Imaging free Zn²⁺ in living cells

While total zinc levels in cells are in the mm range (Table 1)⁽ Reference Maret ^{4 )}, labile i.e. free concentrations are many orders of magnitude lower (pm–nm) in the range likely to regulate physiological targets. However, these concentrations are likely to differ between cell types, different intracellular organelles and in response to environmental perturbation or stimulation.

A deeper understanding of the role of zinc in cell biology and cell signalling has required the development of sensitive and non-invasive tools which provide both spatial and temporal resolution. Fluorescence microscopy is usually adopted and uses one of two probe types⁽ Reference Dean, Qin and Palmer ^{16 )}. First, low molecular weight compounds which display a massive change in fluorescence intensity upon zinc chelation can be loaded into the cell. The second class are genetically-encoded zinc sensors, usually expressed from an introduced cDNA and which bind Zn²⁺ ions through a defined metal-binding protein domain to influence the fluorescence of protein-based fluorophores to which it is fused. These zinc sensors/probes differ according to their affinity for zinc (or sensitivity, reflected by the dissociation constant K _d), their selectivity for zinc against other metals ions and their dynamic range i.e. the change in fluorescence intensity triggered by zinc binding.

Chemically synthesised, low molecular weight probes can be divided in two categories: intensity-based and ratiometric (the latter will not be described here). Intensity-based probes are ‘turn-on’ fluorophores, and display a chelation-dependent increase in fluorescence intensity of up to 100-fold. Most of these probes are based on the modulation in the photon-induced electron transfer phenomenon⁽ Reference Dean, Qin and Palmer ^{16 –} Reference de Silva, Moody and Wright ^{18 )}. Briefly, the probes are composed of a fluorophore, a spacer domain and an electron-rich metal chelate. Photoexcitation of the fluorophore is relaxed through electron transfer with the chelate and this fluorescence quenching is supressed upon metal–ion binding. A variety of fluorescent probes with varying affinities and excitation/emission wavelengths have been described⁽ Reference Carter, Young and Palmer ^{19 )}. The first generation, including Zinquin⁽ Reference Zalewski, Forbes and Betts ^{20 )}, was derived from quinoline, a UV-excitable fluorophore. To resolve problems related to UV-excitation the next generation ZinPyr family (from ZP1 to ZP10), ZnAF, etc. was developed using fluorophores excitable with visible light. The widely-used FluoZin-3, originally based on a Ca²⁺ probe⁽ Reference Gee, Zhou and Qian ^{21 )}, displays a high affinity for zinc (K _d = 15 nm). Trappable versions based on the intracellular cleavage of acetoxymethyl esters⁽ Reference Tsien ^{22 )} and improved in terms of brightness and photostability, have also been developed⁽ Reference Carter, Young and Palmer ^{19 )}.

Compartmentalisation of these probes is not readily controlled and can vary depending on the cell type⁽ Reference Carter, Young and Palmer ^{19 ,} Reference Qin, Miranda and Stoddard ^{23 ,} Reference Thompson, Dockery and Horobin ^{24 )}. Biological targeting is, however, possible e.g. to mitochondria with the addition of a thiamine pyrophosphate group⁽ Reference Chyan, Zhang and Lippard ^{25 )}. Attachment to the cell surface to allow sampling of extracellular Zn²⁺ has also been achieved, and used to measure co-release of Zn²⁺ alongside insulin⁽ Reference Li, Chen and Bellomo ^{26 ,} Reference Pancholi, Hodson and Kobe ^{27 )}. The latter ‘click-SnAr-click’ strategy was also used to add targeting moieties to other intracellular localisations (e.g. mitochondria and lysosomes)⁽ Reference Pancholi, Hodson and Kobe ^{27 )}.

