Liver-derived Neuregulin1α stimulates compensatory pancreatic β cell hyperplasia in insulin resistance

Abstract

Compensatory pancreatic islet hyperplasia is an adaptive response to increased systemic insulin demand, although factors meditating this response remain poorly understood. Here, we show that a liver-derived secreted protein, Neuregulin1α, promotes compensatory proliferation of pancreatic β cells in type 2 diabetes. Liver Neuregulin1α expression and serum Neuregulin1α levels increase in male mice fed an obesity-inducing diet. Male mice lacking either Neuregulin1 in liver or its receptor, ErbB3, in β cells deteriorate systemic glucose disposal due to impaired β cell expansion with reduced insulin secretion when fed the obesity-inducing diet. Mechanistically, Neuregulin1α activates ERBB2/3-ERK signaling to stimulate β cell proliferation without altering glucose-stimulated insulin secretion potential. In patients with metabolic dysfunction-associated steatotic liver disease (MASLD) and obesity but without type 2 diabetes serum Neuregulin1α levels increase, while in patient with MASLD and type 2 diabetes show markedly reduced levels of Neuregulin1α. These results suggest that Neuregulin1α serves as a hepatokine that can expand functional β cell mass in type 2 diabetes.

Fig. 1. Hepatic Neuregulin1 expression correlates with pancreatic islet size in obese diabetic mice.

a Identification of Neuregulin1 by DNA microarray analysis. Genes encoding putative hepatic factors that either increased > 4-fold (118 genes) or decreased by at least 75% (22 genes) after 15 weeks’ exposure to a high-fat/high-sucrose (HFHS) diet. Overall, these genes showed little change (upregulation < 2-fold or downregulation < one-half) after 10 weeks of treatment compared to levels seen in mice before feeding. Expression levels of genes altered in mice fed an HFHS for 15 weeks were calculated as a log2 ratio with reference to levels seen mice prior to manipulation of feeding and plotted in order of increasing change in expression. The bolded black bar at left corresponds to Neuregulin1. b, c Relative quantification of hepatic type I Neuregulin1 mRNA (b) and plasma Neuregulin1α (NRG1α) (c) levels in mice fed an HFHS for the indicated times. Data represent mean ± SEM; n = 8 (0 W), 6 (5 W), 7 (10 W) and 6 (15 W) mice per group in (b), n = 8 (0 W), 5 (5 W), 9 (10 W) and 6 (15 W) mice per group in (c). d–f Correlation of hepatic type I Neuregulin1 (Nrg1) mRNA levels (d) or pancreatic islet size (e) with plasma NRG1α levels and pancreatic islet size (f) with type I Nrg1 mRNA levels of mice fed an HFHS for 0–15 weeks. Pearson or Spearman’s correlation analysis. n = 8 (0 W), 5 (5 W), 8 (10 W) and 4 (15 W) mice per group in (d), n = 8 (0 W), 5 (5 W), 9 (10 W) and 4 (15 W) mice per group in (e), n = 8 (0 W), 5 (5 W), 8 (10 W) and 2 (15 W) mice per group in f A One-way ANOVA with post hoc Dunnett’s multiple comparison test (b) and a Kruskal-Wallis test with post hoc Dunnett’s multiple comparison test (c) was performed, and two-tailed Spearman’s correlation coefficient (d, e) and the two-tailed Pearson correlation coefficient (f) were calculated.

