Differential Effects of Furin Deficiency on Insulin Receptor Processing and Glucose Control in Liver and Pancreatic β Cells of Mice

The insulin receptor (IR) is critically involved in maintaining glucose homeostasis. It undergoes proteolytic cleavage by proprotein convertases, which is an essential step for its activation. The importance of the insulin receptor in liver is well established, but its role in pancreatic β cells is still controversial. In this study, we investigated the cleavage of the IR by the proprotein convertase FURIN in β cells and hepatocytes, and the contribution of the IR in pancreatic β cells and liver to glucose homeostasis. β-cell-specific Furin knockout (βFurKO) mice were glucose intolerant, but liver-specific Furin knockout (LFurKO) mice were normoglycemic. Processing of the IR was blocked in βFurKO cells, but unaffected in LFurKO mice. Most strikingly, glucose homeostasis in β-cell-specific IR knockout (βIRKO) mice was normal in younger mice (up to 20 weeks), and only mildly affected in older mice (24 weeks). In conclusion, FURIN cleaves the IR non-redundantly in β cells, but redundantly in liver. Furthermore, we demonstrated that the IR in β cells plays a limited role in glucose homeostasis.


Introduction
The IR is a key regulator of glucose homeostasis. The binding of insulin to its receptor in metabolic organs such as liver, pancreas, muscle, brain, and adipose tissue leads to the activation of a signaling pathway mostly aimed at maintaining glucose homeostasis in humans and animals [1].
The IR is synthesized as an inactive precursor protein (proIR) that needs post-translational modifications to be activated. First, it requires endoproteolytic cleavage at the tetrabasic sequence RKRR 752 , yielding disulfide-linked α and β subunits [2]. Patients carrying a mutation in this cleavage site present with severe insulin-resistant diabetes [3,4]. The tetrabasic sequence is most likely removed afterward by carboxypeptidase D, a process that has been shown to be required for full activation of the insulin-like growth factor 1 receptor [5]. The endoproteolytic cleavage step has been shown to be performed by the proprotein convertase (PC) family members FURIN, PACE4, PC5/6 (isoforms A and B), and PC7 in overexpression experiments [6]. However, FURIN was put forward as the physiological PC for processing of the proIR in the secretory pathway and PACE4 at the cell surface when FURIN activity is reduced [6]. In colorectal cell lines and in mouse mammary gland tissue, genetic ablation of Furin indeed blocked proIR processing, suggesting 2 of 13 non-redundant cleavage by FURIN [7,8]. However, the proIR was normally processed in liver of an inducible Furin knockout mouse model, revealing tissue-specific redundancy [9]. Cleavage of the proIR by PCs in pancreatic β cells has not been investigated.
Insulin signaling has been studied extensively in many tissues. In liver, it regulates glucose metabolism by suppressing hepatic glucose production and promoting insulin clearance and glycogen storage [10,11]. Liver-specific IR knockout mice show severe glucose intolerance together with hyperinsulinemia, and subsequent hepatic dysfunction [12]. The physiological consequences of reduced IR signaling in β cells remain controversial. Notably, studies in transgenic mice have suggested an important role of the IR in pancreatic β cells for the regulation of peripheral glucose homeostasis [13][14][15]. However, these studies made use of a Cre driver line shown to have off-target expression in the brain [16], or were affected by the presence of the human growth hormone (hGH) minigene [17][18][19]. Recent studies, using Cre drivers lacking these pitfalls, showed that in β cells of adult islets, the IR controls insulin release and β-cell physiology [20,21].
In this study, we used conditional knockout mouse models and CRISPR-generated knockout cell lines to investigate the impact of Furin deficiency on proIR cleavage and insulin signaling in liver and β cells.

IR Processing and Signaling Are Severely Affected in FurKO β Cells
To determine the role of FURIN in the proteolytic activation of the proIR in pancreatic β cells, we first analyzed the murine β-cell line βTC3. This cell line showed a similar gene expression pattern of PCs active in the constitutive secretory pathway at mRNA levels compared to mouse islets ( Figure 1A). Furin, Pcsk6, and Pcsk7, encoding FURIN, PACE4, and PC7, respectively, were highly expressed, while Pcsk5, encoding PC5/6, was not. In βTC3 cells, the lack of Furin (FurKO) resulted in severely impaired, if not blocked, proIR cleavage ( Figure 1B), and transfection with recombinant Furin rescued the cleavage of mature IR ( Figure 1C). Importantly, the uncleaved proIR in FurKO β cells was unable to properly respond to insulin based on lack of phosphorylation of IRS1 and AKT after insulin stimulation ( Figure 1D,E).

