Optimal Elevation of β-Cell 11β-Hydroxysteroid Dehydrogenase Type 1 Is a Compensatory Mechanism That Prevents High-Fat Diet–Induced β-Cell Failure

All experiments conformed to local ethical guidelines and the Home Office (U.K.) Animals Scientific Procedures Act (1986). Male C57BL/KsJ, C57BL/6J (The Jackson Laboratory, Bar Harbor, ME), BKS.Cg+Leprdb/+Leprdb/OlaHsd (Lep^db/db) (Harlan, Bicester, U.K.), and inbred C57BL/6J-11β-HSD1^−/− mice were housed as previously described (12). Mice were fed an HF diet (D12331; Research Diets, New Brunswick, NJ) or control diet (CD) (D12328) for 12 weeks from 4 weeks of age.

An 8.3-kb (HindIII [−8280]/NotI [+12]) fragment of the mouse insulin gene promoter I (MIP) (gift of Mark Magnusson, Vanderbilt University, Nashville, TN) (10) and the 1.265-bp EcoRI fragment of the rat 11β-HSD1 (rHSD1) were cloned into pGEM11Zf(+) to create the MIP-rHSD1 transgene. The MIP-rHSD1 (10.2 kb) transgene was removed with HindIII/SfiI and microinjected into the pronuclei of fertilized oocytes from C57BL/KsJ mice by standard techniques (Genetic Intervention and Screening Technologies, University of Edinburgh). The transgene was detected by PCR on tail genomic DNA using primer sequences: MIP-rHSD1, 5′-GGAACTGTGAAACAGTCCAAGG-3′ and 5′-TTTGCTGGCCCCAGTGACAAGCTTT-3′; HSD1, 5′-AAAGCTTGTCACTGGGGCCAGCAAA-3′ and 5′-AGGATCCAGAGCAAACTTGCTTGCA-3′; and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′GCCAAGGTCATCCATGACAAC and 5′-AGTGTAGCCCAAGATGCCCTT-3′. The mice were genotyped by Southern blotting analysis of genomic DNA using AflIII digestion and ^[α-32P]d-CTP–labeled rHSD1 cDNA. Transgenic heterozygotes MIP-HSD1^tg/+ were maintained on the C57BL/KsJ genetic background, and littermate wild-type C57BL/KsJ mice were used as controls.

11β-HSD1 activity was tested in rat INS1E cells (a gift of C. Wollheim, University of Geneva, Geneva, Switzerland) and mouse MIN6 cells (a gift of G. Rutter, Imperial College London, London, U.K.) 48 h after MIP-rHSD1 lipofection (Invitrogen, Paisley, U.K.).

Pancreata were digested with collagenase XI (Sigma-Aldrich, Dorset, U.K.), and islets were hand-picked under a stereomicroscope in Hanks’ balanced salt solution and 10% FBS (Lonza, Berkshire, U.K.). For 11β-HSD1 activity assay (13), batches of 100 islets/cell lines were incubated in RPMI 1640 medium, 10% FBS, and 2.8, 16.8, or 5.5 mmol/L glucose for 24 h. For RNA extraction, islets were directly homogenized in TRIzol reagent (Life Technologies, Paisley, U.K.). For insulin secretion studies, batches of 10 islets were allowed to rest (2 h) in basal Krebs (1% weight for volume [w/v] BSA fraction V, 2.8 mmol/L glucose) on 8-μm inserts (Millipore, Billerica, MA) suspended in 12-well culture dishes. Samples were collected every 5 or 10 min for discrimination of the basal and glucose-stimulated (16.8 mmol/L glucose) insulin secretion (GSIS). A return to 2.8 mmol/L verified specificity of the response. For membrane-proximal secretion analysis, islets were preincubated (2 h) with 0.5 μg/ml brefeldin A (BFA) and 15 μmol/L cycloheximide (CHX) (Applichem, Darmstadt, Germany) and then incubated for 30 min in basal Krebs. For potassium-stimulated insulin secretion (PSIS), islets were preincubated (2 h) in basal Krebs for 30 min and then for 30 min with 40 mmol/L KCl. Acute (2 h) steroid effects (20 or 200 nmol/L 11-dehydrocorticosterone [11-DHC]) were investigated using Krebs and the 11β-HSD1 inhibitor UE2316 (1 μmol/L) or the glucocorticoid receptor (GR) antagonist RU486 (1 μmol/L). Fatty acid effects were investigated using RPMI 1640 medium, 5.5 mmol/L glucose, 10% FBS, 1% BSA, 20 nmol 11-DHC, with or without 0.5 mmol/L palmitate (22 h) followed by equilibration in Krebs (2 h) and then 2.8 mmol/L (30 min) and 16.8 mmol/L (30 min) incubations. Insulin was measured by radioimmunoassay (Linco Research, St. Charles, MO).

