β Cell-specific deletion of guanylyl cyclase A, the receptor for atrial natriuretic peptide, accelerates obesity-induced glucose intolerance in mice

All animal studies were approved by the Animal Care and Use Commitee of Würzburg University and were in accordance with the Guide for the Care and Use of Laboratory Animals (NIH Publications No. 8023, revised 1978). To obtain mice with restricted ablation (KO) of GC-A in β-cells, mice with two floxed alleles of GC-A (GC-A^fl/fl [21]) were intercrossed with RipCre^tg mice expressing Cre recombinase in β-cells under the control of the rat insulin 2 promoter. RipCre^tg mice were a gift from Pablo Herrera, Dept. of Genetic Medicine and Development, University of Geneva, Switzerland [22, 23]. Importantly, all metabolic parameters including β-cell function and morphology in this specific RipCre^tg line are unaltered [23].

Genotypings were performed by PCR of tail tip and tissue DNA using primers GC-A-1 (5′-TCCTGTCTCCCGTGACCTTCC), GC-A-2 (5′-ATCAGAGAATAACCAGCCAGAG) and GC-A-3 (5′-GCATGTAGTTTGTAGTCTCATAC), which amplify a 186-bp fragment for the GC-A (Npr1) wild-type allele, a 343-bp fragment for the floxed allele, and a ~ 700 bp fragment for the knockout allele [21]. Presence of the Rip-Cre transgene was detected by PCR using primers Cre1 (5′-GCTGCCACGACCAAGTGACAGCAA) and Cre2 (5′-GTAGTTATTCGGATCATCAGCTACAC) (400-bp band). The hereby generated β GC-A KO mice (GC-A^fl/fl; RipCre^tg) and their littermate controls (GC-A^fl/fl) were kept under a 12-h light/12-h dark cycle at constant temperature (23 °C) with unlimited access to food and water. All studies were performed with control and KO littermates of mixed background (C57BL/6J and 129SV).

For studies of diet-induced obesity a 60% high-fat diet (HFD, D12492, Research Diets) or the corresponding normal diet with 10% kcal from fat (ND, D12450B, Research Diets) were provided to 5 weeks old male β GC-A KO and control littermates and maintained for 18 weeks [14]. At the defined time points, mice were euthanized under deep (3%) isoflurane anesthesia, and pancreata were processed for histology or dissection of islets.

All in vivo studies were performed between 0800 and 1000 h. Samples for determinations of blood glucose (Accu-check Mobile, Roche Diagnostics, Mannheim, Germany) and plasma insulin (Ultra Sensitive Mouse Insulin ELISA, Crystal Chem, Downers Grove, IL) were obtained from the tail vein [14]. For the oral (oGTT) or intraperitoneal (IP) glucose tolerance tests, mice were previously fasted for ~ 16 h. Mice received 2 g of glucose/kg body weight (BW) in a volume of 10 μl/g BW and blood glucose concentrations were measured at 0, 15, 30, 60, 90 and 120 min [14]. In oGTTs, the samples obtained at 0, 15 and 30 min additionally served to determine basal and glucose-stimulated plasma insulin levels (GSIS). For the insulin tolerance test (ITT), fed mice were starved 4 h before the experiment. Insulin (1 IU/g BW; Actrapid Penfill Insulin human, Novo Nordisk, Bagsværd, Dänemark) was injected intraperitoneally, and blood glucose concentrations were measured at 0, 15, 30 and 60 min. For glucose tolerance (GTT) and sensitivity (ITT), we measured the total area under the curve (AUC). The AUC was calculated from the baseline (0 min) and divided by the period of time [14].

Arterial blood pressure levels of mice on ND or HFD were determined by tail-cuff plethysmography [21]. Mice were conditioned by placing them in the holding devices on three consecutive days. Subsequently, blood pressure values per mouse were determined during 15 min each day and averaged for a total of 3 subsequent consecutive days.

Mouse BNP (Bachem) was infused via subcutaneously implanted osmotic minipumps (model 1007D; Alzet Corporation) at a dose of 2 ng/h/g BW over 7 days [9]. The control group received pumps containing saline only.

