Pannexin 1 is required for full activation of insulin-stimulated glucose uptake in adipocytes

Abstract

Objective: Defective glucose uptake in adipocytes leads to impaired metabolic homeostasis and insulin resistance, hallmarks of type 2 diabetes. Extracellular ATP-derived nucleotides and nucleosides are important regulators of adipocyte function, but the pathway for controlled ATP release from adipocytes is unknown. Here, we investigated whether Pannexin 1 (Panx1) channels control ATP release from adipocytes and contribute to metabolic homeostasis.

Methods: We assessed Panx1 functionality in cultured 3T3-L1 adipocytes and in adipocytes isolated from murine white adipose tissue by measuring ATP release in response to known activators of Panx1 channels. Glucose uptake in cultured 3T3-L1 adipocytes was measured in the presence of Panx1 pharmacologic inhibitors and in adipocytes isolated from white adipose tissue from wildtype (WT) or adipocyte-specific Panx1 knockout (AdipPanx1 KO) mice generated in our laboratory. We performed in vivo glucose uptake studies in chow fed WT and AdipPanx1 KO mice and assessed insulin resistance in WT and AdipPanx1 KO mice fed a high fat diet for 12 weeks. Panx1 channel function was assessed in response to insulin by performing electrophysiologic recordings in a heterologous expression system. Finally, we measured Panx1 mRNA in human visceral adipose tissue samples by qRT-PCR and compared expression levels with glucose levels and HOMA-IR measurements in patients.

Results: Our data show that adipocytes express functional Pannexin 1 (Panx1) channels that can be activated to release ATP. Pharmacologic inhibition or selective genetic deletion of Panx1 from adipocytes decreased insulin-induced glucose uptake in vitro and in vivo and exacerbated diet-induced insulin resistance in mice. Further, we identify insulin as a novel activator of Panx1 channels. In obese humans Panx1 expression in adipose tissue is increased and correlates with the degree of insulin resistance.

Conclusions: We show that Panx1 channel activity regulates insulin-stimulated glucose uptake in adipocytes and thus contributes to control of metabolic homeostasis.

Keywords: Adipocyte; Extracellular ATP; Glucose uptake; Pannexin 1.

Graphical abstract

Figure 1

Pannexin 1 channel function in adipocytes is regulated by alpha-adrenergic stimulation or via caspase-mediated C-terminal cleavage during apoptosis. (A) 3T3-L1 adipocytes were treated with indicated concentrations of phenylephrine (PE) for 30 min and then stained with 1 μM YO-PRO^® and 1 μg/mL Hoechst for 10 min. Experiment was performed in triplicate. Total cells were quantitated by counting Hoechst-positive cells. Cells positive for YO-PRO^® indicate cells in which Panx1 channels have been activated and opened, allowing dye to enter. Data are expressed as mean ± s.e.m. *p = 0.0087 by Student's t-test. (B) 3T3-L1 adipocytes were treated with 5  μM PE with or without pretreatment with a Panx1 intracellular loop peptide (IL2) or a C-terminal peptide (CT2). The IL2 peptide corresponds to a region of the Panx1 intracellular loop (aa 191–200) while the CT2 peptide is from a region of the C-terminal tail corresponding to aa 381–390 (inset) . ATP release into the media was measured using cell-titer glo assay (Promega). Experiment was performed in triplicate. Data are expressed as mean ± s.e.m. *p < 0.05 by Student's t-test. (C) Adipocytes were isolated from perigonadal adipose tissue of WT or Panx1 KO mice, and ATP release was measured upon stimulation with phenylephrine (PE, 5 μM, 15 min) or no treatment (NT). Experiment was performed in triplicate. Data are expressed as mean ± s.e.m. *p < 0.04 by 2-way ANOVA with Sidak's multiple comparison test. (D) Whole cell patch clamping of 3T3-L1 adipocytes reveals a carbenoxolone-sensitive current when active caspase-3 is present in the pipette indicating that adipocyte Panx1 is activated by the caspase-cleavage mechanism. Arrows indicate time at which active caspase 3 (CASP3) or carbenoxolone (CBX) was added. Control dialysis shown in black. Current–voltage relationship (IV) curves are shown at right. (E) 3T3-L1 adipocytes were exposed to 400 mJ/cm³ UV irradiation (Stratalinker) to induce apoptosis and incubated with or without 400 μM carbenoxolone (CBX) or zVAD, a pan-caspase inhibitor (50 μM) for 1 h. ATP was measured in the supernatant by cell-titer glo assay (Promega). n = 8. Data are expressed as mean ± s.e.m. *p < 0.0001 by ANOVA with Tukey's multiple comparison test.

