β-Cell–Specific Deletion of HMG-CoA (3-hydroxy-3-methylglutaryl-coenzyme A) Reductase Causes Overt Diabetes due to Reduction of β-Cell Mass and Impaired Insulin Secretion

3-Hydroxy-3-methylglutaryl-CoA (HMG-CoA) reductase (HMGCR) is the rate-limiting enzyme in cholesterol biosynthesis, which catalyzes the conversion of HMG-CoA to mevalonic acid (1). Inhibitors of HMGCR, statins, are widely used to prevent the occurrence of coronary heart disease and other atherosclerotic diseases primarily by reducing blood cholesterol levels (2).

However, statins have some adverse effects, including myopathy and hepatotoxicity and an increased risk for diabetes (3). Genetic analyses have confirmed the association between HMGCR inhibition and new-onset type 2 diabetes (4–6). Experiments using cells or animals have shown that statins can increase the risk for diabetes by compromising both insulin secretion and insulin sensitivity. Statins might inhibit insulin secretion by in several ways, such as by inhibiting membrane association of small guanosine-5′-triphosphate–binding proteins (7), by inhibiting glucose-induced cytosolic Ca²⁺ signaling (8), by decreasing cell viability (9) or by decreasing SNARE (soluble N-ethylmaleimide sensitive factor attachment receptor) proteins (10). Meanwhile, statins might induce insulin resistance by inhibiting either expression of GLUT4 in white adipose tissues (11) or phosphorylation of insulin receptor substrate 1 (IRS-1) in muscle cells (12). However, it remains possible that some of these pharmacological effects are mediated by off-target effects of statins, as was reported in the case of myopathy (13). Moreover, inhibition of HMGCR might be diabetogenic by increasing adiposity (4–6).

The specific effects of inhibiting HMGCR can be investigated by genetic manipulation. We previously reported that Hmgcr knockout (KO) mice die in a relatively early stage of embryonic development (before E8.5) (14), precluding their use for further studies. Subsequently, we have generated mice lacking Hmgcr in specific tissues, including liver (15), skeletal muscle (16) and myeloid cells (17,18). To investigate whether deletion of Hmgcr is diabetogenic by compromising insulin secretion, we generated mice lacking Hmgcr in β-cells of pancreatic islets (β-KO) by crossing a transgenic mouse overexpressing Cre recombinase under the promoter of rat insulin2 (RIP-Cre) with floxed Hmgcr mice.

We found that β-KO mice developed severe hypoinsulinemic hyperglycemia and attributed it primarily to failure to proliferate β-cells immediately after birth. Transdifferentiation of β-cells to α-cells around the perinatal period may also be involved in the loss of β-cell mass. Glucose-stimulated insulin secretion (GSIS) was also impaired in pancreatic islets from β-KO mice.

All animal experiments were approved by the Jichi Medical University Institutional Animal Care and Research Advisory Committee (Tochigi, Japan). HMGCR floxed mice were generated as described previously (15). RIP-Cre (Cre) mice, which express Cre recombinase under a short fragment of the rat insulin II gene promoter, were obtained from The Jackson laboratory (Bar Harbor, ME) (19).

β-KO mice were obtained by crossing heterozygous floxed Hmgcr mice with Cre mice. All mice had a C57BL/6J genetic background. In this study, male mice were used for all experiments unless stated otherwise. Cre mice were used as controls instead of floxed Hmgcr mice, because Cre mice themselves have impaired glucose tolerance compared with wild-type mice (20). Mice were fed a normal chow diet containing 4.8% (w/w) fat and 25.1% (w/w) protein (CE-2; Japan CLEA). Tissues were collected at 20:00, shortly after the start of dark the phase.

RT-PCR was performed to detect the disrupted allele for Hmgcr using cDNA in various organs. The primer sequences were forward primer, 5′-GTAAGCGCAGTTCCTTCCG-3′ and reverse primer, 5′-TGCTAATGCACTCGCTCTAGA-3′.

