Introduction

Study of H2S has gained momentum in recent years as its role as an important gasotransmitter with multifaceted biological significance has become recognised. Gasotransmitters are a family of endogenous gaseous signalling molecules involved in multi-level regulation of physiological and pathological functions [1]. Following the identification of nitric oxide (NO) and carbon monoxide (CO) as the first two gasotransmitters, mounting evidence has demonstrated that H2S is the third one.

H2S is produced in various types of cells via the enzymatic function of two enzymes: cystathionine β-synthase (CBS) and cystathionine γ-lyase (CSE) [1]. CSE has been shown to be the predominant H2S-generating enzyme in pancreatic beta cells [2] and freshly isolated rat islets [3].

Activation of ATP-sensitive K+ (KATP) channels is demonstrated to be responsible for the inhibitory effect of H2S on glucose-induced insulin release from INS-1E [2] and HIT-T15 cell lines [4]. As opening of KATP channels leads to beta cell membrane hyperpolarisation, insulin release from pancreatic islets would be inhibited because of reduced Ca2+ influx. Exogenously applied H2S or overproduction of CSE has been reported to induce apoptosis of INS-1E cells, suggesting an inhibitory effect of H2S on insulin production by reduction of beta cell mass [2]. Glucose level has been seen to regulate various gene expressions in beta cells such as Glut2 (also known as Slc2a2) [5], acetyl-CoA carboxylase (Acc [also known as Acaca]) [6] and connexin 36 [7]. Previous research suggested a suppressive role of glucose on endogenous H2S production in INS-1E cells [2]. Considering the critical role of glucose level on gene regulation in pancreatic beta cells and the effect of H2S on insulin secretion, we planned to investigate if the production of CSE is regulated by glucose level in insulin-secreting cells and the underlying mechanism. Our results indicated that CSE production is inhibited by glucose at high concentrations. We further demonstrated that glucose-induced downregulation of CSE production requires SP1 and p38 mitogen-activated protein kinase (MAPK) phosphorylation. The effect of high glucose on CSE production may constitute a novel regulatory mechanism for the fine control of insulin secretion from pancreatic beta cells.

Methods

Cell culture

INS-1E cells were cultured as previously described by Yang et al. and used between passages 50 and 80 [2]. The experiments were performed when cultured cells reached 70–80% confluence. Before treatment with different concentrations of glucose, INS-1E cells were pre-incubated overnight in RPMI 1640 medium containing 1% FBS and 5 mmol/l glucose at 37°C in a humidified mixture of 5% CO2 and 95% O2 (vol./vol.) and then subjected to different concentrations of glucose or other compounds together with 10% FBS.

Preparation of intact islets

Pancreatic islets were isolated from male Sprague–Dawley rats (8–12 weeks) by the collagenase digestion method. The solution used for the isolation was KRB supplemented with 0.5–1 mg/ml collagenase type V, 0.5–1 μl/ml DNase I and 0.5 mg/ml bovine serum albumin. Animal experiments were conducted in compliance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85–23, revised 1996) and approved by the Animal Use Committee of Lakehead University, ON, Canada.

Measurement of endogenous H2S production

H2S production rate was measured as previously described [8–10]. Briefly, INS-1E cells were incubated with either 5 or 20 mmol/l glucose for 24 h, then collected and lysed in 50 mmol/l ice-cold potassium phosphate buffer (pH 6.8). The cell lysates were first incubated with l-cysteine (10 mmol/l) for 90 min at 37°C, and then trichloroacetic acid was added to stop the reaction. The level of Methylene Blue generated by the interaction of H2S with N,N-dimethyl-p-phenylenediamine sulphate was determined at 670 nm with a FLUOstar OPTIMA microplate spectrophotometer (BMG LABTECH, Offenburg, Germany). H2S content in the culture medium was measured as previously described by Yang et al. [11].

