipocytes by measuring neutral lipid accumulation in MDI-induced adipocyte-like cells as a readout of differentiation status (Figure S8). In combination with the above analysis of key Wnt/β-catenin-linked adipogenic differentiation markers (C/EBPα, PPARγ, c-Myc, and CCND1), these experiments showed that Panx1 overexpression or deletion did not produce appreciable changes in any of these measures, suggesting that Panx1 had no major impact on terminal adipogenic differentiation in this system. These conclusions were based on analyses performed at a single late differentiation time point in 3T3-L1 cells using a defined marker panel, and more detailed, time-resolved studies would be required to clarify whether Panx1 exerted subtler or context-dependent effects on adipogenesis.

Pannexin 1 promotes the phosphorylation of glycogen synthase kinase-3β

The destruction complex (DC), comprising scaffold proteins and kinases such as axis inhibition protein 1 (Axin1), adenomatous polyposis coli (APC), the Ser/Thr kinases GSK-3β and CK1α, and the E3 ubiquitin ligase β-TrCP, plays a critical role in phosphorylating and targeting β-catenin for degradation in the absence of Wnt signaling.¹⁸ To determine whether Panx1 influences this complex, we analyzed the expression of key DC components in Panx1 overexpressing and knockout 3T3-L1 preadipocytes cell. Compared to control cells, Panx1-OE cells exhibited significantly elevated levels of phosphorylated GSK-3β (Ser9) and adenomatous polyposis coli (APC), which is a central scaffolding protein that enhances GSK-3beta activity within the DC (p-GSK3β, p < 0.05; p-GSK-3β/GSK-3β, p < 0.01; APC, p < 0.05) (Figure 6A). Conversely, these levels were reduced in Panx1^−/− cells (p-GSK3β, p < 0.05; p-GSK-3β/GSK-3β, p < 0.01; APC, p < 0.05) (Figure 6A). The total protein levels of GSK-3β and other components showed no change in either cell line (Figure 6A). Given that the phosphorylation of GSK-3β at Ser9 inhibits its ability to bind substrates,²¹ our findings suggest that Panx1 is involved in regulating the phosphorylation of GSK-3β and its subsequent action on β-catenin.

Figure 6.

Panx1 forms complexes with β-Catenin and GSK-3β

(A) Western blot of destruction complex proteins in Panx1-OE and Panx1^−/− 3T3-L1 cells (GAPDH as loading control).

(B) Co-IP of Panx1 with β-catenin or destruction complex proteins in HEK293T cells expressing Panx1-HA.

(C) Co-IP of β-catenin (total or phosphorylated) with GSK-3β in Panx1-OE and Panx1^−/− cells (GAPDH as loading control for input).

(D) Western blot showing β-catenin knockdown efficiency in Panx1-OE cells (GAPDH as loading control).

(E) Co-IP of Panx1 with GSK-3β after β-catenin knockdown (GAPDH as loading control for input).

(F) Predicted 3D structures of Panx1, β-catenin, and GSK-3β complexes showing conformational changes.

(G) Ranking scores of binding affinities among Panx1, β-catenin, and GSK-3β. Data are mean ± SD, statistics in (A): ∗p < 0.05 and ∗∗p < 0.01, two-sided Student’s t test.

Protein-protein interactions among Pannexin 1, β-catenin, and glycogen synthase kinase-3β

We further examined the protein-protein interaction (PPI) between Panx1 and the β-Catenin-associated regulatory complex. The results confirmed that Panx1 binds directly to both GSK-3β and β-Catenin (positive control), but not to other components of the destruction complex (DC) (Figure 6B). Furthermore, the interaction between GSK-3β and total β-Catenin was reduced in Panx1-OE cells, while its binding to phosphorylated β-catenin (Ser33/37/Thr41) remained unchanged (Figure 6C). Conversely, in Panx1^−/− cells, GSK-3β exhibited enhanced binding to both total and phosphorylated β-catenin (Figure 6C). Upon knockdown of β-catenin in Panx1-OE cell lines using siRNA-1 (selected based on its knockdown efficiency), densitometric analysis revealed only a modest reduction in the co-immunoprecipitated GSK-3β signal, and the Panx1-GSK-3β association was largely preserved (Figures 6D and 6E). Together with the co-immunoprecipitation data in Panx1-OE and Panx1^−/− cells, these findings suggest that Panx1 can form a complex with β-catenin and GSK-3β and may contribute to regulating their protein-protein interactions, but β-catenin is not required as an obligatory bridge for Panx1–GSK-3β binding.

Structural insights into the Pannexin 1, glycogen synthase kinase-3β, and β-catenin complex by AlphaFold multimeric modeling

To investigate potential protein-protein interactions and binding mechanisms, we performed structural modeling of the ternary complex using AlphaFold. A structural model of the ternary complex, highlighting the interactions between the three proteins, is presented in Figure 6F. The results showed that specific regions of Panx1, including residues 166–179 and 372–422, were identified as binding sites that interact with distinct regions of β-catenin. These include β-catenin residues 112–123 (located near the N-terminus), 292–307, 425–430, and 559–568 (near the C-terminal). In contrast, Panx1 interacted with GSK-3β through non-covalent binding involving specific segments, namely residues 7–18, 158–172, and 361–372 of Panx1, which engaged with GSK-3β at residues 56–71, 85–93, 127–128, 216, 228–229, 260–275, and 285–294. When bound to GSK-3β, β-catenin primarily formed non-covalent interactions through regions near its N-terminus (residues 132–151 and 182–193) as well as the segment spanning residues 224–226. These regions interacted with GSK-3β residues 63–66, 93–96, 140–148, 180–185, 205–221, and 249–271. Collectively, these findings demonstrate that the N-terminal region of β-catenin, which contains key phosphorylation sites such as Ser33/Ser37/Thr41, could form non-covalent interactions with both Panx1 and GSK-3β (Figure 6F).

Additionally, β-catenin showed significant conformational changes in its disordered regions near the N-terminus (residues 1–146) and the C-terminus (residues 686–781) upon binding to GSK-3β (Figure 6F). The flexibility of these disordered regions likely promotes phosphorylation by enabling major conformational changes upon GSK-3β binding. In contrast, binding to Panx1 did not induce significant conformational changes in the N- or C-terminal regions of β-catenin compared to its unbound state, thereby contributing to the stabilization of the β-catenin protein structure (Figure 6F). Analysis of relative binding strengths indicated that the interaction between Panx1 and β-catenin exhibited the highest affinity, with a ranking score of 0.41, compared to those between GSK-3β and β-catenin (0.37), and between Panx1 and GSK-3β (0.36) (Figure 6G).

These data suggest that Panx1 enhances the stabilization of β-Catenin protein conformation through high-affinity binding, thereby affecting the binding or catalytic efficiency of GSK-3β to β-Catenin.