Isolated islet total RNA was purified using PureLink RNA micro kit (Invitrogen, Paisley, U.K.). Random-primed cDNA was synthesized using Superscript III reverse transcriptase (Invitrogen). For real-time PCR, cDNA was mixed with light cycler 480 probes master mix (Roche Diagnostics, Indianapolis, IN) and probes of interest: mouse β-actin, 4352933E; Cdkn1a, Mm00432448_1; Prkacb, Mm00440840_1; Braf, Mm01165837_1; Hspa1b, Mm03038954_s1; HSD11b1, Mm00476182-m1; TBP, Mm00446973-m1 (Applied Biosystems, Warrington, U.K.).

Total islet RNA was processed through standard Affymetrix Mouse Genome 430-2.0 GeneChip protocols (ARK Genomics, Roslin Institute, Edinburgh, U.K.). Data were extracted through the GCOS software, and CEL files were imported into Bioconductor, normalized by RMA in the Affy module, and statistically analyzed with the Limma and Rank products (RankProd) packages. Gene ontology and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis was done with the DAVID tool. Data are available in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE23161.

Mice were anesthetized by intraperitoneal injection of 1 mg/mL medetomidine (Domitor) and 100 mg/mL ketamine (Vetalar) (0.1 mL/10 g body weight). The right jugular vein was cannulated with microrenathane tubing (0.25 inch outer diameter × 0.012 inch inner diameter; Sandown Scientific, Hampton, U.K.). The cannula was exteriorized on the back of the neck and mice were allowed to recover for 2 days. Mice were fasted for 4 h before saline (0.2 μl/min, 2 h) and then glucose (25% w/v) was infused (10 μl/min/30 g body weight) into the mice (CMA/102 microdialysis pump; CMA Microdialysis, Solna, Sweden). Glucose infusion was stopped at 60 min. At 0, 1, 3, 5, 10, 15, 30, 60, and 120 min, blood glucose was measured using a OneTouch Ultra (Johnson and Johnson, Bucks, U.K.). Insulin was measured by ELISA (CrystalChem, Chicago, IL).

Mice were fasted for 6 h and then injected intraperitoneally with 2 mg/g body wt. d-glucose (25% v/w). Blood was sampled at 0, 2, 5, 10, 15, 30, 60, and 120-min intervals after glucose bolus. Blood glucose was determined using Infinity reagent (Thermotrace, Leeds, U.K.) and insulin by ELISA (CrystalChem).

Singly housed mice were sampled by venesection at 8:00 a.m. within 1 min of cage disturbance. Plasma corticosterone levels were measured by radioimmunoassay as previously described (12).

Triglyceride (Infinity; Thermo Fisher Scientific, Middletown, VA) and nonesterified fatty acid (NEFA) (Wako, Richmond, VA) were measured in plasma of fed or 6-h fasted mice after tail venesection.