Genetically-encoded zinc sensors use Fӧrster Resonance Energy Transfer as the sensing modality and consist of a donor and acceptor fluorescent protein linked by a zinc-binding peptide sequence⁽ Reference Hessels and Merkx ^{28 )}. Two families have been designed by the Palmer group based on zinc fingers: Zif- and Zap-sensors. The low (μm) affinity Zif family is derived from the mammalian transcription factor Zif268, and contains either a wild type zinc fingers (ZifCY1), or a mutated (ZifCY2) domain⁽ Reference Dittmer, Miranda and Gorski ^{29 ,} Reference Qin, Dittmer and Park ^{30 )}. The Zap sensors, based on the Saccharomyces cerevisiae transcriptional regulator Zap1, have a very high (pm) affinity for zinc⁽ Reference Qin, Dittmer and Park ^{30 )}. The first member of the family ZapCY1 showed a K _d of 2·5 pm, and was saturated when expressed in HeLa cells. ZapCY2 has a decreased affinity (Kd = 811 pm) and is suitable for in cellulo measurements.

The eCALWY sensors⁽ Reference Vinkenborg, Nicolson and Bellomo ^{7 )} developed by Merkx and colleagues, consist of two cysteine-containing metal binding domains (ATOX1 and WD4) connected by a long, flexible glycine−serine linker and flanked by modified cerulean and citrine fluorescent proteins. Shortening of the linker length between the metal binding domains and/or mutation of one of the metal binding cysteines in the WD4 domain yielded a series of sensor variants showing affinities (pm–nm)⁽ Reference Vinkenborg, Nicolson and Bellomo ^{7 )}.

Taking advantage of simple fusion with a targeting sequence, several zinc sensors have been addressed to different organelles such as the mitochondria (mito-eCALWY-4, mito-ZapCY1)⁽ Reference Chabosseau, Tuncay and Meur ^{8 ,} Reference Park, Qin and Galati ^{31 )}, the endoplasmic reticulum (ER-eCALWY-4, ER-ZapCY1)⁽ Reference Chabosseau, Tuncay and Meur ^{8 ,} Reference Qin, Miranda and Stoddard ^{23 )}, Golgi apparatus (golgi-ZapCY1)⁽ Reference Qin, Dittmer and Park ^{30 )} nucleus (NLS-Zaps)⁽ Reference Miranda, Weaver and Qin ^{32 )} and insulin-secreting vesicles (vamp2-eCALWYs, vamp2-eZinCh1)⁽ Reference Vinkenborg, Nicolson and Bellomo ^{7 )}.

A summary for the results obtained with these probes is presented in Fig. 2. While similar values were returned for all probes when located in the cytosol, i.e. 0·1–1·5 nm ⁽ Reference Vinkenborg, Nicolson and Bellomo ^{7 ,} Reference Chabosseau, Tuncay and Meur ^{8 ,} Reference Hessels and Merkx ^{28 ,} Reference Qin, Dittmer and Park ^{30 ,} Reference Bellomo, Meur and Rutter ^{33 )} (Fig. 2), in line with results using FluoZin-3⁽ Reference Li and Maret ^{11 )}, much more variation exists for mitochondria: (0·1 pm for mito-ZapCY1⁽ Reference Park, Qin and Galati ^{31 )} and 300 pm for mito-eCALWY-4⁽ Reference Chabosseau, Tuncay and Meur ^{8 )}) and endoplasmic reticulum (about 1pm ⁽ Reference Qin, Dittmer and Park ^{30 )} and above 5 nm ⁽ Reference Chabosseau, Tuncay and Meur ^{8 )}). The reasons for these variations are unclear and may involve differences in intracellular pH on which the probes are steeply dependent. Red-shifted variant have been created for Zap and eCALWY sensors⁽ Reference Miranda, Weaver and Qin ^{32 ,} Reference Lindenburg, Hessels and Ebberink ^{34 )}.

Fig. 2. (Colour online) Measurement of free Zn²⁺ concentrations in subcellular compartments in mammalian cells using genetically-encoded sensors. Green: ZapCY1/2⁽ Reference Park, Qin and Galati ^{31 )}, white: eCALWY-4⁽ Reference Vinkenborg, Nicolson and Bellomo ^{7 )}, yellow: eZinCh-1⁽ Reference Vinkenborg, Nicolson and Bellomo ^{7 )}.

Hybrid probes include both genetically-encoded and small molecular elements. These include a probe based on a carbonic anhydrase variant covalently bound to a chemical fluorophore or fused with a red fluorescent protein⁽ Reference Bozym, Thompson and Stoddard ^{35 ,} Reference Hurst, Wang and Thompson ^{36 )}.

The above zinc sensors and probes are thus precious tools with which to decipher the link between changes in intracellular zinc levels, diabetes risk and pathogenic mechanisms.

