Synaptically Released Zinc Triggers Metabotropic Signaling via a Zinc-Sensing Receptor in the Hippocampus

A metabotropic response is triggered by Zn²⁺

To determine if Zn²⁺ activates metabotropic signaling in the brain, we used coronal hippocampal slices. Cells were loaded with Fura-2 and treated with Zn²⁺ by direct delivery into the ACSF perfusate. Application of 200 μm Zn²⁺ triggered a rapid increase in Fura-2 fluorescence in the pyramidal cell layer of the CA3 hippocampal region (Fig. 1A,B) in the presence or absence of extracellular Ca²⁺. This response was similar to that induced by ATP (300 μm, see inset), shown previously to trigger a metabotropic purinergic response (Yamazaki et al., 2002). After application of Zn²⁺, the decline to resting Ca²⁺ levels was somewhat slower in the presence of extracellular Ca²⁺, possibly due to Ca²⁺ influx via the store-operated Ca²⁺ channel, activated after store depletion (Wu et al., 2004). Because Zn²⁺ can permeate into neurons and because Fura-2 has a high affinity for this ion (K_d = 2 nm), it was important to establish that the fluorescence signal observed was related to changes in intracellular Ca²⁺ and not Zn²⁺. Endoplasmic reticulum (ER) Ca²⁺ stores were depleted, using the ER Ca²⁺ pump inhibitor, thapsigargin (TG, 1 μm), applied with the purinergic agonist ATP (300 μm) to enhance intracellular Ca²⁺ release. Subsequent applications of Zn²⁺ (200 μm) (Fig. 1C,D) failed to elicit an increase in Fura-2 signal (10 ± 2% of response in control), indicating that the fluorescence response is not mediated by Zn²⁺ permeation, but rather by release of Ca²⁺ from TG-sensitive stores. We then assessed the Zn²⁺ concentration dependence of the observed Ca²⁺ rise (Fig. 1E). The Zn²⁺-dependent Ca²⁺ response was observed over a wide range of Zn²⁺ concentrations (i.e., from tens to hundreds micromolar) with an apparent K_m of 146 ± 26 μm. Together, these results indicate that Zn²⁺ triggers an intracellular Ca²⁺ signal in the CA3 region, which is extracellular Ca²⁺ independent, and mediated by Ca²⁺ release from TG-sensitive stores (Fig. 1D).

