Bone morphogenetic protein 4 inhibits insulin secretion from rodent beta cells through regulation of calbindin1 expression and reduced voltage-dependent calcium currents
- First Online:
- Cite this article as:
- Christensen, G.L., Jacobsen, M.L.B., Wendt, A. et al. Diabetologia (2015) 58: 1282. doi:10.1007/s00125-015-3568-x
- 747 Downloads
Type 2 diabetes is characterised by progressive loss of pancreatic beta cell mass and function. Therefore, it is of therapeutic interest to identify factors with the potential to improve beta cell proliferation and insulin secretion. Bone morphogenetic protein 4 (BMP4) expression is increased in diabetic animals and BMP4 reduces glucose-stimulated insulin secretion (GSIS). Here, we investigate the molecular mechanism behind this inhibition.
BMP4-mediated inhibition of GSIS was investigated in detail using single cell electrophysiological measurements and live cell Ca2+ imaging. BMP4-mediated gene expression changes were investigated by microarray profiling, quantitative PCR and western blotting.
Prolonged exposure to BMP4 reduced GSIS from rodent pancreatic islets. This inhibition was associated with decreased exocytosis due to a reduced Ca2+ current through voltage-dependent Ca2+ channels. To identify proteins involved in the inhibition of GSIS, we investigated global gene expression changes induced by BMP4 in neonatal rat pancreatic islets. Expression of the Ca2+-binding protein calbindin1 was significantly induced by BMP4. Overexpression of calbindin1 in primary islet cells reduced GSIS, and the effect of BMP4 on GSIS was lost in islets from calbindin1 (Calb1) knockout mice.
We found BMP4 treatment to markedly inhibit GSIS from rodent pancreatic islets in a calbindin1-dependent manner. Calbindin1 is suggested to mediate the effect of BMP4 by buffering Ca2+ and decreasing Ca2+ channel activity, resulting in diminished insulin exocytosis. Both BMP4 and calbindin1 are potential pharmacological targets for the treatment of beta cell dysfunction.
KeywordsBeta cells BMP4 Calb1 Calbindin1 Diabetes Exocytosis Insulin secretion
Bone morphogenetic protein
False discovery rate
Glucose-stimulated insulin secretion
Hanks’ balanced salt solution
Voltage-dependent Ca2+ channel
Pancreatic beta cells initially compensate for insulin resistance by increasing their mass and activity to maintain normal blood glucose levels . In individuals who progress to develop type 2 diabetes, beta cells eventually fail to adapt and progressive loss of functional beta cells starts . Although current type 2 diabetes drugs targeting insulin resistance or insulin secretion are initially efficient, beta cell mass and function tend to decline over time [2, 3]. Unidentified factors may limit the ability of the beta cells to adapt to insulin resistance. In our search for factors with inhibitory effects on beta cells, we recently described the inhibition of beta cell proliferation and insulin secretion by bone morphogenetic protein 4 (BMP4) .
BMPs belong to the TGF-β protein family, members of which are known to play central roles in pancreas and islet development [5, 6, 7, 8, 9, 10]. While the role of BMPs in the developing pancreas and beta cells has gained much attention [8, 11, 12], less is known about their function in the postnatal pancreas. There is increasing evidence that BMP2/4 are inflammatory markers in various tissues under diabetic conditions [13, 14, 15]. We recently showed that BMP2/4 are expressed in pancreatic islets and upregulated during diabetes progression in islets from db/db mice and by proinflammatory cytokines in vitro . Although the culture of pancreatic islets in the presence BMP2/4 negatively affects beta cell function, it appears that the effect of BMPs in vivo is more complex. Beta cell specific deletion of BMP receptor 1A (Bmpr1a) resulted in impaired glucose-stimulated insulin secretion (GSIS), whereas BMP4 transgene overexpression increased GSIS . In contrast, deficiency of inhibitor of differentiation 1 (Id1), encoding a central BMP-regulated transcription factor, resulted in enhanced insulin secretion and protection from diet-induced glucose intolerance, suggesting that BMPs and ID1 normally inhibit adult beta cell function . BMPs are both released from and affect several metabolically relevant tissues, including fat, liver and kidney, thus adding to the complexity of the role of BMPs in metabolism [14, 18, 19]. To characterise the direct effects of BMPs on beta cells, we recently reported BMP4-mediated inhibition of beta cell proliferation and repression of GSIS from mouse, rat and human islets . Here, we further characterise the effects of BMP4 on insulin secretion using single cell electrophysiological measurements, and evaluate gene regulation to identify the molecular mechanism for the observed inhibition of insulin secretion.