Zinc and diabetes

A role for zinc in diabetes aetiology has been known since 1930, when the zinc concentration was reported to be reduced by about 50 % in the pancreas of diabetic compared with non-diabetic cadavers⁽ Reference Scott and Fisher ^{37 )}. Epidemiological studies also suggest that whole body zinc status might be associated with diabetes⁽ Reference el-Yazigi, Hannan and Raines ^{38 ,} Reference Garg, Gupta and Goyal ^{39 )}. Studies on patients with type 2 diabetes (T2D) revealed that the serum concentration of zinc was decreased compared with healthy control subjects⁽ Reference Basaki, Saeb and Nazifi ^{40 ,} Reference Jansen, Rosenkranz and Overbeck ^{41 )}, a finding associated with increased urinary zinc loss⁽ Reference el-Yazigi, Hannan and Raines ^{38 )}.

Adequate levels of Zn²⁺ are essential not only to ensure appropriate synthesis, storage and structural stability of insulin⁽ Reference Dunn ^{42 )} but also to protect against oxidative stress in T1 and T2 diabetes and their associated pathologies⁽ Reference Cruz, de Oliveira and Marreiro ^{43 )}. Thus, zinc is a pro-antioxidant, and a cofactor of superoxide dismutase (isoforms 1 and 3), that regulates the expression of metallothioneins and glutamate-cysteine ligase, thus affecting glutathione levels⁽ Reference Cruz, de Oliveira and Marreiro ^{43 )}.

The idea that zinc supplementation might improve the symptoms of T2D have been examined not only in animal models but also in diabetic patients. Studies on obese ob/ob mice showed that high zinc supplementation attenuates fasting hyperglycaemia and hyperinsulinaemia⁽ Reference Begin-Heick, Dalpe-Scott and Rowe ^{44 )}. Later studies provided similar results, and also revealed reduced body weight following treatment with lower zinc concentrations, in db/db mice⁽ Reference Simon and Taylor ^{45 )}. Moreover, recent studies on streptozotocin-induced diabetic rats revealed that zinc supplementation improves symptoms such as polydipsia and high HDL cholesterol levels⁽ Reference Wang, Li and Fan ^{46 )}. Amelioration of the diabetic phenotype in T2D patients has also been observed upon zinc supplementation (see Table 2. Data reviewed in⁽ Reference Vardatsikos, Pandey and Srivastava ^{47 )}).

Table 2. Zinc supplementation studies in Type 2 diabetic patients (modified from⁽ Reference Vardatsikos, Pandey and Srivastava ^{47 )})

HbA1c, Glycosylated haemoglobin A (a time-averaged measure of blood glucose).

Zinc imaging to assess beta cell mass in vivo?

Pancreatic β-cells have an exceptionally high total Zn²⁺ content (about 10–20 mm), with the majority of β-cell Zn²⁺ being located within dense core insulin secretory granules⁽ Reference Hutton, Penn and Peshavaria ^{48 )}. Zinc ions are relatively less abundant in the exocrine pancreas and in other tissues⁽ Reference Sondergaard, Stoltenberg and Doering ^{49 )}. Although type 1 diabetes involves immune-mediated destruction of insulin-secreting pancreatic β-cells, T2D usually involves defects in both insulin release, and in hormone action⁽ Reference Kahn ^{50 )}. Both impaired glucose sensing by the β-cell⁽ Reference Rutter and Parton ^{51 )}, and a loss of overall β-cell mass⁽ Reference Butler, Janson and Bonner-Weir ^{52 )}, are involved in the decline in β-cell function in T2D, though the relative contributions of each are contested⁽ Reference Rahier, Guiot and Goebbels ^{53 )}. Classical in vivo imaging approaches such as magnetic resonance and positron emission tomography are challenging due to the relative paucity of islets within the pancreas (about 1 % of the total pancreatic volume) and the absence of suitable contrast reagents, which are thus eagerly sought. Dual-modal zinc probes, based on transition metal chelates capable of binding these ions, may therefore be of clinical value as imaging tools in the future⁽ Reference Stasiuk, Minuzzi and Sae-Hen ^{54 )}.