Total RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA). For gene expression analysis, total RNA was reverse transcribed using a high-capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) and underwent real-time quantitative (q)PCR using StepOnePlus (Applied Biosystems). mRNA expression levels were normalized to those of Actb mRNA. The primer and probe sets are listed in Supplementary Table 1.

Immunoblot analyses of the liver were described previously (15).

After acclimation for 5 days, locomotor activity was measured using the ACTIMO-100 activity monitoring system (Shinfactory Co., Ltd., Fukuoka, Japan). VO₂ and VCO₂ were synchronously measured using ARCO-2000 (ARCO System, Chiba, Japan). The respiratory quotient (RQ) was determined by the VCO₂-to-VO₂ ratio.

Plasma glucose level was measured using a glucose CII-Test Wako (Wako Pure Chemical Industries, Ltd., Osaka, Japan). Insulin level was measured using a mouse insulin kit (Morinaga Institute of Biological Science, Kanagawa, Japan). Glucagon level was measured using a glucagon ELISA kit (Mercodia, Uppsala, Sweden). Plasma total cholesterol (TC) was measured using a Determiner L TC II (Kyowa Medex, Tokyo, Japan). Plasma triglyceride (TG) and free fatty acid (FFA) levels were measured using an L-Type TG M kit and a nonesterified fatty acid (NEFA) C-test kit (Wako), respectively. Liver TG and TC content were measured as described previously (15).

After a 16-h fast, a 2.0 g/kg glucose solution was given to the mice via a tube inserted into the esophagus. After a 4-h fast, 0.75 IU/kg insulin (Novolin R; Novo Nordisk, Bagsværd, Denmark) was given to the mice intraperitoneally. Blood samples were collected from the tail, and blood glucose levels were measured using a glucose CII-Test Wako kit.

The mice were sacrificed by cervical dislocation. The pancreases were rapidly dissected and fixed in 10% neutral phosphate buffer formalin overnight. The tissues were subsequently dehydrated through a graded ethanol and xylene and finally embedded in paraffin, sectioned, and stained with hematoxylin and eosin. For immunohistochemistry, sections were deparaffinized using xylene and ethanol. Sections were incubated with 2% normal goat serum for 20 min and then incubated overnight with polyclonal guinea pig anti-insulin antibody (1:800 dilution) (Dako, Santa Clara, CA), polyclonal rabbit anti-glucagon antibody (1:800 dilution) (Progen Biotechnik, Heidelberg, Germany), or monoclonal rabbit anti-Ki67 antibody (1:100 dilution; Abcam, Cambridge, U.K.) at 30°C. Sections were then incubated with Vectastain ABC systems (Vector Laboratories, Burlingame, CA) for 30 min, followed by the addition of diaminobenzidine tetrahydrochloride. Insulin and glucagon areas were estimated by Adobe Photoshop software (Adobe Systems). Four pairs of 5-µm-thick serial sections, 150 µm apart were analyzed. The number of islets was calculated containing at least five nucleuses.

For fluorescence staining, sections were incubated overnight with polyclonal guinea pig anti-insulin antibody (1:800 dilution) (Dako), polyclonal rabbit anti-glucagon antibody (1:800 dilution) (Progen Biotechnik), or monoclonal rabbit anti-Ki67 antibody (1:100 dilution) (Abcam) at 30°C. Sections were then incubated with Alexa Fluor 568 goat anti-guinea pig IgG for insulin (1:200 dilution) (Thermo Fisher Scientific, Waltham, MA) and Alexa Fluor 488 goat anti-rabbit for glucagon and Ki67 (1:200 dilution) (Thermo Fisher Scientific). Apoptosis was determined by TUNEL staining using the In situ Apoptosis Detection Kit (Takara Bio, Shiga, Japan), according to the manufacturer’s instructions. Sections were counterstained with DAPI dihydrochloride solution (Dojindo, Kumamoto, Japan). Sections were observed with a confocal laser microscope (FV1000; Olympus). Insulin and glucagon doubly positive cells were counted in random at least five fields at original magnification ×400. Insulin and Ki67 doubly positive cells were counted similarly. Approximately 100–400 cells were counted.

