Pancreatic alpha cell glucagon–liver FGF21 axis regulates beta cell regeneration in a mouse model of type 2 diabetes

Aims/hypothesis Glucagon receptor (GCGR) antagonism ameliorates hyperglycaemia and promotes beta cell regeneration in mouse models of type 2 diabetes. However, the underlying mechanisms remain unclear. The present study aimed to investigate the mechanism of beta cell regeneration induced by GCGR antagonism in mice. Methods The db/db mice and high-fat diet (HFD)+streptozotocin (STZ)-induced mice with type 2 diabetes were treated with antagonistic GCGR monoclonal antibody (mAb), and the metabolic variables and islet cell quantification were evaluated. Plasma cytokine array and liver RNA sequencing data were used to screen possible mediators, including fibroblast growth factor 21 (FGF21). ELISA, quantitative RT-PCR and western blot were applied to verify FGF21 change. Blockage of FGF21 signalling by FGF21-neutralising antibody (nAb) was used to clarify whether FGF21 was involved in the effects of GCGR mAb on the expression of beta cell identity-related genes under plasma-conditional culture and hepatocyte co-culture conditions. FGF21 nAb-treated db/db mice, systemic Fgf21-knockout (Fgf21−/−) diabetic mice and hepatocyte-specific Fgf21-knockout (Fgf21Hep−/−) diabetic mice were used to reveal the involvement of FGF21 in beta cell regeneration. A BrdU tracing study was used to analyse beta cell proliferation in diabetic mice treated with GCGR mAb. Results GCGR mAb treatment improved blood glucose control, and increased islet number (db/db 1.6±0.1 vs 0.8±0.1 per mm2, p<0.001; HFD+STZ 1.2±0.1 vs 0.5±0.1 per mm2, p<0.01) and area (db/db 2.5±0.2 vs 1.2±0.2%, p<0.001; HFD+STZ 1.0±0.1 vs 0.3±0.1%, p<0.01) in diabetic mice. The plasma cytokine array and liver RNA sequencing data showed that FGF21 levels in plasma and liver were upregulated by GCGR antagonism. The GCGR mAb induced upregulation of plasma FGF21 levels (db/db 661.5±40.0 vs 466.2±55.7 pg/ml, p<0.05; HFD+STZ 877.0±106.8 vs 445.5±54.0 pg/ml, p<0.05) and the liver levels of Fgf21 mRNA (db/db 3.2±0.5 vs 1.8±0.1, p<0.05; HFD+STZ 2.0±0.3 vs 1.0±0.2, p<0.05) and protein (db/db 2.0±0.2 vs 1.4±0.1, p<0.05; HFD+STZ 1.6±0.1 vs 1.0±0.1, p<0.01). Exposure to plasma or hepatocytes from the GCGR mAb-treated mice upregulated the mRNA levels of characteristic genes associated with beta cell identity in cultured mouse islets and a beta cell line, and blockage of FGF21 activity by an FGF21 nAb diminished this upregulation. Notably, the effects of increased beta cell number induced by GCGR mAb were attenuated in FGF21 nAb-treated db/db mice, Fgf21−/− diabetic mice and Fgf21Hep−/− diabetic mice. Moreover, GCGR mAb treatment enhanced beta cell proliferation in the two groups of diabetic mice, and this effect was weakened in Fgf21−/− and Fgf21Hep−/− mice. Conclusions/interpretation Our findings demonstrate that liver-derived FGF21 is involved in the GCGR antagonism-induced beta cell regeneration in a mouse model of type 2 diabetes. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00125-022-05822-2.