Pannexin 1 deficiency inhibits preadipocyte proliferation and differentiation

The Wnt/β-Catenin signaling pathway, along with its downstream effectors such as c-Myc, CCND1, C/EBPα, and PPARγ, is known to be critically involved in adipocyte differentiation.²² To investigate whether Panx1 influences this process, we examined the effects of Panx1 overexpression and knockout on the proliferation of 3T3-L1 preadipocytes using CCK-8 assay and flow cytometry. The CCK-8 assay demonstrated that Panx1 overexpression significantly promoted the proliferation of 3T3-L1 preadipocytes compared to the EV group (p < 0.001), whereas Panx1 deficiency markedly inhibited proliferation (p < 0.001) (Figure 7A). Flow cytometry analysis revealed that Panx1 overexpression led to a significant decrease in the proportion of cells in the G1 phase (36.6% vs. 40.6% in EV group, p < 0.01) and a notable increase in the G2/M phase (23.5% vs. 18.2%, p < 0.01), with no significant change in the S phase (Figures 7B and 7D). In contrast, Panx1 knockout resulted in increased cell accumulation in the G1 (37.5% vs. 35.4%) and S phases (44.2% vs. 38.3%), along with a reduced proportion of cells in the G2/M phase (18.5% vs. 26.8%) compared to controls (Figures 7C and 7D). In all analyses, comparisons were restricted to Panx1-OE versus EV and Panx1^−/− versus Control within each independently generated pair of stable lines, rather than between the two different control groups.

Figure 7.

Effect of Panx1 on cell proliferation

(A) Cell viability of Panx1-OE and Panx1^−/− cells.

(B–D) Flow cytometry analysis of cell cycle distribution (PI staining) with the quantification of G1, S, and G2/M phases.

(E) Western blot confirms β-catenin knockdown and effects on downstream proteins in Panx1-OE cells.

(F and G) Flow cytometry and quantification of cell cycle distribution after β-catenin knockdown. Each panel represents an independently generated pair of stable lines, and statistical analyses are performed within each matched pair. Data are mean ± SD, statistics in (A), (D), and (G): ∗∗p < 0.01 and ∗∗∗p < 0.001, two-sided Student’s t test.

In addition, we also investigated cell cycle changes after knockdown of β-Catenin in Panx1-OE 3T3 cells (Figure 7E). Knockdown of β-Catenin significantly reverses the effect of Panx1 overexpression on the cell cycle of 3T3 cells, as evidenced by an increase in the distribution of cells in the G1 phase (43.1%) and slightly enhances in the S phase (39.8%), along with a decrease in the G/M phase (16.0%) compared to the siCtrl (G1:39.5%; S:39.2%; G/M:19.0%) (Figures 7F and 7G). This pattern is similar to that observed in Panx1^−/− cells. These results suggest that Panx1 promotes cell proliferation by enhancing mitosis (from G2 phase to M phase), whereas the lack of Panx1 reverses this process, resulting in the retention of cells at the DNA synthesis stage (S phase), which appears to be dependent on the Wnt/β-Catenin pathway.

The above results demonstrate that Panx1 deficiency inhibits adipocyte hyperplasia by disrupting the Wnt/β-catenin signaling pathway.

Discussion

Although Panx1 deficiency has been linked to obesity, its precise role within adipocytes remains unclear. Here, we examined how loss of Panx1 influences adipocyte hyperplasia and hypertrophy. Under prolonged exposure to a diet rich in fat, fructose, and cholesterol, Panx1^−/− mice exhibited a pronounced obese phenotype, marked by greater body weight gain and enlarged adipocytes. Mechanistically, Panx1 knockout disrupted lipid metabolism in eWAT, as demonstrated by accumulated TG and TC, enhanced fatty acid synthesis, and impaired fatty acid oxidation, all of which collectively promoted adipocyte hypertrophy. Furthermore, we found that Panx1, β-catenin, and GSK3β form a protein complex via PPI. Within this complex, Panx1 contributes to the stabilization of β-catenin conformation. Conversely, the deficiency of Panx1 promoted β-catenin degradation and downregulated its downstream target genes, resulting in G1 phase cell-cycle arrest and suppression of adipocyte hyperplasia, which indirectly exacerbated adipocyte hypertrophy.

Panx1 plays a crucial role in paracrine and autocrine signaling by acting as a channel for the release of inorganic ions and bioactive metabolites, such as calcium, chloride, and particularly ATP, thereby facilitating communication between cells and their environment, as well as among adjacent cells.²³ Previous studies have demonstrated the involvement of pannexins and extracellular ATP in various physiological and pathological processes.²⁴ In the context of liver disease, Panx1 knockout has been shown to protect mice from acetaminophen-induced acute liver injury and to ameliorate carbon tetrachloride-induced liver fibrosis by suppressing the expression of α-SMA and COL1A1.²⁵ In our GAN and HFD models, Panx1 deficiency did not aggravate overall liver injury. Under GAN feeding, Panx1^−/− mice showed only a modest increase in hepatic lipid droplet accumulation together with higher hepatic TG and TC levels, whereas liver size, histological and molecular markers of fibrosis, inflammatory cell infiltration, and cytokine profiles were largely comparable between genotypes. Importantly, serum TG and TC concentrations remained unchanged, indicating that Panx1 deficiency does not cause overt systemic dyslipidemia in this setting. At the transcriptional level, classical lipogenic and fatty-acid oxidation genes were not significantly altered in Panx1-deficient livers, whereas CD36 and FABP1 were selectively upregulated, and MTTP expression was preserved. Because WT and Panx1^−/− mice were exposed to the same prolonged GAN diet, this pattern suggests that Panx1 deficiency modulates diet-induced hepatic lipid handling primarily by enhancing CD36⁻and FABP1-dependent fatty-acid uptake and intracellular trafficking rather than by driving de novo lipogenesis, impairing β-oxidation, or compromising VLDL export. This interpretation is consistent with previous reports that high-fat and Western-style diets increase hepatic CD36 and FABP1 expression as adaptive responses to increased dietary lipid flux, and that loss of FABP1 protects against Western diet-induced obesity and steatosis.^26,27,28,29 Together with our observations in eWAT, these findings support the notion that Panx1 exerts distinct, tissue-specific effects on lipid metabolism, promoting adipocyte hypertrophy while only modestly influencing hepatic lipid homeostasis in this model.

A pronounced obesity phenotype was observed exclusively in mice that fed the GAN diet, but not in those subjected to the HFD. Consistent with our results, a previous study also reported that Panx1 KO mice did not exhibit increased fat mass when fed a high-fat diet.³⁰ These findings suggest that specific dietary components may influence the development of obesity associated with Panx1 deficiency. The GAN diet—characterized by high levels of fat, fructose, and cholesterol, has been widely used to establish mouse models of MASLD and obesity.³¹ Epidemiological studies have further revealed a positive association between the consumption of fructose-sweetened beverages and the increasing prevalence of obesity in Western populations.³² Moreover, chronic fructose exposure has been shown to enhance insulin secretion in human islets via a mechanism dependent on Panx1-mediated ATP release, an effect that can be reversed by Panx1 blockade.³³ Taken together, these reports imply that high dietary fructose and its ability to activate Panx1 might underline the obesity phenotype observed in Panx1^−/− mice under the GAN diet.