Pancreata were fixed with 4% paraformaldehyde, embedded in paraffin and cut in 8-μm sections. Insulin and glucagon stainings were performed with antibodies from Dako (Santa Clara, CA, USA; AB-1001362) and Sigma (Darmstadt, Germany; AB-259852), respectively [14]. Secondary antibodies (Jackson Immuno Research, West Grove, PA, USA) were Alexa Fluor 488 or CY3 coupled and were detected by immunofluorescence. DAPI was used to label nuclei (Dianova). Pictures of three whole sections of the pancreatic corpus per mouse were taken on an Olympus microscope. Fluorescent areas were measured with ImageJ software. The total islet area, area of insulin-positive β-cells and number of β-cells/islet were calculated [14]. The experimentators were blinded regarding genotype and treatment.

Pancreatic islets were isolated by collagenase digestion and cultured overnight in RPMI 1640 medium (Gibco) containing 10% FCS [14]. Thereafter, islets were incubated at 37 °C in groups of 35 islets (for cGMP) or 10 islets (for insulin) in a buffer solution containing (in mM): 120 NaCl, 25 NaHCO₃, 5 KCl, 2.5 CaCl₂, 1 MgCl₂, and 3 d-glucose (final pH 7.35). After 2 h, the solution was replaced by fresh solution containing 6 mM d-glucose with or without ANP and the islets were incubated for additional 15 min (for cGMP responses) or 60 min (for insulin secretion). For determination of ANP/cGMP effects, the islets were pretreated with the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (0.5 μM IBMX (Sigma), 10 min). cGMP production was stopped by placing islets on ice; after centrifugation (100g, 5 min, 4 °C), the islets were lysed with ice-cold 70% (v/v) ethanol. cGMP was determined by radioimmunoassay (RIA) and normalized to sample protein contents (Bradford) [14, 21]. To evaluate the insulinotropic effects of ANP, insulin concentrations in supernatants were determined by ELISA [14]. To study the effects of ANP on mRNA expression levels of cyclin-D2, islets were serum-starved (1% FCS) for 24 h and thereafter ANP was added for another 24 h. The supernatants were discarded and the islets (40 islets per condition) were lysed in TRIzol reagent (Invitrogen).

Total RNA was isolated from cultured or freshly picked mouse islets and cardiac left ventricles using TRIzol reagent (Life Technologies). After reverse transcription (Transcriptor First Strand cDNA Synthesis Kit, Roche), real-time RT-PCR was performed using a LightCycler Instrument (Roche) [24]. The following primers and probes from Roche were used: for GC-A, TGGAGACACAGTCAACACAGC (forward primer); CCGAAGACAAGTGGATCCTG (reverse primer) and probe #71 (cat. no. 04688945001); for BNP, AAG CTG CTG GAG CTG ATA AGA (forward); GTT ACA GCC AAA CGA CTG AC (reverse), with SybrGreen; for cyclin-D2, GCT GTG CAT TTA CAC CGA CA (forward), ACA CTA CCA GTT CCC ACT CCA (reverse) and probe #45 (cat. no. 04688058001). β2 Microglobulin (cat. no. 301208, FAM, Probe 117) or GAPDH (cat. no. 05046211001, Yellow) served as reference genes [24].

Plasma BNP levels were measured by enzyme immunoassay according to the manufacturer’s protocol (MyBioSource, Inc., San Diego, CA, USA). BNP was measured instead of ANP because circulating BNP levels showed better correlation with cardiac changes [3].