Zinc transporter 8 in insulin secretion and type 2 diabetes risk

During its biosynthesis in β-cells insulin is initially packaged as Zn²⁺-coordinated hexamers⁽ Reference Arvan and Halban ^{55 )} before its concentration and crystallisation into dense cores within the secretory granule. The hormone is subsequently released into the circulation upon stimulation by glucose⁽ Reference Rutter, Pullen and Hodson ^{56 )}. The latter process involves enhanced mitochondrial metabolism⁽ Reference Rutter, Pralong and Wollheim ^{57 )}, closure of ATP-sensitive K⁺ channels⁽ Reference Tarasov, Ravier and Semplici ^{58 )}, Ca²⁺ influx⁽ Reference Rutter, Theler and Li ^{59 )} and granule fusion at the plasma membrane⁽ Reference Rutter ^{60 )}.

ZnT8 (encoded by the SLC30A8 gene), a member of the zinc transporter family (see Fig. 1 and Introduction) is highly and almost exclusively expressed in the β- and α-cells of the endocrine pancreas, where it facilitates the import of cytosolic Zn²⁺ into secretory granules (Fig. 3)⁽ Reference Chimienti, Devergnas and Favier ^{61 )}. Autoantibodies to ZnT8 have been found in individuals with type 1 diabetes and are a relevant prognostic feature of the disease⁽ Reference Wenzlau, Juhl and Yu ^{62 )}. Importantly, genome-wide association studies have provided a potential link between ZnT8 activity and T2D development. Thus, Sladek et al. ⁽ Reference Sladek, Rocheleau and Rung ^{63 )} found that an SNP, rs13266634, within exon 13 of the SLC30A8 gene was enriched in individuals with TD2. The variant at rs13266634 results in a missense mutation whereby an arginine residue is replaced by a tryptophan at position 325 (R325W). Risk allele carriers (R325) present with impaired insulin secretion during an intravenous glucose tolerance test⁽ Reference Staiger, Machicao and Stefan ^{64 )} increased proinsulin: insulin ratio⁽ Reference Kirchhoff, Machicao and Haupt ^{65 )} and lower β-cell function (as assessed through Homeostatic model assessment-B⁽ Reference Dimas, Lagou and Barker ^{66 )}, suggestive of impairments in both insulin secretion and processing. Four SNP located in the 3′-untranslated region of SLC30A8, two of which are in strong linkage disequilibrium with rs13266634, have also been associated with an increased risk of developing T2D. However, there are conflicting reports as to whether or not possession of these SNP is associated with changes in certain parameters of glucose homeostasis (e.g. fasting blood glucose, glucose tolerance etc.)⁽ Reference Staiger, Machicao and Stefan ^{64 –} Reference Cauchi, Del and Choquet ^{68 )}. Interestingly, the effects of dietary zinc supplementation to lower T2D risk are dependent on SLC30A8 genotype⁽ Reference Kanoni, Nettleton and Hivert ^{69 )}. More recent studies in human populations have reported twelve rare loss-of-function SNP, resulting in the production of a truncated protein that is associated with a 65 % decrease in T2D risk⁽ Reference Flannick, Thorleifsson and Beer ^{70 )}. This result was unexpected given the action of the common variant, which is likely to lower ZnT8 activity (see later), to increase disease risk.

Fig. 3. (Colour online) Schematic of the genome-wide association studies (GWAS)-identified zinc transporter ZnT8 and localisation of the risk associated polymorphic amino acid R/W 325.

Fig. 4. (Colour online) Possible sites of insulin action on intracellular Zn²⁺ homeostasis. Intracellular free zinc concentration could be impacted by: – action on zinc transporter (Zip and ZnT) on the release and uptake, respectively, of zinc ions from intracellular stores (1) or the extracellular medium (2) or by modulating the buffering of Zn²⁺by metallothioneins (3)⁽ Reference Jansen, Rosenkranz and Overbeck ^{41 )}.