Pancreatic islets were isolated as described previously (21). In brief, 2.5 mL of collagenase (3 mg/mL) (C7657; Sigma-Aldrich, St. Louis, MO) was injected into the bile duct. The perfused pancreas was subsequently dissected, placed into 50-mL tubes, and incubated at 37°C for 15 min. After the addition of 30 mL of a Hanks’ balanced salt solution (HBSS) containing 20% BSA, the digested tissues were collected by centrifugation at 290g for 1 min. A 20-mL syringe with an 18-gauge needle was filled with 10 mL of HBSS containing 20% BSA and then used to aspirate the pellets. After centrifugation at 330g for 2 min, the pellets were resuspended in 10 mL of cold 1100 Histopaque (240 mL 1119 Histopaque + 200 mL 1077 Histopaque). Ten mL HBSS containing 20% BSA was overlaid on the histopaque solution containing the islets and centrifuged at 900g for 18 min. Then, 20 mL of the supernatant was harvested, filtered through a 70-μm filter, and washed with islet medium (RPMI, 10% FBS, 1% penicillin/streptomycin). The islets were then handpicked under a microscope.

GSIS was examined as described previously (22). In brief, groups of 10 islets of similar size were incubated for 1 h in media with 2.8 or 20 mmol/L glucose. At the end of the incubation period, aliquots of the media were withdrawn to measure the insulin concentration. Total insulin was extracted from the islets with a cold acid ethanol mixture (75% ethanol with 0.2 mol/L HCl). Insulin secretion was normalized to the total insulin contents in the islets. Islet insulin contents were obtained by normalizing the insulin contents to the total protein contents in the islets.

Measurements of cytosolic Ca²⁺ concentrations ([Ca²⁺]_i) were done as described previously (22). In brief, islets on coverslips were mounted in an open chamber and superfused in HEPES-added Krebs-Ringer bicarbonate buffer. [Ca²⁺]_i was measured at 33°C by dual-wavelength fura-2 microfluorometry with excitation at 340/380 nm and emission at 510 nm using a cooled charge-coupled device camera.

All data are presented as means ± SD. GraphPad Prism software was used for data analyses. Mann-Whitney U tests or repeated-measures ANOVA with the Bonferroni multiple-comparison test were used for comparison as appropriate. Differences were considered significant for P values <0.05.

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

To determine whether Hmgcr was deleted appropriately in a β-cell–specific manner, we performed RT-PCR to semiquantify the amounts of transcripts of the floxed and disrupted alleles of the Hmgcr gene (Fig. 1A). We detected a transcript of the disrupted allele in the pancreatic islets but not in other organs, including the liver, intestine, adrenal glands, and hypothalamus. The size of the disrupted allele was identical to the disrupted allele in the liver from liver-specific Hmgcr-KO mice (15). Disruption of the Hmgcr gene was only partial in the pancreatic islets from β-KO mice, while disruption was almost complete in the liver from liver-specific Hmgcr KO mice. Next, we performed real-time–qPCR to quantify the mRNA expression of Hmgcr (Fig. 1B). mRNA expression of Hmgcr in β-KO mice was significantly lower in the pancreatic islets (78% lower), liver (57% lower), and intestine (59% lower) but not in the adrenal glands or hypothalamus.

β-KO mice were viable, with no obvious change in appearance (Supplementary Fig. 1). The growth curve indicates that β-KO mice were significantly lighter than control mice at 3 of 17 time points examined (Fig. 1C). Although there was no difference in the weight of the pancreas (Supplementary Fig. 2A), the weight of the liver was heavier in β-KO mice (Supplementary Fig. 2B). β-KO mice had less epididymal fat and less mesenteric fat than control mice (Supplementary Fig. 2C and D, respectively).

β-KO mice ate significantly more (50% more) food than did the controls (Fig. 1D), indicating that the loss of body weight did not result from reduced food intake. Rectal temperatures of β-KO mice were not different from those of controls (Fig. 1E). To determine whether β-KO mice had more total energy expenditure (TEE) than the controls, we measured locomotor activity (Fig. 2A and B), VO₂ (Fig. 2C and D), and VCO₂ (Fig. 2E and F). β-KO mice exhibited 8% and 16% higher VO₂ than the controls during light and dark phases, respectively. There were no differences in VCO₂ between the two groups. Consequently, β-KO mice exhibited 8% and 11% lower RQ than the controls during light and dark phases, respectively (Fig. 2G and H), indicating preferential consumption of fatty acids to glucose.

