figure a

Introduction

Type 2 diabetes results from the complex interplay of genetic and environmental risk factors, among which obesity plays a predominant role. Interestingly, most obese individuals compensate for insulin resistance by increasing insulin secretion, thereby maintaining normoglycaemia at the price of hyperinsulinaemia. However, this compensation can be sidestepped by a phase of decompensation in which beta cells fail to uphold an adequate secretory response [1,2,3]. This leads to hyperglycaemia and subsequent glucotoxic alterations of beta cell mass and function [4, 5]. Identifying genes involved in beta cell compensation in obesity, and in beta cell failure in type 2 diabetes may provide new insights into beta cell pathophysiology and reveal novel therapeutic targets to preserve beta cell function in individuals with (pre) type 2 diabetes.

Metallothioneins (MTs) are a family of low molecular mass, cysteine-rich, metal-binding proteins, the (patho)physiological functions of which have not been fully characterised. Their predominant roles are heavy metal detoxification, metal ion (including zinc) homeostasis, and the regulation of cellular redox status and antioxidant defences. Among the four different murine genes encoding MTs, Mt1 and Mt2 are expressed ubiquitously, Mt3 is mainly expressed in neurons and Mt4 is expressed in squamous epithelium cells. In humans, in addition to MT2 (also known as MT2A), MT3 and MT4, there are eight MT1 isoforms for a total of 11 functional MT genes [6, 7]. Interestingly, polymorphisms in MT1A and MT2A have been associated with increased risk for type 2 diabetes and diabetic complications [8,9,10]. Moreover, MT1E, MT1M, MT1X and MT2A mRNA levels were significantly upregulated in islets obtained by laser-capture microdissection (LCM) of pancreatic sections from type 2 diabetes donors [11]. However, the role of MTs in the pathophysiology of type 2 diabetes remains unclear. We, therefore, verified whether changes in Mt1 and/or Mt2 gene expression plays a role in the modulation of beta cell function.

Methods

Reagents

Fura-2 leakage resistant (Fura-2 LR), diazoxide (Dz), carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) and ZnCl2 were from Sigma (St Louis, MI, USA). N,N,N′,N′-tetrakis(2-pyridinylmethyl)-1,2-ethanediamine (TPEN) was from Abcam (Cambridge, UK). Fluozin-3, control non-targeting small interfering RNA (siRNA), ON-TARGETplus SMARTpool siRNA and DharmaFECT3 (transfection reagent) were from Thermo Fisher Scientific (Lafayette, CO, USA).

Mice

Six- and 16-week-old male and female C57BL/KsJ db/db mice and age-matched lean control mice (C57BL/KsJ), 16-week-old male and female C57BL/6J ob/ob mice and age-matched lean control mice, and the 16-week-old wild-type (WT) male mice (C57BL/6JAusb) used in diet-induced obesity experiments were from the Garvan Institute breeding colonies (Australian BioResources, Moss Vale, NSW, Australia). Male Mt1-Mt2 double-knockout (KO) mice at 4–5 months of age (129S7/SvEvBrd-Mt1tm1BriMt2tm1Bri/J; herein referred to as KO mice) and their sex- and age-matched controls (129S1/SvImJ) originated from the Jackson Laboratory (Bar Harbor, MA, USA). Male transgenic mice overexpressing Mt1 under the control of its natural promoter at 3 and 9 months of age (B6.Cg-Tg(Mt1)174Bri/J; referred to as Tg-Mt1) and their sex- and age-matched controls (C57BL/6 J) were also from the Jackson Laboratory. All animals were bred in the local animal facility of the health sciences sector at UCLouvain. Mice were housed under a controlled temperature of 22°C and a 12 h light cycle, with ad libitum access to water and chow (8% energy from fat, 21% energy from protein and 71% energy from carbohydrate [wt/wt]; 10.9 kJ/g; Gordon’s Specialty Stock Feeds, Yanderra, NSW, Australia) or a high-fat diet (HFD; 23% energy from fat, 19.4% energy from protein and 57.6% energy from carbohydrate and fibres [wt/wt for all]; 20.1 kJ/g; catalogue no. SF03-020; Specialty Feeds, Glen Forest, WA, Australia). Mice were randomly assigned to experimental groups using an odd/even number method by unblinded experimenters. All experiments were approved by the Institutional Committee on Animal Experimentation of the Health Sciences Sector at UCLouvain (Project 2013/UCL/MD/016) and the Garvan Institute/St Vincent’s Hospital Animal Experimentation Ethics Committee.

Human islets

Human islets were obtained from 24 non-diabetic and 12 diabetic individuals at the Tom Mandel Islet Transplant Program, Melbourne [12]. Human islets were isolated from heart-beating, brain-dead donors and approved for use under the ethics reference HREC011/04 (St Vincent’s Hospital Human Research Ethics Committee). Characteristics of donors and islet preparations are listed in the electronic supplementary material (ESM) Table 1. To evaluate the effects of glucose on MT gene expression, islets were obtained from three non-diabetic donors through the JDRF award 31-2008-416 (European Consortium for Islet Transplantation [ECIT] Islet for Basic Research program) and approved for use under the ethics reference B403/2017/05JUL/355 (Comité d’éthique hospitalo-facultaire Saint-Luc, UCLouvain). Characteristics of these donors are indicated in ESM Table 2. Written informed consent was obtained from next of kin and all experiments were performed in accordance with relevant guidelines and regulations.

Glucose and insulin tolerance tests

Intraperitoneal (i.p.) glucose tolerance tests (i.p.GTTs) and fasting-refeeding tests were performed on overnight-fasted mice and i.p. insulin tolerance tests (i.p.ITTs) on 4 h-fasted mice. For fasting-refeeding tests, blood glucose levels were measured in fasted mice and 1 h after refeeding. For i.p.GTTs and i.p.ITTs, mice were injected i.p. with glucose (2 g/kg of body weight) or insulin (0.75 U/kg of body weight; Actrapid, Novo Nordisk, Bagsværd, Denmark) and blood glucose levels were measured at the indicated time points using a FreeStyle Precision Neo glucometer (Abbott, Wavre, Belgium).

Tissue collection and histological analysis

Mice were killed by cervical dislocation and trunk blood collected and centrifuged for measurement of plasma insulin levels by ultra-sensitive ELISA (Crystal Chem, Downers Grove, IL, USA). The pancreas, liver, white adipose tissue (WAT) fat pads (epidydimal, retroperitoneal and inguinal), interscapular brown adipose tissue (BAT), skeletal muscles of the right leg (tibialis anterior, extensor digitorum longus [EDL], gastrocnemius and soleus) and heart of each animal were dissected, weighed, and frozen or fixed in 4% (wt/vol.) paraformaldehyde for 24 h at 4°C and embedded in paraffin for further histological analysis. Liver and epidydimal fat pad sections were counter stained with haematoxylin and eosin (H&E) to assess morphology. Adipocyte size was analysed using Visiopharm software (Author module, Version 6.4.1; Hørsholm, Denmark). Insulin and glucagon immunostaining of pancreatic sections was performed as previously described [13, 14]. Beta cell mass was computed based on the relative cross-sectional beta cell area and pancreas weight. Three pancreatic sections at three different levels were analysed for each mouse. Pancreatic sections immunoprobed for insulin, glucagon and a nuclear stain (DAPI) were scanned using the panoramic 250 Flash III digital slide scanner (3DHISTECH, Budapest, Hungary) and analysed with Visiopharm.

Islet isolation and culture

Islets were isolated by collagenase digestion, separated by density gradient centrifugation and handpicked under a stereomicroscope. Islets were cultured in standard RPMI medium (Invitrogen, Carlsbad, CA, USA) supplemented with 2 mmol/l glutamine, 5 g/l BSA, 100 U/ml penicillin and 100 μg/ml streptomycin.

Cell culture and treatment

MIN6 cells (passage 26–43; mycoplasma negative), originally provided by J. I. Miyazaki (Osaka University Medical School, Osaka, Japan) [15], were grown in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen) containing 25 mmol/l glucose, 10 mmol/l HEPES, 10% FCS (vol./vol.), 50 U/ml penicillin and 50 μg/ml streptomycin. Cells were transfected with 100 nmol/l control, Mt1 or Mt2 siRNA using DharmaFECT3 transfection reagent 24 h before experiments.

Glucose-stimulated insulin secretion (GSIS) tests

Isolated islets in batches of five, or MIN6 cells seeded in 24-well plates at 2 × 105 cell per well, were preincubated for 1 h in KRB buffer (120 mmol/l NaCl, 4.8 mmol/l KCl, 2.5 mmol/l CaCl2, 1.2 mmol/l MgCl2, 24 mmol/l NaHCO3 and 1 g/l BSA) containing 3 mmol/l glucose and then incubated for 1 h in KRB buffer containing 3 mmol/l, 15 mmol/l or 30 mmol/l glucose. At the end of the incubation, the buffer was collected for measurement of insulin by RIA, and islets were collected and disrupted by sonication in 10 mmol/l Tris, 0.2 mol/l NaCl and 10 mmol/l EDTA for measurement of their DNA and insulin content. All secretion experiments were carried out in duplicate.

Live-cell imaging

NAD(P)H autofluorescence (excitation/emission wavelength [λex/em], 360/470 nm) was measured every 5 s and expressed as the percentage of the fluorescence level measured after 15–20 min of treatment with 10 μmol/l FCCP in KRB buffer containing 30 mmol/l glucose. For measurements of intracellular Ca2+ concentrations, islets were loaded for 2 h with 2 μmol/l Fura-2 LR acetoxymethyl ester and the fluorescence ratio (λex/em, 340/510 to 380/510 nm) was measured every 5 s. For measurements of intracellular free Zn2+ levels ([Zn2+]i), islets were loaded for 2 h with 2 μmol/l Fluozin-3 acetoxymethyl ester and fluorescence (λex/em, 490/510 nm) was measured every 10 s. TPEN (a zinc chelator) was used as a negative control to lower [Zn2+]i and used at 50 μmol/l. Islets from control and KO/Tg-Mt1 mice were simultaneously perifused side by side with KRB buffer continuously gassed with 5% CO2 air mix to maintain pH 7.4 at a flow rate of 1 ml/min at 37°C on the stage of an inverted microscope.

RNA analysis

Total RNA was extracted and reverse transcribed as previously described [16, 17]. Real-time RT-PCR was performed using the SYBR Green method and a 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) or a CFX96 optical cycler detection system (Bio-Rad, Hercules, CA, USA). Primer sequences are listed in ESM Table 3. The value obtained for a specific gene product was normalised to the control gene cyclophilin A or TBP and expressed as fold change of the value in the control condition. TaqMan assays were used to assess the mRNA levels of MT1E (Hs01938284_g1), MT1X (Hs00745167_sH), MT2A (Hs02379661_g1) and the control gene 18S rRNA (Hs03003631_g1) in human islets from control and type 2 diabetes donors (Applied Biosystems). AmpliTaq Gold reagents (Thermo Fisher Scientific) were used on a Light Cycler 480 Instrument II (Roche, Risch-Rotkreuz, Switzerland).

Protein analysis

Islet MT1 and MT2 protein levels were quantified by ELISA (Frontier Institute, Ishikari, Japan). Absorbance values were normalised to total protein content measured with the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific).

Statistical analysis

Results are means ± SEM for the indicated number of experiments. Statistical significance was assessed by unpaired two-tailed Student’s t test, one-way ANOVA with a Newman–Keuls post hoc test, or two-way ANOVA with a Bonferroni post hoc test.

Results

Mt1 and Mt2 mRNA levels were differentially regulated between conditions of beta cell compensation and failure

Islet Mt1 and Mt2 mRNA levels were markedly downregulated in mice fed for 6 weeks with a HFD vs a chow diet. This was associated with increased body-weight gain and plasma insulin levels (compensation) (Fig. 1a,b and ESM Fig. 1). Similarly, islet Mt1 and Mt2 mRNA levels were significantly downregulated in 16-week-old ob/ob mice (Fig. 1c,d), a model of successful beta cell compensation [17, 18], vs control mice. In db/db mice, which progress from successful beta cell compensation to beta cell failure between 6 and 16 weeks of age [17], islet Mt1 and Mt2 mRNA levels were downregulated in 6-week-old mice (Fig. 1e,f) and upregulated in 16-week-old mice (Fig. 1g,h). These results reveal that, in vivo, beta cell compensation is associated with Mt1 and Mt2 downregulation, whereas beta cell failure is associated with Mt1 and Mt2 upregulation.

Fig. 1
figure 1

Mt1 and Mt2 mRNA levels are downregulated in the islets of obese compensating mice and upregulated in the islets of decompensating diabetic mice. (a–h) Changes in the mRNA levels of Mt1 and Mt2 in the islets of (a, b) chow-fed and HFD-fed WT mice, (c, d) control and ob/ob mice at 16 weeks of age, (e, f) control and db/db mice at 6 weeks of age, and (g, h) control and db/db mice at 16 weeks of age. Data are means ± SEM; (a, b) n = 3, (c, d) n = 7–9, (e, f) n = 8–10, (g, h) n = 15–17 animals per group. *p < 0.05, **p < 0.01, ***p < 0.001 vs chow-fed or control mice, unpaired two-tailed Student’s t test. Ctrl, control

Full size image

Glucose stimulation downregulated the expression of Mt1 and Mt2

WT mouse islets were cultured at various glucose concentrations, ranging from low, non-stimulating concentrations (2–5 mmol/l) to the optimal concentration for culture of rodent islets (10 mmol/l) to a very high glucose concentration (30 mmol/l) [16, 19, 20]. Mt1 and Mt2 mRNA and MT1/2 protein levels were markedly downregulated after culture at glucose at 10 mmol/l vs 2 and 5 mmol/l, with little or no further decrease at 30 mmol/l vs 10 mmol/l glucose (Fig. 2a–c). These effects were anti-paralleled by the stimulation of insulin secretion and the upregulation of the antioxidant genes Mt3, Srxn1 and Gpx2 (Fig. 2d–g). These findings show that, compared with other antioxidant genes, Mt1 and Mt2 have a specific gene expression pattern in response to glucose stimulation. They also show an ex vivo association between islet Mt1 and Mt2 downregulation and the stimulation of insulin secretion.

Fig. 2
figure 2

Mt1 and Mt2 mRNA and MT1/MT2 protein levels are downregulated by glucose stimulation in a concentration-dependent manner. Isolated islets from WT mice were cultured for 24 h (for mRNA analysis) or 48 h (for protein analysis) in the presence of increasing glucose concentrations: 2 mmol/l, 5 mmol/l, 10 mmol/l and 30 mmol/l. (a–c) Changes in mRNA levels of (a) Mt1 and (b) Mt2, and (c) MT1/MT2 protein levels. (d–g) Changes in mRNA levels of (d) Mt3, (e) Srxn1 and (f) Gpx2 and (g) in insulin secretion. Changes in mRNA levels were normalised to cyclophilin A and expressed relative to levels with treatment with 10 mmol/l glucose. Data are means ± SEM of n = 3–4 experiments. *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA with Newman–Keuls post hoc test

Full size image

MT1X mRNA levels were upregulated in human islets from individuals with type 2 diabetes and were affected by glucose stimulation

MT1E, MT1X and MT2A were the most upregulated MT genes in islets obtained by LCM from pancreases of individuals with type 2 diabetes vs islets from non-diabetic donors [11]. Among them, MT1X was the only isoform showing significantly upregulated mRNA levels in islets isolated from type 2 diabetes vs non-diabetic donors. Besides, MT1X displayed higher mRNA levels than MT1E and MT2A in islets isolated from non-diabetic donors (Fig. 3a–d). When islets from non-diabetic donors were cultured in the presence of a low, non-stimulating glucose concentration (2.2 mmol/l), the optimal glucose concentration for culture of human islets (5.5 mmol/l), a high glucose concentration (11.1 mmol/l) and a very high glucose concentration (22 mmol/l) [21, 22], MT1E, MT1X and MT2A mRNA levels were all downregulated between treatment with 2.2 mmol/l and 11.1 mmol/l glucose, while, with 22 mmol/l glucose, MT1E and MT1X mRNA levels returned to a similar level as with 5.5 mmol/l glucose (Fig. 3e–g). In parallel, treatment with glucose at 11.1 and 22 mmol/l significantly increased insulin secretion during culture vs treatment with 2.2 and 5.5 mmol/l glucose (Fig. 3h).

Fig. 3
figure 3

MT1X mRNA levels are upregulated in the islets of human diabetic donors and MT gene isoforms are affected by glucose stimulation. (a–c) Changes in the mRNA levels of (a) MT1E, (b) MT1X and (c) MT2A in the islets of non-diabetic individuals (N) and type 2 diabetic donors (T2D). The mRNA levels of each gene were normalised to 18S RNA. (d) Comparison of the mRNA levels of MT1E, MT1X and MT2A in the islets of non-diabetic individuals. The mRNA levels of each gene were normalised to 18S RNA and the absolute ratios compared. (e–h) Human islets from non-diabetic donors were cultured for 24 h in the presence of increasing glucose concentrations. (e–g) Changes in the mRNA levels of (e) MT1E, (f) MT1X and (g) MT2A during culture. The mRNA levels of each gene were normalised to TBP and expressed relative to levels with treatment with 5.5 mmol/l glucose. (h) Changes in insulin secretion during culture. Data are means ± SEM; (a–d) n = 24 non-diabetic and n = 12 type 2 diabetic donors; (e–h) n = 3–4 experiments. *p < 0.05, **p < 0.01, ***p < 0.001 vs non-diabetic donors or as shown; in (a–c), unpaired two-tailed Student’s t test; in (d–h), one-way ANOVA with Newman–Keuls post hoc test

Full size image

These results show that MT1X upregulation is associated with beta cell failure in human type 2 diabetes. They also reveal a specific gene expression pattern of human MT genes in response to glucose stimulation that partly resembles that of Mt1 and Mt2 in mouse islets.

Deletion of Mt1/Mt2 improved glucose tolerance

The potential role of Mt1 and/or Mt2 in the modulation of glucose homeostasis and beta cell function was investigated in a global Mt1-Mt2 double-KO mouse model [23]. Compared with WT mice, KO mice displayed higher body weight and daily food intake (ESM Fig. 2a,b). They also showed increased liver weight (ESM Fig. 2c) without macroscopical or histological signs of steatosis (ESM Fig. 2d), and increased weight of different leg muscles, including tibialis anterior, EDL and gastrocnemius (ESM Fig. 2e–g). However, soleus muscle and heart weights were similar (ESM Fig. 2h,i). Interestingly, the weight of epidydimal, inguinal and retroperitoneal fat pads and the sum of the three fat pads (WAT), thereof, were reduced in KO mice (ESM Fig. 2j–m), whereas BAT weight was similar between KO and WT mice (ESM Fig. 2n). Histological sections of epidydimal fat pads also revealed reduced adipocyte surface in KO mice (ESM Fig. 2o,p).

Fed blood glucose levels were similar in KO and WT mice (Fig. 4a), overnight-fasted blood glucose levels were slightly higher in KO mice (Fig. 4b) and fed and fasted plasma insulin levels were not significantly different (Fig. 4c,d). Interestingly, glucose tolerance during i.p.GTT was markedly improved in KO mice, together with significantly increased plasma insulin levels (30 min following i.p.GTT) (Fig. 4e–g). Similarly, during an overnight fasting/1 h refeeding test, blood glucose levels after refeeding were lower in KO mice (Fig. 4h). On the other hand, insulin sensitivity during i.p.ITT was similar in KO and WT mice (Fig. 4i,j). Together, these findings suggest that Mt1-Mt2 deletion leads to improved glucose tolerance due to increased insulin secretion rather than changes in insulin action.

Fig. 4
figure 4

Mt1-Mt2 deletion improved glucose tolerance in vivo. (a–d) Changes in (a) fed and (b) fasted blood glucose levels, and (c) fed and (d) fasted plasma insulin levels in WT and KO mice. (e) Changes in blood glucose levels and (f) AUC during an i.p.GTT in overnight-fasted WT and KO mice. (g) Plasma insulin levels 30 min following i.p.GTT. (h) Changes in blood glucose levels in WT and KO mice at the fasted state and 1 h after refeeding. (i) Changes in blood glucose levels and (j) AUC during an i.p.ITT in 4 h-fasted WT and KO mice. WT, white bars/circles; KO mice, blue bars/circles. Mice were used between 4 and 5 months of age. Data are means ± SEM; (a) n = 9, (b) n = 13, (c) n = 8–9, (d) n = 9–10, (e, f) n = 7, (g) n = 4, (h) n = 7–9, (i, j) n = 5 animals per group. *p < 0.05, ***p < 0.001 vs WT. In (h) ††p < 0.01 as shown; ***p < 0.001 for the effect of refeeding. In (b), (f) and (g), unpaired two-tailed Student’s t test; in (e) and (h), two-way ANOVA with Bonferroni post hoc test

Full size image

MT deletion potentiated GSIS

The mechanism underlying improved glucose tolerance in KO mice was further investigated in isolated islets. In WT islets, MT1/2 protein levels were expressed under control culture conditions and upregulated by treatment with ZnCl2 (a potent inducer of MT expression). In contrast, MT1/2 proteins were not detected in KO islets, even after treatment with ZnCl2, confirming the lack of Mt1-Mt2 expression (Fig. 5a). Interestingly, GSIS was potentiated in KO vs WT islets after acute stimulation with 15 mmol/l glucose, and to a stronger extent after stimulation with 30 mmol/l glucose (Fig. 5b), while islet insulin content was unchanged in KO vs WT mice (Fig. 5c). In agreement, MT1 but not MT2 knockdown in MIN6 cells potentiated GSIS (Fig. 5d–f), highlighting the role of Mt1, rather than Mt2, in the negative regulation of insulin secretion.

Fig. 5
figure 5

Mt1 deletion potentiated GSIS. (a) Changes in MT1/MT2 protein levels after 24 h culture of WT and Mt1-Mt2 double-KO islets in the absence or presence of 100 μmol/l ZnCl2. (b) GSIS in WT (white bars) and KO (blue bars) islets and (c) islet insulin content (n = 24). (d, e) Changes in the mRNA levels of Mt1 and Mt2 and (f) GSIS in MIN6 cells transfected with either control siRNA (siC; white bars), Mt1 siRNA (siMt1; blue bars) or Mt2 siRNA (siMt2; grey bars). GSIS experiments were done in duplicate. (g) Changes in NAD(P)H autofluorescence normalised for each experiment to the fluorescence level after 15 min treatment with FCCP (n = 4). (h) Changes in Fura-2 LR fluorescence ratio after 2 h loading with 2 μmol/l of the Ca2+ probe (n = 5). WT and KO islets were perifused simultaneously in the same chamber. WT, black traces; KO, blue traces. Data are means ± SEM of n = 3–4 experiments. *p < 0.05, ***p < 0.001 for the effect of ZnCl2 or glucose. †p < 0.05, ††p < 0.01, †††p < 0.001 for the effect of genotype or siRNA. In (d, e), one-way ANOVA with Newman–Keuls post hoc test; in (a), (b) and (f), two-way ANOVA with Bonferroni post hoc test. Dz, diazoxide; K30, 30 mmol/l KCl; Gn, n mmol/l glucose

Full size image

We also measured the effects of acute stepwise increases in glucose concentration on intracellular NAD(P)H and Ca2+ levels and found no differences between WT and KO islets (Fig. 5g,h). These findings indicate that the potentiation of GSIS in KO islets results from an effect downstream of the stimulation of glucose metabolism and Ca2+ influx.

MTs are known for their metal-binding properties and proposed to play a role in metal ion, including zinc, homeostasis. As zinc plays a key role in beta cell biology, we used Fluozin-3 to compare the dynamic changes in [Zn2+]i in response to glucose and zinc supplementation/chelation in islets from WT and KO mice (ESM Fig. 3a). In islets from both mouse types, [Zn2+]i was increased upon supplementation of 3 mmol/l glucose-KRB buffer with 10 μmol/l ZnCl2, slightly decreased upon subsequent stimulation with 30 mmol/l in the continued presence of 10 μmol/l ZnCl2, rapidly decreased upon ensuing zinc chelation using TPEN, and markedly increased upon final addition of 1 mmol/l ZnCl2 to the medium. These effects were almost identical in WT and KO islets. In agreement, the mRNA levels of the key beta cell zinc transporters Zip6 (also known as Slc39a6), Zip7 (Slc39a7), Znt1 (Slc30a1) and Znt8 (Slc30a8) were similar between KO and WT islets (ESM Fig. 3b–e). These results rule out a potential role of changes in [Zn2+]i levels in the potentiation of GSIS in KO islets.

Assessment of the pancreas morphology revealed no difference in islet architecture between islets from WT and KO mice (ESM Fig. 4a). There were no significant changes in pancreatic weight, beta and alpha cell masses, or the percentage of alpha/beta cells per islet area between KO and WT mice (ESM Fig. 4b–e). Moreover, islets from WT and KO mice displayed no difference in the mRNA levels of the beta cell-enriched genes preproinsulin, Pdx1, Glut2 (also known as Slc2a2), Pcx and Gpd2 (ESM Fig. 4f–j). Similarly, there were no changes in the mRNA levels of the endoplasmic reticulum (ER) stress-response genes Hspa5 and Ddit3 (ESM Fig. 4k,l). Importantly, the deletion of Mt1 and Mt2 was not compensated for by upregulation of other MT genes. Thus, Mt3 mRNA levels were unchanged in islets from KO vs WT mice (ESM Fig. 4m), while Mt4 mRNA levels were undetected after 40 cycles of PCR amplification in both islet types.

Altogether, these findings strongly support a role for Mt1 as a negative modulator of GSIS.

Mt1 overexpression attenuated GSIS

To further assess the implication of Mt1 in the negative regulation of GSIS, we examined islets isolated from global transgenic mice overexpressing mouse Mt1 under the control of its natural promoter (Tg-Mt1) [24, 25]. In comparison with WT mice, Tg-Mt1 mice exhibited similar body and liver weights (ESM Fig. 5a–c). They displayed no significant difference in the weight of epidydimal (p = 0.0519), inguinal and retroperitoneal fat pads or the sum of WAT, thereof (ESM Fig. 5d–g). There was also no difference in BAT weight (ESM Fig. 5h). Tg-Mt1 mice also displayed similar fed blood glucose levels (Fig. 6a) and plasma insulin levels (Fig. 6b) to WT mice, and a fasting-refeeding test revealed no significant differences between the two groups (Fig. 6c). Similarly, i.p.GTT tests were similar in 3-month-old and 9-month-old animals (Fig. 6d,e). In contrast, i.p.ITT tests revealed a notable difference between WT and Tg-Mt1 mice, with Tg-Mt1 mice displaying lower blood glucose levels during the test (Fig. 6f,g). This effect may stem from an impact of Mt1 overexpression on peripheral tissues.

Fig. 6
figure 6

Mt1 overexpression did not affect glucose tolerance but affected insulin tolerance in vivo. (a) Changes in fed blood glucose levels and (b) fed plasma insulin levels in WT and Tg-Mt1 mice. (c) Changes in blood glucose levels in WT and Tg-Mt1 mice at the fasted state and 1 h after refeeding. (d, e) Changes in blood glucose levels during an i.p.GTT in overnight-fasted WT and Tg-Mt1 mice at the age of (d) 3 months and (e) 9 months. (f) Changes in blood glucose levels during an i.p.ITT in 4 h-fasted WT and Tg-Mt1 mice and (g) respective AUC. WT, white bars/circles; Tg-Mt1, red bars/circles. Data are means ± SEM; (a, b) n = 5–6, (d) n = 4, (e) n = 6–8, (f, g) n = 4 animals per group. *p < 0.05 for the effect of genotype (f); ***p < 0.001 for the effect of refeeding (c), two-way ANOVA with Bonferroni post hoc test. Tg, Tg-Mt1 mice

Full size image

In isolated islets, MT protein levels were markedly upregulated in Tg-Mt1 islets, confirming the overexpression of Mt1 (Fig. 7a). Interestingly, in contrast with its potentiation in KO vs WT islets, GSIS was significantly attenuated in Tg-Mt1 vs WT islets after acute stimulation with 30 mmol/l glucose (Fig. 7b), while islet insulin content was similar between the two islet types (Fig. 7c). This attenuation in GSIS occurred despite similar rises in intracellular NAD(P)H and Ca2+ levels in islets from WT and Tg-Mt1 mice in response to stepwise increases in glucose (Fig. 7d,e), indicating that the alteration of GSIS in Tg-Mt1 mice vs WT mice lies at a step downstream of the stimulation of glucose metabolism and Ca2+ influx.

Fig. 7
figure 7

Mt1 overexpression attenuated GSIS. (a) Changes in MT1/MT2 protein levels in WT and Tg-Mt1 islets. (b) GSIS in islets from WT (white bars) and Tg-Mt1 (red bars) mice and (c) islet insulin content (n = 18). (d) Changes in NAD(P)H autofluorescence normalised for each experiment to the fluorescence level after 15 min treatment with FCCP (n = 4). (e) Changes in Fura-2 LR fluorescence ratio after 2 h loading with 2 μmol/l of the Ca2+ probe (n = 3). WT and Tg-Mt1 islets were perifused simultaneously in the same chamber. WT, black traces; Tg-Mt1, red traces. Data are means ± SEM; (a, b, d) n = 4, (e) n = 3 experiments. **p < 0.01, ***p < 0.001 for the effect of glucose; †p < 0.05, ††p < 0.01 for the effect of genotype. In (a), unpaired two-tailed Student’s t test; in (b), two-way ANOVA with Bonferroni post hoc test . Dz, diazoxide; K30, 30 mmol/l KCl; Gn, n mmol/l glucose; Tg, Tg-Mt1 mice

Full size image

Similarly, changes in [Zn2+]i in response to glucose and zinc supplementation/chelation were not different between islets from WT and Tg-Mt1 mice (ESM Fig. 6a). In agreement, the mRNA levels of zinc transporters Zip6, Zip7, Znt1 and Znt8 were unchanged in islets from Tg-Mt1 mice vs WT mice (ESM Fig. 6b–e).

Importantly, in comparison with islets from WT mice, Tg-Mt1 islets exhibited a similar expression level of the beta cell-enriched genes preproinsulin, Pdx1, Glut2 and Pcx (Fig. 8a–d). In addition, the mRNA levels of ER stress-response genes Hspa5 and Ddit3 were also not different between the two mouse models (Fig. 8e,f). On the other hand, the overexpression of Mt1 (Fig. 8g) resulted in reduced mRNA levels of Mt2 (p < 0.05) and Mt3 (p = 0.0704) in comparison with WT islets (Fig. 8h,i), while Mt4 mRNA levels were not detected after 40 cycles of PCR amplification in both types of islets.

Fig. 8
figure 8

Mt1 overexpression did not affect the expression of beta cell-enriched genes and ER stress-response genes. Changes in the mRNA levels of (a–d) preproinsulin, Pdx1, Glut2 and Pcx, and (e, f) ER stress-response genes Hspa5 and Ddit3 and (g–i) MT gene isoforms Mt1, Mt2 and Mt3. Data are means ± SEM of n = 3–4 animals per group. *p < 0.05 vs WT, unpaired two-tailed Student’s t test. Tg, Tg-Mt1 mice

Full size image

Collectively, these findings further support the implication of Mt1 in the negative regulation of insulin secretion.

Discussion

We have unveiled a novel role of Mt1 in beta cells as a negative regulator of insulin secretion. The key findings of the study are: (1) Mt1 and Mt2 islet gene expression in obese mice was downregulated with beta cell compensation and upregulated with beta cell failure; (2) MT1X islet mRNA levels were upregulated in human type 2 diabetes donors; (3) physiological and supraphysiological glucose stimulation downregulated mouse and human MT islet gene expression; (4) deletion of Mt1 and Mt2 improved glucose tolerance in vivo and potentiated GSIS in isolated islets; (5) knockdown of MT1, but not MT2, potentiated GSIS in MIN6 cells; and (6) Mt1 overexpression attenuated GSIS in isolated islets (Table 1). These cumulative findings strongly support the implication of Mt1 in the negative regulation of beta cell function.

Table 1 Overview of the principal findings of the study
Full size table

Mt1 and Mt2 exhibit an atypical gene expression pattern in comparison with other antioxidant genes

MTs are known for their protective antioxidant properties [26,27,28]. Oxidative stress plays an important role in beta cell demise and islets of humans with diabetes and animal models display upregulated expression of many antioxidant genes and markers of oxidative damage [4]. Interestingly, antioxidant genes like Hmox1, Gpx1, Gpx2, Sod1 and Nrf2 (also known as Nfe2l2) were also upregulated in the islets of compensating young db/db mice and mice fed an HFD [17, 29, 30], in sharp contrast with Mt1 and Mt2 downregulation (Fig. 1). Similarly, in isolated islets, treatment with glucose upregulated Mt3, Srxn1 and Gpx2, while it downregulated Mt1 and Mt2 (Fig. 2 and [31, 32]). Noteworthy, our previous studies in Wistar rats revealed that islet Mt1 expression was upregulated by fasting and downregulated upon refeeding (J-C Jonas, unpublished data). Furthermore, a recent study exploring beta cell heterogeneity by single-cell transcriptomics revealed that high activity of the insulin gene promoter was associated with low expression of Mt1 and Mt2 and vice versa [33]. These observations demonstrate that Mt1 and Mt2 behave differently from other oxidative stress-response genes. Besides, Mt1-Mt2 deletion does not affect antioxidant/stress-response gene expression (ESM Fig. 4 and ESM Fig. 7), indicating no evident impact on islet redox status under physiological conditions. Thus, Mt1 and/or Mt2 may play a role in beta cell (patho)physiology that goes beyond their known antioxidant function.

Mt1 negatively regulates GSIS

Deletion of Mt1 and Mt2 improved glucose tolerance and potentiated GSIS in isolated islets from KO mice. These findings contrast with a previous study using islets from Mt1-Mt2 KO mice [34]. Although differences in genetic backgrounds may contribute to this discrepancy, our study is more comprehensive than the previous investigation. Thus, besides our complementary in vivo and ex vivo results, the knockdown of MT1 in MIN6 cells reproduced the enhanced insulin secretory phenotype of the KO islets. Furthermore, in our study, islets overexpressing Mt1 exhibited the reciprocal phenotype.

Our results underscore the role of Mt1, rather than Mt2, in the modulation of insulin secretion. Although these genes exhibit similar expression patterns, Mt1 displays a higher expression level in comparison with Mt2 and Mt3 in primary mouse islets (ESM Fig. 8). In addition, mRNA sequence alignment analysis showed that Mt1 and Mt2 are only ~80% identical. Thus, the ~20% sequence difference may also underlie functional specificities. Such specificities may stem from different metal affinities and binding properties or from specific protein–protein interactions [6, 35].

How could Mt1 impact on insulin secretion?

We did not detect significant changes in classical metabolic (NAD(P)H) and ionic parameters (cytosolic free Ca2+ and Zn2+) in response to glucose stimulation between islets from WT mice and KO or Tg-Mt1mice. The expression of antioxidant and other stress-response genes was not different between WT and KO islets, thereby ruling out a potential impact on redox status (ESM Fig. 4 and ESM Fig. 7). Furthermore, Mt1-Mt2 deletion or Mt1 overexpression had no impact on cell death in islets cultured under control conditions (10 mmol/l glucose; ESM Fig. 9). Nevertheless, insulin secretion in response to high potassium was also potentiated in islets from KO vs WT mice (ESM Fig. 10). This result strongly suggests a potential impact of Mt1 on the beta cell exocytotic machinery. Interestingly, MT3 was previously shown to interact with Rab3A GTPase in neurons, thereby playing a role in presynaptic vesicle trafficking [36]. One may, therefore, hypothesise that MT1 interacts with a yet-to-be-identified protein of the beta cell exocytotic machinery to modulate insulin secretion.

The mechanism(s) of induction of Mt1 in diabetes

Hyperglycaemia, per se, may not be the upstream factor involved in increased expression of Mt1 in diabetes. Indeed, although Mt1 mRNA expression transiently increases in rat islets cultured overnight with glucose at 30 mmol/l vs 10 mmol/l [16], prolonged exposure of mouse and rat islets to high glucose (30 mmol/l) had little or no impact (vs 10 mmol/l) on Mt1 and Mt2 mRNA and MT1/MT2 protein levels (Fig. 2 and [32]). In human islets, culture in the presence of the already supraphysiological glucose concentration of 11.1 mmol/l vs 2.2 and 5.5 mmol/l also downregulated the mRNA levels of MT genes. Upregulation beyond this concentration may result from differences in glucose sensitivity and metabolism between human and mouse islets and may involve the activation of glucotoxicity-related pathways [4]. We have previously shown that rat islet Mt1 expression is markedly induced by exogenous H2O2, the sarcoendoplasmic reticulum Ca2+-ATPase (SERCA) pump inhibitor thapsigargin, the cytokine IL-1β and hypoxia [37]. However, the islet expression of genes induced by oxidative stress, ER stress and inflammatory stress are observed in the prediabetic stage in db/db mice [17], suggesting alternate mechanisms. On the other hand, evidence from several studies implicates a possible role of hypoxia: Mt1 mRNA expression is upregulated by hypoxia in mouse and human islets (ESM Fig. 11 and [38]), and a clear temporal in vivo association is observed between MT expression and an hypoxic gene expression signature in islets of db/db mice [39, 40].

Limitations of the study and perspectives

In this study, we used global KO animals. Since Mt1-Mt2 deletion may affect other metabolic tissues, one may argue that the observed secretory phenotype involves the effect of systemic factors. However, this is unlikely as we systematically precultured isolated islets for 1 week before GSIS tests. We also combined different models to demonstrate that Mt1 negatively regulates insulin secretion, i.e. by confirming the secretory phenotype after MT1 knockdown in MIN6 cells. Additionally, the attenuation of GSIS by Mt1 overexpression further supports our hypothesis. Although the Tg-Mt1 model is also global, it presents two important advantages: (1) the overexpression of the mouse Mt1 gene rather than human MT2A gene [41]; and (2) the control of Mt1 by its natural promoter rather than the insulin promoter, thereby avoiding the ER stress and oxidative stress observed in MT2A-transgenic mice [41]. Indeed, our Tg-Mt1 islets displayed normal expression of ER stress-response genes (Fig. 8). Nevertheless, development of an Mt1-floxed mouse model is warranted for further exploration of these novel roles of Mt1 in beta cell biology.

Conclusion

Mt1 negatively regulates insulin secretion. Downregulation of islet Mt1 in obesity may, thus, contribute to beta cell compensation, and its upregulation in type 2 diabetes may contribute to beta cell failure. Inhibition of Mt1 may, therefore, represent an attractive therapeutic target to preserve and restore insulin secretion in type 2 diabetes.