IR Proteolytic Cleavage Is Not Altered in Liver-Specific FurKO Mice
At mRNA level, Furin was the most abundant PC in mouse liver, although Pcsk5 and Pcsk6 were highly expressed as well ( Figure 2A). In vivo, conditional knockout of Furin in mouse hepatocytes using the Alb-Cre mice (LFurKO; Figure 3A) induced a modest, non-significant increase of Pcsk5 and Pcsk6 gene expression ( Figure 2A). We subsequently evaluated the IR processing in hepatocytes from LFurKO and control mice by Western blot. As shown in Figure 2B,C, the cleavage of proIR was not significantly reduced in LFurKO mice compared to the controls, indicating almost complete redundancy for proteolytic cleavage by other PCs. and PC7, respectively, were highly expressed, while Pcsk5, encoding PC5/6, was not. In βTC3 cells, the lack of Furin (FurKO) resulted in severely impaired, if not blocked, proIR cleavage ( Figure 1B), and transfection with recombinant Furin rescued the cleavage of mature IR ( Figure 1C). Importantly, the uncleaved proIR in FurKO β cells was unable to properly respond to insulin based on lack of phosphorylation of IRS1 and AKT after insulin stimulation ( Figure 1D,E). , and Pcsk7 gene expression in the islets of Fur fl/fl mice (n = 3-5 per genotype) and wild-type βTC3 cells. *** p < 0.001 determined by two-way ANOVA with Sidak's multiple comparisons test. All the RT-qPCR data were normalized to Gapdh. (B) Western blot analysis of the IR protein level in whole-cell lysate of WT and FurKO IR-transfected and non-transfected βTC3 cells. GAPDH was used as loading control. (C) Western blot analyses of IR protein levels in whole cell lysates from βTC3 cells transfected with Insr or co-transfected with Insr and Furin. GAPDH was used as loading control. (D) Western blot analysis of phosphorylated and total IRS1, and Akt in whole cell lysates from βTC3 cells treated with 100 nM insulin or vehicle for 5 min. GAPDH was used as loading control. (E) Schematic representation of a suggested mechanism of IR signaling in FurKO βTC3 cells compared to a physiological condition. mouse hepatocytes using the Alb-Cre mice (LFurKO; Figure 3A) induced a modest, nonsignificant increase of Pcsk5 and Pcsk6 gene expression (Figure 2A). We subsequently evaluated the IR processing in hepatocytes from LFurKO and control mice by Western blot. As shown in Figure 2B,C, the cleavage of proIR was not significantly reduced in LFurKO mice compared to the controls, indicating almost complete redundancy for proteolytic cleavage by other PCs. ) mice, n = 5 per genotype. *** p < 0.001 determined by two-way ANOVA with Sidak's multiple comparisons test. All RT-qPCR data were normalized to Gapdh. (B) Western blot analysis of IR processing from liver membrane fractions of control and LFurKO mice, n = 3 animals per genotype. GAPDH was used as loading control. (C) Protein-level quantification reported as % of cleaved IR over the total IR (ProIR + cleaved IR). ProIR data are represented as the mean ± SEM. Non-significant differences were detected by unpaired t-test. ) mice, n = 5 per genotype. *** p < 0.001 determined by two-way ANOVA with Sidak's multiple comparisons test. All RT-qPCR data were normalized to Gapdh. (B) Western blot analysis of IR processing from liver membrane fractions of control and LFurKO mice, n = 3 animals per genotype. GAPDH was used as loading control. (C) Protein-level quantification reported as % of cleaved IR over the total IR (ProIR + cleaved IR). ProIR data are represented as the mean ± SEM. Non-significant differences were detected by unpaired t-test. The area under the curve was expressed as g/dL/120 min; n = 5-6 mice/group. * p < 0.05 determined by unpaired t-test. (E) Intraperitoneal glucose tolerance test (IPGTT) on 16-week-old male LFurKO and control (Fur fl/fl ) mice fed for 8 weeks on a high-fat diet (HFD, 45% kJ from fat). No significant differences were found by two-way ANOVA with Sidak's multiple comparisons test. (F) The area under the curve was expressed as g/dl/120min; n = 5-6 mice/group. No significant differences were observed by unpaired t-test. (G) IPGTT on 16-week-old male βFurKO and control (Fur fl/fl ) mice fed for 8 weeks on a HFD (45% kJ from fat). *** p < 0.001 determined by two-way ANOVA with Sidak's multiple comparisons test. (H) The area under the curve was expressed as g/dl/120min; n = 6 mice/group. *** p < 0.001 determined by unpaired t-test. All data were presented as mean ± SEM.

