Spontaneous, synchronous zinc spikes oscillate with neural excitability and calcium spikes in primary hippocampal neuron culture

Abstract

Zinc has been suggested to act as an intracellular signaling molecule due to its regulatory effects on numerous protein targets including enzymes, transcription factors, ion channels, neurotrophic factors and postsynaptic scaffolding proteins. However, intracellular zinc concentration is tightly maintained at steady levels under natural physiological conditions. Dynamic changes in intracellular zinc concentration have only been detected in certain types of cells that are exposed to pathologic stimuli or upon receptor ligand binding. Unlike calcium, the ubiquitous signaling metal ion that can oscillate periodically and spontaneously in various cells, spontaneous zinc oscillations have never been reported. In this work, we made the novel observation that the developing neurons generated spontaneous and synchronous zinc spikes in primary hippocampal cultures using a fluorescent zinc sensor, FluoZin-3. Blocking of glutamate receptor-dependent calcium influx depleted the zinc spikes, suggesting that these zinc spikes were driven by the glutamate-mediated spontaneous neural excitability and calcium spikes that have been characterized in early developing neurons. Simultaneous imaging of calcium or pH together with zinc, we uncovered that a downward pH spike was evoked with each zinc spike and this transient cellular acidification occurred downstream of calcium spikes, but upstream of zinc spikes. Our results suggest that spontaneous, synchronous zinc spikes were generated through calcium influx-induced cellular acidification, which liberates zinc from intracellular zinc binding ligands. Given that changes in zinc concentration can modulate activities of proteins essential for synapse maturation and neuronal differentiation, these zinc spikes might act as important signaling roles in neuronal development.

Graphical Abstract

1. Introduction

As the second richest trace metal in organisms (Al-Fartusie & Mohssan 2017), zinc performs a variety of critical biological functions that have been usually attributed to its structural and catalytic roles in a large amount of proteins (Christianson 1991; Andreini et al. 2006). There is a growing body of evidence that zinc ions might exert the biological effects as signaling molecules which requires a dynamic change in the concentration of labile zinc ions (Maret 2017; Hirano et al. 2008). Such dynamic changes have been reported as “zinc waves” in mast cells as a result of inflammatory responses during extracellular stimulation (Yamasaki et al. 2007). Zinc elevations can also be induced during T cell receptor activation (Yu et al. 2011), cell differentiation, or changes of redox homeostasis (Maret & Krezel 2007). In neurons, increases in intracellular zinc concentration have been associated with neurotoxicity and neurodegeneration in cerebral ischemia (Galasso & Dyck 2007) and oxidative stress (McCord & Aizenman 2014). However, unlike the well-established signaling metal ion calcium, whose concentration can fluctuate periodically, repetitive rhythmic zinc elevations have never been documented previously.

The crosstalk between zinc and calcium has been revealed at different levels. Zinc can regulate intracellular calcium signals, and vice versa, calcium can also influence zinc signals. Extracellular zinc signals are detected by a “zinc-sensing receptor” (ZnR/GPR39) on the plasma membrane, triggering G protein dependent signaling and calcium liberation from endoplasmic reticulum (Hershfinkel et al. 2001; Hershfinkel 2018). In addition, zinc was found to directly increase the permeability of ryanodine receptor and release calcium from the sarcoplasmic reticulum in cardiomyocytes (Woodier et al. 2015). Another mechanism by which zinc affects calcium is through modulating the gating of a variety of calcium permeable channels such as Transient Receptor Potential Ankyrin 1 (TRPA1) channel (Hu et al. 2009), T-type voltage-gated calcium channels (VGCCs) (Traboulsie et al. 2007), N-methyl-d-aspartate (NMDA) receptors (Harrison & Gibbons 1994), and purinergic receptor (Harrison & Gibbons 1994; Kovacs et al. 2018). The influence of calcium on zinc is exemplified in calcium-stimulated exocytotic events, resulting in extracellular zinc release from neurons (Frederickson & Bush 2001) and pancreatic beta cells (Qian et al. 2003), or “zinc sparks” during oocyte activation (Que et al. 2015). Here, we present the discovery of a new connection between zinc and calcium in primary cultured hippocampal neurons.