Providing a link between increased T2D risk and rs13266634 inheritance has proved challenging. Unlike most genome-wide association studies identified SNP to date, rs13266634 is located in a coding region (exon) of the genome and, therefore, affects the primary sequence of ZnT8 and potentially the tertiary structure of the protein and its function (Fig. 3). However, the crystal structure of ZnT8 is yet to be elucidated and structure–function relationships, as well as the effects of SNP upon these, have been modelled on the bacterial homologue YiiP, which only shares 51·8 % sequence homology⁽ Reference Chao and Fu ^{71 )}. Position 325 is located towards the ‘tip’ of the ZnT8 molecule within regions involved in homodimerisation, but as both R- and W-side chains point away from the dimerization interface itself, it is unlikely that R325W substitution would have an effect on ZnT8 dimerisation or Zn²⁺ binding⁽ Reference Nicolson, Bellomo and Wijesekara ^{72 ,} Reference Weijers ^{73 )}. It is nonetheless conceivable that the positive charge of the R-side chain may hinder inter- and intra-molecular interactions, for example with presently unidentified Zn²⁺-binding proteins which deliver the ions to the mouth of the channel to facilitate Zn²⁺ transport. Providing evidence for this are zinc uptake studies⁽ Reference Nicolson, Bellomo and Wijesekara ^{72 ,} Reference Kim, Toyofuku and Lynn ^{74 )}, which showed that R325 is a less active Zn²⁺ transporter than the non-risk W325 in β-cells.

To further elucidate the role of ZnT8 in glucose homeostasis, several groups including our own have produced animal models harbouring either global ZnT8 deletion⁽ Reference Nicolson, Bellomo and Wijesekara ^{72 ,} Reference Lemaire, Ravier and Schraenen ^{75 –} Reference Gerber, Bellomo and Hodson ^{77 )} or deletion restricted to the β-cell⁽ Reference Wijesekara, Dai and Hardy ^{78 ,} Reference Tamaki, Fujitani and Hara ^{79 )}, with some recombination in the hypothalamus⁽ Reference Wicksteed, Brissova and Yan ^{80 )}. Each mouse model shows variations in certain phenotypic traits (see Table 3)⁽ Reference Rutter ^{81 )}, which are attributed to differences in genetic background, deletion strategy and housing conditions. While the majority of animal models displayed impairments in glucose tolerance upon ZnT8 deletion, albeit with subtle age and sex-differences between the colonies, no investigators reported improvements in glucose tolerance upon ZnT8 deletion. Using electron microscopy, marked changes were seen in insulin granule morphology, with a large proportion of insulin secretory granules lacking a dense core of crystallised insulin, but containing either ‘empty’ granules or granules possessing ‘rod like’ structures. Islet Zn²⁺ was also decreased in both global and β-cell specific ZnT8 null animals⁽ Reference Nicolson, Bellomo and Wijesekara ^{72 ,} Reference Wijesekara, Dai and Hardy ^{78 )}. Although circulating insulin levels were significantly lowered in ZnT8 null mice compared with controls, glucose-stimulated insulin release was unchanged or slightly increased in islets isolated from ZnT8 null mice. Providing an elegant explanation for these apparently contradictory results, Tamaki et al.⁽ Reference Tamaki, Fujitani and Hara ^{79 )} recently demonstrated a role for Zn²⁺ co-secreted with insulin from granules in regulating the rate of hepatic insulin clearance mediated by clathrin-dependent internalisation of the insulin receptor. As Zn²⁺ does not affect the uptake of C-peptide or proinsulin, this mechanism could potentially explain the impairments in both circulating insulin and the increased proinsulin:insulin ratio seen in risk allele carriers⁽ Reference Rutter and Chimienti ^{82 )}. Interestingly, cytosolic free Zn²⁺, as measured with the eCALWY4 probe, was reduced in the global knockout (KO) mouse for ZnT8⁽ Reference Gerber, Bellomo and Hodson ^{77 )}, alongside granule zinc concentrations as estimated by the release of zinc during exocytosis⁽ Reference Li, Chen and Bellomo ^{26 )}. These findings imply a more complex role for this transporter in the regulation of zinc fluxes than has previously been appreciated.

Table 3. Glycemic phenotpye of ZnT8 null mouse lines

RIP2Cre, rat insulin promoter 2-driven Cre recombinase; GSIS, glucose-stimulated insulin secretion; IGT, impaired glucose tolerance; NR, not recorded (unpublished results).