GC-A^fl/fl mice with and without the Rip-Cre^tg were born in the expected Mendelian and sex ratios. PCR analysis of genomic DNA demonstrated that Cre-mediated complete recombination of the floxed GC-A gene only occurred in pancreatic islets (Fig. 1a). No deletion was detected in white adipose tissue, skeletal muscle, heart (Fig. 1a) or other tissues of the doubly transgenic (GC-A^fl/fl;Rip-Cre) mice. To characterize the impact of this gene recombination event we studied GC-A mRNA expression and receptor function in islets isolated and cultured from both genotypes, ex vivo. Figure 1b shows that GC-A mRNA expression levels in islets prepared from GC-A^fl/fl;Rip-Cre^tg mice were reduced by ~ 70% (as compared to islets from their GC-A^fl/fl littermates). As sensitive assay of ANP/GC-A signaling, we studied cGMP responses of isolated islets to ANP. As shown in Fig. 1c, ANP increased the cGMP contents of control islets (prepared from GC-A^fl/fl mice) in a concentration-dependent manner. We then compared the responses of islets from GC-A^fl/fl and GC-A^fl/fl; Rip-Cre^tg littermates to a maximal ANP concentration (100 nM). As shown in Fig. 1d, the stimulatory effects of ANP on islets cGMP contents were markedly reduced in islets prepared from the double (GC-A^fl/fl; Rip-Cre) transgenic mice as compared to islets from GC-A^fl/fl littermates. It is not surprising that the GC-A mRNA expression and cGMP activity were not fully abolished in the islets from the GC-A^fl/fl; RipCre^tg mice since the GC-A receptor is ablated in β-cells but preserved in other cell types of the islets. In particular, capillary endothelial cells have high GC-A expression levels [6, 24]. To overcome this limitation and dissect specific effects of ANP on β-cell function, we studied islet’s insulin release [14, 15]. As illustrated in Fig. 1e, ANP (10 nM during 1 h) enhanced glucose-dependent insulin secretion in islets isolated from control mice (GC-A^fl/f) but not in islets obtained from their GC-A^fl/fl; RipCre^tg littermates. Together, these experiments demonstrate the efficient inactivation of GC-A in β-cells from GC-A^fl/fl; RipCre^tg mice. Even more, whereas global GC-A KO mice are hypertensive [6], GC-A^fl/fl; RipCre^tg mice have unaltered arterial blood pressure (Fig. 1f), confirming that apart from β-cells, the effects of ANP/BNP on other target cells are preserved. Hence, we refer such mice as β-cell specific GC-A knockout (β GC-A KO) mice. Their GC-A^flox/flox littermates without Cre were used as control mice in all subsequent experiments.

Since exogenous ANP enhanced glucose-dependent insulin secretion from control but not from β GC-A KO islets in vitro, we expected that β GC-A KO mice (in vivo) would display diminished insulin levels and impaired glucose tolerance. However, fasting or random fed plasma insulin concentrations and oral glucose-stimulated insulin levels (GSIS) were not different between genotypes (Fig. 2a, b). Consistently, the oGTT and ipGTT revealed that glucose handling was similar in β GC-A KO mice as compared to their control littermates (Fig. 2c, d). Together these studies demonstrate that the endogenous cardiac NP/GC-A signaling pathway does not critically modulate resting or glucose-stimulated secretory functions of β-cells under physiological, normal dietary conditions in vivo.

Infusion of ANP increases circulating insulin in human probands [17]. To dissect whether this effect is mediated by the GC-A receptor on β-cells, we compared the actions of infused BNP on plasma insulin levels in control and β GC-A KO mice. We used BNP instead of ANP because it has almost the same affinity for GC-A as ANP, but higher stability and plasma half life [6, 9]. Osmotic minipumps filled with vehicle or synthetic BNP (to deliver 2 ng BNP/h/g BW [9]) were implanted subcutaneously for 7 days. It was previously shown that this protocol results in a 2-fold increase of baseline BNP plasma levels [9]. In the control mice, fasted (basal) and oral glucose-stimulated plasma insulin levels were mildly but significantly enhanced by BNP (Fig. 3a). This was accompanied by a mild decrease in fasted plasma glucose levels and slightly improved oral glucose tolerance (Fig. 3a, bottom). Notably, such effects of infused, exogenous BNP on plasma insulin levels and oral glucose tolerance were fully abolished in the β GC-A KO mice (Fig. 3b). These results extend the observations in isolated islets showing that elevated NP levels can enhance glucose-dependent insulin release. They demonstrate that this effect is mediated by GC-A signaling in β-cells.

To study whether endogenous NP/GC-A signaling improves β-cell functions under pathological conditions such as obesity-induced insulin resistance, control and β GC-A KO mice were fed a HFD starting at 5 weeks of age. Compared to control littermates, β GC-A KO mice had similar body weight gain (Fig. 4a) and food consumption (not shown). After 15 weeks of HFD the mean BWs were 33 ± 1.8 g (control mice) and 32.6 ± 1.8 g (KO mice), representing a weight gain of ~ 23% as compared to mice under ND. In both genotypes, obesity was associated with mild but significant increases of systolic and diastolic blood pressure levels (Fig. 4b). Notably cardiac left ventricular BNP mRNA expression levels were increased under HFD (Fig. 4c). Despite, plasma BNP levels were mildly but significantly decreased under HFD (Fig. 4d). Although the circulating BNP levels were similar in both genotypes (Fig. 4d), the cGMP contents of islets frehsly prepared from the β GC-A KO mice were markedly diminished as compared to controls (Fig. 4e). This corroborates abolished GC-A signaling in β-cells.