Under a fasting condition, β-KO mice had markedly higher levels of plasma glucose than controls at 5 and 10 weeks of age (Fig. 3A), while they had significantly lower levels of plasma insulin (Fig. 3B). Under an ad libitum condition, they were even more hyperglycemic (Fig. 3C) than under a fasting condition, with a similar degree of hypoinsulinemia (Fig. 3D). β-KO mice had significantly higher (2.9-fold higher) levels of plasma glucagon under an ad libitum condition (Fig. 3E). While β-KO mice had plasma glucose and insulin levels indistinguishable from those of the controls at postnatal day 0 (P0) (Fig. 3F and G), β-KO mice became hyperglycemic at P9, P14, and P21.

Female β-KO mice had higher plasma levels of glucose but lower plasma levels of insulin than controls (Supplementary Fig. 3A and B). Plasma levels of either glucose or insulin in mice heterozygous for the disrupted Hmgcr allele were indistinguishable from those in controls (Supplementary Fig. 3C and D).

Consistent with the hyperglycemia under both fasting and ad libitum conditions, oral glucose tolerance test (OGTT) results showed hyperglycemia (Fig. 4A and B) and impaired insulin levels (Fig. 4C and D) in β-KO mice compared with the controls. β-KO mice were as sensitive to insulin as controls (Fig. 4E and F), suggesting that β-KO mice became hyperglycemic largely because of impaired insulin secretion.

Under a fasting condition, plasma levels of TC and FFA were not different between the two groups, while plasma TG levels were significantly lower in β-KO mice than in controls (Table 1). Under an ad libitum condition, there were no differences in plasma TC and TG levels between the two groups.

To examine the changes in hepatic lipid metabolism, we measured liver lipid contents. β-KO mice had 13% higher liver TC than the controls, while there was no difference in the liver TG content (Supplementary Fig. 4A and B). β-KO mice had 21% higher mRNA expression of Ldlr than the controls, while β-KO mice had 28% lower mRNA expression of Pcsk9 than the controls (Supplementary Fig. 4C). Although β-KO mice had lower mRNA expression of Hmgcr than the controls, no significant difference was detectable in the protein expression of HMGCR between the two groups (Supplementary Fig. 4D and E).

To determine how insulin secretion was impaired in β-KO mice, we first examined pancreatic sections. The number and size of islets in β-KO mice were both significantly decreased (Fig. 5A). Morphometric analyses of pancreatic sections showed that β-KO mice had 46% fewer islets (Fig. 5B) and 65% less islet area per pancreas area (Fig. 5C). The relative mass of β-cells and α-cells was quantified by estimating the areas positive for insulin and glucagon, respectively, per total pancreatic areas in pancreatic sections. The β-cell areas of β-KO mice were reduced by 83% compared with those of the controls (Fig. 5D), while the α-cell areas were not changed (Fig. 5E). Moreover, the islets from β-KO mice lost their typical spherical shape, and the islet cores contained α-cells intermingled with β-cells (Fig. 6A).

To determine how the deletion of Hmgcr decreased the β-cell mass, we performed immunostaining for Ki67-positive and TUNEL-positive cells to estimate the frequencies of proliferation and apoptosis of β-cells, respectively, at P0 and P9. Although insulin and Ki67 doubly positive cells were not detectable at P0, the percentage of insulin and Ki67 doubly positive cells in β-KO mice was lower than the controls at P9 (Fig. 6B and C). Thus, we considered that the reduction of β-cell mass was related to the cell proliferation disorder immediately after birth. By contrast, virtually cells doubly positive for both insulin and TUNEL were not detected in control and β-KO mice (data not shown), indicating that change in cell death made negligible contributions to the reduction of β-cell mass.