consecutive times. The mice without diabetic condition were omitted from the study.
Mice were randomised into groups having similar distributions based on their body weight and blood glucose level.
To determine the effects of GCGR mAb on beta cell regeneration, db/db mice or HFD+STZ-induced mouse models of type 2 diabetes in C57BL/6N mice were treated for 6 weeks via weekly intraperitoneal injection of REMD 2.59 (a human antagonistic GCGR mAb, 5 mg/kg; REMD Biotherapeutics, Camarillo, CA, USA), or human IgG (5 mg/kg, as control). There were 5 mice in each group. To clarify the involvement of FGF21 in GCGR mAb-induced beta cell regeneration, HFD+STZ-induced mouse models of type 2 diabetes in Fgf21 −/− mice and WT mice were treated with 5 mg/kg REMD 2.59 or human IgG for 6 weeks. There were 6 mice in each group. Besides, db/db mice were divided into four groups: 1) control group: treated with human IgG and rabbit IgG; 2) GCGR mAb group: injected with REMD 2.59 and rabbit IgG; 3) FGF21 neutralizing antibody (nAb) group: treated with FGF21 nAb (Antibody & Immunoassay Services, Hong Kong, China) and human IgG; 4) GCGR mAb + FGF21 nAb group: injected with REMD 2.59 and FGF21 nAb. REMD 2.59 and human IgG were intraperitoneally injected weekly at a dose of 5 mg/kg, while FGF21 nAb and rabbit IgG were intraperitoneally injected daily at a dose of 6 μg/day for 3 weeks. There were 6 mice in each group. To determine the effects of liver-derived FGF21 in GCGR mAb-induced beta cell regeneration, HFD+STZ-induced mouse models of type 2 diabetes in Fgf21 Hep−/− mice and Flox littermates were treated with 5 mg/kg REMD 2.59 or human IgG for 6 weeks.
The high affinity (a higher affinity for GCGR than the endogenous ligand glucagon) and strong specificity (highly specific antagonistic activity against GCGR, without inhibitory effect on glucagon-like peptide-1 receptor and other receptors with similar structures) of GCGR mAb has been proved previously [2][3][4]. The neutralizing ability of FGF21 nAb has been reported previously [5].

Immunofluorescent staining and quantification
Pancreases were fixed with 10% (vol/vol) neutral-buffered formalin and embedded in paraffin, and 5-µm -thick sections were prepared. For immunofluorescence, the sections were incubated with primary antibodies at 4°C overnight and secondary antibodies for 1 h at room temperature, followed by staining with DAPI. All primary and secondary antibodies were diluted in antibody dilution buffer (Tris-HCl buffer, BSA, sodium azide; Zhongshan Biotechnology, Beijing, China). Images were captured under Leica TCS SP8 confocal fluorescence microscope (Leica Microsystems, Wetzlar, Germany) or an automatic digital slide scanner (Pannoramic MIDI, 3D HISTECH, Budapest, Hungary).
The primary antibodies were as follows: rabbit polyclonal anti-glucagon (1:800; Cell Signaling Technology, Boston, MA, USA; RRID: AB_659831), mouse monoclonal antiinsulin (1:800; Sigma-Aldrich; RRID: AB_260137), rabbit monoclonal anti-insulin For cell quantification in the immunofluorescent staining, 3 to 5 equally spaced sections (which covered the entire pancreas) per pancreas with 3 to 6 mice per group were imaged, and the spacing between the two adjacent sections was 200 μm. The islet number per section, and the alpha cell number and beta cell number per islet were counted manually with the positive staining of glucagon and insulin. The islet area (the glucagon-positive and insulin-positive region) was analysed by Fiji software (National Institutes of Health, Bethesda, MD, USA) [6]. For panoramic scanning, fluorescence was imaged using an automatic digital slide scanner (Pannoramic MIDI).

Mouse beta cell line culture
The mouse pancreatic beta cell line Min6 cells, kindly gifted by Prof. Yiming Mu from the General Hospital of the People's Liberation Army (Beijing, China), were cultured in Dulbecco's modified eagle medium (DMEM, 25 mmol/l glucose; Invitrogen) supplemented with 15% (vol/vol) FBS, 2 mmol/l GlutMax and 55 μmol/l βmercaptoethanol (Thermo Fisher Scientific). Min6 cells were verified to be of mouse origin and negative for inter-species contamination from rat or human. Mycoplasma was tested as negative using Mycoplasma PCR detection kit (Beyotime Biotechnology, Shanghai, China).