Western blots

Cultured cells were harvested and lysed in a Tris-EDTA sucrose lysis buffer plus protease inhibitors as previously described by Yang et al. [2]. Protein extracts were separated by SDS-PAGE and blotted onto nitrocellulose membranes (Pall Corporation, Pensacola, FL, USA) [12]. The primary antibodies used were: anti-CSE antibody (Abnova, Taiwan, Republic of China), anti-SP1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti-phospho SP1 (threonine 453) (Abcam, Cambridge, MA, USA), anti-phospho-p38 MAPK, anti-p38 MAPK (Cell Signaling Technology, Beverly, MA, USA) and β-actin (Sigma, Oakville, ON, Canada). Immunoreactions were visualised by ECL Western Blotting System (GE Healthcare, Amersham, UK). Densitometric quantification was performed using Alpha Digi Doctor software (Richardson, TX, USA) and normalised against the quantity of β-actin.

Short interfering RNA (siRNA) transfection

Pre-designed Sp1-targeted siRNA (Sp1-siRNA) and control siRNA were purchased from Santa Cruz. INS-1E cells were seeded in six-well plates at a density of 1 × 105 cells per well in the presence of 5 mmol/l glucose. Transfection of siRNA into INS-1E cells was achieved using Lipofectamine 2000 Transfection Reagent (Invitrogen, Burlington, ON, USA). Briefly, the cells were transfected with 20 nmol/l Sp1-siRNA or control-siRNA in Opti-MEM I culture medium (Invitrogen) without antibiotics for 4 h. Fresh normal growth medium was then added and the cells were incubated for another 44 h.

Determination of mRNA level by real-time PCR

INS-1E cells were harvested and the total RNA was isolated using TriReagent. First-strand cDNA was prepared by reverse transcription using M-MuLV reverse transcriptase and random hexamer primers from a ProtoScript II RT-PCR Kit (New England Biolabs, Pickering, ON, Canada) according to the manufacturer’s protocol. The primers of rat Cse (GenBank accession number AY032875) were 5′-AGCGATCACACCACA-GACCAAG-3′ and 5′-ATCAGCACCCAGAGCCAAAGG-3′. Quantum RNA β-actin internal standards were purchased from Ambion (Foster City, CA, USA). Real-time PCR was performed in an iCycler IQ 5 apparatus (Bio-Rad, Mississauga, ON, Canada) associated with the iCycler optical system software (version 3.1) using SYBR Green PCR Master Mix and relative mRNA quantification was calculated as described previously [2, 11].

Cloning of mouse Cse promoter region and construction of reporter plasmids

The 172 bp genomic DNA fragment upstream from the transcriptional start site (−149 to +23) of the Cse gene, containing canonical TATA and CAAT boxes and the SP1 site, was isolated by PCR from mouse-tail genomic DNA [13]. Briefly, the small fragment was amplified using 1.5 μg mouse genomic DNA as template in a 20 μl volume reaction with 5 pmol/μl of each primer. The sequences of the primers used were as follows: F-Cse-Sp1 (KpnI) (−149/−134), 5′-CGGTACCTCTGTGCCACTGGGAG-3′; R-Cse-Sp1 (HindIII) (+7/+23), 5′-GAAGCTTGAGTGCGAGGTGTTGCT-3′. The underlined sequence is the restriction site for KpnI or HindIII. The cloned fragments were subcloned into the promoterless expression vector pGL3 basic (Promega, Madison, WI, USA) to obtain the reporter plasmid pGL3 (−149/+23)-Cse-Prom-Luc. Another plasmid was constructed, which was derived from the pGL3 (−149/+23)-Cse-Prom-Luc, and contained a mutated SP1 site. The SP1 response element (5′-GAGGCGGGGC-3′) was mutated into (5′-GATTCGGGGC-3′) by using QuikChange Site-Directed Mutagenesis Kit (Stratagene, Mississauga, ON, USA), with oligonucleotide (−144 to −117) 5′-GCCACTGGGATTCGGGGCAGGAACGATC-3′ and its complementary oligonucleotide according to the manufacturer’s recommendations. The mutations are underlined and in italic letters. All plasmid constructs were verified by DNA sequencing.