Rat islet isolation and culture
Neonatal rat islets of Langerhans were isolated from 4-day-old Wistar rat pups (Taconic, Lille Skensved, Denmark), as previously described . All mice were housed according to the Principles of Laboratory Care. The isolated islets were pre-cultured for 7–10 days in RPMI 1640 with UltraGlutamine (Lonza, Vallensbaek, Denmark) and supplemented with 10% newborn calf serum (Biological Industries, Kibbutz Beit Haemek, Israel), 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco, Life Technologies, Taastrup, Denmark) in 5% CO2 at 37°C. For experiments, rat islets were cultured as intact and free-floating in medium supplemented with 2% human serum (BioWhittaker, Lonza) or as single cells following digestion with 0.2% trypsin (Gibco) and 10 mmol/l EDTA (Gibco) in Hanks’ balanced salt solution (HBSS). Dispersed islets were cultured on coverslips coated with bovine corneal extracellular matrix (ECM; Biological Industries) in the medium described above containing 2% human serum.
Mouse islet isolation and culture
Pancreatic islets from 10–12-week-old female NMRI mice were isolated by collagenase digestion, as previously described . Islets were handpicked into albumin-coated Petri dishes (1 mg albumin/ml HBSS) and cultured for 1 day in medium I (RPMI 1640 containing 10 mmol/l glucose and 10% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin) and then for a further 3 days in medium II (RPMI 1640 containing 10 mmol/l glucose and 2% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin) in the absence or presence of BMP4 (50 ng/ml). After culture, mouse islets were fixed for transmission electron microscopy (see electronic supplementary material (ESM) Fig. 1 for further details) or dispersed into single cells using Ca2+-free buffer for electrophysiological experiments.
Pancreatic islets from 12–19-week-old 129SV/C57/6crl wild type (WT) and calbindin1 (Calb1) knockout (KO) mice (calbindinD28k−/−)  were isolated by bile duct perfusion of the pancreas with Liberase (Roche, Hvidovre, Denmark). Digestion was stopped by the addition of HBSS buffer containing Mg2+ and Ca2+ (Gibco), 3 g/l BSA and 0.5 g/l d-glucose. Islets were filtered through a 400 μm mesh, followed by 100 μm and 70 μm mesh strainers (BD Falcon, Albertslund, Denmark). Retained islets were handpicked under a dissection microscope and cultured for 1 day in medium I. The following day, islets were transferred to medium II, in which they were kept throughout stimulation. Female mice were used for electrophysiological experiments and male mice for insulin secretion assays.
A total of 800 intact, free-floating rat islets were cultured in 5.5 cm Petri dishes for 5–10 days and then exposed to BMP4 for 96 h. Total RNA was extracted using TRIzol (Gibco). One microgram of total RNA was labelled using a One-Cycle Target labelling kit (Affymetrix, Santa Clara, CA, USA) following the manufacturer’s instructions. Hybridisation cocktails were hybridised to Rat Genome 230 2.0 GeneChip arrays (Affymetrix) at 45°C for 17 h (at 60 rpm) in a Hybridization Oven 640 (Affymetrix). GeneChips were then washed and stained in a GeneChip fluidics station 450 using the EukGE-WS2v5_450 fluidics protocol (Affymetrix), and then scanned in a GeneChip scanner 3000 (Affymetrix). Microarray data were normalised and gene expression measures derived using the RMA algorithm and the ‘Affy’ Bioconductor package (www.bioconductor.org). A Custom chip definition file obtained from brainarray.mbni.med.umich.edu was used. Qlucore Omics Explorer 3.0 (Qlucore, Lund, Sweden) was used for statistical analysis of the normalised data. For comparing BMP4 to vehicle treatment, microarray data were variance filtered (σ/σ(max) > 0.1), and the two groups compared using the Student’s t test (false discovery rate [FDR] = 5%; Benjamini Hochberg correction for multiple testing).