To determine when the β-KO mice developed morphological changes of pancreatic islets before weaning, we examined the histology of the pancreatic islets at P0, P9, and P14. The islets from β-KO mice showed loss of typical structure characteristic of normal islets at P9 but not at P0 (Fig. 6D), suggesting that the mass of β-cells was reduced immediately after birth. Cells doubly positive for both insulin and glucagon, which were not detected in 5-week-old β-KO mice (Fig. 6A), were detected in P9 and P14 β-KO mice (Fig. 6D and E), suggesting that the transdifferentiation from β-cells to α-cells also may contribute to the reduction of β-cell mass, at least around the perinatal period.

The insulin contents of β-KO islets were consistently and significantly lower than those of control islets (Fig. 7A and Supplementary Fig. 5A and B). To evaluate the effects of Hmgcr deficiency on insulin secretion, we examined GSIS from islets. The basal insulin secretion (at 2.8 mmol/L glucose) was not different between the two groups, but GSIS (at 20 mmol/L glucose) was severely impaired in β-KO islets (Fig. 7B). Similarly, glucose-induced [Ca²⁺]i was impaired by 20 mmol/L glucose in β-KO islets (Fig. 7C and D).

To further characterize the changes in β-KO islets, we measured mRNA expression of genes specific to β-cells in islets isolated from 5-week-old mice by real-time–qPCR (Fig. 7E). Expression of insulin 2 (Ins2) was significantly decreased in the β-KO islets. Expression of transcription factors important for the maintenance of β-cell differentiation, such as Pdx1, neurogenic differentiation 1 (Neurod1), and Mafa were also remarkably decreased in β-KO islets. On the other hand, the expression level of Ngn3, which is transiently expressed in endocrine progenitors of the embryonic pancreas, was not decreased in β-KO islets. Expression of MAF BZIP transcription factor B (Mafb) and Aristaless-related homeobox (Arx), which are critical for the development of the α-cell, was not different between the two groups (Fig. 7F). The TC content of β-KO islets was reduced because of HMGCR deficiency (Fig. 7G). The mRNA expression of sterol regulatory element binding transcription factor 2 (Srebf2), which positively regulates the expression of genes involved in cholesterol biosynthesis, LDL receptor (Ldlr), which is a target gene of Srebf2, ATP-binding cassette transporter subfamily A Member 1 (Abca1), ATP-binding cassette subfamily G Member 1 (Abcg1), both of which mediate the efflux of cholesterol, nuclear receptor subfamily 1 group H Member 3 (Nr1h3), and nuclear receptor subfamily 1 group H Member 2 (Nr1h2), both of which regulate Abca1 and Abcg1, was decreased, while the mRNA expression of HMG-CoA synthase (Hmgcs) and Farnesyl-diphosphate farnesyltransferase 1 (Fdft1) was not changed (Fig. 7H).

In this study, we showed that deletion of Hmgcr in β-cells caused overt diabetes as early as P9 in mice by markedly reducing β-cell mass as well as insulin secretion from each islet. β-Cell proliferation disorder immediately after birth was considered to be the cause of the β-cell mass reduction. Transdifferentiation of β-cells to α-cells might also contribute to the β-cell mass reduction.

Complete inhibition of HMGCR by pharmacological doses of statins inhibits cell replication in many types of cells, including β-cell–like MIN6 cells (9,23). Indeed, we showed that deletion of Hmgcr in the liver leads to fatal hypoglycemia associated with severe liver dysfunction (15) and that deletion of Hmgcr in the skeletal muscles leads to rhabdomyolysis (16). Our analyses of myeloid-specific Hmgcr-KO mice also suggested that most of their myeloid cells were haploinsufficient for the Hmgcr locus (17). However, it is more likely that the β-cells in the β-KO mice were homozygous for the disrupted allele, because mice heterozygous for the disrupted Hmgcr allele did not develop hyperglycemia (Supplementary Fig. 3C). Unfortunately, the amounts of islets were too small to collect β-cells for genotyping and/or measurements of mRNA expression.