*Pound (mixed)⁽ Reference Pound, Sarkar and Benninger ^{76 )}, Pound (BL6)⁽ Reference Pound, Sarkar and Ustione ^{109 )}, Nicolson⁽ Reference Nicolson, Bellomo and Wijesekara ^{72 )}, Lemaire⁽ Reference Lemaire, Ravier and Schraenen ^{75 )}, Wijesekara⁽ Reference Wijesekara, Dai and Hardy ^{78 )}, Tamaki⁽ Reference Tamaki, Fujitani and Hara ^{79 )}, Mitchell⁽ Reference Mitchell, Hu and Meur ^{86 )}.

Exposing ZnT8 null animals to the diabetogenic effects of a high-fat diet led to varying results between global and β-cell specific mouse models. High-fat diet-fed β-cell specific null mice displayed similar bodyweights but displayed impaired secretion and were glucose intolerant compared with littermate controls whereas global deletion of ZnT8 resulted in increased weight gain and subsequently displayed higher levels of insulin resistance⁽ Reference Hardy, Wijesekara and Genkin ^{83 )}.

Notably, ZnT8 is also expressed at low but detectable levels in rat-insulin promoter-expressing neurons residing within hypothalamic appetite centres. Therefore, ectopic deletion in both global and rat-insulin promoter2Cre-driven ZnT8 mouse models may explain the phenotype observed in some of these mouse models. To overcome this issue, as well as other issues with Cre strains that express a human growth hormone minigene⁽ Reference Brouwers, de and Osipovich ^{84 )}, we deleted ZnT8 using a highly-specific β-cell Ins1Cre driver line, which produces no detectable recombination in the hypothalamus⁽ Reference Thorens, Tarussio and Maestro ^{85 )}. Confirming previous results, Ins1Cre-mediated ZnT8 deletion results in impaired glucose tolerance, abnormal insulin granule morphology and reduced cytosolic Zn²⁺ concentrations⁽ Reference Mitchell, Hu and Meur ^{86 )}. This model also shows reduced liberation of Zn²⁺ from isolated islets, supporting the view of Tamaki et al. that effects of Zn²⁺ on the liver may limit the levels of bioavailable insulin to impair glucose tolerance. Furthermore, we have recently generated a mouse line in which the protective variant (W325) of human ZnT8 is expressed selectively in the β-cell driven by an insulin promoter-controlled tetracycline responsive promoter⁽ Reference Pullen, Sylow and Sun ^{87 )}. In contrast to both global and β-cell-specific ZnT8 null mice, increased ZnT8 expression significantly improved glucose tolerance compared with wild type littermate controls, as expected. However, the improved glucose tolerance did not appear to be due to enhanced insulin secretion, which was reduced in islets isolated from these animals. This is likely instead to be secondary to elevations in secreted Zn²⁺, acting to inhibit insulin secretion through an autocrine/paracrine loop⁽ Reference Mitchell, Hu and Meur ^{86 ,} Reference Slepchenko, James and Li ^{88 )}. Whether the enhanced Zn²⁺ secretion has any effects on neighbouring α-cells, or on hepatic insulin clearance⁽ Reference Tamaki, Fujitani and Hara ^{79 )}, is yet to be investigated.

Genome-wide association studies and animal data have provided conflicting reports regarding the role of ZnT8 in T2D risk (see⁽ Reference Rutter and Chimienti ^{82 )} for a detailed discussion). Collectively, these data would suggest that both inheritance of rs13266634 and deletion of ZnT8 in mice are detrimental in maintaining glucose homeostasis. Given the highly-restricted expression profile of ZnT8, there may be some promise in therapeutically-targeting ZnT8 in the treatment of T2D.

A role for zinc transporter 8 in glucagon secretion?

In addition to expression in β-cells, there is evidence that ZnT8 is also present in α-cells, at least in mouse and human pancreata. Its expression in these cells was confirmed by immunocytochemistry in pancreatic slices and dissociated islets and by gene expression analysis on a fluorescence-activated cell sorting purified mouse population⁽ Reference Nicolson, Bellomo and Wijesekara ^{72 )}. Similarly to the β-cell, ZnT8 appears to be the most highly expressed member of the ZnT family in the α-cell. In the porcine pancreas, however, ZnT8 is exclusively expressed in the β-cell⁽ Reference Schweiger, Steffl and Amselgruber ^{89 )} implying clear species differences and suggesting that in pig islets zinc homeostasis is regulated differently from human and rodent islets. Knockdown of ZnT8 in glucagonoma-derived αTC1·9 cells resulted in an increase of glucagon mRNA, and decreased regulated glucagon secretion by 70 %. Overexpression of the ZnT8 R325 or W325 variants in these cells reciprocally led to reduced glucagon content and 50 % lower glucagon secretion⁽ Reference Souza, Qui and Inouye ^{90 )}.