Analysis of Calb1 and Ins1 mRNA expression by real-time quantitative PCR
A total of 800 intact neonatal rat islets cultured for 5 days were exposed to 50 ng/ml BMP4 for the indicated time periods, and then total RNA was extracted using TRIzol. cDNA synthesis was performed using TaqMan Reverse Transcription Reagents (Applied Biosystems, Carlsbad, CA, USA). TaqMan Gene Expression probes against rat Calb1 (Rn00583140_m1), Ins1 (Rn02121433_g1) and Ppia (Rn_00690933_m1) were obtained from Applied Biosystems. Samples were run on an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). Each sample was run in duplicate or triplicate and expression was normalised to the internal control, Ppia.
Analysis of calbindin1 protein expression by western blotting
A total of 1,000 intact neonatal rat islets were cultured for 5–10 days prior to exposure to 50 ng/ml BMP4 for 24, 48, 72 or 96 h. SDS-PAGE and western blotting were performed as described previously . Primary antibodies were rabbit anti-calbindin1 (Cell Signaling Technology, AH Diagnostics, Aarhus, DK) and mouse anti-β-actin (Abcam, Cambridge, UK). Chemiluminescence was detected using Lumi-GLO (Cell Signaling Technology) and visualised using Las 3000 (Fujifilm, GE Healthcare Europe, Brondby, Denmark). Densitometric scanning was performed using ImageJ freeware (National Institutes of Health, Bethesda, MD, USA).
On day 7–10 after isolation, 40 neonatal rat islets were transferred to medium containing 2% human serum, 100 U/ml penicillin and 100 μg/ml streptomycin, and then stimulated with 50 ng/ml BMP4 for 0–96 h. Mouse islets were stimulated the day after isolation. For each condition, 20 islets were transferred to Krebs-Ringer HEPES buffer (KRHB; 115 mmol/l NaCl, 4.7 mmol/l KCl, 2.6 mmol/l CaCl2, 1.2 mmol/l KH2PO4, 1.2 mmol/l MgSO4, 10 mmol/l HEPES, 0.2% BSA, 2 mmol/l glutamine, 5 mmol/l NaHCO3 and 1% each penicillin and streptomycin, pH 7.4) containing 2 mmol/l glucose and incubated for 90 min prior to the GSIS experiment. Islets were sequentially exposed to 2 mmol/l glucose, 20 mmol/l glucose, and 20 mmol/l glucose plus 10 μmol/l forskolin (Sigma-Aldrich, Brondby, Denmark) for 30 min. The buffer was then collected from each experimental group and insulin content was determined using an in-house insulin ELISA. Results were corrected for DNA content using Quant-IT PicoGreen dsDNA Reagent and Kit (Invitrogen, Life Technologies, Taastrup, Denmark).
Production of Calb1 lentivirus
Mouse Calb1 cDNA in the pENTR221 entry vector (Invitrogen) was transferred to the pLenti6.2/v5DEST Gateway vector (Invitrogen). Lentivirus was produced in HEK293ft cells using the ViraPower Lentiviral Expression System (Invitrogen, Carlsbad, CA, USA) and lentivirus particles were harvested by ultracentrifugation. Viral titer was determined in HT1080 cells (Invitrogen). The virus was used at a multiplicity of infection of 5 for 6 h.
