Berberine is an insulin secretagogue targeting the KCNH6 potassium channel

Abstract

Coptis chinensis is an ancient Chinese herb treating diabetes in China for thousands of years. However, its underlying mechanism remains poorly understood. Here, we report the effects of its main active component, berberine (BBR), on stimulating insulin secretion. In mice with hyperglycemia induced by a high-fat diet, BBR significantly increases insulin secretion and reduced blood glucose levels. However, in mice with hyperglycemia induced by global or pancreatic islet β-cell-specific Kcnh6 knockout, BBR does not exert beneficial effects. BBR directly binds KCNH6 potassium channels, significantly accelerates channel closure, and subsequently reduces KCNH6 currents. Consequently, blocking KCNH6 currents prolongs high glucose-dependent cell membrane depolarization and increases insulin secretion. Finally, to assess the effect of BBR on insulin secretion in humans, a randomized, double-blind, placebo-controlled, two-period crossover, single-dose, phase 1 clinical trial (NCT03972215) including 15 healthy men receiving a 160-min hyperglycemic clamp experiment is performed. The pre-specified primary outcomes are assessment of the differences of serum insulin and C-peptide levels between BBR and placebo treatment groups during the hyperglycemic clamp study. BBR significantly promotes insulin secretion under hyperglycemic state comparing with placebo treatment, while does not affect basal insulin secretion in humans. All subjects tolerate BBR well, and we observe no side effects in the 14-day follow up period. In this study, we identify BBR as a glucose-dependent insulin secretagogue for treating diabetes without causing hypoglycemia that targets KCNH6 channels.

Fig. 1. BBR increases high-glucose-dependent insulin secretion from pancreatic islets.

a Berberine (BBR) is a plant alkaloid isolated from the Coptidis rhizoma, the dried rhizome of Coptis chinensis. b Viability of the indicated cells incubated with different concentrations of BBR for 30 min. n = 5. c Mouse islets (15 islets per batch) were incubated with low (2.8 mM) or high (25 mM) glucose concentrations in the absence or presence of the indicated concentrations of BBR. The secreted insulin level in the supernatant was normalized to the corresponding insulin content of the islets. n = 5. P = 0.0417 for 5 μM BBR, P = 0.0297 for 25 μM BBR under 25 mM glucose concentration. Statistical significance was assessed using one-way ANOVAs coupled with Dunnett’s post-hoc test (two-sided). d In the absence (control) or presence of 5 μM BBR, ~50 islets from male C57BL/6 J mice were perfused with a low (2.8 mM) glucose concentration for 10 min followed by a high (16.7 mM) glucose concentration for 30 min. The amount of insulin secreted was normalized to the total insulin content. First phase (0–5 min), second phase and total insulin secretion were calculated as the area under the curve (AUC). n = 4. P = 0.042 for first phase, P = 0.036 for total insulin secretion. Statistical significance was assessed using the Mann–Whitney U-test (two-sided). e Approximately 50 islets were perfused with a low (2.8 mM) glucose concentration for 10 min and then with a high (16.7 mM) glucose concentration for 120 min. Afterwards, during the continuous perfusion of 16.7 mM glucose, the islets were treated with 5 μM BBR or vehicle (control) for 20 min followed by 30 mM KCl, which served as the internal control for islet viability, for 30 min. The amount of insulin secreted was normalized to the total insulin content. The dotted box represents insulin secretion from islets stimulated with 5 μM BBR or vehicle under high (16.7 mM)-glucose conditions. n = 4. P = 0.029. Statistical significance was assessed using the Mann–Whitney U-test (two-sided). f, g Mice islets were treated with or without 5 μM BBR or BBR plus 1 μM glibenclamide (f) /100 μM verapamil (g) in the presence of 25 mM glucose for 30 min. Insulin secretion was normalized to that of the control group. n = 6. P = 0.042 for glucose, P = 0.015 for glucose plus glibenclamide in f. P = 0.025 for glucose, P = 0.171 for glucose plus verapamil in g Statistical significance was assessed using the Mann–Whitney U-test (two-sided). The values are presented as means ± s.e.m. *P < 0.05.