Ectopic lipid accumulation-induced lipotoxicity, inflammatory cytokine secretion from macrophages infiltrating visceral adipose tissue, insulin resistance, and increased circulating levels of non-esterified fatty acids, glycerol, and hormones collectively drive the development of obesity.^34,35 In the present study, we found that Panx1^−/− mice had markedly fewer F4/80-positive macrophages in eWAT than WT controls, accompanied by reduced levels of inflammatory cytokines including IL-1α, IL-6, and TNF-α. These results suggest that Panx1 deficiency disrupts the conventional link between obesity and inflammation. Previous studies have demonstrated that Panx1 plays a dual role in regulating both acute and chronic inflammation, primarily through ATP and calcium-dependent signaling pathways.³⁶ Inhibition of Panx1 channels has been shown to ameliorate inflammatory responses.³⁶ Additionally, Panx1 channel activation during apoptosis promotes the assembly of NLRP3 inflammasomes and facilitates IL-1β signaling.³⁷ Conversely, genetic or pharmacological inhibition of Panx1 suppresses NF-κB activation and subsequently reduces IL-1β expression.³⁸ In line with these reports, the reduced inflammatory cytokine levels observed in our study may be attributed to the regulatory function of Panx1 in inflammatory processes. Furthermore, obesity in Panx1-deficient GAN-fed mice occurred independently of inflammation and eWAT browning, indicating that dysregulated lipid metabolism and suppressed adipocyte hyperplasia primarily drove the weight gain in these animals.

In this study, we observed significant decreases in both total and non-phosphorylated β-Catenin, as well as GSK-3β phosphorylation levels at the Ser9 site in Panx1^−/− mice and 3T3-L1 preadipocytes. Conversely, these levels were significantly increased in Panx1 overexpressing cells. The activation of the canonical Wnt signal is generally considered to have an anti-adipogenic effect, and blocking this pathway at multiple levels of signal cascades promotes adipogenesis.^39,40 In the absence of Wnt ligands, β-Catenin binds to the DC, which includes APC, AXIN, GSK-3β, CK1α, and β-TRCP.⁴¹ It is phosphorylated by GSK-3β at the Ser37, Ser33 and Thr41 sites, as well as by CK1α at the Ser45 site, and then degraded through the ubiquitination process regulated by β-TrCP.⁴² The catalytic activity of GSK-3β is regulated by two phosphorylation sites.⁴³ Phosphorylation at the Ser9 site in the N-terminal region of GSK-3β negatively regulates its kinase activity by inhibiting the binding of substrate, while phosphorylation at the Tyr216 site in the activation loop of this protein increases its catalytic activity.^44,45 Protein kinase A, Akt, and protein kinase C have all been implicated in regulating the phosphorylation of GSK3β at the Ser9 residue.²¹ Ji-Chun Yang and colleagues recently reported that Panx1 overexpression activates the Akt pathway, leading to reduced Foxo1 protein levels and downregulation of its gluconeogenic target genes in HFD-fed mice.⁴⁶ These observations suggest that Panx1 may promote the Ser9 phosphorylation of GSK3β via Akt signaling, although the precise mechanisms remain to be clarified.

Silvia Penuela et al. reported that Panx1 binds directly to the N-terminus of β-Catenin proteins through its C-terminal region, and that knockdown of Panx1 reduces the abundance of β-Catenin in melanoma cells.⁴⁷ They also propose a hypothesis that the connection between Panx1 and β-Catenin likely enhances the stability of the β-Catenin protein, but there is still insufficient evidence to support this.⁴⁷ In contrast to their study, we have demonstrated that Panx1 can interact with both GSK3 and β-Catenin proteins. Overexpressing Panx1 reduces the molecular interaction between GSK-3β and β-Catenin, while knocking out Panx1 can increase this association. In Panx1^−/− cells, we observed increased levels of phosphorylated β-Catenin protein interacting with GSK-3β, suggesting that Panx1 deficiency promotes the enzymatic activity of GSK-3β on β-Catenin. In addition, GSK-3β primarily binds to the N-terminal residues of the β-Catenin protein. This interaction leads to conformational changes of β-Catenin, which may facilitate the phosphorylation process. In contrast, Panx1 binds to the N- and C-terminal of β-Catenin, respectively, without altering the conformation of either protein. Most importantly, it was found that Panx1 has a stronger relative binding affinity and lower energy with β-Catenin within the complex. All the data we have obtained indicate that Panx1 has a strong affinity for β-Catenin, leading to increased protein stability and preventing its conformational change induced by GSK-3β and subsequent phosphorylated degradation. In line with these findings, our data further show that this Panx1-GSK-3β-β-catenin axis operates largely independently of extracellular Wnt ligands, highlighting Panx1 as an intrinsic upstream modulator that can tune basal β-catenin activity even when Wnt ligand availability is low or fluctuating.

Numerous studies have demonstrated that the Wnt/β-Catenin pathway plays a key regulatory role in the differentiation of adipocyte progenitors to mature adipocyte-like cells.^48,49 In this study, we did not observe any changes in Wnt/β-Catenin signaling pathway-related proteins or genes in mature 3T3 adipocyte-like cells. Meanwhile, Panx1-OE 3T3-L1 preadipocytes exhibited a significant decrease in cell distribution in the G1 phase and an increase in cell distribution in the G2/M phase. Conversely, Panx1 knockdown resulted in more cells being arrested in the G1 and S phases, while decreasing cellular distribution in the G2 phase, which is consistent with β-Catenin knockdown in Panx1-OE 3T3 cells. These findings demonstrate that Panx1’s regulation of the cell cycle through the Wnt/β-Catenin signaling pathway primarily occurs during the differentiation process of adipocyte progenitor cells.

Sex is a well-recognized biological variable in obesity and metabolic disease, and males and females differ in adipose tissue distribution, hormonal regulation, and susceptibility to liver injury. In the present study, we therefore focused our main analyses on male mice to reduce biological variability and to maximize statistical power for mechanistic experiments. Nevertheless, an independent female cohort subjected to the same HFD and GAN protocols showed qualitatively similar responses to Panx1 deficiency, with no detectable obesity phenotype under HFD and increased body weight and adiposity under GAN feeding, albeit with a slightly delayed onset compared with males (Figure S2). These findings support the generalizability of our conclusions across sexes.