Insulin release in single cells overexpressing calbindin1
Dispersed rat islet cells were cultured on coverslips coated with bovine corneal ECM in 4-well plates for 4 days prior to transduction with Calb1- or GFP-expressing lentivirus for 6 h at a MOI of 5. After 96 h, GSIS was evaluated as described above.
Capacitance measurements and ion current measurements were performed on single beta cells in a mixture of dispersed islets cells using the patch-clamp technique, as previously described . Beta cells where identified by their size and the inactivation properties of the voltage-dependent Na+ channel [24, 25]. Exocytosis was evoked by a train of 10 500 ms depolarisation events from −70 mV to 0 mV at a frequency of 1 Hz and measured as changes in membrane capacitance. The voltage-dependent Ca2+ channel (VDCC) current to voltage relationship was determined using a protocol in which the membrane was depolarised from −70 mV to voltages between −40 mV and +40 mV for 50 ms.
Live cell Ca2+ imaging
Islets were loaded with 4 μmol/l Fura-2-acetoxymethyl ester (TefLabs, Austin, TX, USA) for 40 min, followed by 30 min de-esterification in imaging buffer at pH 7.4 (3.6 mmol/l KCl, 0.5 mmol/l MgSO4, 2.5 mmol/l CaCl2, 140 mmol/l NaCl, 2 mmol/l NaHCO3, 0.5 mmol/l NaH2PO3 and 5 mmol/l HEPES). High K+ buffer was identical except for 70 mmol/l NaCl and 70 mmol/l KCl. Imaging was performed using a Polychrome V monochromator (TILL Photonics, Graefeling, Germany) and an Eclipse Ti Microscope (Nikon, Tokyo, Japan) with a ER-BOB-100 trigger on an iXON3 camera and iQ2 (Andor Technology, Belfast, UK) software. Recording was performed at 1 frame/s at 37°C under perfusion at 1 ml/min. A region was marked around each islet and the light intensity was recoded in that region to obtain the integrated light intensity per unit area (in μm2) at 340 nm (150 ms exposure) and 380 nm (100 ms exposure). These measured intensities were then used to calculate the ratio of Fura-2 bound (340 nm) and unbound (380 nm) to Ca2+ at 1 frame/s.
BMP4 inhibits GSIS
BMP4 inhibits exocytosis and VDCC current
To gain further insight into the mechanism of BMP4 inhibition of exocytosis, we performed live cell Ca2+ imaging of primary adult mouse islets. Evaluation of Ca2+ fluctuations was determined by differences in the Fura-2 340 nm:380 nm ratio (representative traces shown in Fig. 2i, j; all traces are shown in ESM Fig. 2). Temporal fluctuation of the intracellular Ca2+ in response to glucose were organised into two classes: class 1 islets have separate first and second phases, in which a defined first phase peak is followed by a lowering of Ca2+ and a second phase with distinct regularly spaced Ca2+ oscillations; class 2 islets exhibit a rise in Ca2+ with no clear first and second phases and no distinct oscillations in the second phase. In all, 64% of control islets and 26% of BMP4-treated islets had a distinct first peak and second phase oscillations in the presence of high glucose (see ESM Table 1). In addition, there was a significant reduction in the first lowering of Ca2+ (Fig. 2j, k) and the first phase peak amplitude at 16.7 mmol/l glucose (Fig. 2j, m) and in the response to depolarising K+ (Fig. 2j, o). The response time after increased glucose concentration was reduced by BMP4 treatment (Fig. 2j, l), whereas the response time to glucose lowering was increased (Fig. 2n). We did not observe significant changes in the amplitude or frequency of Ca2+ oscillations.