Batches of 60 islets were incubated in RPMI 1640 medium (11.1 mmol/L glucose) and 10% FBS for 24 h with or without 1 μmol/L UE2316 11β-HSD1 inhibitor or 5 μmol/L KT5720 cAMP-dependent protein kinase (PKA) phosphorylation inhibitor (Sigma-Aldrich). Islets were homogenized in lysis buffer (50 mmol/L Tris, pH 7.4, 0.27 mol/L sucrose, 1 mmol/L sodium orthovanadate, pH 10, 1 mmol/L EDTA, 1 mmol/L EGTA, 10 mmol/L sodium β-glycerophosphate, 50 mmol/L NaF, 5 mmol/L sodium pyrophosphate, 1% [w/v] Triton X-100, 0.1% [v/v] 2-mercaptoethanol, one tablet of complete TM protease inhibitor [Roche, Burgess Hill, U.K.]) and subjected to Western blotting. Primary antibodies were anti-PKAβcat (1:1,000) (Santa Cruz Biotechnology, Santa Cruz, CA), anti–phospho-ERK thr²⁰² tyr²⁰⁴ (1:1,000), anti-ERK1/2 native (1:1,000), and anti–β-actin (1:1,000) (Cell Signaling Technologies, Beverley, U.K.). HRP–anti-rabbit IgG (1:1,000) was used as a secondary antibody with enhanced chemiluminescence (Amersham, Little Chalfont, U.K.) and exposure to Amersham HyperfilmTH ECL autoradiographic film (Amersham). Blots were quantified using ImageJ software (www.ncbi.nlm.nih.gov).

Optical projection tomography (OPT) was performed on whole pancreata (14) using an OPT scanner (Bioptonics 3001; Skyscan, Kontich, Belgium) and images were analyzed with Data Viewer version 1.3.2 (Bioptonics).

Whole pancreata were fixed (4% paraformaldehyde), wax embedded, and sectioned (5 μm) before being immunostained with sheep anti-HSD1 antibody (13). For chromogen labeling with diaminobenzidine (DakoCytomation, Inc, Carpinteria, CA), biotinylated anti-sheep (Abcam, plc, Cambridge, U.K.) was used as secondary antibodies. Images were processed using KS3000 software (version 3.0; Carl Zeiss Vision, GmbH).

Frozen sections (5 μm) of whole pancreata or single cells from trypsin-digested islets were fixed in acetone:methanol (90:10 v/v) and immunostained using guinea pig anti-insulin, sheep anti–11β-HSD1 antibody (13), rabbit anti-glucagon, somatostatin, pancreatic polypeptide (Chemicon; Millipore, Billerica, MA), and Cy2-conjugated AffiniPure F(ab’)₂ fragment donkey anti–guinea pig IgG, rhodamine red X–conjugated AffiniPure F(ab’)₂ fragment donkey anti-sheep, fluorescein isothiocyanate–conjugated AffiniPure F(ab’)₂ fragment donkey anti-sheep, and rhodamine red X–conjugated AffiniPure F(ab’)₂ fragment donkey anti-rabbit (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) were used as secondary antibodies. Images were taken using a Leica confocal microscope linked to LSM Image Browser software.

Whole pancreata were homogenized in lysis buffer, and insulin was extracted with acid ethanol and measured by radioimmunoassay (Linco Research). Insulin was normalized to total protein content (Bio-Rad Laboratories, Hercules, CA).

Data are expressed as mean ± SEM and were analyzed by unpaired Student t test, one-way ANOVA Newman-Keuls, or two-way ANOVA.