After 4 weeks on HFD, fasted blood glucose levels and glucose tolerance (oGTT) were still unaltered (not shown). Notably, after 8 weeks on HFD, only the β GC-A KO mice had significantly enhanced fasted glucose levels (Fig. 5a). The oGTT revealed that glucose tolerance was only mildly changed in the control mice but markedly impaired in their KO littermates on HFD (Fig. 5b). Concordantly, the AUC for the oGTT was significantly greater for the β GC-A KO mice compared to control mice on HFD or to KO mice on ND (Fig. 5c). These genotype-dependent differences vanished after 12 and 18 weeks on HFD, with both genotypes now showing similar pathological oGTTs (Fig. 5c). In agreement with a selective abrogation of NP/GC-A signaling in β-cells, the ITT showed similar insulin sensitivities in control and β GC-A KO mice on HFD (Fig. 5d). Figure 5e illustrates that the fasted (basal) and glucose-stimulated plasma insulin levels (GSIS) were markedly higher under HFD (right panel) as compared to ND (left panel), corroborating HFD-induced insulin resistance. However, surprisingly we did not observe genotype-dependent differences.

Previous studies showed that ANP enhances β-cell cyclin-D2 expression and proliferation of cultured β-cells in vitro [16]. Accordingly, in the present studies ANP (1–100 nM, 24 h) evoked significant and concentration-dependent increases of cyclin-D2 mRNA expression (as marker for β-cell proliferation) in islets isolated from control mice (Fig. 6a). We then compared the responses of islets from control and β GC-A KO littermates to the threshold ANP concentration (1 nM). As shown in Fig. 6b, the stimulatory effect of ANP on cyclin-D2 expression was abolished in islets prepared from the KO mice and is therefore mediated by the GC-A receptor.

To follow the hypothesis that the NP/GC-A system contributes to the initial adaptative increase in β-cell mass which occurs in clinical and experimental type 2 diabetes [25, 26], we performed studies of islets morphology and gene expression in control and β GC-A KO mice after 8 weeks of HFD or the corresponding ND. Morphometrical analyses of pancreatic sections stained for insulin, glucagon and DAPI demonstrated that in control mice a HFD provoked significant increases of islet areas, i.e. of the area occupied by β-cells, which was due to an increase in their numbers (Fig. 6c). Notably, this adaptative β-cell proliferation was attenuated in β GC-A KO littermates on a HFD (Fig. 6c). Concordantly, quantitative RT-PCR revealed that the expression of cyclin-D2 was enhanced in islets from HFD-fed control mice, although the difference to ND was not statistically significant (P = 0.09; Fig. 6d). Notably, cyclin-D2 levels in islets prepared from HFD-fed β GC-A KO mice were clearly and significantly lower in comparison to islets from HFD-fed control littermates (Fig. 6d). Together these observations indicate that NPs support adaptative β-cell proliferation during early stages of increased insulin demand, here provoked by HFD as experimental model of type 2 diabetes.

As illustrated, ANP enhanced glucose-dependent insulin release from control but not from β GC-A KO islets, which is consistant with other published in vitro studies [14, 15]. However, under baseline, physiological conditions glucose-induced insulin secretion (GSIS) and oral or IP glucose tolerance were not different between β GC-A KO and control littermates (in vivo). The fact that oral and IP glucose had similar effects on plasma glucose and insulin levels in both genotypes largely excludes the possibility that the endogenous cardiac NPs, at physiological plasma levels, augment stimulus-secretion coupling or the modulatory effects of incretines such as GLP-1 in vivo, for instance after a meal. Also, β-cell size and numbers were not different between the two genotypes. Hence, although β-cells express GC-A receptors [14‐16; present study], NP/GC-A signaling is dispensable for baseline or postprandial insulin release under resting, physiological conditions.