How did not the Hmgcr-deficient β-cells expand normally and secrete insulin in response to glucose? We initially hypothesized that deletion of Hmgcr stimulates the proteolytic conversion of SREBPs to active nuclear forms to compensate the deficit of cellular cholesterol. The upregulated active forms of SREBPs might suppress the biogenesis of β-cells by competing with upstream stimulatory factors (USFs) for a binding site in the promoter of PDX-1 (24). SREBPs and USFs both belong to the family of loop basic helix-loop-helix transcription factors. Indeed, β-KO mice phenotypically resemble β-cell–specific transgenic mice overexpressing Srebp-1a (25), Srebp-1c (26), or Srebp2 (27). The mice in those studies (24–26) had impaired GSIS and reduced islet mass due to decreases in the number and size of β-cells. However, this hypothesis is not supported by our finding that genes targeted by SREBP2, such as Hmgcs, Ldlr, and Fdft1, were not upregulated in the islets (Fig. 7H). On the other hand, β-KO mice had marked reduction of the mRNA expression of Abca1, Abcg1, Nr1h3, and Nr1h2. The expression of Nr1h3 might be decreased because of deficiency of insulin, as was shown in the liver (28). The decreased expression of Nr1h3 and Nr1h2, along with reduced supply of their ligands, oxysterols, due to loss of cholesterol synthesis, might contribute to the reduced expression of Abca1 and Abcg1. In this context, it is of note that β-cell–specific deletion of Abca1 caused insulinopenic diabetes (29,30).

The downstream effectors of the Hippo pathway Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ) are required for regeneration in different organs (31). It is interesting to note that YAP/TAZ are controlled by the mevalonate pathway (32). Nuclear localization of YAP/TAZ is positively regulated by geranylgeranylation of RhoA, which is a downstream pathway of HMGCR. YAP/TAZ has been implicated to expand β-cell mass without negatively affecting the differentiation or functional state of β-cells (33). Therefore, it is tempting to speculate that deletion of Hmgcr suppresses the activation of YAP/TAZ by inhibiting RhoA activation. This may restrict β-cell proliferation, which is very active during the perinatal period (34,35), because the Ins2 promoter is active as early as E11.5 in the developing pancreas (36). Further studies are clearly needed to verify this hypothesis.

After extensive proliferation of β-cells during a perinatal period, two maturation steps are required to attain normal function of β-cells. The first step occurs around P14 and regulated by transcriptional factors MafA, Pdx1, Neurod1, and Urocortin3 (Ucn3) (37–40). The second step takes place around weaning (about P25), when fat-rich milk is changed to carbohydrate-rich chow (41). The mRNA expression of MafA, Pdx1, and Neurod1 was extremely low at 5 weeks of age in β-KO mice (Fig. 7E), which is consistent with maturation defect. However, the hyperglycemia (Fig. 3F) and structural abnormality of pancreatic islets (Fig. 6D) were apparent as early as P9, which indicates that the low β-cell mass of β-KO mice resulted from an initial defect in development rather than from a defect in maturation.

Transdifferentiation of β-cells to α-cells has emerged as a novel mechanism by which certain genetic and/or environmental disorders cause β-cell failure. For example, β-cell–specific deletion of Foxo1 (42) or Pdx1 (43) leads to hyperglycemia through the reprogramming of β-cells toward α-cells. β-KO mice may lead to a reduction of β-cell mass through transdifferentiation toward other endocrine cells. In particular, cells doubly positive for insulin and glucagon were detected at P9 and P14 in β-KO mice (Fig. 6D and E). In addition, mRNA expression of Ngn3 was maintained despite the striking reduction of the expression of Ins2, Pdx1, and MafA (Fig. 7E), which is very similar to the findings in the mice with β-cell–specific deletion of Foxo1 (42). Moreover, plasma glucagon levels in the β-KO mice were three times those in the controls (Fig. 3E), and the α-cell area–to–pancreas area was not changed despite the reduction of islets (Fig. 5B and E). These findings are suggestive of transdifferentiation of β-cells to α-cells. Hyperglucagonemia, which is known to be associated with type 1 diabetes (44), in β-KO mice might result from the impaired action of insulin on α-cells (45).