The rat 11β-HSD1 cDNA was cloned downstream of the mouse insulin I promoter (Fig. 1A) that drives highly β-cell–specific expression (10,15). The construct was validated by demonstrating increased 11β-HSD1 activity after transfection into clonal INS1E and MIN6 cells (Supplementary Fig. 1A and B). Transgenic mice were generated on the β-cell, failure-susceptible C57BL/KsJ (KsJ) strain background (11). Southern blotting suggested integration of a low copy number (approximately two) of the transgene (Fig. 1B). In MIP-HSD1^tg/+ islets, 11β-HSD1 mRNA was increased eightfold (Fig. 1C). 11β-HSD1 protein was elevated specifically in β-cells of MIP-HSD1 transgenics (Fig. 1D and Supplementary Fig. 2). Note that 11β-HSD1 is expressed in β-cells of normal islets, as shown by the colocalization with insulin-positive cells and in other islet cell subtypes (Supplementary Fig. 3), as found by others (16). 11β-HSD1 reductase (glucocorticoid reactivation) activity was approximately twofold higher in isolated primary MIP-HSD1^tg/+ islets than littermate control islets and was glucose inducible (Fig. 1E). Hexose-6-phosphate dehydrogenase level, which drives 11β-reductase activity (17), was not limiting. 11β-HSD1 dehydrogenase (glucocorticoid inactivating) activity was 10-fold lower than reductase in islets and was unaffected by transgene or high glucose (Supplementary Fig. 1C and D). MIP-HSD1^tg/+ mice were viable and appeared grossly normal (Table 1). Fasting plasma insulin, glucose triglyceride, and NEFA levels were unaffected by genotype (Table 1), indicating similar peripheral insulin sensitivity and β-cell exposure to lipotoxicity with HF.

To determine transgene impact in vivo, mice were fed an HF diet for 12 weeks to induce β-cell failure (11). Plasma corticosterone levels were not significantly affected by diet or genotype (Table 1), indicating that transgene effects were β-cell specific and not due to hypothalamic-pituitary-adrenal disturbance. To accurately assess insulin secretion, especially over the rapid response of the first phase (1–10 min), glucose was infused through a jugular cannula. CD-fed animals had comparable responses to glucose infusion (Fig. 2A and C). HF-fed KsJ mice, as expected (11), showed markedly attenuated GSIS indicative of β-cell failure and an associated prolonged elevation of plasma glucose levels in vivo (Fig. 2B and C). In contrast, HF-fed MIP-HSD1^tg/+ mice exhibited profoundly elevated first- and second-phase GSIS that completely normalized glucose levels (Fig. 2B and C).

Protective MIP-HSD1^tg/+ compensatory hyperinsulinemia can occur through two major pathways: increased β-cell number (neogenesis, β-cell replication, and reduced apoptosis) and increased β-cell function (insulin output per β-cell). To address β-cell/islet number, whole pancreas islet number was quantified by OPT (14) (Fig. 3A and B) and validated by conventional immunohistochemistry in pancreas sections (Fig. 3C). MIP-HSD1^tg/+ had more islets (Fig. 3A–C) of smaller dimensions compared with littermates on control and HF diet (Fig. 3D). Average islet size increased in KsJ mice with chronic HF, indicating selective loss of small functional islets (Fig. 3A–D), a process driven at least in part by apoptosis (18), whereas MIP-HSD1^tg/+ mice showed attenuation of β-cell loss. β-Cell replication (as indicated by Pdx1 and Ki67 costaining) was unaltered in MIP-HSD1^tg/+ transgenics (Supplementary Fig. 4). The insulin promoter is expressed at an early stage of pancreatic islet development (19) when increased glucocorticoid exposure reduces β-cell mass (20). This would predict, if anything, lower islet mass in MIP-HSD1^tg/+ mice. However, the MIP-HSD1 transgene had no discernible effect on the development of the endocrine pancreas (OPT in fetal day E18 islet bodies and newborn pancreas) (Supplementary Fig. 5), suggesting that a postnatal regulatory mechanism drives increased islet number in MIP-HSD^tg/+ mice.

To test the additional possibility of enhanced islet insulin secretory function, we measured GSIS in carefully size-matched (average diameter 100 μm) isolated islets in vitro. In agreement with the findings in vivo, isolated islets from CD-fed KsJ and MIP-HSD1^tg/+ mice showed similar GSIS (Fig. 3E, left), whereas islets from HF-fed MIP-HSD1^tg/+ mice exhibited augmented GSIS (Fig. 3E, right).