Collectively, our study demonstrates that Panx1 deficiency exacerbates obesity and promotes adipocyte hypertrophy in mice fed a GAN diet. Mechanistically, Panx1 deficiency facilitates the phosphorylation-dependent degradation of β-catenin and downregulates its downstream target genes, leading to suppressed adipocyte hyperplasia and ultimately accelerating hypertrophic progression (Graphical Abstract). These results establish Panx1 as a key regulatory protein in obesity pathogenesis. Our findings provide a deeper understanding of the mechanisms underlying obesity and offer potential insights for developing targeted interventions for metabolic diseases.

Limitations of the study

This study has several limitations. First, the use of an adipose tissue-specific Panx1 knockout mouse model would provide more direct evidence for the adipose-intrinsic role of Panx1 in obesity. Second, dietary interventions using single components, such as fructose or other individual nutrients, may help to more precisely identify which elements of the GAN diet drive the observed phenotype. Third, owing to the limited availability of clinical samples, we were unable to directly examine the association between Panx1 expression in human adipose tissue and obesity. Fourth, chow-fed wild-type and Panx1-deficient mice were assessed primarily for body weight, and more subtle metabolic effects under standard chow conditions cannot be excluded. Finally, although an independent female cohort showed qualitatively similar responses to Panx1 deficiency under HFD and GAN feeding, adequately powered studies will be required to define potential sex-specific roles of Panx1 in adipose tissue and systemic metabolism.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the correspondence contact, Yang Zhang (yang518@mail.ccmu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Data: The RNA-seq data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject accession number SRA: PRJNA1335784 and are publicly available.
• •Code: This article does not report original code.
• •Other: Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (grant number: 82070627), the Beijing Municipal Natural Science Foundation (grant number: 7222090, 7262063); the Beijing Municipal Institute of Public Medical Research Development and Reform Pilot Project (grant number: 2021-10), the Capital’s Funds for Health Improvement and Research (grant number: 2024-2-1151), the Beijing Institute of Hepatology Reform and Development Project (grant number: 2024Y-KF-03); Beijing high-level Public Health Technical Talent Construction Project (grant number: 03-23), the CAMS Innovation Fund for Medical Science (grant number: 2021-I2M-1-042), the CAMS Innovation Fund for Medical Sciences (CIFMS) (grant number: 2022-I2M-2-002), the Scientific Research Project of Beijing Youan Hospital (BJYAYY-YN2024-15), the Natural Science Foundation of Capital Medical University (PYZ24153, PYZ25152), the Beijing High-Level Innovation and Entrepreneurship Talent Support Program, and Young Backbone Talent Projects (G202532308).