Impact of Conditional Furin Deletion in Liver and Pancreatic β Cells on Glucose Homeostasis
β-cell-specific Furin knockout (βFurKO) mice were generated by breeding Fur fl/fl mice [9] with RIP-Cre +/− ( Figure 3A). A comparative gene expression analysis of all PCs active genotype. * p < 0.05; *** p < 0.001 determined by two-way ANOVA with Sidak's multiple comparisons test. All the data were normalized to Gapdh. (C) Intraperitoneal glucose tolerance tests (IPGTTs) on 8week-old male LFurKO and control mice on normal chow diet (NCD) (n = 5-6 mice/group). * p < 0.05 determined by two-way ANOVA with Sidak's multiple comparisons test. (D) The area under the curve was expressed as g/dL/120 min; n = 5-6 mice/group. * p < 0.05 determined by unpaired t-test. (E) Intraperitoneal glucose tolerance test (IPGTT) on 16-week-old male LFurKO and control (Fur fl/fl ) mice fed for 8 weeks on a high-fat diet (HFD, 45% kJ from fat). No significant differences were found by two-way ANOVA with Sidak's multiple comparisons test. (F) The area under the curve was expressed as g/dL/120 min; n = 5-6 mice/group. No significant differences were observed by unpaired t-test. (G) IPGTT on 16-week-old male βFurKO and control (Fur fl/fl ) mice fed for 8 weeks on a HFD (45% kJ from fat). *** p < 0.001 determined by two-way ANOVA with Sidak's multiple comparisons test. (H) The area under the curve was expressed as g/dL/120 min; n = 6 mice/group. *** p < 0.001 determined by unpaired t-test. All data were presented as mean ± SEM.

Impact of Conditional Furin Deletion in Liver and Pancreatic β Cells on Glucose Homeostasis
β-cell-specific Furin knockout (βFurKO) mice were generated by breeding Fur fl/fl mice [9] with RIP-Cre +/− ( Figure 3A). A comparative gene expression analysis of all PCs active in the constitutive secretory pathway, Furin, Pcsk5, Pcsk6, and Pcsk7 in liver and pancreatic islets of Fur fl/fl mice is displayed in Figures 3B and 4. Overall, mRNA expression levels of all tested PCs were significantly higher in mouse liver with respect to isolated mouse islets. We subsequently studied how FURIN deficiency affected glucose homeostasis in vivo in both tissue-specific knockout models. As expected, we observed that LFurKO mice remained glucose tolerant both on chow and a HFD ( Figure 3C-F) with normal insulin sensitivity on HFD ( Figure S1B). In contrast, βFurKO mice were severely glucose intolerant, with significantly higher fasting blood glucose levels on HFD ( Figure 3G-H and Figure S1B), even on a chow diet, as we described in a previous study [22]. We also did not observe changes in fasting blood glucose and body weight in LFurKO mice compared to controls ( Figure S1C,D). in the constitutive secretory pathway, Furin, Pcsk5, Pcsk6, and Pcsk7 in liver and pancreatic islets of Fur fl/fl mice is displayed in Figure 3B and Figure 4. Overall, mRNA expression levels of all tested PCs were significantly higher in mouse liver with respect to isolated mouse islets. We subsequently studied how FURIN deficiency affected glucose homeostasis in vivo in both tissue-specific knockout models. As expected, we observed that LFurKO mice remained glucose tolerant both on chow and a HFD ( Figure 3C-F) with normal insulin sensitivity on HFD ( Figure S1B). In contrast, βFurKO mice were severely glucose intolerant, with significantly higher fasting blood glucose levels on HFD ( Figure  3G-H and Figure S1B), even on a chow diet, as we described in a previous study [22]. We also did not observe changes in fasting blood glucose and body weight in LFurKO mice compared to controls ( Figure S1C,D). . Schematic overview of the IR action on glucose and relative abundance of each PC in liver and pancreatic islets of mice. FURIN non-redundantly cleaves the proIR in pancreatic beta cells of mice, but the IR exerts only a minor effect on glucose homeostasis. In contrast, proprotein convertases other than FURIN can process proIR in the mouse liver, but here this receptor plays a pivotal role in glucose homeostasis. Image created with BioRender.com.