We next asked if the Zn²⁺-induced Ca²⁺ signal is mediated through a Gq-coupled receptor and via the IP₃ pathway, both characteristic of the described previously ZnR-mediated responses in non-neuronal cells (Hershfinkel et al., 2001). Application of 200 μm Zn²⁺ after treatment with the phospholipase C (PLC) inhibitor U73122 (1 μm) resulted in a Ca²⁺ rise that was inhibited by 77 ± 3% compared with control. In the presence of the inactive analog of this inhibitor U73343 (1 μm) the Zn²⁺-dependent response was similar (92 ± 11%) to control (Fig. 2A,B). The Gαq inhibitor YM-254890 (1 μm) (Sharir and Hershfinkel, 2005) inhibited the Zn²⁺-dependent Ca²⁺ response by 68 ± 4% compared with control (Fig. 2B). Since mGluRs mediate metabotropic signaling in this region, we compared the Zn²⁺-dependent response in the presence or absence of mGluR inhibitors. To ascertain that the mGluRs were blocked, saturating concentrations of the inhibitors were used (Nakamura et al., 2000), and their efficacy was also verified by monitoring their ability to block the Ca²⁺ response triggered by the group I mGluR agonist DHPG ((S)-3,5-dihydroxyphenylglycine) (Bianchi et al., 1999). Application of the mGluR5 inhibitor MPEP (2-methyl-6-(phenylethynyl)-pyridine; 5 μm), with the mGluR1 inhibitor AIDA (1-aminoindan-1,5-dicarboxylic acid; 500 μm), attenuated the Ca²⁺ signal induced by 100 μm DHPG by 62 ± 6% (n = 8), while 10 μm MPEP in the presence of 1 mm AIDA reduced the DHPG-dependent response by 69 ± 5% (n = 8). In contrast, the Zn²⁺-dependent Ca²⁺ signal was not reduced at either concentration of the mGluR inhibitors (p > 0.3) (Fig. 2B), suggesting that the Zn²⁺-triggered response is not related to mGluRs, but is mediated via a distinct metabotropic pathway. The metal selectivity of the metabotropic response was then assessed, focusing on Cd²⁺ and Pb²⁺. Cd²⁺ is a group IIB metal, like Zn²⁺, which binds with high affinity to Zn²⁺ binding sites on proteins such as metallothionein (Henkel and Krebs, 2004). Pb²⁺ is a divalent metal that has been shown to alter synaptic function (White et al., 2007). To avoid background effects mediated by permeation of the ions, and to resolve if these ions trigger the metabotropic Ca²⁺ response, we determined the difference between the maximal Fura-2 signal, triggered by Cd²⁺, Pb²⁺ or Zn²⁺, in the presence or absence of the Gαq inhibitor (ΔR_{Gq inhibitor}, see Materials and Methods). As shown in Figure 2C, the Zn²⁺-dependent Fura-2 response was largely blocked by the Gαq inhibitor (1 μm), yielding a ΔR_{Gq inhibitor} = 0.65 ± 0.05. In contrast, the inhibitor did not significantly reduce Cd²⁺ or Pb²⁺-dependent fluorescence (ΔR_{Gq inhibitor} = 0.05 ± 0.09 or 0.1 ± 0.05, respectively). This indicates that Cd²⁺ and Pb²⁺ do not induce a metabotropic Ca²⁺ rise, and that the ZnR response is selective for Zn²⁺.

Previous work in our lab has shown that the epithelial ZnR undergoes profound functional desensitization after prolonged exposure to Zn²⁺ at concentrations which can activate the ZnR (Azriel-Tamir et al., 2004; Sharir and Hershfinkel, 2005). Such a regulatory mechanism would be especially relevant in the CNS, considering the pathophysiological implications of sustained activation of the IP₃ pathway (Tang et al., 2003), and the potential excessive release of synaptic Zn²⁺ during intense excitatory activity (Weiss et al., 2000). To determine if Zn²⁺-dependent signaling in neurons undergoes desensitization, slices were pretreated with 75 μm Zn²⁺ for 15 min and then washed in ACSF (see Materials and Methods). We subsequently applied 300 μm Zn²⁺ while monitoring the Ca²⁺ response. The Zn²⁺ and ATP-dependent Ca²⁺ response observed in the control slices (pretreated with ACSF) is smaller than that shown in Figure 2B, and may result from the prolonged interval before imaging. The Zn²⁺-dependent Ca²⁺ response in slices that were pre-exposed to Zn²⁺ was only 15 ± 2% compared with control (Fig. 2D) (ΔR = 0.33 ± 0.02 in control compared with 0.05 ± 0.01 in the Zn²⁺ pretreated slices), while application of ATP triggered a robust Ca²⁺ rise (Fig. 2D). As shown, ATP-induced Ca²⁺ responses were somewhat larger in Zn²⁺-desensitized preparations, likely due to the fact that intracellular stores had not been depleted by a ZnR-mediated response. Together, these results indicate that the Zn²⁺-dependent metabotropic response triggered in the CA3 region is mediated through a Gq-coupled receptor via the IP₃ pathway, and undergoes a Zn²⁺-dependent desensitization similar to that described previously for the epithelial ZnR.