Author contributions

Y.Z. and H.X. conceived and designed the experiments. H.X., S.L., Q.W., and R.S. performed the experiments. H.X., Y.Z., S.L., Y.G., J.Z., F.X., Y.W., and R.C. collected data and performed statistical analyses. X.L., B.K., and M.C. were involved in the design and methodology of mouse experiments. X.Y. contributed to reagents and materials. H.X. and Y.Z. interpreted the data and drafted the article. Y.F. and J.Z. conducted the data analysis and revised the article. Y.Z. and B.L. supervised the work. All authors read and approved the final article.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Pannexin-1 (D9M1C) Rabbit mAb	Cell Signaling Technology	Cat# 91137S; RRID: AB_2800184
DYKDDDDK Tag (D6W5B) Rabbit mAb	Cell Signaling Technology	Cat# 14793S; RRID: AB_2572291
GAPDH (D16H11) XP Rabbit mAb	Cell Signaling Technology	Cat# 5174S; RRID: AB_10622025
β-Actin (13E5) Rabbit mAb	Cell Signaling Technology	Cat# 4970S; RRID: AB_2223172
Anti-PRDM16 antibody	Abcam	Cat# ab106410; RRID: AB_10861873
F4/80 Rabbit mAb	ABclonal	Cat# A23788
Anti-UCP1 antibody	Abcam	Cat# ab10983; RRID: AB_2241462
ATGL antibody	Cell Signaling Technology	Cat# 2138S; RRID: AB_2167955
Anti-ABHD5/CGI-58 antibody	Abcam	Cat# ab183739
c-Myc antibody	Cell Signaling Technology	Cat# 9402S; RRID: AB_2151827
C/EBPα (D56F10) XP Rabbit mAb	Cell Signaling Technology	Cat# 8178S; RRID: AB_11178525
Anti-Lamin B1 antibody	Abcam	Cat# ab16048; RRID: AB_443298
Phospho-β-Catenin (Ser33/37/Thr41) antibody	Cell Signaling Technology	Cat# 9561S; RRID: AB_331129
Phospho-β-Catenin (Ser45) antibody	Cell Signaling Technology	Cat# 9564S; RRID: AB_331130
Cyclin D1 (E3P5S) XP Rabbit mAb	Cell Signaling Technology	Cat# 55506S; RRID: AB_2799427
c-Jun (60A8) Rabbit mAb	Cell Signaling Technology	Cat# 9165S; RRID: AB_2130165
β-TRCP (D13F10) Rabbit mAb	Cell Signaling Technology	Cat# 4394S; RRID: AB_10545756
Axin1 (C76H11) Rabbit mAb	Cell Signaling Technology	Cat# 2087S; RRID: AB_10544575
β-Catenin (D10A8) XP Rabbit mAb	Cell Signaling Technology	Cat# 8480S; RRID: AB_11127855
GSK-3β (D5C5Z) XP Rabbit mAb	Cell Signaling Technology	Cat# 12456S; RRID: AB_2636978
CK1α polyclonal antibody	Proteintech	Cat# 55192-1-AP
APC polyclonal antibody	Proteintech	Cat# 19782-1-AP
Phospho-GSK-3β (Ser9) XP Rabbit mAb	Cell Signaling Technology	Cat# 5558S; RRID: AB_331405
Non-phospho (Active) β-Catenin (Ser33/37/Thr41) Rabbit mAb	Cell Signaling Technology	Cat# 8814S; RRID: AB_11127856
Non-phospho (Active) β-Catenin (Ser45) XP Rabbit mAb	Cell Signaling Technology	Cat# 19807S; RRID: AB_2798342
ANTI-FLAG® M2 Affinity Gel	Millipore	Cat# A2220; RRID: AB_10063035
Anti-rabbit IgG, HRP-linked antibody	Cell Signaling Technology	Cat# 7074S; RRID: AB_2099233
Anti-mouse IgG, HRP-linked antibody	Cell Signaling Technology	Cat# 7076S; RRID: AB_330924
Bacterial and viral strains
Lentivirus: GV492 (Ubi-MCS-3FLAG-CBh-gcGFP-IRES-puro)	GeneChem	Cat# GV492
Lentivirus: GV708 (U6-sgRNA-EF1a-Cas9-FLAG-CMV-EGFP-P2A-puro)	GeneChem	Cat# GV708
Chemicals, peptides, and recombinant proteins
TRIzol reagent	Life Technologies	Cat# 15596018
HiScript III All-in-one RT SuperMix	Vazyme	Cat# R323-01
ChamQ SYBR qPCR Master Mix	Vazyme	Cat# Q311-03
DMEM	Gibco	Cat# 11965092
Fetal bovine serum	Gibco	Cat# 10099141
Lipofectamine 3000	Invitrogen	Cat# L3000001
Deposited data
RNA-seq data	NCBI SRA	PRJNA1335784
Experimental models: Cell lines
3T3-L1 preadipocytes	Cyagen	Cat# M5-0101
HEK293T cells	Chinese Academy of Sciences Cell Bank	N/A
Experimental models: Organisms/strains
Panx1 global knockout (Panx1−/−) mice	GemPharmatech	N/A
C57BL/6 WT mice	GemPharmatech	N/A
Oligonucleotides
mouse c-Jun-F	This paper	5′-ACTCGGACCTTCTCACGTC-3′
mouse c-Jun-R	This paper	5′-GGTCGGTGTAGTGGTGATGT-3′
mouse ATGL-F	This paper	5′-TCCGTGGCTGTCTACTAAAGA-3′
mouse ATGL-R	This paper	5′-TGGGATATGATGACGTTCTCTCC-3′
mouse CGI-58-F	This paper	5′-GGGTGGCTTCCTACCTGGT-3′
mouse CGI-58-R	This paper	5′-CAGCGTCCATATTCTGTTTCCAT-3′
mouse C/EBPα-F	This paper	5′-AGGTGCTGGAGTTGACCAGT-3′
mouse C/EBPα-R	This paper	5′-CAGCCTAGAGATCCAGCGAC-3′
mouse CCND1-F	This paper	5′-CACAACGCACTTTCTTTCCA-3′
mouse CCND1-R	This paper	5′-ACCAGCCTCTTCCTCCACTT-3′
mouse c-Myc-F	This paper	5′-GCTCGCCCAAATCCTGTA-3′
mouse c-Myc-R	This paper	5′-AGGACTCGGAGGACAGCA-3′
mouse β-Catenin-F	This paper	5′-GCTGCTCATCCCACTAATGT-3′
mouse β-Catenin-R	This paper	5′-CCGCGTCATCCTGATAGTTAAT-3′
mouse PPARγ-F	This paper	5′-GGAAGACCACTCGCATTCCTT-3′
mouse PPARγ-R	This paper	5′-GTAATCAGCAACCATTGGGTCA-3′
mouse COL3A1-F	This paper	5′-CCTGGCTCAAATGGCTCAC-3′
mouse COL3A1-R	This paper	5′-GACCTCGTGTTCCGGGTAT-3′
mouse ACACA-F	This paper	5′-AATGAACGTGCAATCCGATTTG-3′
mouse ACACA-R	This paper	5′-ACTCCACATTTGCGTAATTGTTG-3′
mouse ACLY-F	This paper	5′-AATCCTGGCTAAAACCTCGCC-3′
mouse ACLY-R	This paper	5′-GCATAGATGCACACGTAGAACT-3′
mouse FASN-F	This paper	5′-AGAGATCCCGAGACGCTTCT-3′
mouse FASN-R	This paper	5′-GCTTGGTCCTTTGAAGTCGAAGA-3′
mouse SCD1-F	This paper	5′-CCGGAGACCCCTTAGATCGA-3′
mouse SCD1-R	This paper	5′-TAGCCTGTAAAAGATTTCTGCAAACC-3′
mouse ACOX1-F	This paper	5′-CAGACCCTGAAGAAATCATGTGG-3′
mouse ACOX1-R	This paper	5′-CAGGAACATGCCCAAGTGAAG-3′
mouse SREBP2-F	This paper	5′-TGGGCGATGAGCTGACTCT-3′
mouse SREBP2-R	This paper	5′-CAAATCAGGGAACTCTCCCAC-3′
mouse DGAT2-F	This paper	5′-CGAGACACCATAGACTACTTGCT-3′
mouse DGAT2-R	This paper	5′-GCGGTTCTTCAGGGTGACTG-3′
mouse DGAT1-F	This paper	5′-GTGCCATCGTCTGCAAGATTC-3′
mouse DGAT1-R	This paper	5′-GCATCACCACACACCAATTCAG-3′
mouse ACTA2-F	This paper	5′-CCCAGACATCAGGGAGTAATGG-3′
mouse ACTA2-R	This paper	5′-TCTATCGGATACTTCAGCGTCA-3′
mouse CPT1A-F	This paper	5′-AGATCAATCGGACCCTAGACAC-3′
mouse CPT1A-R	This paper	5′-CAGCGAGTAGCGCATAGTCA-3′
mouse PPIA-F	This paper	5′-CAGTGCTCAGAGCTCGAAAGT-3′
mouse PPIA-R	This paper	5′-CACCGTGTTCTTCGACATCA-3′
mouse MTTP-F	This paper	5′-ATACAAGCTCACGTACTCCACT-3′
mouse MTTP-R	This paper	5′-TCTCTGTTGACCCGCATTTTC-3′
mouse CD36-F	This paper	5′-AGATGACGTGGCAAAGAACAG-3′
mouse CD36-R	This paper	5′-CCTTGGCTAGATAACGAACTCTG-3′
mouse Panx1-F	This paper	5′-CCACCGAGCCCAAGTTCAA-3′
mouse Panx1-R	This paper	5′-GGAGAAGCAGCTTATCTGGGT-3′
mouse FABP1-F	This paper	5′-AAGGCAGTCGTCAAGCTGG-3′
mouse FABP1-R	This paper	5′-CATTGAGTTCAGTCACGGACTT-3′
mouse Ang-1-F	This paper	5′-TGCACTAAAGAAGGTGTTTTGCT-3′
mouse Ang-1-R	This paper	5′-CCGGTGTTGTATTACTGTCCAA-3′
mouse Ang-2-F	This paper	5′-AGAATAAGCAAGTCTCGCTTCC-3′
mouse Ang-2-R	This paper	5′-TGAACCCTTTAGAGGCTCGGT-3′
mouse Pdgfra-F	This paper	5′-ATGAGAGTGAGATCGAAGGCA-3′
mouse Pdgfra-R	This paper	5′-CGGCAAGGTATGATGGCAGAG-3′
mouse VEGFA-F	This paper	5′-GCACATAGAGAGAATGAGCTTCC-3′
mouse VEGFA-R	This paper	5′-CTCCGCTCTGAACAAGGCT-3′
mouse VEGFB-F	This paper	5′-GCCAGACAGGGTTGCCATAC-3′
mouse VEGFB-R	This paper	5′-GGAGTGGGATGGATGATGTCAG-3′
Panx1 gRNA #1	GemPharmatech	5′-CAATAGCGTACCCTAGTCT-3′
Panx1 gRNA #2	GemPharmatech	5′-TCATGTCACATCATCGCTCA-3′
sgRNA #1	This paper	5′-CACTGACAGACACCTGCCTG-3′
sgRNA #2	This paper	5′-CCTCTGGTCTGCTCTGTGTC-3′
sgRNA #3	This paper	5′-ATCGGCCTTCTATCGCCTTCT-3′
β-catenin siRNA #1 (sense)	This paper	5′-GAUGGUGUCUGCUAUUGUATT-3′
β-catenin siRNA #1 (antisense)	This paper	5′-UACAAUAGCAGACACCAUCTT-3′
β-catenin siRNA #2 (sense)	This paper	5′-GCAUAACCUUUCCCAUCAUTT-3′
β-catenin siRNA #2 (antisense)	This paper	5′-AUGAUGGGAAAGGUUAUGCTT-3′
Scrambled siRNA (siCtrl) sense	This paper	5′-UUCUCCGAACGUGUCACGUTT-3′
Scrambled siRNA (siCtrl) antisense	This paper	5′-ACGUGACACGUUCGGAGAATT-3′
Software and algorithms
GraphPad Prism	GraphPad	Version 8.3
R software	CRAN	Version 4.2
ImageJ	NIH	Version 2.2.0
xPonent	Luminex	Version 2.3.0
AF2-Multimer	DeepMind	N/A
AlphaFold3	DeepMind	N/A
PyMOL	Schrödinger	N/A
PLIP	Open source	N/A
Other
ChemiDoc MP Imaging System	Bio-Rad	N/A
Flow cytometer (FACSCalibur)	BD Biosciences	N/A
Leica TCS SP8 confocal microscope	Leica	N/A
QMR nuclear magnetic resonance system	NIUMAG	QMR06-090H