IR Deficiency in Pancreatic β Cells Does Not Induce Severe Glucose Intolerance
IR signaling has distinct roles in liver and pancreas [12,14]. Since the role of the IR in pancreatic β cells is still controversial, we re-evaluated the effect of Insr knockout in β cells of mice, using a Cre-driver line without the hGH minigene. We intercrossed RIP-Cre mice [23] with Insr-floxed (Insr lox/lox ) animals to generate βIRKO mice ( Figure 5A); these mice showed a 68% reduction of Insr mRNA in the islets ( Figure 5B), consistent with (near) complete inactivation in the β cells. The remaining mRNA expression level was likely linked to the expression of Insr in the other cell types of the islets (α, δ cells, and blood vessel endothelial cells), in which the Cre transgene is not expressed. In contrast to earlier observations [13,24], glucose tolerance was not significantly altered in either 12-or 20week-old mice ( Figure 5C-F), and only mildly affected in 24-week-old animals ( Figure  5G,H). In addition, fasting blood glucose levels and body weight were unaltered in βIRKO mice ( Figure S1E,F). Moreover, a previous study reported that Insr deficiency in mouse islets and β cell lines led to an induction of ATF4-dependent genes (i.e., Trib3) [25], similar to our previous results using the βFurKO [22]. However, we observed a non-significant reduction in the gene expression levels of Trib3, Chop, and Atf4 in isolated islets from the βIRKO mice ( Figure 5I). This showed that the ATF4 pathway was not altered in these βIRKO mice ( Figure 5J). . Schematic overview of the IR action on glucose and relative abundance of each PC in liver and pancreatic islets of mice. FURIN non-redundantly cleaves the proIR in pancreatic beta cells of mice, but the IR exerts only a minor effect on glucose homeostasis. In contrast, proprotein convertases other than FURIN can process proIR in the mouse liver, but here this receptor plays a pivotal role in glucose homeostasis. Image created with BioRender.com.

IR Deficiency in Pancreatic β Cells Does Not Induce Severe Glucose Intolerance
IR signaling has distinct roles in liver and pancreas [12,14]. Since the role of the IR in pancreatic β cells is still controversial, we re-evaluated the effect of Insr knockout in β cells of mice, using a Cre-driver line without the hGH minigene. We intercrossed RIP-Cre mice [23] with Insr-floxed (Insr lox/lox ) animals to generate βIRKO mice ( Figure 5A); these mice showed a 68% reduction of Insr mRNA in the islets ( Figure 5B), consistent with (near) complete inactivation in the β cells. The remaining mRNA expression level was likely linked to the expression of Insr in the other cell types of the islets (α, δ cells, and blood vessel endothelial cells), in which the Cre transgene is not expressed. In contrast to earlier observations [13,24], glucose tolerance was not significantly altered in either 12-or 20-weekold mice ( Figure 5C-F), and only mildly affected in 24-week-old animals ( Figure 5G,H). In addition, fasting blood glucose levels and body weight were unaltered in βIRKO mice ( Figure S1E,F). Moreover, a previous study reported that Insr deficiency in mouse islets and β cell lines led to an induction of ATF4-dependent genes (i.e., Trib3) [25], similar to our previous results using the βFurKO [22]. However, we observed a non-significant reduction in the gene expression levels of Trib3, Chop, and Atf4 in isolated islets from the βIRKO mice ( Figure 5I). This showed that the ATF4 pathway was not altered in these βIRKO mice ( Figure 5J). RT-qPCR analysis of Insr in the islets of control (Insr lox/lox ) and βIRKO mice, n = 4 mice/group. All data were normalized to Gapdh and to the controls. *** p < 0.001 determined by unpaired t-test. (C-H) IPGTT on male βIRKO and control mice of 12 weeks (n = 9-14 mice/group) (C), 20 weeks (n = 7-13 mice/group) (E), and 24 weeks (n= 8-16 mice/group) (G). Quantification of the area under the curve for (C,E,G) was expressed as g/dL/120 min, and plotted in (D,F,H). * p < 0.05,** p < 0.01 determined by two-way ANOVA with a Sidak's multiple comparisons test.* p < 0.05 determined by unpaired t-test. All data were presented as mean ± SEM. (I) RT-qPCR analysis of Trib3, Chop, and Atf4 gene expression in the islets of control (Insr lox/lox ) and βIRKO mice, n = 3-4 mice/group. No significant differences were detected by two-way ANOVA with Sidak's multiple comparisons test. All the RT-qPCR data were normalized to the Gapdh. (J) Schematic of mTORC1/ATF4 pathway, which was upregulated in βFurKO but not in βIRKO mice.

(B) RT-qPCR analysis of
Insr in the islets of control (Insr lox/lox ) and βIRKO mice, n = 4 mice/group. All data were normalized to Gapdh and to the controls. *** p < 0.001 determined by unpaired t-test. (C-H) IPGTT on male βIRKO and control mice of 12 weeks (n = 9-14 mice/group) (C), 20 weeks (n = 7-13 mice/group) (E), and 24 weeks (n= 8-16 mice/group) (G). Quantification of the area under the curve for (C,E,G) was expressed as g/dL/120 min, and plotted in (D,F,H). * p < 0.05, ** p < 0.01 determined by two-way ANOVA with a Sidak's multiple comparisons test. * p < 0.05 determined by unpaired t-test. All data were presented as mean ± SEM. (I) RT-qPCR analysis of Trib3, Chop, and Atf4 gene expression in the islets of control (Insr lox/lox ) and βIRKO mice, n = 3-4 mice/group. No significant differences were detected by two-way ANOVA with Sidak's multiple comparisons test. All the RT-qPCR data were normalized to the Gapdh. (J) Schematic of mTORC1/ATF4 pathway, which was upregulated in βFurKO but not in βIRKO mice.

Discussion
One of the main goals of this study was to demonstrate that the IR is processed by FU-RIN in a tissue-specific way, and that the lack of insulin signaling has differential functions in both hepatocytes and pancreatic β cells. Our results revealed that in pancreatic β cells, proIR is non-redundantly cleaved by FURIN. Similar results were observed in colorectal cell lines and mouse mammary gland tissue lacking Furin [7,8]. In contrast, in liver of FurKO mice, we show a complete redundancy of proIR cleavage, which is in agreement with our previous work in which an inducible promoter was used to inactivate Furin [9]. This tissue specificity in cleaving proIR is likely due to the differential abundance of the other PCs transcripts in the analyzed tissues. For instance, the overall mRNA level of Pcsk6, which is the most likely PC to provide redundancy as described in certain cellular models [6], is higher in mouse liver than in mouse pancreatic islets. This supports the hypothesis that the cleavage of a specific substrate in a given tissue can be linked to the availability of compensatory PCs [26]. However, the differential abundance of the PC transcripts in these tissues is only one of the possible explanations of the redundant cleavage of proIR in liver. For instance, the presence of a regulated secretory pathway in β cells but not in hepatocytes likely has an influence on the trafficking of the PCs and proIR. In particular, the process of post-Golgi trafficking through direct constitutive secretion in the liver is distinctly different from the constitutive-like secretion pathway described in β cells [27]. In addition to the tissue-specific enzyme expression, and the cell-specific differences in trafficking of substrates and enzymes, several factors are known to contribute to the enzyme-substrate interaction. The substrate specificity of PCs is overlapping but not identical, and hence is responsible for part of the substrate selectivity. Another crucial factor is the spatiotemporal co-expression of enzyme and substrate. For instance, this is relevant for the precursor of pituitary adenylate cyclase-activating polypeptide (proPACAP), which is cleaved by PC4 in germ cells [28,29], and by PC1/3 and PC2 in neuroendocrine cells [30]. Additional factors that might contribute to tissue-dependent processing are only partly known but can include, for instance, certain post-translational modifications. Thus, glycosylation or phosphorylation near the cleavage site can dynamically affect processing, as was recently shown for FGF23 [31,32]. Overall, this finding of tissue-specific processing adds an extra layer of regulation, allowing the spatiotemporal fine-tuning of IR activity.