Transient transfection and luciferase assay

INS-1E cells were transfected with 900 ng of the reporter plasmid pGL3 (−149/+23)-Cse-Prom-Luc DNA (CSE-Sp1 vector), mutated-Sp1 promoter or pGL3-basic vector, mixed with 100 ng pRL-TK vector (Promega) as an internal control using Lipofectamine 2000. Six hours after the transfection, the OptiMEM medium was replaced by RPMI 1640 medium with 5 or 20 mmol/l glucose in the presence or absence of 10 μmol/l p38 MAPK inhibitors SB203580, SB202190 or their non-active analogue SB202474 and incubated for another 24 h before harvesting with passive lysis buffer (Promega). Samples of the lysates were assayed for luciferase activities in a Fluostar Luminometer (BMG LABTECH, Germany) using the Dual-Luciferase Reporter Assay System (Promega). Both firefly and Renilla luciferase activities were measured as luminescence intensities and the promoter activity was expressed as ratios between firefly and Renilla luciferase activities.

Measurement of insulin secretion from INS-1E cells

INS-1E cells were washed and pre-incubated with glucose-free KRB (pH 7.4) plus 0.1% BSA in 24-well plates. After 30 min pre-incubation, cells were incubated for another 30 min at 37°C in the presence of different glucose concentrations with or without 5 mmol/l dl-propargylglycine (PPG), 20 mmol/l sodium pyruvate or 20 mmol/l 2-deoxy-d-glucose (2-DG). Where sodium salts of pyruvate were added, the Na+ content of the KRB was correspondingly decreased. At the end of each incubation period, the medium was collected and immediately stored at −20°C until insulin determination was completed by using the rat insulin ELISA kit (Mercodia AB, Sylveniusgatan, Uppsala, Sweden).

Materials and data analysis

Chemicals were all obtained from Sigma (Oakville, ON, Canada) unless otherwise mentioned. Unless otherwise specified, ‘glucose’ refers to d-glucose in this communication. The data are expressed as mean ± SEM from at least three independent experiments. Statistical analyses were performed using Student’s t test on paired data or one-way ANOVA. Statistical significance was considered at p < 0.05.

Results

High glucose suppressed H2S release from INS-1E cells and pancreatic islets

High glucose (20 mmol/l) lowered H2S content of the culture media by 42.1 ± 2.8 % (n = 6, p < 0.05) for INS-1E cells (Fig. 1a) and 76.4 ± 1.8 % (n = 3, p < 0.05) for freshly isolated rat pancreatic islets (Fig. 1b) compared with that treated with 5 mmol/l glucose. High glucose also decreased H2S production rate in INS-1E cells (n = 4; p < 0.05; Fig. 1c).

Fig. 1
figure 1

High glucose inhibited H2S production from INS-1E cells and freshly isolated rat islets. INS-1E cells and freshly isolated rat islets were cultured overnight in 5 mmol/l glucose medium with 1% FBS, and then changed to 5 or 20 mmol/l glucose medium with 10% FBS for another 24 h. High glucose significantly inhibited H2S production in the culture medium for INS-1E cells (a) and pancreatic rat islets (b). H2S production rate in INS-1E cells (c) was also significantly decreased by high glucose. Data in (a) and (b) were from at least three independent experiments; data in (c) were from four independent experiments. *p < 0.05

Full size image

High glucose decreased Cse mRNA expression and CSE protein production in INS-1E cells

After incubation with high glucose for 24 h, CSE protein production was significantly decreased (Fig. 2a). Cse mRNA level in INS-1E cells treated with 20 mmol/l glucose was 36.5 ± 9.9% of that with 5 mmol/l glucose (Fig. 2b). To test if the decreased Cse gene expression occurred at the transcriptional level, INS-1E cells pre-incubated for 24 h in 5 or 20 mmol/l glucose medium were then exposed to actinomycin D (5 μmol/l) for 0.5 or 2 h (Fig. 2c). Inhibition of RNA synthesis with actinomycin D in the presence of 20 mmol/l glucose did not lead to further reduction in the Cse mRNA level compared with 5 mmol/l glucose incubation, suggesting that glucose may not affect Cse mRNA stability. Therefore, it repressed the expression of Cse mRNA at the level of transcription.