Experimental model and study participant details

C57BL/6 global Panx1 knockout (Panx1^−/−) mice were generated by GemPharmatech Co., Ltd. (Nanjing, China) using CRISPR/Cas9 technology. The targeting strategy deleted a critical coding region of the Panx1 gene to produce a global loss-of-function allele. Genomic DNA was isolated from mouse tail biopsies using the HotSHOT alkaline lysis method as described by Truett et al.,⁵⁰ and 1–5 μL of lysate were used for PCR genotyping. Genotyping was performed by allele-specific PCR with a common forward primer and WT- or KO-specific reverse primers. The WT and targeted alleles yielded products of 530 bp and 718 bp, respectively (Figure S1A). Panx1 protein expression was further examined by Western blotting in brain lysates, where Panx1 is normally highly expressed, and Panx1 was detected in WT but not in Panx1^−/− brain (Figure S1B), confirming a functional knockout.

Cell lines

3T3-L1 preadipocytes (M5-0101, Cyagen Bioscience, Suzhou, China) were cultured in OriCell Basal Medium (BHDM-03011, OriCell, Suzhou, China) supplemented with 10% bovine calf serum (FBSST-01033, OriCell, Suzhou, China) and 1% penicillin-streptomycin (15140122, Gibco, USA). HEK293T cells, sourced from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China), were cultured in Dulbecco’s Modified Eagle Medium (DMEM, 11965092, Gibco, USA) containing 10% fetal bovine serum (10099141, Gibco, USA) and 1% penicillin-streptomycin. All cell lines were authenticated by the suppliers and tested negative for mycoplasma contamination.

Method details

Animal husbandry and dietary interventions

Eight-week-old male and female Panx1^−/− mice were used for all primary experiments. An independent cohort of age-matched female WT and Panx1^−/− mice was studied under the same HFD and Gubra Amylin (GAN) diet protocols, and these data are presented in Figure S4. In addition, a small cohort of age-matched male WT and Panx1^−/− mice was maintained on a standard chow diet, and terminal body weight was recorded at 24 weeks. All animals were housed under temperature-controlled conditions (23 ± 2 °C) with a 12 h light/dark cycle.

For diet-induced obesity, mice were fed a conventional high-fat diet (HFD; 20% protein, 60% fat and 20% carbohydrate; #D12942, Research Diets, New Brunswick, USA) for 8 weeks. To impose a more severe obesogenic and metabolic challenge, a separate cohort of WT and Panx1^−/− mice was fed the GAN diet⁵¹ (40 kcal% fat without trans-fat, 22% fructose, 10% sucrose and 2% cholesterol; SPF Biotechnology, Beijing, China) for 20 weeks. Thus, the HFD was used as a classical model of diet-induced obesity and insulin resistance with relatively modest liver involvement over this time frame, whereas the GAN diet, which combines high fat, fructose and cholesterol, was used to induce more pronounced obesity, adipose tissue expansion and metabolic stress, and has also been described as a preclinical NASH diet model.

During the treatment period, body weight was monitored weekly, and fasting blood glucose was measured biweekly after a 4–6 h fast. Fat and lean mass were assessed by nuclear magnetic resonance (QMR06-090H, NIUMAG, Nanjing, China) before the endpoint. Blood and tissue samples were collected at sacrifice.

All animal procedures were approved by the Animal Ethics Committee of Capital Medical University (AEEI-2023-003) and the Animal Care and Use Committee of Beijing You’an Hospital and were conducted in accordance with relevant institutional guidelines and regulatory requirements.

Generation of stable cell lines

The Panx1 (NM_ 019482) sequence was cloned into the GV492 vector (Ubi-MCS-3FLAG-CBh-gcGFP-IRES-puromycin) for establishing lentiviral vectors with overexpression of Panx1 (Panx1-OE). The Panx1 knockout cell lines (Panx1^−/−) were generated by the CRISPR/Cas9 system. Single-guide RNA sequences targeting the genomic Panx1 sequence were designed using the CRISPR design tool and then inserted into the GV708 plasmid (U6-sgRNA-EF1a-Cas9-FLAG-CMV-EGFP-P2A-puro). These packed lentiviruses were produced from Genechem company (Shanghai, China). Empty vector (GV492 or GV708) lentivirus was used as a negative control. To establish stable Panx1 overexpression or knockout cell lines, 3T3-L1 preadipocytes cells were infected with the lentivirus at a multiplicity of infection of 20 and cultured in complete medium for 24 h, followed by selecting with 2 μg/mL puromycin for pooled lentiviral screens. The overexpression and knockout cell lines were verified by Western blotting assays. Meanwhile, the empty vector as a control for Panx1 overexpression. Because the Panx1 overexpression and knockout cell lines were generated using different plasmid constructs, the corresponding control groups were named differently to avoid confusion: the control for Panx1-overexpressing cells was designated as the empty-vector group (EV), whereas the control for Panx1^−/− cells was designated as the Control group. To investigate the effect of exogenous activation of the Wnt/β-catenin pathway on the active form of β-catenin protein in Panx1 overexpression and knockout cell lines, we treated the cells with 40 ng/mL Wnt3a (ab81484, abcam, UK) for 6 h. After protein extraction, western blot was performed to detect changes in the active form of β-catenin protein.