As a result of its co-secretion with insulin, zinc can reach high local concentrations within the islet, with the potential of acting as a mediator of paracrine signalling. Thus, in the perfused rat pancreas, the zinc chelator Ca²⁺-EDTA led to a stimulation of secretion when using the mitochondrial substrate monoethyl-succinate as a secretagogue, an agent which normally acts as an inhibitor of glucagon release⁽ Reference Ishihara, Maechler and Gjinovci ^{91 )}. Correspondingly, zinc diminished pyruvate- or glucose-induced glucagon secretion from isolated rat α-cells through the reversible activation of K_ATP channels and a subsequent decrease in electrical activity⁽ Reference Franklin, Gromada and Gjinovci ^{92 )}. Likewise, studies in the mouse demonstrated an inhibitory effect of Zn²⁺ on glucagon secretion⁽ Reference Gyulkhandanyan, Lu and Lee ^{93 )}. However, in the latter species, this effect of zinc could not be attributed to K_ATP channel activation but to uptake of the ion by calcium channels and an alteration in redox state⁽ Reference Gyulkhandanyan, Lu and Lee ^{93 )}. Of note, fasting plasma glucagon levels were normal in global ZnT8 null mice and glucagon secretion from isolated islets was not higher in KO mice than controls. However, exogenous zinc did lead to a reduction of glucagon secretion under stimulatory conditions⁽ Reference Hardy, Serino and Wijesekara ^{94 )}. Assuming that granular zinc was almost entirely depleted in the KO mice, these data suggest that zinc secreted from β-cells is not responsible for the inhibition of glucagon secretion at high glucose.

Although there have been extensive studies on the role of ZnT8 in the β-cell (see preceding section and references therein), its function in the α-cell has not been investigated in detail. One study looked briefly at an α-cell-specific ZnT8 KO and did not detect any differences in fasting plasma glucagon levels or glucose homeostasis compared with control mice⁽ Reference Wijesekara, Dai and Hardy ^{78 )}. Further examination of islets from these mice is necessary as well as investigating changes at the single cell level.

In line with our earlier findings⁽ Reference Wijesekara, Dai and Hardy ^{78 )}, we have more recently observed no effect of ZnT8 deletion selectively in α-cells on glucose homeostasis or fasting glucagon⁽ Reference Solomou, Meur and Bellomo ^{95 )}. However, we observed in these more recent studies that KO mice displayed enhanced responses to hypoglycaemia in vivo and increased glucagon secretion from isolated islets, implying cell-autonomous roles for ZnT8 in the α-cell.

Zinc and insulin action

Insulin binding to the α-subunit of the insulin receptor leads to enhanced intrinsic protein tyrosine kinase activity of the β-subunit and phosphorylation of the receptor on multiple tyrosine residues⁽ Reference Lee and Pilch ^{96 )}. The activated insulin receptor then phosphorylates several scaffolding proteins, including the insulin receptor substrates, which subsequently bind and activate other signalling proteins to trigger two main different pathways (Fig. 5): the activation of the serine/threonine kinases Ras and Raf by the SOS/Grb2/SHC complex leading to extracellularly-regulated kinases1/2 activation and cellular proliferation, and activation of phosphatidyl inositol 3′-kinase⁽ Reference Kanzaki ^{97 )}. The latter triggers the translocation of phoshoinositide-dependent kinase-1 to the plasma membrane together with protein kinase B and activates downstream protein kinases such as p70 ribosomal S6 kinase. The latter stimulate glucose transport, glycogen synthesis, lipogenesis and other processes (Fig. 5).