Fig. 2
figure 2

High glucose downregulated Cse mRNA and CSE protein production in INS-1E cells. After INS-1E cells were incubated with the indicated concentration of glucose for 24 h, the cells were collected and subjected to western blots (a) and real-time PCR (b) analyses. (c) Glucose had no effect on the stability of the Cse transcript. INS-1E cells were pre-incubated with 5 or 20 mmol/l glucose for 24 h, and then treated in the absence (white bars) or presence of 5 μmol/l actinomycin D (striped bars) for 0.5 or 2 h. The data were normalised to β-actin production and are presented relative to production at 5 mmol/l glucose. In (a) and (b) the data are from four independent experiments; in (c) the data are from three independent experiments. *p < 0.05

Full size image

Effects of pharmacological treatments on the production of CSE

Western blots results showed that 20 mmol/l 2-DG mimicked the inhibitive effect of 20 mmol/l glucose on CSE production in INS-1E cells (Fig. 3a) and rat islets (Fig. 3b). Pyruvate at 20 mmol/l significantly decreased CSE production in both INS-1E cells (Fig. 3a) and isolated rat islets (Fig. 3b), whereas 20 mmol/l l-glucose and mannitol were ineffective. To explore whether beta cell depolarisation affects glucose-induced downregulation of CSE production, INS-1E cells were incubated with 5 mmol/l glucose in the presence of KCl (20 mmol/l). KCl treatment did not alter CSE production (Fig. 3a).

Fig. 3
figure 3

Effect of pharmacological treatments on the production of CSE. INS-1E cells (a) and pancreatic rat islets (b) were incubated with 5 or 20 mmol/l d-glucose, 20 mmol/l pyruvate, 20 mmol/l 2-DG, 20 mmol/l KCl in the presence of 5 mmol/l d-glucose, 20 mmol/l mannitol or 20 mmol/l l-glucose for 24 h, respectively. After that, the cells or islets were collected and subjected to western blot analysis. The data are from three independent experiments. *p < 0.05 vs 5 mmol/l d-glucose. Glu, glucose; Pyr, pyruvate

Full size image

High-glucose-modulated CSE production requires the phosphorylation of p38 MAPK and SP1

The phosphorylation of p38 MAPK in INS-1E cells was evident within 15 min of high glucose treatment and peaked at 1 h (Fig. 4a). High-glucose-induced phosphorylation of SP1 protein without changing the amount of total SP1 protein (Fig. 4b). SB203580 and SB202190 (p38 MAPK inhibitor) but not SB202474 at 10 μmol/l abolished the inhibitive effect of high glucose on CSE production. Moreover, inhibition of the SP1 binding to the GC boxes of promoter by 100 nmol/l mithromycin A (MA) induced a greater decrease of CSE production (Fig. 4c).

Fig. 4
figure 4

Glucose stimulated phosphorylation of p38 MAPK and SP1. (a) Glucose induced phosphorylation of p38 MAPK. INS-1E cells were incubated with glucose at 5 or 20 mmol/l for the indicated time. After that, the cells were collected and subjected to western blots analysis. *p < 0.05 vs. 5 mmol/l glucose treatment group at the same time point. (b) Glucose stimulated SP1 phosphorylation. INS-1E cells were pretreated with 5 mmol/l glucose RPMI-1640 medium containing 1% FBS overnight and then incubated with 5 mmol/l or 20 mmol/l glucose for 24 h. After that, the cells were collected and subjected to western blot analysis. *p < 0.05. (c) The p38 MAPK-SP1 pathway mediated the effect of glucose on CSE production. INS-1E cells were pretreated with or without SB203580, SB202190 or SB202474 (control reagent) for 1 h, and then incubated with glucose at 5 or 20 mmol/l for 24 h in the presence or absence of 100 nmol/l MA. *p < 0.01 vs. 5 mmol/l glucose alone group; † p < 0.01 vs. 5 mmol/l glucose with SB202474 group. The data are from three independent experiments

Full size image

Role of p38 MAPK in glucose-induced phosphorylation of SP1

To confirm the observation that p38 MAPK is involved in SP1-mediated glucose deactivation of CSE, we examined the effects of p38 MAPK inhibitors on glucose-induced change in production of phospho and total SP1. Pretreatment with two specific p38 MAPK inhibitors, SB203580 and SB202190 (both 10 μmol/l), drastically decreased glucose-induced phosphorylation of SP1 without changing the amount of total SP1 compared with the control reagent SB202474, DMSO (1%) or high glucose alone (Fig. 5a). Furthermore, the effects of Sp1-specific siRNA on Cse expression were determined at protein levels (Fig. 5b). Transfection of Sp1-specific siRNA significantly reduced CSE production. These results indicated an important role of SP1 in CSE production.