Histopathological analysis

Tissue samples such as the liver and WAT were fixed in 10% buffered-neutral formalin. Paraffin-embedded liver sections were subjected to hematoxylin and eosin (H&E) staining and Sirius Red staining using a commercial kit (G1472, Solarbio, Beijing, China), performed by Servicebio (Wuhan, China) according to standard histological procedures.

The glucose tolerance test and insulin tolerance test

Glucose tolerance test (GTT) and insulin tolerance test (ITT) were conducted at two time points: 7 days before study initiation (baseline) and 7 days prior to the experimental endpoint. For GTT, mice were fasted for 5 h and injected intraperitoneally with glucose (2 g/kg, G8150, Solarbio, China). For ITT, mice were fasted for 4 h and injected intraperitoneally with insulin (1 U/kg, P3376, Beyotime, China). Blood glucose levels were measured at 0, 15-, 30-, 60-, and 120-min post-injection.

Multiplex cytokine bead array assay

Cytokine and chemokine levels in serum, liver, and epididymal white adipose tissue (eWAT) tissues were quantified using the MILLIPLEX MAP Mouse Cytokine/Chemokine Magnetic Bead Panel (MCYTMAG-70 KPX32; Millipore) according to the manufacturer’s instructions. The targets measured included Eotaxin/CCL11, interleukin 1 alpha (IL-1α), interleukin 1 beta (IL-1β), interleukin 6 (IL-6), interleukin 10 (IL-10), interferon gamma-induced protein 10 (IP-10), tumor necrosis factor alpha (TNF-α), monocyte chemoattractant protein-1 (MCP-1), interferon-gamma (IFN-γ), macrophage inflammatory protein-1 alpha (MIP-1α), macrophage inflammatory protein (MIP-2), vascular endothelial growth factor (VEGF), and so on. Data was collected using xPonent software (Luminex), and statistical analysis as described below.

Generation of stable cell lines and treatment

The Panx1 (NM_ 019482) sequence was cloned into the GV492 vector (Ubi-MCS-3FLAG-CBh-gcGFP-IRES-puromycin) for establishing lentiviral vectors with overexpression of Panx1 (Panx1-OE). The Panx1 knockout cell lines (Panx1^−/−) were generated by the CRISPR/Cas9 system. Single-guide RNA sequences targeting the genomic Panx1 sequence were designed using the CRISPR design tool and then inserted into the GV708 plasmid (U6-sgRNA-EF1a-Cas9-FLAG-CMV-EGFP-P2A-puro). These packed lentiviruses were produced from Genechem company (Shanghai, China). Empty vector (GV492 or GV708) lentivirus was used as a negative control. To establish stable Panx1 overexpression or knockout cell lines, 3T3-L1 preadipocytes cells were infected with the lentivirus at a multiplicity of infection of 20 and cultured in complete medium for 24 h, followed by selecting with 2 μg/mL puromycin for pooled lentiviral screens. The overexpression and knockout cell lines were verified by Western blotting assays. Meanwhile, the empty vector as a control for Panx1 overexpression. Because the Panx1 overexpression and knockout cell lines were generated using different plasmid constructs, the corresponding control groups were named differently to avoid confusion: the control for Panx1-overexpressing cells was designated as the empty-vector group (EV), whereas the control for Panx1^−/− cells was designated as the Control group. To investigate the effect of exogenous activation of the Wnt/β-catenin pathway on the active form of β-catenin protein in Panx1 overexpression and knockout cell lines, we treated the cells with 40 ng/mL Wnt3a (ab81484, abcam, UK) for 6 h. After protein extraction, western blot was performed to detect changes in the active form of β-catenin protein.

3T3-L1 adipocyte differentiation and BODIPY staining

3T3-L1 preadipocytes (passage 10 or below) were seeded in six-well plates at a density of 4 × 10⁴ cells/well and cultured in OriCell Basal Medium until they reached approximately 70% confluence. To induce differentiation, the medium was replaced with 2 mL MDI induction medium per well, containing 0.5 mM methylisobutylxanthine (HY-12318, MedExpress, USA), 1 μM dexamethasone (HY-14648, MedExpress, USA), 1 μM rosiglitazone (HY-17386, MedExpress, USA), and 10 μg/mL insulin (P3376, Beyotime, China) (Day 0). On Day 2, the MDI medium was replaced with 2 mL of insulin medium (1 μM rosiglitazone and 10 μg/mL insulin) per well. On Day 4, the insulin medium was replaced with fresh OriCell Basal Medium. Fully differentiated adipocyte-like cells were typically obtained by Day 6–8.

Differentiation was monitored by BODIPY 493/503 (D3922, Invitrogen, USA) staining to assess lipid accumulation. Detailed, cells were fixed with 4% paraformaldehyde for 15 min, followed by three washes with PBS. Cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min. The cells were then stained with 1 μg/mL BODIPY 493/503 for 30 min at room temperature, protected from light, and washed three times with PBS. The cells were mounted with a DAPI-containing mounting medium (ProLong Gold Antifade Mountant with DAPI) and visualized using a fluorescence microscope (TCS SP8, Leica, Wetzlar, Germany) with appropriate excitation and emission wavelengths (495 nm and 510–540 nm for BODIPY).

Gene expression analysis

Total RNA was isolated from cultured cells and tissue samples by TRIzol reagent (15596018; Life Technologies, Carlsbad, CA) and then subjected to complementary DNA (cDNA) synthesis using HiScript III All-in-one RT SuperMix Perfect kit (R323-01; Vazyme, Nanjing, China) for qPCR. Quantitative real-time PCR (qRT-PCR) were performed by ChamQ SYBR qPCR Master Mix (Q311-03; Vazyme, Nanjing, China) and a QuantStudioTM Real-Time PCR system (Applied Biosystems ViiA 7 Real-Time PCR System). The targeting mRNA expression levels were normalized to peptidylprolyl isomerase A (PPIA). Fold change of gene expression level was calculated based on the threshold cycle method (2^−ΔΔCT).

RNA-sequencing

A total of 1 μg of RNA isolated from the eWAT tissue of mice was used for library preparation with the MGIEasy RNA Library Prep Kit (MGI Tech, Shenzhen, China) according to the manufacturer’s instructions. Gene expression analysis was performed by a BGISEQ-500 (MGI Tech Co., Ltd, Shenzhen, China) with a single end 50 bp module. The differential expression genes (DEGs) were identified using DEseq2.⁵² based on the read’s matrix, with a filter criterion of fold change (FC) > 1 and p < 0.05. The gene set predefined in the GO pathway was analyzed using Gene set enrichment analysis (GSEA), and the genes were ranked based on the strength of expression difference. A p values less than 0.05 and False Discovery Rate less than 0.25 were statistically significant.