BMP4 mediates upregulation of calbindin1
Genes regulated by at least twofold in neonatal rat islets exposed to 50 ng/ml BMP4 for 96 h
DNA-binding protein inhibitor ID-3
DNA-binding protein inhibitor ID-1
Ly6/PLAUR domain containing protein 8 precursor
Iroquois-class homeodomain protein IRX-3
DNA-binding protein inhibitor ID-2
MICAL C-terminal-like protein
Family with sequence similarity 101. member A (Fam101a)
BMP and activin membrane-bound inhibitor homologue
Protein atonal homologue 8
Activity-regulated cytoskeleton-associated protein
Calcium/calmodulin-dependent protein kinase type 1G
Carbohydrate sulfotransferase 10
Protein delta homologue 1
Tubulin polymerisation-promoting protein family member 3
Transmembrane protein 100
Creatine kinase B-type
Protein phosphatase 1 regulatory subunit 36
Transmembrane protein 100
Reticulon-4 receptor-like 1
Suppression of tumorigenicity 5 protein
A-kinase anchor protein 12
Probable ATP-dependent RNA helicase DDX31
Fatty acid desaturase 1
Bone morphogenetic protein 3
5-hydroxytryptamine (serotonin) receptor 5B
G protein-coupled receptor 6
Overexpression of Calb1 impairs GSIS
BMP4 inhibition of GSIS and exocytosis is dependent on Calb1 upregulation
We further investigated the effect of BMP4 on GSIS from pancreatic islets from Calb1 KO mice and WT littermates. As for neonatal rat islets, we observed a significant inhibition of GSIS by BMP4 treatment in WT adult mouse islets (Fig. 4c), but none in islets from Calb1 KO mice (Fig. 4c). In accordance with this, BMP4 did not reduce exocytosis in islets from Calb1 KO mice, but rather showed a nonsignificant increase in exocytosis (Fig. 4d–f).
Here, we investigate the mechanism behind BMP4-mediated inhibition of GSIS in neonatal rat and adult mouse islets of Langerhans. This effect was not caused by reduced Ins1 mRNA expression or protein content, or by changes in the number, size or localisation of insulin granules, but instead seems to be due to diminished Ca2+ influx through VDCCs resulting in reduced exocytosis. Metabolism of glucose increases intracellular ATP levels, resulting in closure of ATP-dependent K+ channels and membrane depolarisation. Subsequent opening of VDCCs triggers Ca2+ entry which induces insulin exocytosis (reviewed in [30, 31]). Hence, Ca2+ entry is essential for the amount of insulin released. We observed a BMP4-dependent decrease in the VDCC current, leading to a reduction in depolarisation-induced exocytosis and GSIS. The Ca2+ sensitivity of exocytosis was not decreased but was instead increased, indicating that the reduced Ca2+ current is the main determinant of reduced exocytosis. Reduced Ca2+ influx was also confirmed by a lack of response to K+ in live cell Ca2+ measurements. Both observations are indicative of reduced depolarisation-evoked Ca2+ influx or increased Ca2+ buffering after BMP4 treatment. An increase in Ca2+ buffering would agree with the continued and more pronounced reduction in late exocytosis evoked by the latter depolarisations and the second train (Fig. 2d, e). Interestingly, our microarray analysis identified Calb1, the gene encoding the Ca2+-binding protein calbindin1, as being upregulated by BMP4. We confirmed this regulation in independent rat islet samples and mouse islets. Although it has been suggested that neonatal and adult islets respond differently to glucose and other stimuli, BMP4 inhibited GSIS and upregulated calbindin1 in both model systems. Overexpression of Calb1 reduced GSIS. The effect of BMP4 on GSIS was lost in islets from Calb1 KO mice (Fig. 4c). The exocytotic response to BMP4 in WT islets showed the same trend as previously observed (Figs 2a–e and 4d), whereas this regulation was lost in islets from Calb1 KO mice. There was a trend towards increased exocytosis in response to BMP4 in Calb1 KO mice, but this was nonsignificant and was not reflected in GSIS. BMP inhibition of GSIS was not observed in islets from Calb1 KO mice, suggesting that increased calbindin1 expression may cause the observed BMP4-mediated decline in GSIS. Indeed, the BMP4-induced decrease in VDCC influx was remarkably similar to the effect observed upon overexpression of calbindin1 in a pancreatic beta cell line (Fig. 2b–h and ). In addition to a Ca2+-scavenging effect, accumulating evidence suggests that calbindin1 reduces Ca2+ currents via an association with L-type VDDCs . Glucose- and Ca2+-dependent translocation of calbindin1 to the plasma membrane has previously been suggested to facilitate the interaction with L-type Ca2+ channels [28, 29]. This could explain the lack of effect on the capacitance increase evoked by the first depolarisation (Figs 2c and 4e), because the first influx of Ca2+ would induce translocation of calbindin1 to the plasma membrane.