During early brain developmental stages, neuronal populations self-organize into networks and generate spontaneous and synchronized neuronal activity (Luhmann et al. 2016; Spitzer 2006). Such rhythmic synchronizations have been termed giant depolarizing potentials (Ben-Ari et al. 1989; Canepari et al. 2000), and are driven by both glutamatergic and GABAergic transmission (Ben-Ari et al. 1989; Blankenship & Feller 2010) as well as dendrodentritic gap junctions (Opitz et al. 2002). These rhythmic signals have been detected in various brain areas, including the immature hippocampus (Kasyanov et al. 2004), the neocortex (Allene et al. 2008), the thalamus (Pangratz-Fuehrer et al. 2007), the embryonic spinal cord (Czarnecki et al. 2014), the retina (Galli & Maffei 1988), and the cerebellum (Watt et al. 2009). Membrane depolarization activates VGCCs and NMDA receptors in developing neurons, allowing calcium influx into the cells to form calcium spikes (Leinekugel et al. 1995; Leinekugel et al. 1997). These calcium signals play vital roles in the Hebbian modification of synapse plasticity (Leinekugel et al. 1997) and the formation of the neuronal network (Luhmann et al. 2016). Consistent with previous findings, we confirmed that hippocampal neurons can fire large synchronous calcium spikes ~14 to 25 days in vitro (DIV) after initial plating in culture dishes. Surprisingly, these calcium spikes can produce subsequent periodic changes in zinc concentration (referred as zinc spikes) that are coordinated with pH dynamics. We reveal a novel signaling transduction mechanism in which changes in calcium concentration can trigger dynamic cellular acidification, leading to zinc spikes in neurons.

2. Materials and Methods

The experimental design of this study was not pre-registered and the study was exploratory.

2.1. Animals.

All experiments were conducted in strict compliance with the Institutional Animal Care and Use Committee (IACUC)-approved animal protocols from the University of Denver (approval number: 1363214–1). The timed, pregnant female SAS Sprague Dawley rats (strain code: 400) (RRID:MGI: 5651135) were purchased from Charles River and housed in the animal facility until gestational day 18. The rats were identified by a cage card and maintained under a 12:12 h light-dark cycle with food and water provided ab libitum. The pregnant rat was euthanized using CO₂, which is a rapid, safe, and humane method for euthanasia. To prevent spontaneous recovery after CO₂ exposure, secondary physical euthanasia via bilateral pneumothorax was used. Surgically removed rat fetuses (within the uterus) are immediately placed on ice to induce hypothermic anesthesia, and are kept on ice until euthanasia by decapitation. Sixteen timed, pregnant female SAS Sprague Dawley rats were used in this study. About 8–13 fetal rats can be obtained from each pregnant rat. The total number of pups used in this study is 128–208.

2.2. Primary rat hippocampal neuron culture.

Fetal rats were obtained from timed, pregnant Sprague Dawley rats at a gestational age of 18 days (E18) in order to isolate the hippocampal neurons according to the previously described method (Minckley et al. 2019). The hippocampi were removed from the brains in dissection medium containing 1X HBSS (Thermo Fisher Scientific Cat# 14185052), 10 mM HEPES buffer (pH 7.3) (Thermo Fisher Scientific Cat# 15630080) and 5 μg/mL gentamycin (Thermo Fisher Scientific Cat# 15750060). The hippocampi were minced and treated with dissection solution containing 20 U/mL papain (Worthington Cat# LS003126), and dissociated by 50 μg/mL DNAase I (Sigma-Aldrich Cat# D4527). Neurons were plated on 1 mg/mL poly-L-lysine-coated (Sigma-Aldrich, Cat# P0899) round glass coverslip-covered imaging dishes (MatTek Corporation, Part No: P35G-1.5–10-C). Around 1 × 10⁵ cells/cm² were plated in each dish in neuron plating medium containing MEM (Thermo Fisher Scientific Cat# 11095080) supplemented with glucose (Sigma-Aldrich Cat# G8769) and 5% FBS (Thermo Fisher Scientific Cat# 16000044). Cells were maintained in neuron culture medium containing Neurobasal medium (Thermo Fisher Scientific Cat# 21103049), GlutaMAX (Thermo Fisher Scientific Cat# 35050061) and B-27 (Thermo Fisher Scientific Cat# 17504001) after adhering. Hippocampal cultures were maintained at 37 °C, 5% CO₂ and fed every three days.

2.3. Recording of intracellular zinc, calcium and pH.

FluoZin-3 AM (Thermo Fisher Scientific Cat# F24195), Fura Red AM (Thermo Fisher Scientific Cat# F3020) and pHrodo Red AM (Thermo Fisher Scientific Cat# P35372) were used following the manufacturer’s instructions. Cell culture dishes were stained and imaging fields with healthy cells from each dish were arbitrarily chosen. For recording intracellular zinc and calcium, neurons were stained with 1 μM FluoZin-3 AM and 2 μM Fura Red AM in 1 mL neuron culture medium for 15 min at 37 °C, 5% CO_2. For recording intracellular zinc and pH, neurons were stained with 1 μM FluoZin-3 AM for 10 min and then stained with 1.25 μM pHrodo Red AM for 5 min at 37 °C, 5% CO₂. After staining for 15 min, 1 mL neuron culture medium containing dyes was replaced with 1mL prewarmed neuron culture media and neurons were incubated for another 15 min at 37 °C, 5% CO₂. Neurons were washed with phosphate-free HHBSS immediately before imaging. All imaging experiments were performed on the Nikon/Solamere CSUX1 spinning disc microscope. Images were collected with a 40 X 1.4 NA oil immersion objective. The recording of intracellular zinc dynamics by FluoZin-3 AM was acquired every 20 sec (1 sec for recording the sensor kinetics and 5 sec for measuring the duration of calcium and zinc spikes) with a 200 ms exposure of 488 nm laser excitation at 10 mW power. The ratiometric measurements of intracellular calcium dynamics by Fura Red have been made using 445 nm (calcium bound state) and 488 nm (calcium free state) with 100 ms exposure at 10 mW power. Each frame of Fura Red signals was acquired every 20 sec (1 sec for recording the sensor kinetics and 5 sec for measuring the duration of calcium and zinc spikes). The recording of intracellular pH dynamics by pHrodo Red AM was acquired every 20 sec, with a 200 ms exposure of 561 nm laser excitation at 10 mW power.

2.4. Drugs and buffers.

Neurons were imaged under phosphate-free HEPES-Buffered Hanks Balanced Salt Solution (HHBSS) buffer. The HHBSS buffer was prepared from Chelex-treated water to ensure there is no zinc in the buffer and only contained 1.26 mM CaCl₂ (Sigma-Aldrich, Cat# 21115), 5.4 mM KCl (Sigma-Aldrich Cat# P9541), 1.1 mM MgCl₂.6H₂O (Sigma-Aldrich Cat# M2393), 137 mM NaCl (Sigma-Aldrich Cat# S3014), 16.8 mM D-Glucose (Sigma-Aldrich Cat# G7528) and 30 mM HEPES (Sigma-Aldrich Cat# H4034). Drugs were prepared as follows: TPEN (Sigma-Aldrich Cat# P4413) was prepared as 25 mM stock solution in DMSO (Sigma-Aldrich, Cat# D8418). TPA (Sigma-Aldrich Cat# 723134) was prepared as 25 mM stock solution in DMSO. Nifedipine (Sigma-Aldrich Cat# N7634) was prepared as 100 mM stock solution in DMSO. APV (Tocris Cat# 0106) was prepared as 20 mM stock solution in H₂O. NBQX (Tocris, Cat# 1044) was prepared as 10 mM stock solution in H₂O. Bafilomycin (Sigma-Aldrich Cat# B1793) was prepared as 110 μM stock solution in DMSO. All solutions were diluted in the HHBSS buffer to their respective working concentrations during imaging experiments.

2.5. Data analysis.

Imaging data were collected through MicroManager software and analyzed with Fiji (Image J). Raw data output from Fiji were analyzed by Excel and KaleidaGraph program (Version 4.5). Shapiro-Wilk normality test and Jack-Knife outlier analysis were performed by JMP Pro (Version 14.2). For single wavelength sensors, FluoZin-3 and pHrodo Red, changes in fluorescence intensity (ΔF=F-F₀) were normalized to the baseline fluorescence immediately preceding 0 s (F₀) indicated as ΔF/F₀. For the ratiometric sensor, Fura Red, the indicator fluorescence ratio signal, R, were calculated to indicate calcium levels (equal to the emission intensity measured with 445 nm excitation divided by that measured with 488 nm excitation). The changes in excitation ratio (ΔR=R-R₀) were normalized to the baseline ratio immediately preceding 0 s (R₀) indicated as ΔR/R₀. The time-lapse imaging line plots were generated by connecting the sequential acquisition time point together in a time-lapse imaging experiment. Zinc, calcium, and pH spikes were identified as having ΔF/F₀ or ΔR/F₀ greater than or equal to 0.5. Integrated zinc, calcium, and pH signals were calculated in Excel to integrate all fluorescence signals for each spike from baseline to spike peak and back to baseline again. Statistical analysis was performed using student’s t-test. The student’s t-tests were preceded by an F-test to identify variance equality. Paired t-test was performed by pairing the zinc spike with the corresponding calcium or pH spike or pairing zinc spikes before and after drug treatments. Error bars indicate standard error of the mean (SEM). Significance levels are defined with p value less than 0.05. Group data are presented as mean ± SEM unless otherwise noted. All statistical tests were two-tailed. “N” numbers refer to the number of cells from at least 3 independent imaging dishes. No blinding or sample size calculations were performed. No exclusion criteria were pre-determined. Two data points were excluded in Fig.3h after performing Jack-Knife outlier analysis.

Figure 3. Generation of zinc spikes depends on calcium influx.

(a) Representative traces of zinc and calcium concentrations in the presence and absence of extracellular calcium. Zinc spikes were unobservable when extracellular calcium was removed. (b) Representative traces of zinc and calcium concentrations in neurons treated with 10 μM nifedipine, an inhibitor of VGCCs. Blocking VGCCs depleted both zinc and calcium spikes. (c) Representative traces of zinc and calcium concentrations in neurons treated with 20 μM APV to block NMDA receptors. Treatment with APV eliminated both zinc and calcium spikes. (d) Representative traces of zinc and calcium concentrations in neurons treated with 10 μM NBQX to block AMPA receptors. NBQX abolished both zinc and calcium spikes. (e) Representative traces of zinc and calcium concentrations in neurons treated with 5 μM DMSO. DMSO alone did not deplete zinc or calcium spikes. (f) Representative traces of zinc and calcium concentrations in neurons during 1-hour imaging experiment. (g) The effects of nifedipine, APV, NBQX, DMSO, and no drug treatment on the quantity of zinc and calcium spikes within 30 minutes (nifedipine treated: n = 18 cells from at least 3 independent imaging dishes; APV treated: n = 9 cells from at least 3 independent imaging dishes; NBQX treated: n = 9 cells from at least 3 independent imaging dishes; DMSO treated: n = 22 cells from at least 3 independent imaging dishes; No treatment: n = 10 cells from at least 3 independent imaging dishes ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05, n.s., not significant, paired t-test). (h) The correlation plot between integrated signals of zinc spikes and calcium spikes.

3. Results

3.1. Cultured hippocampal neurons fire spontaneous and synchronous zinc spikes

Intracellular zinc and calcium signals can be detected by various fluorescent probes. In order to record these two ions simultaneously, we utilized the small molecule sensors FluoZin-3 and Fura Red. FluoZin-3 (K_d = 15 nM) shows specific response to zinc and the signals are not affected by calcium, magnesium (Zhao et al. 2008), or pH changes (Devinney Ii et al. 2005). We confirmed the zinc specificity of FluoZin-3 by both in vitro and in situ experiments. Without the addition of zinc, the fluorescence intensity of 1 μM FluoZin-3 was not affected by 1 μM CaCl₂, 10 μM CaCl₂, 1 mM MgCl₂, 1 μM MnCl₂, or 1 μM CuCl₂ (Fig. S1a). When FluoZin-3 was bound with 1 μM ZnCl₂, its fluorescence was not changed by addition of 1 μM CaCl₂, 10 μM CaCl₂, 1 μM MgCl₂, or 1 μM MnCl₂, but was reduced by 35% with 1 μM CuCl₂ (Fig. S1b), which is consistent with previous finding that high concentration of copper (> 50 nM) can quench the fluorescence of FluoZin-3 when FluoZin-3 was bound with zinc (Zhao et al. 2009). In neurons stained with FluoZin-3 AM, the sensor fluorescence was not changed after neurons were treated with 1.26 mM CaCl₂, 100 μM MnCl₂, or 100 μM CuCl₂ for 5 minutes, but showed robust increased response to ZnCl₂ (Fig. S2). In addition, the FluoZin-3 sensor signals were unchanged from pH 9 to pH 6, with only slight reduction when pH is lower than 6 (Fig. S3). The ratiometric calcium indicator Fura Red displays distinct emission spectrum compared with FluoZin-3 so that the two probes can be used together to record zinc and calcium simultaneously. Fura Red has very high calcium-binding affinity (K_d = 140 nM) and its fluorescence is not perturbed by potentially competing ions such as magnesium and protons (Kurebayashi et al. 1993). With these two sensors, we found that some neurons in the hippocampal culture fired synchronized zinc spikes along with calcium spikes among axons, dendrites, and soma in the absence of extracellular zinc (Fig. 1a & 1b, and Movie S1). We tracked 1463 neurons in 335 dishes, where 48% of neurons displayed zinc spikes. These zinc spikes can be observed for more than 7 days and the critical window that we can find these spikes varied among different preparations of primary culture, usually between DIV 15 and DIV 21. Around 22% of neurons at DIV15 can fire zinc spikes, and the percent of neurons with zinc spikes can increase to 64% at DIV 21 (Fig. S4).

Figure 1. Spontaneous synchronous zinc spikes occur in primary hippocampal neurons.

(a) Representative fluorescent images of zinc (FluoZin-3) and calcium (Fura Red) dynamics in primary hippocampal neurons recorded at 260 sec, 320 sec, and 580 sec. Scale bar = 20 μm. (b) The time-lapse line plots of zinc spikes (green trace) and calcium spikes (magenta trace) illustrated in (a). (c) A representative trace of one single zinc spike preceded by one calcium spike and quantification of average zinc and calcium spikes duration. A zinc spike is generated after a calcium spike and lasts longer (n = 16 cells from at least 3 independent imaging dishes, ***p <0.001, paired t-test by paring the zinc spike with the corresponding calcium spike). Each frame of FluoZin3 and Fura Red signal was acquired every 5 sec in (c).

A close analysis of the spike onsets revealed that zinc spikes are always preceded by calcium spikes (Fig. 1c). There is a time delay of ~ 25 seconds between the initiation of a calcium spike and a zinc spike and there is a time delay of ~ 145 seconds between the peak of zinc spikes and calcium spikes. The average duration of a zinc spike was 408 seconds, significantly longer than the duration (266 seconds) of a calcium spike (Fig. 1c). The delayed response in zinc compared to calcium was not caused by differences in sensor kinetics of FluoZin-3 and Fura Red, as they both show similar turn-on kinetics in neurons upon with high calcium or zinc influx (Fig. S5).

3.2. Zinc spikes are preceded by calcium spikes

To further examine whether the calcium spikes are upstream of zinc spikes, we utilized different metal ion chelators to distinguish the two different cations. The membrane-permeable, heavy metal chelator N,N,N’,N’-tetrakis(2-pyridylmethyl)ethylenediamine (TPEN) is commonly utilized to reduce cellular zinc concentration due to its high affinity for zinc (Cho et al. 2007). TPEN can also chelate calcium, but with lower affinity, and it has been used to chelate high concentration of calcium in the Endoplasmic Reticulum (ER) (Morgan et al. 2012; Sztretye et al. 2007). We found that TPEN treatment depleted both zinc spikes and calcium spikes and reduced cytosolic zinc and calcium concentration below baseline levels, suggesting that TPEN can chelate both zinc and calcium (Fig. 2a & 2c). Compared with TPEN, the chelator Tris(2-pyridylmethyl)amine (TPA) demonstrates lower affinity for zinc (K_d = 10 pM), but with faster kinetics (Huang et al. 2013). TPA selectively abolished zinc spikes without depleting the calcium spikes (Fig. 2b & 2c), indicating that TPA can selectively chelate zinc without affecting calcium. The results also showed that chelation of zinc spikes would not affect calcium spikes, suggesting that zinc spikes occur downstream of calcium spikes. Previous studies have discovered that depolarization of hippocampal neurons induced calcium-dependent copper redistribution and enhanced labile copper concentration (Dodani et al. 2011). Chelation of intracellular copper (Cu⁺) by copper chelator neocuproine didn’t affect the frequency of dynamic changes in FluoZin-3 fluorescence intensity, suggesting that intracellular copper did not affect zinc spikes (Fig. S6).

Figure 2. Zinc spikes are downstream of calcium spikes.

(a) Representative line traces of zinc and calcium signals in neurons treated with 100 μM TPEN. TPEN abolished both cellular zinc and calcium spikes. (b) Representative line traces of zinc and calcium signals in neurons treated with 10 μM TPA. TPA selectively eliminated zinc spikes without influencing the calcium spikes. (c) The effects of TPEN and TPA treatment on the quantity of zinc and calcium spikes (TPEN treated: n = 12 cells from at least 3 independent imaging dishes. TPA treated: n = 9 cells from at least 3 independent imaging dishes. ****p<0.0001, **p<0.01, n.s., not significant, paired t-test).

3.3. Generation of zinc spikes depends on calcium influx

Because a calcium spike always occurs prior to a zinc spike, we hypothesize that zinc spikes are produced by calcium spikes. To test this hypothesis, we examined the zinc spikes and calcium spikes in calcium free buffer. As expected, both zinc spikes and calcium spikes disappeared in calcium-free buffer and resumed in buffers supplied with calcium (Fig. 3a), suggesting that the emergence of zinc spikes and calcium spikes hinged on extracellular calcium environment and hence calcium influx. Given that calcium influx can be mediated through the VGCCs and NMDA receptors (Chen et al. 1999; Leinekugel et al. 1997; Catterall 2011), we first utilized various drugs in neurons to block calcium influx. Both zinc spikes and calcium spikes vanished when neurons were exposed to the VGCC inhibitor (Helton et al. 2005; Saliba et al. 2009), nifedipine (Fig. 3b). Furthermore, we tested if NMDA receptors on the postsynaptic membrane are responsible for maintaining zinc spikes in neurons. Inhibiting NMDA receptors with DL-2-amino-5-phosphonovaleric acid (APV) (Anderson et al. 1987) depleted both the spontaneous zinc spikes and calcium spikes (Fig. 3c). Since activation of NMDA receptors require neuron depolarization which is usually caused by opening of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor, we also blocked the AMPA receptor by 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide (NBQX) (Van Damme et al. 2003). Inhibition of AMPA receptors also abolished the zinc spikes and calcium spikes in neurons (Fig. 3d). In the control experiments treated with DMSO, the amount of zinc and calcium spikes within 30 minutes didn’t change (Fig. 3e & 3g). During the 1-hour imaging period, the amplitude of zinc spikes tended to decrease even without any drug treatments (Fig. 3f), which might be caused by zinc efflux in the zinc free imaging buffer or FluoZin-3 dye leakage; however, the frequency of zinc spikes was unchanged (Fig. 3f & 3g). Hoechst staining showed that neurons were still healthy after 1-hour imaging of zinc spikes (Fig. S7). To provide further evidence about the crosstalk between zinc spikes and calcium spikes, we quantified and compared the two types of spikes. We defined spikes as any elevated signals with ΔF/F₀ greater than or equal to 0.5. The integrated signals for each zinc spike were plotted against the integrated calcium signals of the corresponding calcium spike fired prior to the zinc spike (Fig. 3h). A larger calcium spike was accompanied with a larger zinc spike, demonstrating a positive correlation between zinc spikes and calcium spikes. Our data not only provided evidence supporting spontaneous calcium spikes in developing neurons are evoked by glutamate activation and membrane depolarization, but also revealed that zinc spikes depend on calcium influx and neural excitability.