We next investigated whether the insulin secretory machinery was specifically enhanced in MIP-HSD1^tg/+ islets. To assess this, insulin translation and Golgi trafficking was inhibited with CHX and brefeldin A (BFA), revealing relatively enhanced basal (unregulated) insulin secretion from islets of MIP-HSD1^tg/+ mice on HF compared with control-fed mice (Fig. 3F). In addition, PSIS (glucose independent, membrane proximal) was significantly higher from islets of HF-fed MIP-HSD1^tg/+ compared with HF-fed KsJ mice (Fig. 3G), consistent with a larger pool of secretory granules and/or a faster rate of exocytosis in islets of MIP-HSD1^tg/+ mice on HF.

To confirm glucocorticoid and 11β-HSD1 dependence of the transgene, islets were incubated acutely (2 h) in 20 nmol/L (∼physiological) or 200 nmol/L (supraphysiological) 11-DHC. GSIS was significantly higher from islets from HF MIP-HSD1^tg/+ mice exposed to 20 nmol/L 11-DHC, and this was abolished by the 11β-HSD1 inhibitor UE2316 (Supplementary Table 1) and the GR antagonist RU486. Supraphysiological 11-DHC completely blunted GSIS, as mentioned by others (4,6,16). Inhibition was also reversed in islets of HF-fed MIP-HSD1^tg/+ mice by 11β-HSD1 and GR inhibitors.

Fatty acids augment insulin secretion in the short term. To investigate the role of elevated β-cell 11β-HSD1 in this process, palmitate-stimulated GSIS was assessed in isolated islets from MIP-HSD1^tg/+ after 24 h exposure to 20 nmol/L 11-DHC. Notably, long-term static exposure of islets to 20 nmol/L 11-DHC alone impaired insulin release in vitro (Supplementary Fig. 8). Palmitate-stimulated GSIS from MIP-HSD^tg/+ was reversed by 11β-HSD1 and GR inhibitors (Supplementary Fig. 8), suggesting a role for 11β-HSD1 in fatty acid–stimulated GSIS.

To better understand the mechanisms of improved islet function in HF-fed MIP-HSD1^tg/+ mice, we performed a comparative transcriptomic analysis (Supplementary Table 2). MIP-HSD1^tg/+ islets had higher mRNA levels for genes of the small GTPase pathway (Arf6, Prkacb, and PKA catalytic subunit β [PKAβcat]) linked to enhanced insulin secretion and islet survival (21–23), protein trafficking, turnover, and vesicle docking (VapB and C, Arf6, cplx2, and stx11), consistent with the enhanced secretory responses (Fig. 3E, right, and F and G) and the mitogen-activated protein kinase/extracellular signal–related kinase (ERK) (B-raf, Max, and Cdkn1a/p21) and Jak/Stat pathways (Ccnd2 and Jak1), indicative of maintained differentiation versus proliferation commitment (24–27) and protection from apoptosis (28,29). MIP-HSD1^tg/+ islets also exhibited higher mRNA levels of genes related to protection from cellular stress (hspa1a, -a1b, -a4, and trib3) (30). Key gene expression changes were validated by real-time PCR (Fig. 4A–D). Interestingly, Cdkn1a gene expression was downregulated in islets of KsJ mice on HF (Fig. 4D), consistent with HF-induced cellular stress, apoptosis (29), and islet loss (Fig. 3A–C), whereas Cdkn1a was upregulated by HF in MIP-HSD1^tg/+, consistent with maintained functionality with reduced cellular stress.

To test whether the validated gene expression changes had a functional impact upon islet responses, we investigated two key pathways linked to enhanced insulin secretion (PKA) and downstream cell differentiation/survival pathways (ERK). Protein levels of the catalytic subunit of PKA (PKAβcat) were markedly induced in islets from HF diet–fed MIP-HSD1^tg/+ mice (Fig. 4E), and this was reversed with the 11β-HSD1 inhibitor UE2316 (Fig. 4F) but not the PKA phosphorylation inhibitor KT5720 (Fig. 4G). Likewise, activation (phosphorylation) of the ERK signaling pathway was enhanced in MIP-HSD1^tg/+ islets (Fig. 4H). Induction of ERK activation in MIP-HSD1^tg/+ islets was abolished by UE2316 (Fig. 4I), confirming enzyme specificity. The increased ERK phosphorylation in MIP-HSD1^tg/+ islets was also abolished in the presence of the PKA inhibitor KT5720 (Fig. 4J), as expected.