It is of note that β-KO mice showed higher VO₂ and lower RQ than the controls (Fig. 2). Similar increased energy expenditure was reported in patients with insulin-deficient type 1 diabetes primarily due to hyperglucagonemia (46).

Finally, it is worth discussing the reduced mRNA expression of Hmgcr in the liver and intestine of β-KO mice (Fig. 1B), although we did not detect a parallel decrease of HMGCR protein in the liver (Supplementary Fig. 4D). Because similar decreases in Hmgcr expression in the liver were reported in hypoinsulinemic hyperglycemic rats (47,48), it is reasonable to speculate that the reduced mRNA expression of Hmgcr in the liver and intestine was a consequence of the hypoinsulinemic hyperglycemia of β-KO mice. The increase in food intake (Fig. 1D), which is usually observed in insulin-deficient diabetes in rodents (49), might also have contributed to the downregulation of the Hmgcr gene by supplying the mice with excess cholesterol. β-KO mice showed a significant reduction of plasma TG levels under a fasting but not under an ad libitum condition (Table 1). The mRNA expression of Ldlr was increased, while that of Pcsk9 was reciprocally decreased in β-KO mice compared with the controls. These changes might result from the hyperglucagonemia, which was reported to decrease plasma LDL-cholesterol by downregulating Pcsk9 expression (50). The increased liver TC content and lower plasma TG levels might result from the increased uptake of lipoproteins via the increased Ldlr, because the expression of genes involved in de novo lipogenesis and VLDL assembly were not changed (Supplementary Fig. 4C).

In conclusion, deletion of Hmgcr in β-cells caused overt diabetes as early as P9 in mice by markedly reducing β-cell mass as well as insulin secretion from each islet. The β-cell mass reduction was mainly caused by impaired proliferation of β-cells immediately after birth, and transdifferentiation of β-cells to α-cells might also contribute. These findings underscore the importance of the mevalonate pathway in the maintenance of β-cells and glucose homeostasis, particularly in the developmental phase.

Acknowledgments. The authors thank Mika Hayashi, Nozomi Takatsuto, Mihoko Sejimo (Division of Endocrinology and Metabolism, Department of Internal Medicine, Jichi Medical University), and Megumi Yatabe (Division of Histology and Cell Biology, Department of Anatomy, Jichi Medical University) for excellent technical assistance. The authors thank Professor Hitoshi Shimano (Department of Internal Medicine [Endocrinology and Metabolism], Faculty of Medicine, University of Tsukuba) for critical reading of the manuscript.

Funding. This study was supported by Japan Society for the Promotion of Science KAKENHI (grant numbers JP16K09789, JP19H03712, and JP17390266) and the Program for the Strategic Research Foundation at Private Universities (2011–2015) Cooperative Basic and Clinical Research on Circadian Medicine and Non-communicable diseases from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Duality of Interest. This study was supported by unrestricted grants from Astellas Pharma, Daiichi Sankyo Co, Shionogi Co, Boehringer Ingelheim Japan, Ono Pharma, Mitsubishi Tanabe Pharma, Takeda Pharma Co, Toyama Chemical Co, Teijin, Sumitomo Dainippon Pharma, Sanofi K.K., Novo Nordisk Pharma, MSD K.K., Pfizer Japan, Novartis Pharma, and Eli Lilly and Company. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. S.T. and A.T. designed and performed experiments, maintained mice, performed data analysis, contributed to the discussion, and wrote, reviewed, and edited the manuscript. S.N. designed the experiments, maintained the mice, performed RT-PCR, and contributed to the discussion. D.Y., T.W., A.M., M.I., H.Y., C.E., M.T., K.E., T.O., K.F., and T.Y. contributed to the discussion. K.D. measured [Ca²⁺]_i in islets and contributed to the discussion. Y.T. measured locomotor activity, VO₂ and VCO₂, and contributed to the discussion. S.I. designed the experiments, contributed to the discussion, and wrote the manuscript. S.I. is the guarantor of this work and, as such, had full access to all the data in this study and takes responsibility for the integrity of the data and the accuracy of the data analysis.