Fig. 5
figure 5

Role of SP1 in glucose-induced regulation of CSE. INS-1E cells were cultured in 5 mmol/l glucose RPMI-1640 medium containing 1% FBS overnight, and were changed on the next day to fresh medium (5 mmol/l glucose), pretreated with or without 0.1% DMSO (vehicle), 10 μmol/l SB202474, SB203580 or SB202190 for 1 h, respectively, and then incubated with 20 mmol/l glucose for an additional 24 h. Production of SP1 and phosphorylated SP1 was detected by western blot using specific antibodies. (a) p38 MAPK inhibitors SB203580 and SB202190 but not non-active analogue SB202474 decreases high-glucose-induced SP1 phosphorylation in INS-1E cells without changing total SP1 production. *p < 0.05 vs control at 20 mmol/l glucose alone. (b) INS-1E cells were transfected with either control siRNA (black bars) or Sp1-siRNA (white bars) for 48 h at 5 mmol/l glucose RPMI-1640 medium. Western blot analysis showed successful Sp1 knockdown and reduced CSE production in INS-1E cell lines. The data were normalised and are presented as ratio to β-actin production. The data are from three independent experiments. *p < 0.05

Full size image

p38 MAPK-SP1 signalling pathway mediated high-glucose-induced decrease of Cse promoter activity

Transfection of INS-1E cells with Cse core promoter (Cse-Sp1 promoter) resulted in a 39-fold increase in luciferase activation in the presence of 5 mmol/l glucose (n = 5; *p < 0.01 vs cells transfected with the promoterless pGL3 basic vector). However, Cse-Sp1 promoter activity was inhibited significantly by high glucose (20 mmol/l). Mutating the consensus SP1 binding site completely abolished Cse core promoter (Cse-Sp1 promoter) activity (Fig. 6a). By inhibiting p38 MAPK, SB202190 and SB203580 completely abolished the inhibitory effect of 20 mmol/l glucose on Cse promoter activity in the presence or absence of 10 μmol/l SB202474 (Fig. 6b).

Fig. 6
figure 6

Effect of glucose on mouse CSE promoter activity. (a) High glucose attenuated CSE promoter activity. INS-1E cells were co-transfected with reporter plasmids Cse-Sp1 vector (black bars), pGL3-basic vector (shaded bars) or mutated-Sp1 construct (white bars) together with pRL-TK vector for 6 h, then incubated in 5 or 20 mmol/l glucose RPMI-1640 medium for an additional 24 h. Relative luciferase activity was normalised with Renilla luciferase from pRL-TK vector. The data were from five independent experiments. *p < 0.01 vs 5 mmol/l glucose alone group. (b) SB203580 and SB202190 but not SB202474 reversed glucose-reduced CSE promoter activity. INS-1E cells were transfected with Cse-Sp1 vector in the presence or absence of the indicated concentration of glucose or inhibitors for 24 h. Relative luciferase activity was normalised with Renilla luciferase from pRL-TK vector. The data were from four independent experiments. *p < 0.01 vs 5 mmol/l glucose alone group; † p < 0.01 vs 5 mmol/l glucose plus SB202474 group

Full size image

PPG, but not pyruvate or 2-DG, enhances high-glucose-induced insulin secretion

High glucose (20 mmol/l) stimulated insulin secretion from INS-1E cells by approximately threefold compared with the basal glucose (5 mmol/l). After treatment with 5 mmol/l PPG to inhibit CSE activity, insulin secretion at basal glucose was not altered. However, high-glucose-stimulated insulin secretion was slightly but significantly increased (Fig. 7a). Conditions of 20 mmol/l 2-DG alone or in the presence of 5 mmol/l glucose did not stimulate insulin secretion. However, in the presence of 20 mmol/l glucose, 2-DG significantly inhibited glucose-induced insulin secretion. Pyruvate (20 mmol/l) had no effect on insulin release in the presence or absence of glucose within 30 min of application (Fig. 7b).

Fig. 7
figure 7

Inhibition of CSE activation enhanced glucose-induced insulin secretion. After 30 min pre-incubation with glucose-free KRB, INS-1E cells were incubated for another 30 min at 37°C in the presence of different glucose concentrations with or without 5 mmol/l PPG (a), 20 mmol/l pyruvate or 2-DG (b). Insulin level in the culture media was measured by ELISA. The data were from four independent experiments. *p < 0.05 vs 5 mmol/l glucose alone group. † p < 0.05 vs 20 mmol/l glucose alone group

Full size image

Discussion

Endogenous H2S level in pancreatic beta cells plays a critical role in regulating insulin release. H2S has been shown to inhibit insulin secretion from insulin-secreting cell lines (INS-1E, MIN6 and HIT-T15) and isolated islets [2, 4, 14, 15]. H2S has also been linked to glucose metabolism and insulin resistance by interaction with methylglyoxal, an intermediate of glucose metabolism [16] and by inhibiting basal and insulin-stimulated glucose uptake in adipocytes [17].

Pancreatic production of H2S is regulated by CSE and/or CBS [2, 4, 14, 15]. Cse gene knockdown by Cse-siRNA largely eliminated H2S production from INS-1E cells [2]. We reported significant production levels of CSE in rat pancreatic islets, but Cbs mRNA expression was extremely low [3]. PPG, as a specific CSE inhibitor, drastically reduced H2S production rate to near zero in pancreatic islets from both Zucker diabetic fatty and Zucker fatty rats. Taken these together, it is believed that CSE is the main enzyme for H2S production in rat pancreatic beta cells and the INS-1E cell line.

High glucose is capable of regulating the expression of many genes in pancreatic beta cells [18]. For example, Glut2 [5] and Acc [6] were upregulated by glucose, whereas Cx36 (also known as Gjd2) [7], Pparα (also known as Ppara) [19] and those encoding the sulfonylurea receptor 1/inwardly rectifying K+ channel 6.2 (Sur/Kir6.2 [also known as Abcc8/Kcnj11]) [20] were downregulated by high glucose. We have previously demonstrated that endogenous H2S production rate in INS-1E cells was modulated by different glucose levels [2]. The underlying molecular mechanism for the interaction of high glucose and H2S production, however, had been unclear. In this study, we found that high glucose inhibited CSE activity represented by decreased H2S production rate within 2 h (data not shown). Also for the first time, we demonstrated CSE production was repressed by high glucose. Therefore, high glucose would both downregulate CSE production and inhibit CSE activity. The latter may be a relatively rapid mechanism for mediating glucose-induced insulin secretion.

We further investigated whether this inhibitory effect of glucose on CSE production is specifically related to cellular metabolism and use of glucose. l-Glucose, an analogue of glucose that cannot enter beta cell [21], had no effect on CSE protein abundance. The inhibitory effect of 20 mmol/l glucose on CSE production is also unlikely to be due to osmolality change as 20 mmol/l mannitol did not alter CSE production. Another analogue of glucose, 2-DG, can be taken up by beta cells through GLUT2 transporters and phosphorylated by glucokinase but cannot be metabolised beyond the level of glucose 6-phosphate in the glycolytic or pentose-phosphate pathway [22]. Interestingly, 2-DG inhibited CSE production (Fig. 3). This phenomenon suggests that glucose 6-phosphate may be key to the glucose-induced downregulation of CSE production. Brun et al. [6] reported previously that a wide variety of metabolisable nutrients were incapable of inducing Acc mRNA in the INS-1 cell line. However, 2-DG was capable of inducing Acc mRNA, suggesting that glucose does not have to be metabolised beyond glucose 6-phosphate in the glycolytic pathway to induce production of this gene product. Further studies are merited to determine the direct effect of glucose 6-phosphate on CSE production to confirm this hypothesis. Pyruvate also significantly decreased CSE protein production even more than high glucose did in rat islets (Fig. 3b). This result indicates that pyruvate, a product of a CSE-catalysed reaction, may inhibit CSE production as a product-based feedback control mechanism.

As 2-DG and pyruvate decreased the production of CSE, it would be expected that the consequently lowered endogenous H2S level might promote insulin release, the same effect as offered by PPG (Fig. 7a). However, as the application of 2-DG and pyruvate in the insulin release experiments was limited to 30 min, CSE protein level would not be changed within this short period of time. We also found that 2-DG and pyruvate alone had no effect on H2S production rate in INS-1E cells, which reflects CSE activity (data not shown). Taken together, the effects of 2-DG and pyruvate on insulin release (Fig. 7b), if any, could not be ascribed to changed CSE production or activity. Indeed, 20 mmol/l 2-DG significantly inhibited the insulin release stimulated by 20 mmol/l glucose. Similar results had been reported previously that 2-DG at 17 mmol/l decreased insulin release from rat pancreas [23]. The underlying mechanism for this interaction of 2-DG and glucose may be related to the competition of 2-DG with glucose for transportation into the cells [24] and for reaction with glucokinase or hexokinase within the cell [25–27].

Glucose metabolism stimulates insulin secretion by closing KATP channels and subsequent calcium influx, eventually leading to exocytosis of insulin granules. To examine if glucose-induced depolarisation of beta cells is responsible for CSE downregulation, INS-1E cells were treated with 20 mmol/l KCl in the presence of basal glucose. This treatment did not alter CSE production (Fig. 3a). Similar results were obtained from freshly isolated islets under the same treatments (Fig. 3b). These results suggested that Ca2+ influx-mediated insulin secretion was not implicated in glucose-repressed CSE production.

The human CSE and rodent Cse genes have been cloned before [13, 28]. The core promoter of mouse Cse contains several putative transcriptional factor-binding sites, including myeloid zinc finger protein 1 (MZF-1) and SP1 [13]. Deletion of Mzf1 (in Cos-7 and HEK-293 cells) or Sp1 consensus sequence (in HEK-293 cells) from Cse promoter significantly decreased the promoter activity, suggesting the involvement of these factors in the basal transcriptional activity of Cse. Bioinformatics analysis of the promoter region of the mouse and rat Cse also revealed the presence of a potential Sp1 consensus sequence (5′-GAGGCGGGGC-3′) located within the −149/+23 region of the mouse Cse promoter and at −182/−173 region in the rat Cse promoter. Previous investigation found that the −137 to +18 sequence conferred the highest promoter activity in HEK-293 cells using the promoter deletion and mutation method [13]. Given that the use of mouse Cse promoter has been examined thoroughly, and SP1 is a ubiquitous transcription factor abundantly expressed in different species with a highly conservative consensus sequence of ‘GAGGCGGGGC’, we generated a pGL3 (−149/+23)-Cse-promoter expression vector (Cse-Sp1 vector) cloned upstream to luciferase and used it to transfect INS-1E cells. We demonstrated that SP1 is a crucial transactivator for Cse gene expression because transfection of INS-1E cells with this Cse core promoter containing the Sp1 consensus sequence (Cse-Sp1 vector) resulted in a 39-fold increase in luciferase activity compared with promoterless vector (Fig. 6a). MA binds to GC-rich regions in chromatin and interferes with the transcription of genes bearing GC-rich motifs in their promoter. It is also known as a major SP1 inhibitor, as SP1 recognises GC-rich sequences [29]. It drastically decreased CSE production in the presence of either 5 mmol/l or 20 mmol/l glucose (Fig. 4c). Sp1 knockdown also abolished a large portion of CSE protein production (Fig. 5b), suggesting a critical role of SP1 in CSE production. A single Sp1 site mutation of Cse-Sp1 vector completely abolished the promoter activity of Cse (Fig. 5a), again suggesting the importance of Sp1 in the basal transcriptional activity of Cse. High-glucose treatment significantly attenuated SP1 transactivation, which provided evidence that a glucose-responsive element is located within the proximal portion of the mouse Cse promoter (Fig. 6a, b).

Several reports have suggested the ubiquitously produced transcription factor SP1 may provide a mechanism for glucose responsiveness [30, 31]. Changes in SP1 phosphorylation result in either facilitation or suppression of DNA binding, promoter activation [31–33] and gene transcription [34–36]. SP1 protein can be phosphorylated at various sites by different kinases including protein kinase A, protein kinase C, MAPK, casein kinases 1 and 2, and calmodulin kinases [37]. It was previously reported that glucose can de-phosphorylate SP1 and increase promoter binding activity on the Acc gene in adipose tissue [38]. The phosphorylation status of SP1 appears to be cell- and gene-type dependent, which is also a major factor in mediating the effects of extracellular stimuli such as insulin [39].

In the present study, we examined possible signal-transduction pathways related to SP1 phosphorylation in high-glucose-induced downregulation of Cse expression. Exposure to high glucose has been reported to lead to the activation of various MAPK cascades in pancreatic beta cell lines [40, 41]. While p38 MAPK phosphorylation became significant 15 min after high-glucose treatment (Fig. 4a), the decreased Cse mRNA and protein levels were detected 24 h later. Whether the downregulation of Cse mRNA expression actually occurred immediately after p38 MAPK phosphorylation was not tested. It is reasoned that, however, p38 MAPK phosphorylation functions as a trigger and the decreased CSE production followed. This effect of high glucose is relatively specific as no significant difference was found in phospho-p44/42 MAPK level between 5 and 20 mmol/l glucose treatments (data not shown). Inhibition of p38 MAPK phosphorylation by SB203580 or SB202190 at 20 mmol/l glucose significantly decreased SP1 phosphorylation without changing the amount of total SP1 protein (Fig. 5a), indicating that high glucose phosphorylates the existing SP1 protein through p38 MAPK. SB203580 and SB202190 but not SB202474 also completely reversed the inhibitory effect of high glucose on CSE production (Fig. 4c). Moreover, inhibition of SP1 phosphorylation by SB203580 or SB202190 abolished the inhibitive effect of glucose to CSE promoter activity (Fig. 6b). These novel observations suggest that high glucose increases phosphorylation of SP1 via p38 MAPK activation in INS-1E cells, resulting in a decreased CSE promoter activity and reduced Cse transcription.

Kaneko et al. [42] reported glucose-induced H2S production by observing stimulated CSE production after 18 h incubation with high glucose in mouse islets and MIN6 cells. The discrepancies between the present study and the findings of Kaneko et al. may be resolved by considering that H2S level is not directly assayed in their study and glucose level in the pre-incubation before treatment was not indicated. The discrepancy may also be related to different responses of different cell types or animal species to H2S and other experimental conditions. The reported effects of CSE/H2S on the survival/apoptosis of pancreatic beta cells are not consistent either [11, 42]. We did not find any reduced cell viability after incubation with high glucose for 24 h compared with basal glucose (data not shown). Our observation was supported by another study in which high glucose promoted pancreatic islet beta cell survival through phosphatidylinositol-3-kinase/Akt signalling at 24 h [43].

Our study did not directly test whether the phosphorylation of SP1 at the Thr453 leads to reduced SP-1 binding to the Cse promoter region. A reduced transcriptional or promoter activity is not always accompanied by reduced DNA binding [44]. Transcription activity could be regulated through changes in the interaction between SP1 and other regulatory factors as a result of SP1 phosphorylation. Thus, altered Cse promoter activity associated with SP1 phosphorylation might be related to other transcription factors, whose activity has been modified via the phosphorylation of SP1.

In conclusion, we report a novel transcriptional mechanism of glucose-induced downregulation of CSE in which SP1 phosphorylation by p38 MAPK acts as one molecular link between glucose level and CSE production as outlined in Fig. 8. A delicate interplay between glucose level and CSE/H2S in pancreatic beta cells would potentially underlie the regulation of glucose metabolism and insulin secretion under both physiological and pathological situations.

Fig. 8
figure 8

Schematic signal transduction pathways underlying glucose-repressed CSE production

Full size image