Figure 2

Full activation of insulin-stimulated glucose uptake in adipocytes requires ATP release by Panx1 channels. (A) Blockade of Pannexin-1 channels with carbenoxolone (100 μM, CBX) or probenecid (1 mM, Prob) significantly decreases insulin-stimulated ³H-glucose uptake in 3T3L1-adipocytes. Data are expressed as mean ± s.e.m. *p < 0.001 by 2 way ANOVA with Tukey's multiple comparisons test. (B) Insulin-stimulated glucose uptake is significantly decreased in adipocytes isolated from perigonadal adipose tissue of adipocyte-specific Pannexin-1 null mice. Addition of exogenous ATP (50 μM) restores insulin-stimulated ¹⁴C-glucose uptake in adipocytes isolated from Panx1 null mice. Data are expressed as mean ± s.e.m. *p < 0.003 by 1 way ANOVA with Tukey's multiple comparisons test. (C) In vivo [³H] 2-deoxy-d-glucose uptake was assessed in perigonadal white adipose tissue (WAT) and gastrocnemius muscle (MUS) in age-matched, male, chow fed WT and AdipPanxKO littermates (n = 6). Data are presented as mean ± s.e.m. *p < 0.05 by paired t-test. Dotted line indicates the 20% decrease in glucose uptake in WAT in AdipPanxKO mice compared to WT.

Figure 3

Insulin induces Panx1 channel activation and ATP release. (A) Adipocytes isolated from perigonadal adipose tissue of WT mice release ATP upon insulin stimulation, which can be blocked by treatment with the Panx1 inhibitor probenecid. Data are expressed as mean ± s.e.m. *p < 0.05 by Student's t-test. (B) Whole cell patch clamp of HEK cells transfected with human insulin receptor and mouse Panx1 reveals a Panx1 dependent current upon treatment with insulin that is abolished by addition of the Panx1 inhibitor carbenoxolone. Current–voltage relationship curve is shown in middle. Insulin treatment significantly increases CBX-sensitive current density (n = 8 cells). (C) Whole cell patch clamp of HEK cells transfected with human insulin receptor and human Panx1 in which the C-terminal caspase cleavage site has been replaced with a TEV protease cleavage site reveals a Panx1 dependent and CBX-sensitive current upon treatment with insulin. Current–voltage relationship curve is shown in middle. Insulin treatment significantly increases CBX-sensitive current density (n = 8 cells).