Fig. 5. (Colour online) Possible sites of action of Zn²⁺ on insulin signalling. PTB1B, protein tyrosine phosphatase 1B; SHC, Src-homology-2-containing; SOS, Son-of-sevenless; MEK, MAP/ERK kinase; MAP, mitogen-activated protein kinase; IRS, insulin receptor substrate; P1-3K, phosphatidylinositol 3’ kinase; PTEN, phosphatase and tensin homologue; PDK, phosphoinositide-dependent kinase; aPKC, atypical protein kinase C; GSK, glycogen synthase kinase; PPI, protein phosphatase inhibitor-1; GLUT4, glucose transporter family member 4.

An action of zinc on insulin-target tissue was described for the first time by Coulston and Dandona, when they observed that treatment of rat adipocytes with high zinc concentrations led to an increased rate of lipogenesis⁽ Reference Coulston and Dandona ^{98 )}. At the cellular level, zinc was found to increase insulin receptor substrates-1 tyrosine phosphorylation in the presence or absence of insulin in skeletal muscle cells⁽ Reference Miranda and Dey ^{99 )}. Moreover, zinc was able to activate protein kinase B in several studies using 3T3-L1 and rat adipocytes: zinc treatment induced not only phosphorylation of the insulin receptor, but also of protein kinase B. While the former is possibly dependent on the inhibition by zinc of protein tyrosine phosphatase 1B⁽ Reference Haase and Maret ^{100 )}, the latter is a consequence of phosphatidyl inositol 3′-kinase inhibition⁽ Reference Tang and Shay ^{101 )}. Phosphatase and tensin homologue is responsible for the negative regulation of phosphatidyl inositol 3′-kinase /protein kinase B activity and appears to be degraded in response to zinc⁽ Reference Wu, Wang and Zhang ^{102 )} at physiological (about 600 pm) concentrations⁽ Reference Plum, Brieger and Engelhardt ^{103 )}.

As mentioned earlier, one of the most studied targets of the insulin-mimetic effects of zinc is protein tyrosine phosphatase 1B and a recent study has elucidated the inhibition constant and the mechanisms of zinc binding to protein tyrosine phosphatase 1B ⁽ Reference Haase and Maret ^{100 ,} Reference Bellomo, Massarotti and Hogstrand ^{104 )}. Zinc also inhibits protein tyrosine phosphatase 1B with an apparent inhibition constant as low as 5 nm. By acting through these two targets, an increase in intracellular zinc signal can be predicted to enhance insulin signalling downstream of the insulin receptor. Zinc also mimics insulin action by triggering nuclear exclusion of the transcription factor FOXO1, and induces glycogen synthesis by lowering the phosphorylation of glycogen synthase kinase-3⁽ Reference Barthel, Ostrakhovitch and Walter ^{105 )}.

The effect of zinc on insulin signalling described above may also explain, at least in part, the impact of the metal ion in the pathogenesis of T2D. Although in vitro analyses showed that insulin leads to an increase in intracellular zinc⁽ Reference Jansen, Rosenkranz and Overbeck ^{41 )}, the exact physiological mechanisms responsible are not entirely understood (Fig. 2). While the free concentration of Zn²⁺ within the endoplasmic reticulum is contested (see earlier) it has been suggested that zinc importer-7-mediated release from this compartment may be involved in the actions of epidermal growth factor, which like insulin acts through a receptor tyrosine kinase⁽ Reference Taylor, Hiscox and Nicholson ^{9 )}. Moreover, ablation of zinc importer-7 in skeletal muscle cells displayed a reduction in GLUT4 protein expression and insulin-stimulated glycogen synthesis⁽ Reference Myers, Nield and Chew ^{106 )}. As discussed earlier, zinc may also inhibit insulin clearance by the liver, leading to elevated circulating levels of the hormone⁽ Reference Tamaki, Fujitani and Hara ^{79 )}.

Finally, zinc may also contribute to the anti-oxidant actions of insulin, serving (i) as a cofactor of superoxide dismutase, (ii) to enhance the expression of metallothioneins and (iii) to stimulate glutamate-cysteine ligase and glutathione synthesis. Interestingly, supplementing diabetic rats with zinc increased superoxide dismutase activity⁽ Reference Zhu, Nie and Li ^{107 )}. Decreased lipid peroxidation, concomitant with an increase in glutathione concentration, was also found in zinc treated diabetic rats⁽ Reference Karatug, Kaptan and Bolkent ^{108 )}. Potential sites of action of Zn²⁺ on insulin signalling are shown in Fig. 5.