To determine whether neurons or glia exhibit metabotropic Zn²⁺-dependent Ca²⁺ release, Zn²⁺ was applied as before and the Ca²⁺ response monitored in a confocal microscope (Fig. 3A). Slices were subsequently stained with the astroglial marker SR101 (Sulforhodamine 101) (Nimmerjahn et al., 2004). As shown in Figure 3B, cells which exhibited a Zn²⁺-dependent Ca²⁺ rise (white arrows) were not labeled with the astrocyte marker, SR101 suggesting that ZnR activity is not found in astroglia. To further address this issue, we asked if ZnR activity is observed in an established neuronal cell line, SH-SY5Y, a subclone of the SK-N-SH cell line. SH-SY5Y cells were previously used to study neurite outgrowth, cell growth and glutamate toxicity (Canals et al., 2005; Smith et al., 2009). As shown in Figure 3C, application of 200 μm Zn²⁺, a concentration which triggered robust ZnR activity in the slices (Fig. 1), was also followed by an intracellular Ca²⁺ rise in the SH-SY5Y cells. When SH-SY5Y cells were treated with the Gαq inhibitor, YM-254890 (1 μm), the Zn²⁺-induced Ca²⁺ response was inhibited by 95 ± 3% compared with control (Fig. 3C) (ΔR = 0.61 ± 0.15 in control and 0.03 ± 0.01 in the presence of the inhibitor), indicating that the Zn²⁺-dependent Ca²⁺ response is mediated via a Gq-dependent pathway.

MAP and CAM kinase pathways are activated by metabotropic Zn²⁺-dependent signaling

Activation of extracellular regulated kinase (ERK1/2) and Ca²⁺/calmodulin-dependent protein kinase (CaMKII) play fundamental roles in hippocampal synaptic plasticity and neuronal survival (Hansen et al., 2003; Giovannini, 2006; Luo and DeFranco, 2006; Cohen-Matsliah et al., 2007). Because intracellular Ca²⁺ rise can activate these pathways (Berkeley and Levey, 2003), we sought to determine if the Zn²⁺-dependent metabotropic response could trigger ERK1/2 or CaMKII phosphorylation in neurons. As shown in Figure 4, A and B, application of Zn²⁺ (100 μm, 90 s) led to a fourfold increase in pERK1/2 in the CA3 region compared with control. Colocalization with DAPI using confocal microscopy indicated that pERK1/2 staining was abundant in the nuclear and perinuclear regions (Fig. 4A, bottom). Application of blockers of the iGluRs (20 μm CNQX and 50 μm AP5) (Fig. 4B), and the Ca²⁺ channel blocker, nimodipine (1 μm), did not attenuate the Zn²⁺-dependent phosphorylation of ERK1/2, indicating that it is mediated by extracellular Zn²⁺ and not by its potential permeation through these pathways. As illustrated in Figure 4, C and D, application of Zn²⁺ also resulted in an increase of ∼50% in phosphorylated CaMKII. Finally, to assess the role of the metabotropic pathway in mediating Zn²⁺-dependent kinase phosphorylation, slices were treated with the Gαq inhibitor YM-254890 (1 μm) before application of Zn²⁺. As shown in Figure 4, application of the Gαq inhibitor reduced the Zn²⁺-dependent phosphorylation of both ERK1/2 and CaMKII to baseline levels. Together, our results indicate that a Gαq-coupled pathway mediates Zn²⁺-dependent ERK1/2 and CaMKII phosphorylation in the CA3 region of the hippocampus.