Glucose-induced Ca2+ oscillations occur in fewer BMP4-treated islets compared with control cells, possibly because of reduced Ca2+ influx through VDCCs. This was further supported by the blunted Ca2+ response to high glucose and depolarising K+ after BMP4 treatment (Fig. 3i–o and ESM Table 1). Loss of K+-induced Ca2+ influx was also observed in a calbindin1-overexpressing cell line, indicating that this effect is caused by increased calbindin1 expression [27, 29]. Finally, Ca2+ oscillations persisted for longer in BMP4-treated islets when glucose was returned to 2.8 mmol/l (Fig. 2n and ESM Table 1), indicating a failure of the beta cells to repolarise to baseline. These data again suggest dysfunctional Ca2+ handling after BMP4 treatment.
Interestingly, calbindin1 expression is increased in pancreatic islets from diabetic rats and mice [32, 33]. Concomitant upregulation of BMP2 (but not BMP4) and calbindin1 in the islets of Langerhans has been observed in a type 2 diabetic mouse model . BMP2 also upregulated the expression of calbindin1 and reduced GSIS in rat islets (data not shown). Generally, BMP2/4 may be considered as inflammatory markers in several metabolic tissues. Under diabetic conditions, expression of BMP2 or BMP4 has been reported to increase in the arteries, kidney, bones and islets [4, 14, 18, 19]. Increased BMP4 expression was reflected by increased circulating levels of BMP4 in type 2 diabetic patients in one study . Thus, systemic inhibition of BMP2/4 appears to represent a possible strategy for broad targeting of a mediator of low-grade inflammation associated with type 2 diabetes. Interestingly, it was recently reported that systemic administration of the natural BMP inhibitor noggin lowered blood glucose levels in db/db mice .
In conclusion, we have provided insight into the mechanisms involved in BMP4-mediated inhibition of insulin secretion and gene regulation in islets of Langerhans, and identified calbindin1 as a mediator of BMP4-induced beta cell dysfunction.
We acknowledge the technical assistance from H. Fjordvang and L.G. Pedersen from the University of Copenhagen, S. Mach from the Ludwig Maximilian University of Munich, and B.–M. Nilsson and A.-M. Veljanovska-Ramsay from the Lund University Diabetes Center.
We are thankful for support from the Novo Nordisk Foundation, Danish Research Council, Danish Diabetes Academy, European Foundation for the Study of Diabetes, A.P. Møller Foundation, Swedish Research Council, Region Skåne (ALF), Albert Påhlsson Foundation and Swedish Diabetes Foundation. LE is a senior researcher at the Swedish Research Council. GLC holds a postdoctoral grant from the Danish Diabetes Academy.
Duality of interest
MLBJ, CB and KSF are employees of Novo Nordisk A/S.
NB, GLC, MLBJ, CB and LE designed the study; GLC, MLBJ, AW, IGM, JF, KSF, MM, CB, NB and LE participated in acquisition, analysis and interpretation of data; GLC, NB and LE drafted the manuscript; and GLC, MLBJ, AW, IGM, JF, KSF, MM, CB, NB and LE revised the manuscript critically for important intellectual content and approved the final version to be published. NB is the guarantor of this work.