We next investigated whether 11β-HSD1 deficiency impacted GSIS with HF feeding. 11β-HSD1^−/− mice exhibited a blunted early-phase GSIS and mild transient glucose intolerance in vivo (Supplementary Fig. 6A and B). Note that glucose/insulin homeostasis in 11β-HSD1^−/− mice is comparable to control C57BL/6J mice when on CD (12), indicating a β-cell–specific defect. However, insulin secretion was markedly lower in HF-fed 11β-HSD1^−/− than C57BL/6J mice in association with markedly improved glucose tolerance (Supplementary Fig. 6C and D), suggesting substantially reduced secretory demand from β-cells due to the dominant protection from glucose intolerance in HF-fed 11β-HSD1^−/− mice.

Glucocorticoids sometimes show complex dose-related effects (31) (Fig. 3H). To address this, we analyzed insulin secretion in homozygous MIP-HSD1^tg/tg mice with ∼16-fold higher islet 11β-HSD1 mRNA and ∼4-fold higher 11β-HSD1 activity (Supplementary Fig. 7A and B). In contrast to MIP-HSD1^tg/+, MIP-HSD1^tg/tg mice exhibited a significant reduction of whole pancreas insulin content and marked suppression of GSIS on normal diet in vivo, although normal glucose tolerance was maintained (Supplementary Fig. 7C, D, and F). However, MIP-HSD1^tg/tg mouse GSIS was comparable to littermates on HF diet (Supplementary Fig. 7E, left, and F, left), reflecting a relative improvement in GSIS versus CD, and this partially normalized glucose tolerance (Supplementary Fig. 7E, right, and F, right). In agreement with this intermediate phenotype, GSIS from isolated islets of HF-fed MIP-HSD1^tg/tg mice showed a higher basal and peak secretion, although this was not sustained over the whole second phase (Supplementary Fig. 7F).

Considering the inverted U-shaped MIP-HSD1 gene dose-response effect on β-cell secretory function, we assessed islet 11β-HSD1 protein and activity levels in contrasting models of β-cell failure (HF-fed KsJ and Lep^db/db mice) and β-cell compensation (C57BL/6J mice). HF induced a modest elevation of 11β-HSD1 in islets from robustly compensating C57BL/6J mice (Fig. 5A and B). In contrast, downregulation of islet 11β-HSD1 was found in β-cell failure–susceptible, HF-fed KsJ mice (Fig. 5C and D). Markedly high 11β-HSD1 levels were found in diabetic Lep^db/db islets (Fig. 5E and F).

Our data show that, for the first time, an unexpectedly beneficial metabolic outcome results from optimal elevation of β-cell 11β-HSD1, in contrast to the diabetogenic effects of elevated 11β-HSD1 in adipose tissue and liver (3). Moreover, this effect may be physiologically relevant, as normal mice that exhibit robust compensatory insulin secretion on HF exhibit a similar modest upregulation of β-cell 11β-HSD1. A higher or lower 11β-HSD1 level suppresses GSIS, consistent with an inverted U-shaped dose-response effect sometimes seen with glucocorticoids (31). This notion is supported by the stimulatory effect of physiological 11β-HSD1 substrate (11-DHC) but suppressive effects of high or indeed long-term exposure to 11-DHC in vitro. In agreement, diabetic Lep^db/db mice (32) and MIP-HSD1^tg/tg on CD have impaired GSIS in vivo associated with more marked elevation of β-cell 11β-HSD1. These data are also consistent with the association between increasing 11β-HSD1 levels and the severity of hyperglycemia in diabetic fa/fa rats (5). Conversely, 11β-HSD1^−/− mice on CD and KsJ mice on HF diet have β-cell secretory defects associated with the absence or decrease of 11β-HSD1 activity, respectively. These findings may have implications for the timing and dose of steroid or 11β-HSD1 inhibitor interventions in the therapeutic treatment of patients predisposed to β-cell failure. Defining the impact of altered islet 11β-HSD1 expression across the progression from glucose intolerance, prediabetes, and frank diabetes in both rodent models and humans, relative to peripheral tissue 11β-HSD1 levels, will be of future importance to assess any adverse risks of therapeutic inhibition. However, the dominant protection from glucose intolerance found in HF-fed 11β-HSD1^−/− mice indicates that beneficial effects of 11β-HSD1 inhibition may outweigh any detrimental effect on β-cells.