Fig. 2. BBR increases insulin secretion in high-fat diet (HFD)-fed hyperglycemic mice but not in Kcnh6 global knockout (KO) or Kcnh6 islet β-cell-specific knockout (βKO) hyperglycemic mice.

a–e HFD-fed mice and Kcnh6 KO mice were orally administered 560 mg/kg BBR or vehicle before being loaded with 1 g/kg glucose for the intraperitoneal glucose tolerance test (IPGTT) and insulin release test (IRT). a Blood glucose (control, n = 8; BBR, n = 7, P = 0.009) and b plasma insulin (control, n = 5; BBR, n = 5, P = 0.016) levels in HFD-fed mice. c Kcnh6 KO mice were generated using the TALEN technique. Two TALENs specifically binding the target sequences in exon 5 of the Kcnh6 gene resulted in the knockout of the gene. Western blot analysis of the KCNH6 protein from WT and KO mouse islets. A representative immunoblot from 3 different experiments is shown. d Blood glucose (control, n = 10; BBR, n = 7) and e plasma insulin (control, n = 9; BBR, n = 8) levels in Kcnh6 KO mice. f–j Chow diet-fed control and Kcnh6 βKO mice were orally administered 560 mg/kg BBR or vehicle before being loaded with 3 g/kg glucose for the IPGTT and IRT. f Blood glucose (control, n = 6; BBR, n = 4, P = 0.038) and g plasma insulin (control, n = 10; BBR, n = 8, P = 0.012) levels in chow diet-fed control mice. h Kcnh6 βKO mice were generated using the CRISPR-mediated Cre-LoxP recombinase system. The endogenous Kcnh6 locus was targeted for the conditional excision of exons 3 and 4. Western blot analysis of the KCNH6 protein from control and βKO mouse islets. A representative immunoblot from three different experiments is shown. i Blood glucose (control, n = 4; BBR, n = 4) and j plasma insulin (control, n = 4; BBR, n = 4) levels in Kcnh6 βKO mice. The values are presented as means ± s.e.m. *P < 0.05 and **P < 0.01. Statistical significance was assessed using the Mann–Whitney U-test (two-sided).

Fig. 3. BBR targets KCNH6 channel proteins on islet β-cells.

a The active derivative of BBR, berberrubine, was immobilized on high-performance affinity beads. b Schematic of the experimental design. BBR-binding proteins were purified from cell extracts using different beads, as indicated. (I) Naked beads were used as a control; (II) BBR-immobilized beads were used to pull down BBR-binding proteins in the eluate; (III) 1 mM BBR was added to the extracts before the incubation with BBR-immobilized beads to reduce the yield of specific binding proteins; (IV) the same amount of vehicle (DMSO) was added to cell extracts before the incubation with the beads as a control for (III). Bound proteins were eluted and subjected to western blot analyses. c BBR-binding proteins from INS-1 β-cell extracts were analyzed with western blotting using an anti-KCNH6 antibody. A representative immunoblot from three different experiments is shown. d KCNH6-null human embryonic kidney 293 T (HEK293T) cells were transfected with plasmids encoding FLAG-tagged human KCNH6. BBR-binding proteins from transfected HEK293T cell extracts were analyzed with western blotting using an anti-FLAG antibody. A representative immunoblot from three different experiments is shown.

Fig. 4. BBR inhibits KCNH6 currents by accelerating channel inactivation.

a KCNH6 currents were recorded from transfected HEK293T cells with majority of KCNH6 K_v channels and very small amount of endogenous K_v channels using a specific stimulus protocol. A representative immunoblot from three different experiments is shown. b–f Activation of the KCNH6 channel. b KCNH6 K_v currents in the untransfected cells and KCNH6-null HEK293T cells transfected with the human KCNH6 gene treated with 10 μM BBR. c I–V curve for step currents measured in the steady state. P = 0.013 for −20 mV, P = 0.001 for −10 mV, P < 0.0001 from 0 to 60 mV. d I–V curve for tail currents, as measured from the peak currents. P = 0.005 for −10 mV, P = 0.0002 for 0 mV, P < 0.0001 from 10 to 60 mV. e Comparison of the steady-state activation curves. The lines represent the best fit to the data according to a Boltzmann curve. f Time constants of activation (n = 8). g–j Inactivation of the KCNH6 channel. g Representative inactivation currents. h Comparison of steady-state inactivation curves. The lines represent the best fit to the data according to a Boltzmann curve (n = 8). i Time constants of inactivation (n = 8). P = 0.026 for −120 mV, P = 0.037 for −100 mV, P = 0.016 for −80 mV. j Representative traces of inactivation tail currents triggered by a −120 mV stimulus. The right panel shows the normalized inactivation traces. The time constants of inactivation are 14.92 ms and 9.19 ms for control and BBR-treated cells, respectively. The values are presented as means ± s.e.m. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Statistical significance was assessed using the Mann–Whitney U-test (two-sided).