Primary mouse hepatocyte isolation and culture
Eight-week-old male C57BL/6N mice on chow-diet were treated for 6 weeks via intraperitoneal administration of GCGR mAb (5 mg/kg) or human IgG (5 mg/kg, as control) once a week. There were 3 mice in each group. Primary hepatocytes were isolated by nonrecirculating collagenase perfusion through the portal vein as previously reported [5] The hepatocytes were plated on 6-well plates which were coated with rat collagen type I (Sigma-Aldrich), and were then cultured in RPMI 1640 medium containing 10% (vol/vol) FBS for 6 h before further co-culture. In the co-culture system, the primary hepatocytes from each mouse were used independently in the experiments.

Co-culture of beta cell line or primary islets with primary hepatocytes
The co-culture system was established by the transwell chamber with 3 μm-pore size (Corning, New York, NY, USA). The primary mouse hepatocytes were plated on the lower chamber. The Min6 cells or primary mouse islets were plated on the upper chamber. The co-culture systems were incubated for 24 h with or without FGF21 nAb (10 μg/ml; Antibody & Immunoassay Services), and the beta cell line or primary islets were collected for further analysis.

Conditional culture of beta cell line or primary islets with mouse plasma
Eight-week-old male C57BL/6N mice on chow-diet were treated for 6 weeks via intraperitoneal injection of GCGR mAb (5 mg/kg) or human IgG (5 mg/kg, as control) once per week. There were 6 mice in each group. Plasma was collected after a 12-h fast before sacrifice, and was stored at −80°C. In the conditional culture system, plasma from two mice in the same treatment group was pooled together. There were 3 batches of plasma in each group for the plasma-conditional culture experiments.
Min6 cells or primary mouse islets were cultured with 10% (vol/vol) mouse plasma and 90% culture medium with or without FGF21 nAb (10 μg/ml), and the beta cell line or primary islets were collected for further analysis.

RNA extraction, reverse transcription and quantitative RT-PCR
Total RNA was extracted with Trizol reagent (Thermo Fisher Scientific) and reversely transcribed to cDNA using a RevertAid First Strand cDNA Synthesis kit (Thermo Fisher Scientific). The quantitative RT-PCR was performed using iQ SYBR Green Supermix (Toyobo Co., Ltd., Osaka, Japan) on a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). Relative quantification for gene expression was calculated using the 2 −ΔΔCT method, which was normalized to the internal reference, Gapdh. The primer sequences were summarized in ESM Table 1.

Protein extraction and western blot analysis
Total proteins from liver tissues were obtained using radioimmunoprecipitation assay lysis buffer, which contained protease inhibitor and phosphatase inhibitor (all from Applygen Technologies, Beijing, China). The denatured proteins (approximately 30 μg) were separated by 12% (wt/vol) SDS-PAGE electrophoresis and transferred to a nitrocellulose membrane. The membranes were incubated overnight at 4°C with the primary antibodies (both at 1:1000 dilution): rabbit anti-FGF21 (Abcam; RRID: AB_2629460) and mouse anti-GAPDH (Zhongshan Biotechnology, Beijing, China; RRID: AB_2107448). The primary antibodies were diluted in TBST with 5% BSA.
Before testing, positive control and negative control were used to verify the antibodies.
After three washes, the blots were incubated for 1 h with RDye 800CW-conjugated goat anti-rabbit IgG or goat anti-mouse IgG (both at 1:10,000 dilutions; LICOR Biosciences, Lincoln, NE, USA). The secondary antibodies were diluted in TBST. Protein bands were visualized with an Odyssey 290 Infrared Imaging System (LICOR Biosciences). GAPDH was used as a loading control.