GPR39 mediates neuronal Zn²⁺-dependent signaling

It was recently suggested that a member of the ghrelin receptor family of GPCRs, namely GPR39, can be activated by serum Zn²⁺ (Yasuda et al., 2007). The physiological significance of the Zn²⁺-dependent signal mediated by GPR39 activation remains unclear, however (Holst et al., 2007; Yasuda et al., 2007). We hypothesized that GPR39 might be the ZnR itself and thus sought to determine if it mediates neuronal Zn²⁺-dependent signaling. The endogenous expression of GPR39 was initially determined by immunoblot analysis of the neuronal SH-SY5Y cells (Fig. 5A). This cell line is particularly useful for ectopic expression of genes and RNA constructs because of the high transfection efficiency that is typically achieved. A short hairpin RNA (shGPR39) construct targeted to silence GPR39 efficiently lowered GPR39 expression in these cells, as determined by immunoblot analysis (Fig. 5A). As a control, a hairpin construct targeted against T1R3, a closely related GPCR, did not alter GPR39 expression in SH-SY5Y cells (Fig. 5A). The Ca²⁺ response to application of Zn²⁺ was then monitored in SH-SY5Y cells transfected with shGPR39 or in cells transfected with either shT1R3 or with an empty vector. As shown in Figure 5, B and C, the Zn²⁺-dependent Ca²⁺ response was similar in vector and shT1R3-expressing cells, but significantly attenuated (by 73 ± 7%) in cells transfected with shGPR39. We then asked whether GPR39 was endogenously expressed in the hippocampus. Immunofluorescent labeling of GPR39 (Fig. 5D) was observed as punctuate staining at the periphery of cells expressing the neuronal nuclear marker (NeuN) in CA3. Some punctuate staining was also observed near dendrites in the CA3 region (Fig. 5D, inset). This suggests that GPR39 is present on the cells that exhibit Zn²⁺-dependent metabotropic activity. These results are consistent with the hypothesis that GPR39 may mediate ZnR signaling in CA3 neurons.

Synaptic Zn²⁺ triggers ZnR-dependent response in the CA3 region

While the experiments described above strongly support the notion that a metabotropic receptor is activated in neurons in the CA3 region by exogenous Zn²⁺, induction of the Ca²⁺ response by synaptically released Zn²⁺ is required to demonstrate the physiological significance of this phenomenon. To address this issue, synaptic Zn²⁺ release was induced by electrical stimulation of the mossy fibers using a protocol consisting of 10 pulses at 66 Hz (Fig. 6A) while monitoring intracellular Ca²⁺ changes in CA3 neurons. As shown in Figure 6B, a robust increase in intracellular Ca²⁺ concentration was triggered after the electrical stimulation. This response was blocked by application of the voltage-gated Na⁺-channel blocker tetrodotoxin (1 μm, 91 ± 3% inhibition compared with control) (Fig. 6B,D). We also applied CdCl (200 μm) at a concentration that did not attenuate a metabotropic signal triggered by ATP (data not shown) yet is known to effectively inhibit the Ca²⁺-channels that govern synaptic release (Hinkle et al., 1987; Rosenmund and Stevens, 1996). Similar to the effect of tetrodoxin (TTX), application of CdCl inhibited the fluorescence response in the CA3 cells after stimulation of the mossy fibers (Fig. 6B,D) (86 ± 5% inhibition compared with control), suggesting that the metabotropic Ca²⁺ response in the CA3 region is triggered by synaptic transmission. Application of the Gαq inhibitor YM-254890 (1 μm) also resulted in a 92 ± 4% inhibition of the responses, compared with control slices (Fig. 6B,D). The inhibitory action of YM-254890 indicates that a Gq-coupled receptor triggers the Ca²⁺ response in the CA3 neurons after mossy fiber stimulation. The observed Ca²⁺ signals occurred over a range of several seconds, characteristic of metabotropic receptor-mediated physiological responses (Fig. 1) (Kapur et al., 2001). Inhibitors of iGluRs, namely CNQX (20 μm) and AP5 (50 μm), failed to attenuate the Ca²⁺ rise (Fig. 6C). Finally, in slices that were treated with AIDA and MPEP, inhibitors, respectively, of mGluR1 and mGluR5, the Ca²⁺ response triggered by the electrical stimulation was reduced by ∼50%, compared with the response in control slices (Fig. 6D). A similar inhibitory effect was observed in slices treated with both the mGluR and iGluR inhibitors (1 mm AIDA, 10 μm MPEP, 20 μm CNQX and 50 μm AP5). Thus, a significant, residual metabotropic Ca²⁺ response persisted in the presence of mGluR inhibitors, while the Gαq inhibitor completely inhibited this response (Fig. 6D). This suggests that after synaptic release a receptor distinct from the mGluRs is responsible for a substantial metabotropic Ca²⁺ response in CA3.