Fig. 2. Loss of hepatic Neuregulin1 impairs compensatory pancreatic islet hyperplasia in obese diabetic mice.

a, b Blood glucose (a) and plasma insulin (b) levels during an oral glucose tolerance test in liver-specific Neuregulin1 knockout (LNrg1 KO) and wild-type (WT) mice exposed 15 weeks to a high-fat/high-sucrose (HFHS) diet. Data represent mean ± SEM; n = 12 (blood glucose) and 11 (plasma insulin) mice per group. c Insulin tolerance test of LNrg1 KO and WT mice fed an HFHS for 15 weeks. Data represent mean ± SEM; n = 6 mice per group. No statistically significant differences were found between the two groups at any time point. d Immunoblotting of total and phosphorylated forms of AKT in indicated tissues of LNrg1 KO and WT mice fed an HFHS for 15 weeks. Tissues were harvested 15 min after insulin (0.75 IU/kg body weight) administration. This experiment was repeated three times independently using 6 mice of each group. e, f Representative H&E (e) and Ki-67 (f) staining of the pancreas of LNrg1 KO and WT mice fed either normal chow (NC) or an HFHS for 15 weeks. Islets are enclosed by dashed lines. Black arrowheads in (f) indicate Ki-67-positive cells. Scale bar, 50 μm. Quantification of islet area (r) and Ki-67-positive cell number (f). Data represent mean ± SEM; n = 5 (NC) and 9 (HFHS) mice per group, respectively. g Plasma Neuregulin1α (NRG1α) levels in LNrg1 KO and WT mice fed either an NC or an HFHS for 15 weeks. Data represent mean ± SEM; n = 6 (NC) and 7 (HFHS) mice per group. Unpaired Student’s t tests (a, b, c, e, and f) and a two-tailed Mann-Whitney U test (g) were performed.

Fig. 3. Administration of rNRG1α enhances systemic glucose disposal with increased insulin secretion.

a, b, e, f Oral glucose tolerance test (a, e) and insulin secretion (b, f) after glucose challenge in normal (a, b) and ob/ob (e, f) mice treated for 4 weeks with recombinant protein encompassing the Neuregulin1 α type EGF-like domain (rNRG1α). Data represent mean ± SEM; n = 10 (PBS) and 9 (rNRG1α) mice per group (a). n = 8 mice per group (b). n = 8 (PBS) and 9 (rNRG1α) mice per group (e). n = 10 (PBS) and 11 (rNRG1α) mice per group (f). c, d, g, h Representative H&E (c, g) and Ki-67 (d, h) staining of the pancreas of normal (c, d) and ob/ob (g, h) mice treated for 4 weeks with rNRG1α. Islets are enclosed by dashed lines. Black arrowheads in (d) and (h) indicate Ki-67-positive cells. Scale bar, 50 μm. (right) Quantification of islet area (c, g) and number of Ki-67-positive cells (d, h). Data represent mean ± SEM; n = 6 mice per group (c, d). n = 8 (PBS) and 9 (rNRG1α) mice per group (g, h). Unpaired Student’s t tests (a–h) were performed.

Fig. 4. β cell-specific ErbB3 deletion impairs systemic glucose disposal and insulin secretion in obese diabetic mice.

a, b Immunoblots (a) and corresponding quantification (b) of total and phosphorylated forms of ERBBs, AKT, and ERK in the pancreas of normal mice injected with either PBS or recombinant protein encompassing the Neuregulin1 α type EGF-like domain (rNRG1α). TUBULIN served as a loading control. The average value of the ratio of phosphorylated to non-phosphorylated proteins in the samples of the PBS-treated group, normalized by the expression level of TUBULIN, was set to 1. Data represent mean ± SEM; n = 5 mice per group (b). The samples derive from the same experiment, and blots were processed in parallel. This experiment was repeated three times independently using5 mice of each group. c, d Immunoblots (c) and corresponding quantification (d) of total and phosphorylated forms of ERBB3 and ERK in the pancreas of normal mice fed a high-fat/high-sucrose (HFHS) diet for 15 weeks. β-ACTIN served as a loading control. The average value of the ratio of phosphorylated to non-phosphorylated proteins in the samples of normal mice fed a normal chow (NC), normalized by the expression level of β-ACTIN, was set to 1. Data represent mean ± SEM; n = 5 mice per group (d). e, f Blood glucose (e) and plasma insulin (f) levels after oral glucose challenge in β cell-specific ErbB3 KO (βErbB3KO) and WT mice fed an HFHS for 15 weeks. Data represent mean ± SEM; n = 12 (e) and 11 (f) mice per group. g, h. Representative H&E (g) and Ki-67 (h) staining of the pancreas of HFHS-treated WT and βErbB3KO mice. Islets were enclosed by the dashed line. Black arrowheads in (h) indicate Ki-67-positive cells. Scale bar, 50 μm. Quantification of islet area (g) and Ki-67-positive cell number (h). Data represent mean ± SEM; n = 11 (g) and 8 (h) mice per group. Unpaired Student’s t tests (b, d, e–h) were performed.