In our pharmacological studies, a low dose of BNP infused during 7 days, resulting in a ~ 2-fold increase of plasma BNP levels [9], mildly but significantly increased baseline and glucose-stimulated insulin levels and improved oral glucose tolerance in control mice. These effects of exogenous BNP were abolished in β GC-A KO littermates demonstrating that they were mediated by β-cell GC-A signaling. Notably human studies have shown an association of a single nucleotide polymorphism rs5068 on the natriuretic peptide precursor A (NPPA) locus gene with higher levels of circulating ANP and lower likelihood of incident diabetes [13]. Our data indicate that this inverse correlation might be partly due to protective effects of sustained elevations of endogenous NP on β-cell functions.

In general, apart from their diverse acute cardiovascular and metabolic effects, NPs exert chronic effects on cell growth and proliferation. They inhibit cardiac myocyte hypertrophy and fibroblast proliferation, but stimulate the proliferation of vascular endothelial cells or renal podocytes (all reviewed in [6]). These studies also demonstrated that such effects of NPs are not relevant under baseline, physiological conditions but markedly impact cardiac remodeling, angiogenesis or podocyte regeneration under disease conditions [32‐34]. As already mentioned, it was also shown that ANP can stimulate the proliferation of cultured rat pancreatic islets β-cells and insulinoma cells (INS-1E) via GC-A/cGMP-dependent activation of phosphatidylinositol 3′-kinase and Akt/Foxo1a/cyclin-D2 signaling [16]. In line with these observations, ANP enhanced cyclin-D2 mRNA expression levels in islets isolated from control but not from β GC-A KO mice (present studies). It is well known that under physiological conditions, β-cells barely proliferate. However, in initial stages of clinical or experimental prediabetes type 2, e.g. in obesity, the endocrine pancreas compensates for insulin resistance by increasing β-cell mass and insulin secretion [25, 26, 35]. Later on β-cell failure caused by a loss of cells and reduced insulin secretion from remaining individual β-cells leads to overt diabetes [35]. To study whether in this situation the endogenous NP/GC-A system modulates β-cell proliferation and function, control and β GC-A KO littermates were fed a HFD, a condition known to provoke obesity, arterial hypertension, insulin resistance and enhanced β-cell proliferation in mice [8, 35, 36]. Control and β GC-A KO mice showed similar weight gain, insulin sensitivity (ITT) and arterial hypertension under HFD. Cardiac BNP mRNA expression levels were doubled, possibly in response to increased cardiac afterload [3]. However, circulating BNP levels were lower under HFD as compared to ND, which is consistant with many published clinical and experimental studies showing that obese and type-2 diabetics display reduced circulating NP levels [5]. This has been linked to enhanced expression of the natriuretic peptide clearance receptor, NPR-C, in adipocytes [5, 9, 37]. Importantly, while these systemic effects of HFD (on blood pressure and BNP expression and circulation) were similar in control and β GC-A KO mice, islets freshly prepared from the later mice had markedly reduced cGMP contents, which is consistant with local ablation of the GC-A receptor.

In control mice, the size of the endocrine islets, the total number of β-cells per islet as well as islet cyclin-D2 expression levels were already increased after 2 months of high-fat feeding. Notably, these adaptative morphological and molecular changes were absent in β GC-A KO mice. Accordingly, these mice showed accelerated glucose intolerance already at 2 months of HFD (shown by oGTT), although plasma insulin levels apparently were unaltered. The mismatch between the increased glucose excursion during the oGTT in β GC-A KO mice and similar glucose-induced insulin levels (GSIS) in comparison to control littermates suggests that islet β-cells lacking GC-A are not able to compensate sufficiently for an increased glucose excursion during the development of an increased insulin demand (here provoked by HFD). Further studies are needed to clarify the exact mechanism.

The molecular pathways that are activated and drive the increases in β-cell mass in obesity and insulin resistance have yet to be fully elucidated [38]. High glucose levels and incretine hormones such as GLP-1 might be involved [38, 39]. Our data suggest that cardiac NPs, via GC-A/cGMP signaling, support these pathways and thereby the adaptative proliferation of β-cells during early stages of obesity-linked insulin resistance and metabolic syndrome.

Our study harbors three limitations: