Introduction
Pancreatic beta cells are highly dependent on oxidative metabolism for ATP synthesis, particularly at elevated glucose concentrations [
1,
2]. Correspondingly, hypoxia has been shown to influence islet survival and function during transplantation [
3,
4]. Moreover, as many as 25% of islets are exposed in vivo to low oxygenation [
5], suggesting that hypoxia acts as a regulator of islet function under physiological conditions. Indeed, glucose-induced oxygen consumption creates intracellular hypoxia sufficient to activate hypoxia-inducible factors (HIFs) in rat beta cells [
6], an effect that is increased in diabetic animals [
7] and which may contribute to defective insulin secretion in some forms of type 2 diabetes.
Hypoxic stress induces genes such as metallothionein (
MT1/
2) as a defence against large changes in free metal ion concentration [
8], which may affect the activity of anti-oxidative enzymes [
9]. Whether genetic factors influence the susceptibility of pancreatic beta cells to hypoxia has not previously been explored. Suggesting this as a possibility, genome-wide association studies (GWAS) have revealed that a non-synonymous single nucleotide polymorphism (rs13266634) in the
SLC30A8 gene (encoding the secretory granule-resident zinc transporter 8 [ZnT8]) is associated with a ~20% increase in disease risk per allele [
10‐
12].
SLC30A8 expression is largely confined to pancreatic beta and alpha cells [
13], and is required for the accumulation of zinc into the secretory granule, where it binds to insulin [
14,
15]. Consequently, mice inactivated systemically for
Slc30a8 display profound changes in insulin crystal formation and secretory granule morphology [
14,
15], consistent with the lower zinc-transporting activity of the transporter isoform encoded by the risk allele [
15]. Defective insulin secretion is seen in global
ZnT8
−/− mice on some [
15], but not all, backgrounds [
14,
16,
17], and mice with selectively deleted beta cell
Slc30a8 display marked changes in insulin secretion and glucose tolerance [
18]. Whether
Slc30a8 influences cytosolic, as well as granular, Zn
2+ concentrations has not previously been examined because of the uncertain subcellular targeting of the probes used in earlier work [
15].
Monitoring cytosolic Zn
2+ with the molecularly targeted recombinant probe eCALWY4 [
19], the present study aimed to explore the impact of hypoxia on Zn
2+ homeostasis, and the expression of
SLC30A8/ZnT8 and other zinc transporters and importers, in human and rodent beta cells.
Discussion
We demonstrate here that hypoxia strongly, but reversibly, regulates the expression of
Slc30a8/ZnT8 in islets and beta cells from human and two rodent species. These findings thus extend the list of pathophysiological factors, currently including cytokines [
31,
37] and fatty acids [
36], which regulate the expression of this type 2 diabetes risk gene in islets. Hypoxia might therefore contribute to the downregulation of
Slc30a8 previously observed in human type 2 diabetes islets [
38].
The present findings extend to the islet beta cell those of a recent report [
39] showing that hypoxia lowers the expression of
Slc30a8/ZnT8 in the retinal pigment epithelium of the eye. The latter studies [
39] provided evidence for control of ZnT8 levels via HIF1 stabilisation. By contrast, in our hands, the HIF1α-stabilising agent DMOG tended only slightly to reduce
ZnT8 mRNA levels. Instead, our results suggest that the effects of hypoxia on
Slc30a8/ZnT8 expression in pancreatic islets are more complex and may conceivably involve changes in the expression of
Pdx1, recently shown to control
Slc30a8 expression in clonal beta cells [
40].
Recently, Lefebvre and colleagues [
36] reported that depletion of intracellular zinc reduced
Slc30a8 expression in human islets, confirming findings in INS-1E cells [
41]. These earlier observations thus raise the possibility that the lowered cytosolic Zn
2+ concentrations reported here during hypoxia may contribute to the lowering of
Slc30a8/ZnT8 expression. Arguing against this view, addition of extracellular Zn
2+ (30 μmol/l ZnCl
2) did not rescue the lowering of
Slc30a8 mRNA levels after hypoxia (ESM Fig.
7). Moreover, any direct or indirect (e.g. by alterations in cytosolic Zn
2+) effect of hypoxia-driven metallothionein gene induction also seems unlikely, as hypoxia-induced changes in
Slc30a8 mRNA levels were not different in
Mt1/
Mt2
−/− mice compared with control animals (Fig.
5a).
We propose instead a model by which hypoxia initially depresses
Slc30a8 transcription and/or mRNA stability by mechanisms which remain to be elucidated (as discussed below). This then leads to a lowering of cytosolic free Zn
2+, and a consequent drop in
Mt1/
Mt2 expression. The above sequence of events is supported by the observation of reduced cytosolic Zn
2+ in
ZnT8
−/− mice compared with controls (Fig.
6d), and by an increased cytosolic Zn
2+ concentration in clonal pancreatic beta cells when
Slc30a8 was overexpressed (Fig.
6f).
The observation that cytosolic free Zn
2+ is lowered by ablating
Slc30a8, a mediator of Zn
2+
uptake into granules, may at first glance appear surprising. We would stress that the use here of a molecularly targeted cytosolic Zn
2+ probe excludes uncertainties over the subcellular compartment in which Zn
2+ concentrations are interrogated. So how might this apparent paradox be explained? First, our data suggest that ZnT8 may, under normal circumstances, catalyse Zn
2+
efflux from granules, in line with bidirectional transport by ZnT5 [
42]. Alternatively, gradual release of Zn
2+ from granules by exocytosis may elevate the extracellular Zn
2+ concentration in the medium surrounding WT, but not
ZnT8
−/−, cells for which granule Zn
2+ content is near zero (see Chimienti et al and Li et al [
13,
43]). The released Zn
2+ may then be recaptured, at least in part, and taken into the cytosol, by plasma-membrane-located Zn
2+-uptake systems (e.g. ZIP1, voltage-gated Ca
2+ channels) [
30]. However, we calculate that release of Zn
2+ ions into the medium is likely to increase total external Zn
2+ concentration to only a miniscule extent (<50 nmol/l) vs a total extracellular concentration of >6 μmol/l [
44], although local concentrations at the cell surface [
14,
43] may be higher.
One of the effects of ZnT8 inhibition and consequently lowered cytosolic Zn
2+ levels was impaired metallothionein induction in response to hypoxia (Fig.
6a, b). However, this did not affect islet cell survival under the conditions examined, despite the fact that islets lacking metallothionein-1 and -2 exhibited a twofold higher frequency of beta cell death compared with control islets (Fig.
5c). The latter finding is consistent with earlier data showing that overexpression of metallothionein protects islets against hypoxia, leading to improved islet cell survival [
45]. We assume, therefore, that residual levels of metallothionein (~50%) in
ZnT8
−/− mouse islets are sufficient to prevent an increase in cell death.
Unexpectedly, in small islets from older mice, we observed a significantly lower rate of cell death in
ZnT8
−/− compared with WT islets. This finding is in line with our previous report on glucose homeostasis in these mice, where a compromising effect of ZnT8 deficiency disappeared with age [
15], and suggests that, with ageing, a deleterious effect of ZnT8 deficiency may revert to one of protection. On the other hand, the lack of a genotype effect in large islets may be due to the variance of oxygen concentration in these islets, where the core is relatively anoxic as a consequence of larger diffusion distance when exposed to hypoxia in vitro. Nonetheless, the expression of hypoxia-inducible genes (with the exception of
Mt1/
Mt2) tended to be enhanced in
ZnT8
−/− mice, possibly reflecting regulation of HIF1α by Zn
2+ [
46]. Importantly, the current results may provide an explanation for the recent finding that rare loss-of-function mutations in the
SLC30A8 gene in man are associated with protection against type 2 diabetes [
47].
Reduced expression of ZnT8 in hypoxia may thus reflect an ‘adaptive’ response of beta cells to permit survival under a hypoxic/oxidative stress in a less differentiated state (as previously described after partial pancreatectomy-induced hyperglycaemia [
48]). The mechanisms behind this observation remain to be elucidated, but may suggest that reduced zinc levels are beneficial in the situation of increased hypoxic stress. Indeed, it has previously been shown that high concentrations of Zn
2+ are able to induce islet cell death in a dose-dependent manner [
49,
50]. A similar rate of cell death after 24 h and 48 h of hypoxia, as observed here, suggests that, once adapted to hypoxia, islets are better able to survive. Further studies will be required, however, to determine whether pharmacological modification of ZnT8 function may be beneficial in islet transplantation or type 2 diabetes mellitus.
Acknowledgements
We thank L. Zhao and E. Oliver Perez (Imperial College London, London, UK) for assistance with the hypoxia experiments. We would also like to thank P. Chabosseau (Imperial College London, London, UK) for assistance with measurements of cytosolic Zn2+ and R. Zullig (University of Zurich, Zurich, Switzerland) for assistance with islet preparation.
Preliminary reports of these findings have previously been published in abstract form [
51] and reported at the 49th EASD Annual Meeting in Barcelona in 2013.