The gene-dose, inverted-U effect of the MIP-HSD1 transgene may reconcile conflicting reports of suppressive (4–8) versus stimulatory effects (33–35) of glucocorticoids on insulin secretion. Thus, the prevailing effect is likely a function of dose and duration. Indeed, although our in vitro data are generally supportive of our in vivo findings, we note that long-term, but not short-term, static incubation with low 11-DHC is suppressive. We contend that static incubation with steroids and isolated islets in vitro does not reflect dynamic glucocorticoid exposure/metabolism patterns found in vivo. This merely serves to underline the importance of studying the impact of altered β-cell 11β-HSD1 in the whole animal chronic disease context. Of note, β-cell–specific GR overexpression (RIP-GR) causes direct suppression of insulin secretion (7). Beyond exposure and dose issues, this suggests that MIP-HSD1 effects could result from additional effects beyond GR activation. For example, physiological corticosterone binds with high affinity to mineralocorticoid receptors, which antagonizes GR-mediated suppression of insulin secretion and may thus contribute to beneficial (36) and dose-sensitive effects. Notably, the synthetic GR-selective glucocorticoid dexamethasone, commonly used in in vitro islet studies, is a poor 11β-HSD1 substrate, perhaps further explaining differences between solely GR- versus 11β-HSD1–mediated effects in islets. 11β-HSD1 reductase activity depletes endoplasmic reticulum (ER) NADPH, but this does not appear to affect the increased cytosolic NADPH levels required to activate GSIS (37). Elevated 11β-HSD1 may also preferentially amplify nongenomic effects of glucocorticoids (38) more than GR overexpression, or indeed non-GR effects due to alternative substrate metabolism (39) by 11β-HSD1. Indeed, alternate 11β-HSD1 metabolites (40) may affect GSIS through oxysterol activation of nuclear liver X receptor (41). Finally, RIP-GR mice may manifest impaired GSIS as a result of developmentally impaired β-cell mass (20), unlike the MIP-HSD1 model. In any case, there is more evidence to suggest that islet 11β-HSD1, rather than GR, levels are elevated in islets from diabetic mice (5,6, and the current study), suggesting the MIP-HSD1 mice are perhaps a more relevant in vivo model. Crucially, the unexpected augmentation of GSIS in the MIP-HSD1 mice due to chronic modest elevation of β-cell 11β-HSD1 in vivo has allowed us to challenge prevailing hypotheses on glucocorticoid-mediated diabetogenic effects in islets. Those hypotheses are derived from arguably artificial systems such as systemic administration of high-dose steroids in vivo, which induces confounding peripheral insulin resistance, as well as the effects of acute and often pharmacological steroid concentrations in islets in vitro.