To assess the role of Zn²⁺ in triggering the synaptically mediated metabotropic Ca²⁺ response, electrical stimulation was performed in the presence of the cell-impermeable Zn²⁺ chelator, CaEDTA, which does not significantly affect extracellular Ca²⁺ concentration (Qian and Noebels, 2005). As shown in Figure 7, A and B, in the presence of 10 μΜ CaEDTA, the Ca²⁺ response within the CA3 layer was only slightly lower than that in control. However, an inhibitory effect was observed in the presence of either 150 μm or 1 mm CaEDTA (Fig. 7B) (inhibition of 51 ± 5% or 54 ± 7%, respectively, compared with control). This indicates that synaptically released Zn²⁺ mediates a major part of the metabotropic Ca²⁺ response. The relatively short time required to induce the metabotropic effect indicates that CaEDTA is unlikely to affect this signaling by regulating the protein expression pattern, as was previously demonstrated for GluR2 activation (Calderone et al., 2004). Finally, we asked if the mGluR inhibitors together with the chelator have an additive effect in blocking the metabotropic Ca²⁺ response. As shown in Figure 7B, the metabotropic signal after electrical stimulation of the mossy fibers in the presence of the mGluR inhibitors (1 mm AIDA and 10 μm MPEP) and CaEDTA (150 μm) was attenuated by 84 ± 6%. Thus, the coapplication of the chelator together with the mGluR inhibitors yielded a significantly smaller response than that monitored in the presence of either CaEDTA (p < 0.05) or the inhibitors alone (p < 0.05).

Mice deficient in ZnT3 lack synaptic Zn²⁺ while maintaining apparently normal synaptic activity, including release of glutamate and GABA (Cole et al., 2000, 2001; Lopantsev et al., 2003). In agreement with previous studies (Qian and Noebels, 2005), we observed that stimulation of the mossy fibers was followed by a significantly lower fluorescence signal by the extracellular Zn²⁺-sensitive dye, Newport green (1 μm), in slices from ZnT3 knock-out (KO) mice compared with wild-type controls (data not shown). These mice can thus serve as a useful model to determine the role of synaptic Zn²⁺ in activating the ZnR and subsequent metabotropic Ca²⁺ signaling. The Ca²⁺ response after stimulation of the mossy fibers in slices from ZnT3 KO mice was 56 ± 7% of the response in slices from wild-type (WT) mice (Fig. 7C). Thus, in slices from mice lacking synaptic Zn²⁺, we observed ∼45% inhibition of the metabotropic signaling in CA3 (Fig. 7D), similar to the reduction of the metabotropic signal by CaEDTA in WT mice (Fig. 7B). Application of exogenous Zn²⁺ to slices from the ZnT3 KO mice, however, was followed by a Ca²⁺ rise (108 ± 7%) similar to WT controls (Fig. 7D). This Ca²⁺ response triggered by exogenous Zn²⁺ in slices from ZnT3 KO mice indicates that the lack of synaptic Zn²⁺, rather than a deficiency of ZnR, is responsible for the attenuated metabotropic response. Application of CaEDTA (150 μm) to slices from ZnT3 KO mice yielded a Ca²⁺ response that did not significantly differ from the response obtained in nontreated ZnT3 KO slices (Fig. 7E,F). Finally, application of the mGluR inhibitors, using the same experimental paradigm described in Figure 6D, was followed by a Ca²⁺ response that was much smaller than the response in the nontreated, ZnT3 KO slices (Fig. 7F) (73 ± 3% inhibition). Thus, the mGluR inhibitors exhibit significantly (p < 0.01) greater inhibition of the response in ZnT3 KO slices than their effect in the WT slices. The inhibitory effect of the extracellular Zn²⁺ chelator in slices from the WT, but not the ZnT3 KO mice, together with the substantially attenuated metabotropic Zn²⁺ response in the ZnT3 KO mice, support the conclusion that synaptic Zn²⁺, released from mossy fiber terminals, activates a ZnR and triggers Ca²⁺ signaling in CA3 neurons of the hippocampus.