Fig. 5. BBR inhibits voltage-dependent potassium (K_v) currents, prolongs action potential durations (APDs), and promotes insulin exocytosis in mouse pancreatic islet β-cells.

a–e Total K_v currents in WT and Kcnh6 KO mouse pancreatic islet β-cells treated with 10 μM BBR or vehicle plus washout. a The total K_v currents of pancreatic islet β-cells expressing endogenous KCNH6 K_v channels and other endogenous K_v channels were measured in voltage-clamp mode. b Representative K_v currents were recorded from the indicated WT and KO β-cells. c–e Summary of the steady-state current-voltage (I–V) curves for K_v currents (upper panels) and the mean K_v current densities at +70 mV (lower panels) (WT, n = 8; WT + BBR, n = 8; WT + BBR + wash, n = 6; KO, n = 7; KO + BBR, n = 7; KO + BBR + wash, n = 4). P = 0.029 in (c), P = 0.023 in (d). f–i Action potentials in WT and KO pancreatic islet β-cells treated with 10 μM BBR or vehicle plus washout. f Representative action potentials were recorded from the indicated WT and KO β-cells in current-clamp mode. g–i Summary of the mean APDs (WT, n = 11; WT + BBR, n = 11; WT + BBR + wash, n = 7; KO, n = 14; KO + BBR, n = 13; KO + BBR + wash, n = 5). P = 0.005 in g, P = 0.008 in h. j–l Total internal reflection fluorescence (TIRF) microscopy analysis of pancreatic β-cells from WT mice. Islet β-cells expressing insulin-EGFP were incubated with 25 mM glucose for 30 min followed by the application of 10 μM BBR or control (DMSO). j TIRF images show live β-cells treated with the control or BBR. Red dots indicate the positions of exocytotic events occurring over 4 min after the application of BBR or the control. The white lines represent the outline of cells. Scale bars, 5 μm. A representative image from three different experiments is shown. k, l The average numbers of fusion events in control (n = 11 cells) and BBR-treated cells (n = 15 cells) at 1 min intervals are shown in k and are summed in l. P = 0.004 for 0–1 min, P = 0.004 for 1–2 min in k, P = 0.0005 in l. The values are presented as means ± s.e.m. *P < 0.05, **P < 0.01, and ***P < 0.001. Statistical significance was assessed using the Mann–Whitney U-test (two-sided).

Fig. 6. BBR increases insulin secretion in humans.

a A hyperglycemic clamp study was performed in 15 healthy male subjects using a crossover study design. Volunteers were examined and recruited according to the criteria within 14 days before the experiment (−14 d to −1 d). All subjects were admitted to the hospital on 0 d and underwent a 160-min hyperglycemic clamp after a single oral administration of 1 g of BBR or placebo on the next morning (1 d). All subjects participated in experiments on two days separated by a 14-day washout period (2–15 d) and changed the drug (BBR or placebo) in the next experiment. After the last experiment, subjects were followed up for 14 days to observe the side effects (18–31 d). b Plasma glucose levels in subjects throughout the clamp study. The dashed lines represent the range of ±5% of the hyperglycemic (basal blood glucose level +6.9 mmol/L) target level. c Insulin levels throughout the clamp study. n = 15. d–f Incremental plasma insulin AUC (insulin iAUC) throughout d the whole period (0–160 min), e the first phase (0–10 min), and f the second phase (10–160 min). g Proinsulin C-peptide levels throughout the clamp study. n = 15. P = 0.024 in d and P = 0.022 in f. h–j Incremental plasma proinsulin C-peptide AUC (C-peptide iAUC) throughout h the whole period (0–160 min), i the first phase (0–10 min), and j the second phase (10–160 min). n = 15. P = 0.007 in h and P = 0.006 in j. The values are presented as means ± s.e.m. *P < 0.05, and **P < 0.01. Statistical significance was assessed using the ratio paired t-test (two-sided).

Fig. 7. Proposed mechanisms by which BBR increases insulin secretion.

(1) By directly binding to the KCNH6 channel and accelerating channel closure, BBR blocks outward K_v channel currents and prolongs APDs in pancreatic islet β-cells. (2) The prolongation of APDs results in an increase in Ca²⁺ influx through voltage-gated calcium channels (VGCCs), (3) resulting in the accumulation of intracellular Ca²⁺. (4) Increased intracellular Ca²⁺ concentrations trigger more insulin secretion. Therefore, BBR may have potential as a new drug treatment for diabetes.