Immunoblotting analysis

For western blot analysis, total protein was extracted from cultured cells and eWAT tissue using RIPA buffer (ab156034, Abcam, UK) supplemented with 1 mM PMSF and Protease/Phosphatase Inhibitor Cocktail (78440, Thermo Scientific, USA). Nuclear and cytoplasmic proteins were isolated with a Nuclear Extraction Kit (ab219177, Abcam, UK) according to the manufacturer’s instructions. Protein concentrations were determined with a BCA Protein Assay Kit (23227, Thermo Scientific, USA). Equal amounts of protein were separated by 10% SDS-PAGE and transferred to PVDF membranes (ISEQ00010, Millipore, USA), which were then blocked with 5% non-fat milk (P0216, Beyotime, China). Membranes were incubated overnight at 4°C with primary antibodies, followed by HRP-conjugated anti-rabbit or anti-mouse secondary antibodies (#7074 or #7076S, Cell Signaling Technology, USA). Signals were detected using a ChemiDoc MP Imaging System (Bio-Rad, USA) and quantified with ImageJ (v2.2.0). The western blot results shown are representative of at least three independent experiments. For all Western blot analyses, band intensities were quantified and normalized to the corresponding loading controls (β-actin or GAPDH for whole-cell and tissue lysates, and Lamin B1 for nuclear extracts).

Cell transfection and immunoprecipitation (IP) assays

IP assays were performed in two cell lines: 1) HEK293T cells and 2) a stable 3T3-L1 cell line with Panx1 overexpression and knockout. HEK293T cells were transfected with either pcDNA3.1-HA (control) or pcDNA3.1-Panx1-HA for 6 h, followed by replacing the medium with DMEM for 24 h. What’s more, Panx1-OE 3T3-L1 preadipocytes were transfected with β-Catenin small interfering RNAs (siRNAs) for 6 h, followed by replacing the medium with normal culture medium for 24 h. The two siRNAs targeting β-Catenin were synthesized by OBiO Technology (Shanghai, China) and siRNA-1 showed the highest knockdown efficiency. Transfection of vectors or siRNAs into melanoma cells was carried out using Lipofectamine 3000 (L3000001, Invitrogen, USA). Then add IP lysis buffer (87787, Thermo Scientific, USA) to the cells and incubate on ice for 30 min. After centrifuging at 4°C and 14000 rpm for 30 min, transfer the supernatant to a new microcentrifuge tube. Then, the lysate was incubated with Anti-FLAG M2 Magnetic Beads (A2220, Sigma, USA) or Anti-HA Affinity Gel (H6908, Sigma, USA) overnight at 4°C on a shaker. The beads were washed by 5 times with cold IP buffer and denatured in 2×loading buffer at 95°C for 3–5 min. Finally, western blotting analysis was performed to detect the interacting proteins.

Immunofluorescence staining assays

Cells and liver sections were fixed with 4% paraformaldehyde for 30 min, permeabilized with 1% BSA in PBS containing 0.4% Triton X-100 for 10 min, and blocked with 10% BSA in PBST at room temperature (RT) for 60 min. Cells were incubated with anti-β-catenin antibody and sections with F4/80 antibody (1:100 in PBS with 1% BSA and 0.4% Triton X-100) at 4°C overnight. After washing with PBST, Alexa Fluor 555-conjugated anti-mouse IgG (1:200, Ab150074, Abcam, UK) was applied at RT for 2 h. Finally, samples were counterstained with DAPI using anti-fade mounting medium for 10 min at RT and mounted with coverslips. Images were acquired using a fluorescence microscope (TCS SP8, Leica, Wetzlar, Germany) and analyzed with ImageJ (V2.2.0).

Cell counting kit-8 (CCK-8) assay

For the proliferation assay of Panx1 overexpression and knockout cells, 2,000 cells were seeded in a 96-well plate per well. CCK-8 assay kit (M4839, AbMole BioScience, USA) was used to determine cell proliferation as instructed by the manufacturer. The absorbance at 450 nm was measured using microplate reader (Bio-Tek, Winooski, VT).

Cell cycle analysis

3T3-L1 preadipocytes cells (2 × 10⁵) were distributed onto a 6-well plate and incubated for 48 h. Fixed in 75% ethanol (in PBS) on ice for 30 min and then resuspended in PBS containing 40 μg/mL propidium iodide (ST511, Beyond, China) and 100 μg/mL RNase A (EN053, ThermoFisher, USA). After incubating for 15 min at 37 °C, the percentage of G0/G1, S, and G2 phases of 3T3-L1 preadipocytes cells were assessed by the flow cytometer (FASCS Calibur; Becton-Dickinson, San Jose, CA, USA) with an argon-ion laser at 488 nm. For all cell-cycle experiments, Panx1-OE cells and their empty-vector (EV) controls, as well as Panx1^−/− cells and their matched Control cells, were analyzed as independent pairs of stable lines, and statistical comparisons were performed only within each matched pair.

AlphaFold 3D modeling

The complete protein sequences of Panx1, GSK-3β and β-Catenin are collected from UniProt database, and three protein-protein sequence pairs are then curated, i.e., Panx1 with GSK-3β, Panx1 with β-Catenin, and GSK-3β with β-Catenin. We directly submitted individual sequence pairs to the AF2-Multimer v2.3.0 server. For each pair, five prediction results were downloaded, and the top one was chosen for down streaming analysis. Each predicted complex was first visualized using Pymol to manually check the binding mode and binding sites, then all non-covalent interactions were detected by PLIP, a rule-based tool to characterize non-covalent interactions involved in the complex structure.

According to a recent benchmark report,⁵³ the predicted metrics of AlphaFold3, including the ranking score, iptm, and ptm, can be used to approximately evaluate the change of binding strengths between protein-protein interaction (PPI). Therefore, we employed the prediction results of AlphaFold3 to evaluate the change of binding energy of three PPI pairs. Protein sequences of each pair were directly submitted to the AlphaFold3 server with default parameters and random seeds. For each PPI, top prediction from all five prediction results were chosen to extract corresponding ranking score. Consequently, the ranking scores of all three PPIs were compared to identifying relative binding strengths.

Quantification and statistical analysis

Data are presented as mean ± SD. Statistical analyses were performed using GraphPad Prism version 8.3.0 (GraphPad Software, USA) and R software (version 4.2.0). For data with normal distribution and homogeneity of variance, comparisons between two groups were performed using a two-tailed unpaired Student’s t test, and comparisons among multiple groups were analyzed by one-way ANOVA followed by Tukey’s post hoc test. For experiments involving two independent variables, two-way ANOVA was applied, followed by Sidak’s multiple comparisons test where appropriate. For datasets with non-Gaussian distribution, nonparametric analysis was performed using the Mann–Whitney U test. For targeted multi-gene comparisons, multiple unpaired tests were conducted, and p values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure. An adjusted q value <0.05 was considered statistically significant. Statistical significance was defined as ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Published: February 20, 2026