Fig. 5. Neuregulin1 promotes β cell proliferation in an ERBB2/3-ERK dependent manner.

a, b Immunoblots (a) and corresponding quantification (b) of total and phosphorylated forms of ERBBs, AKT, and ERK in MIN6 cells treated with either PBS or indicated concentrations of recombinant protein encompassing the Neuregulin1 α type EGF-like domain (rNRG1α) for 30 min. Tubulin served as a loading control. The average value of the ratio of phosphorylated to non-phosphorylated protein in samples from the group treated without rNRG1, normalized by the expression level of TUBULIN, was set to 1. Data represent mean ± SEM; n = 5 (pERBB1/ERBB1, pERBB4/ERBB4, pAKT/AKT, and pERK/ERK) and 6 (pERBB2/ERBB2 and pERBB3/ERBB3) per group. The samples derive from the same experiment, and blots were processed in parallel. This experiment was repeated at least five times independently. c Percentage of Ki-67-positive cells in isolated mouse islets treated with control PBS or rNRG1α in the absence or presence of indicated inhibitors. Inhibitors were added to the medium 30 min before rNRG1α stimulation (10 nM, 24 h). LPT; lapatinib, TX1; TX1-85-1, GEFI; gefitinib, LY; LY294002, SCH; SCH772984. Data represent mean ± SEM; n = 18 (PBS), 19 (rNRG1α), and 5 (rNRG1α plus each inhibitor) islets per group. d Glucose-stimulated insulin secretion. Islets were stimulated for 1 h with either low (2.5 mM) or high (25 mM) glucose in the absence or presence of rNRG1α (10 nM). Data represent mean ± SEM; n = 4 islets per group. No statistically significant differences were found between the two groups. An unpaired Student’s t test (b, d) and a one-way ANOVA (c) and were performed.

Fig. 6. Serum Neuregulin1α levels increase in MASLD human subjects with obesity.

a Immunoblotting of Neuregulin1α in culture medium from isolated hepatocytes. Lanes 1&2, 16, and 8 μl of 100 x concentrated culture medium from mouse hepatocytes (3.5 × 10⁵ cells/mL); Lanes 3&4, 16, and 8 μl of 100 x concentrated culture medium from human hepatocytes (7.6 × 10⁵ cells/mL) isolated from PXB mice. This experiment was repeated twice independently. b, c Correlation of levels of hepatic type I Neuregulin1 (Nrg1) mRNA with serum Neuregulin1α protein (NRG1α) (b), and hepatic Neuregulin1α mRNA with serum NRG1α (c) from metabolic dysfunction associated steatotic liver disease (MASLD) patients. χ² tests were used for statistical analysis. n = 25. d, e Levels of serum Neuregulin1α from MASLD patients. d: comparison of serum NRG1α levels among MASLD patients. The MASLD group was divided into three groups depending on obesity. e: comparison of serum Neuregulin1α levels in MASLD patients with obesity and diabetes. n = 36 (BMI < 25), 36 (25 ≤ BMI < 30), 42 (30 ≤ BMI), and 38 (diabetes) individuals with MASLD per group. Single regression analysis (two-tailed Spearman’s correlation coefficient) (b, c), a Kruskal-Wallis test (d), and a two-tailed Mann-Whitney U test (e) were performed.