Elevated expression of genes involved in differentiation, insulin secretion, and cellular stress protection and survival are associated with augmented MIP-HSD1^tg/+ islet function. Although similar effects of glucocorticoids have been noted in other cellular systems, our data show these effects for the first time, to our knowledge, in pancreatic islets, and highlight novel areas of cross-talk in this cell type. Thus, glucocorticoids curtail proliferation and induce differentiation, in part through the increased cyclin-dependent kinase inhibitor 1A (p21) (24,26,27,42) and enhanced differentiation through ERK1/2 (26). Increased v-raf murine sarcoma viral oncogene homolog B1 (B-Raf) signaling (43) and enhanced PKAβcat and small GTPase activation makes a clear mechanistic link with augmented insulin secretory capacity (21,25,43) as well as islet survival in the face of HF diet–induced ER stress (23); chronic exposure to elevated NEFA was comparable between genotypes, indicating a β-cell–specific protection from lipotoxicity in the MIP-HSD1^tg/+ mice. In contrast, we find a novel permissive role for 11β-HSD1 in short-term, fatty acid–induced GSIS in vitro, suggesting a physiological role for 11β-HSD1 in meal lipid–related GSIS processes. Further, a direct interaction between GR and PKA has been proposed in other cellular systems (40,44), suggesting a novel mechanism for functional augmentation of β-cells that can now be fully explored. Heat shock protein induction is consistent with increased GR activation (45,46) and, along with p21 in islets (29), is associated with cell survival and protection from apoptosis (28,47), echoing critical processes activated in β-cells that successfully adapt to ER stress (30,45). Note the impact of elevated 11β-HSD1 on β-cell apoptosis will need careful future exploration in model systems other than the HF-fed C57BL/KsJ mice where this process is more evident and quantifiable. Notably, we did not find suppression of inflammatory processes linked to islet damage in type 2 diabetes (48), as occurs when islets are exposed to high levels of corticosterone (45), perhaps reflecting the more physiological nature of β-cell protection in our model. The data therefore give insight into a novel mechanistic framework whereby optimally elevated local glucocorticoid action may facilitate safe execution of increased secretory demand, thus avoiding β-cell failure. Of particular note, given the topical nature of replication as a mechanism for compensation (49,50), the novel protective response of MIP-HSD1^tg/+ mice does not involve increased β-cell replication. Intriguingly, despite suppressed MIP-HSD1^tg/tg islet function under basal (CD) conditions, the markedly elevated β-cell 11β-HSD1 of MIP-HSD1^tg/tg elicits an HF diet–responsive β-cell rescue response that is adequate enough to normalize glucose homeostasis. This suggests the potential for mechanistic dissection of therapeutic glucocorticoid effects from those that are undesirable.

Transgenic elevation of β-cell 11β-HSD1 has revealed a novel physiological function for the enzyme in β-cell compensation that may open up new therapeutic opportunities. More broadly, coordinated regulation of local glucocorticoid regeneration in β-cells with dynamic changes in adipose tissue delineates a responsive physiological process that shapes metabolism to protect against metabolic disease. Of more immediate medical significance, our data critically inform imminent therapeutic strategies aimed at manipulation of tissue glucocorticoid action.

These studies were funded by a project grant from Diabetes UK (05/0003083 to N.M.M., J.J.M., and J.R.S.) and supported by a British Heart Foundation Research Excellence Award (BHF REA) (RE/08/001 transition award to S.T. and N.M.M.), a British Endocrine Society Early Career Grant (S.T., N.M., and J.R.S.), a Wellcome Trust Career Development Fellowship (079660/z/06/z to N.M.M.), and a Wellcome Trust Programme Grant (J.R.S., J.J.M., and B.R.W.). D.R.D. was funded by the Wellcome Trust Cardiovascular Research Initiative and by the BHF REA. Development of the 11β-HSD1 inhibitor UE2316 was supported by a Wellcome Trust Seeding Drug Discovery Award (B.R.W., S.P.W., and J.R.S.).

S.P.W., B.R.W., J.R.S., and N.M.M. hold relevant patents on the use of 11β-HSD1 inhibitors for metabolic indications. No other potential conflicts of interest relevant to this article were reported.

S.T. conceived and designed experiments, researched data, and wrote the manuscript. X.L. and L.R. researched data. S.P.W. and B.R.W. reviewed and edited the manuscript. D.R.D. researched data and reviewed and edited the manuscript. J.J.M. reviewed and edited the manuscript. J.R.S. contributed to discussion and reviewed and edited the manuscript. N.M.M. conceived and designed the experiments, researched data, and wrote the manuscript. N.M.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Bernadette Bréant and Amrit Singh-Estivalet (Unité UMRS 872, Paris, France) for helpful advice on islet preparation.