The lack of cleavage of the proIR in βFurKO cells results in an impaired response to insulin, indicating that proIR needs to be proteolytically cleaved by FURIN for a functional insulin signal transduction in pancreatic β cells. These results are in line with previous studies in patients and cellular models overexpressing the IR protein with mutations in the cleavage site [3,4,33,34]. Overall, our results highlight an important role of FURIN activity in the regulation of the autocrine functions in β cells through proteolytic activation of IR, as suggested by the alterations in glucose homeostasis observed in βFurKO mice. On the other hand, the redundancy-hypothesis of proIR cleavage in mouse liver was confirmed by the unaffected hepatic IR processing, and normal response to glucose challenge in hepatocytespecific FurKO mice. Interestingly, these mice showed improved glucose tolerance on NCD, suggesting that inhibition of hepatic Furin may protect against glucose dysregulation in metabolic disorders.
Since FURIN has multiple substrates, the glucose intolerance observed in the βFurKO mice cannot automatically be attributed to lack of processing and signaling of the proIR. Therefore, we studied the effect of IR-deficiency in a βIRKO mouse model using the same Cre driver as in the βFurKO mice. Surprisingly, we only observed a mild glucose intolerance in 24-week-old βIRKO mice, and a normal response to a glucose challenge in younger mice. These results are in sharp contrast with a previous study in which βIRKO mice showed a progressive impairment in glucose tolerance [13]. However, for this study, the authors used the RIP-Cre Mgn driver line, which was later shown to exhibit glucoserelated phenotype itself [35,36] due to ectopic hGH expression [17]. Using the RIP-Cre Herr driver line, which is not affected by the hGH expression, we could establish that glucose homeostasis is normal in βIRKO mice of similar age, indicating that hGH is the most likely cause of the phenotype of the aforementioned mouse model. Accordingly, in a recent study using an inducible MIP1-Cre knock-in mouse line, inactivation of the IR in adult β cells resulted in normal glucose tolerance, insulin release, and β-cell mass when fed a NCD diet. The mice became glucose intolerant only when challenged with a HFD [20].
Finally, we previously found that FURIN deficiency in pancreatic β cells leads to severe glucose intolerance caused by altered lysosomal acidification and aberrant activation of the mTORC1/ATF4 pathway [22]. Crosstalk between IR signaling and the mTORC1 pathway has been reported in several studies [37,38]. However, we did not observe any effect of the lack of IR in the activation of the mTORC1/ATF4 pathway.
In conclusion, our findings establish a critical role of FURIN in proteolytic processing of the proIR, and regulation of insulin signaling in pancreatic β cells, but not in hepatocytes. We have also established that impaired IR signaling in pancreatic β cells does not result in severe glucose intolerance (Figure 4). These results highlight that processing of substrates by members of the PC family can be entirely tissue-dependent, which should be taken into account for therapeutic strategies.

Mice
Alb-Cre (B6.Cg-Tg(Alb-cre)21Mgn/J) mice, bought from Jackson Laboratories (https: //www.jax.org), were bred with Fur fl/fl mice to generate hepatocyte-specific Furin-deficient mice (LFurKO) [9]. RIP-Cre +/− mice (Tg[Ins2-Cre]23Herr) [23] were bred with animals in which the exon 4 of the Insr gene was flanked by loxP sites (IR Lox ) (https://www. jax.org/strain/006955; accessed on 6 September 2016), or with Fur fl/fl mice to generate βIRKO and βFurKO mice, respectively. Mice were backcrossed at least 5 times to a C57Bl6J background as previously described [22]. All the mice were housed in standard cages on a 12 h day/night cycle and fed a standard rodent chow or HFD (45 kJ % from fat (lard), 20 kJ % from proteins, and 35 kJ % from carbohydrates; Ssniff, Germany) in a conventional animal facility of the KU Leuven. Food and water were provided ad libitum. To avoid confounding effects of estrogens on glucose homeostasis [39,40], only male mice were included in this study.

Intraperitoneal Glucose and Insulin Tolerance Test (IPGTT and IPITT)
For IPGTT, mice were fasted overnight or 5 h and injected with 2 mg/g body weight (BW) D-glucose (GTT) or 0.75 IU/g BW human insulin (ITT) in PBS, respectively. Blood glucose levels were monitored at indicated time points using a Contour Glucometer (Roche, Basel, Switzerland).