Figure 4

Weight gain, energy expenditure, serum free fatty acids and insulin are not different between high fat fed WT and AdipPanx1KO mice. (A) Intraperitoneal glucose tolerance test was performed on age-matched, male, WT and AdipPanxKO littermates that had been fed chow or high fat diet (60% fat) for 12 weeks. Mice were injected i.p. with 1 g/kg glucose, and blood glucose was measured in tail vein blood via glucometer (One Touch Ultra). Data are presented as mean ± s.e.m and representative of 3 independent experiments (WT HFD n = 6, AdipPanxKO HFD n = 9, WT chow n = 7, AdipPanxKO chow n = 6). Combined area under the curve (AUC) analysis of glucose tolerance tests reveals that AdipPanxKO mice are significantly more glucose intolerant after high fat feeding compared to WT mice (WT HFD n = 18, AdipPanxKO HFD n = 14, WT chow n = 7, AdipPanxKO chow n = 4); *p = 0.025 by 2-tailed Student's t-test. Box plots represent the 10th to 90th percentile. Three mice in the high fat diet group (WT n = 1, AdipPanxKO n = 2) did not respond to diet as evidenced by AUC not being different from chow groups and thus were excluded from the analysis. One data point (WT n = 1) was greater than 2 standard deviations from the mean and thus was excluded. (B) Intraperitoneal insulin tolerance tests (0.75 U/kg) was performed on age-matched, male, WT and AdipPanxKO littermates that had been fed chow or high fat diet (60% fat) for 12 weeks. n = 6–7 mice per group. Data are expressed as mean + s.e.m. Box plots represent the 10th to 90th percentile. *p < 0.05 by Student's t-test. (C,D) Male WT and AdipPanx1KO littermates were fed a high fat diet (60% fat) for 12 weeks. Mice on a high fat diet were weighed weekly. Adiposity (%fat) and lean mass (%lean) were assessed by echo MRI. Data are representative of 3 independent experiments, n = 6 WT, n = 9 AdipPanx1KO. Data are expressed as mean ± s.e.m. (E) Serum free fatty acids (FFA) were measured after 12 weeks of high fat feeding in WT (n = 5) and AdipPanxKO (n = 9) mice using the colorimetric FFA kit from Wako. Data are expressed as mean ± s.d. (F) Serum insulin in WT and AdipPanxKO mice that were either fed chow diet or HFD for 12 weeks were measured after a 5 h fast. (n = 6 WT chow, n = 7 AdipPanxKO chow, n = 5 WT high fat diet, n = 9 AdipPanxKO high fat diet). Data are expressed as mean ± s.d. (G–I) High fat diet-fed (12 weeks) and chow-fed WT and AdipPanx1KO mice (n = 4 per group) were placed in metabolic cages for 72 h. Average VO₂ consumption, VCO₂, and respiratory exchange ratio (RER) by animal are shown for light and dark periods. The initial 4 h of readings were not part of average light values as this was time for mice to acclimate. Box plots represent the 10th to 90th percentile. (J) Locomotion was recorded as X- and Y-axis beam breaks during light and dark cycle. Total beam breaks were not different between WT and AdipPanx1KO mice indicating no overall difference in locomotion. Box plots represent the 10th to 90th percentile.

Figure 5

Pannexin 1 expression in human adipose tissue is associated with obesity and insulin resistance. (A) Data from NCBI gene array (GDS1498[ACCN]) were analyzed for Panx1 expression in subcutaneous adipose tissue from lean male (n = 10), lean female (n = 10), and obese male (n = 9) and obese female (n = 10) human subjects. Data are normalized to lean samples, each point represents one human subject, error bars indicate s.d. *p < 0.0001 by 2 way ANOVA with Sidak's multiple comparison test. (B) Omental fat samples were obtained from human subjects prior to bariatric surgery and analyzed for Panx1 mRNA levels normalized to 18S mRNA. The average of all samples was set to 1 and log 10 values of Panx1 mRNA were plotted against the blood glucose levels of patients at time of surgery, revealing a positive and significant correlation (r² = 0.181 and p = 0.04 by linear regression). The correlation of Panx1 levels with blood glucose levels was particularly pronounced in Caucasian females (r² = 0.56, n = 11, black line). Each point represents one human subject, female subjects are shown as gray (non-Caucasian) and black (Caucasian) squares, male subjects as white squares. (C) Samples from B were grouped into tertiles based on low (<0.5 fold, n = 8), medium (0.5–1.5 fold, n = 10), and high Panx1 expression (>1.5 fold, n = 5) and plotted against HOMA-IR, a clinical measure of insulin resistance. *p < 0.02 by Student's t-test. Each point represents one human subject; error bars represent s.e.m.