Islet Isolation
Islet were isolated by locally injecting the pancreas with 1 Wünsch unit/mL Liberase (Roche, Basel, Switzerland) in HANKS buffer as previously described [17,22]. Injected pancreata were incubated for 18 min at 37 • C in a shaker (200 rpm), and the islet fraction subsequently was separated from exocrine tissue using a Dextran T70 gradient. Finally, islets were handpicked twice in HANKS buffer under a stereomicroscope to reach a pure islet fraction for further processing.

Cell Culture and Transfection
The mouse insulinoma cell line βTC3 was cultured in DMEM:F12 (1:1) supplemented with 10% heat-inactivated fetal bovine serum. For overexpression experiments, βTC3 cells were transfected with plasmids encoding mouse Furin and/or Insr (pcDNA3.1 backbone) using Lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA) according to the manufacturer's protocol. For insulin-stimulation experiments, βTC3 cells were seeded at a cell density of 80%, and the day after they were starved overnight in serum-free DMEM:F12 medium containing 0.1% bovine serum albumin (BSA). Subsequently, cells were cultured for 2 h in glucose-free modified KRB buffer (125 mM NaCl, 4.74 mM KCl, 1 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 1.2 mM, MgSO 4 , 5 mM NaHCO 3 , 25 mM HEPES (pH 7.4)) with 0.1% BSA to minimize endogenous insulin secretion. The stimulation was performed in the same buffer with 100 nM insulin for 5 min.

Generation of Furin Knockout βTC3 Cells Using the CRISPR-Cas9 Nuclease System
The Furin knockout βTC3 cell line was generated as previously described [22,41].

Microarray Analysis
The microarray analysis was performed in the previous study, and the microarray data were deposited in the Gene Expression Omnibus (GSE150312) [22].

Crude Membrane Fraction from Mouse Liver
Mouse liver samples were homogenized in 5 volumes of sucrose buffer (sucrose 0.25 M, Tris-HCl 50 mM, MgCl 2 5 mM. The post-nuclear supernatant was obtained after centrifugation at 600× g at 4 • C for 10 min, and further centrifugation of the supernatant at 10,000× g at 4 • C for 30 min. The resulting supernatant was ultra-centrifuged at 25,000× g at 4 • C for 1 h (TLA 55 rotor). The pellet was dissolved in 1X cell lysis buffer (Bioke, Leiden, Netherlands) supplemented with complete EDTA-free protease inhibitor cocktails (Sigma, St. Louis, MO, USA) and 2% SDS.

Quantitative RT-PCR (RT-qPCR)
RNA from snap-frozen mouse liver, freshly isolated mouse islets, or βTC3 cells was extracted using the Nucleospin RNA II (Macherey Nagel, Düren, Germany) kit according to the manufacturer's protocol. The cDNA was synthetized from 1 µg RNA in 20 µL of reaction volume using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA). Primers were designed using the ProbeFinder software (Roche Applied Sciences, Basel, Switzerland). The quality of the RNA templates was assessed by using a NanoDrop ND-1000 spectrophotometer (V.3.8.1). RT-qPCR was performed in triplicate with a Bio-Rad CFX96 real-time qPCR detection system (Bio-Rad) using SYBR Green. RT-qPCR was performed with a CFX Connect Real-Time PCR Detection System from Bio-Rad using SYBR Green supermix (Bio-Rad). Data is represented as 2 −∆Ct . Primers for mouse genes are listed in Table S1.

Statistical Analysis
Results are expressed as means ± SEM. Statistical analysis was performed by unpaired Student's t-test or one-way ANOVA with Sidak's multiple comparisons test for grouped analysis, or repeated measure two-way ANOVA for pairwise time-specific differences between genotypes. A value of p < 0.05 was considered significant. * p < 0.05, ** p < 0.01, *** p < 0.001.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/ijms22126344/s1, Figure S1: Blood glucose parameters and body weight records of βFurKO, LFurKO, or βIRKO mice compared to their relative controls respectively; and Table S1: Primer specifications for RT-qPCR.

Data Availability Statement:
The microarray data were deposited in the Gene Expression Omnibus (GSE150312) [22].