3.4. Zinc spikes are accompanied by pH dynamics in neurons

Given that spontaneous zinc spikes are recorded in zinc free buffer, this suggests that zinc spikes are caused by zinc released from intracellular stores rather than influx from extracellular buffer. We sought to determine the mechanism by which increased calcium can enhance intracellular zinc release in neurons. Extensive evidence has shown that calcium can cause cellular acidification in different types of neurons including dorsal root ganglia and hippocampal neurons (Fujita et al. 2008; Dhaka et al. 2009; Hartley & Dubinsky 1993). It has been established that glutamate induced calcium influx can cause cellular acidification, which can liberate zinc from intracellular stores in hippocampal neurons (Kiedrowski 2011; Kiedrowski 2012). Therefore, cytosolic acidification might act as an intermediate signal to transduce the calcium signals to zinc signals. In order to test this, we utilized the intracellular pH indicator pHrodo Red (Arppe et al. 2014) along with FluoZin-3 to track pH and zinc simultaneously. The imaging results clearly demonstrated that a downward pH spike occurred almost simultaneously with a zinc upward spike (Fig. 4a & Movie S2), with only a 2-second delay between the initiation of a pH spike and a zinc spike (n = 23 cells).

Figure 4. Upward zinc spikes are synchronized with downward pH spikes.

(a) Representative traces of zinc and pH signals in neurons. (b) Representative traces of zinc and pH in neurons treated with 200 nM bafilomycin to block v-ATPase. Bafilomycin depleted the large pH spikes and zinc spikes. (c) Representative traces of zinc and pH in neurons treated with 100 μM TPEN. TPEN blocked both zinc and pH spikes in neurons. (d) Representative traces of zinc and pH in neurons treated with 10 μM TPA. TPA abolished zinc spikes without changing pH spikes. (e) Effects of bafilomycin, TPEN, and TPA treatment on quantity of zinc and pH spikes within 30 minutes (bafilomycin treated: n = 26 cells from at least 3 independent imaging dishes; TPEN treated: n = 11 cells from at least 3 independent imaging dishes; TPA treated: n = 15 cells from at least 3 independent imaging dishes. ****p<0.0001, ***p<0.001, n.s., not significant, paired t-test).

The delay between the pH spike and zinc spike suggest that the zinc spike is evoked by the pH spike. In order to provide further evidence about the causative relationship between pH spikes and zinc spikes, we perturbed cellular pH homeostasis using the v-ATPase inhibitor bafilomycin (Cavelier & Attwell 2007). A pH spike was evoked right after addition of bafilomycin, which might be caused by the sudden cellular acidification when v-ATPase is inhibited (Fig. 4b). Additional pH spikes detected were visibly diminished in the presence of bafilomycin and the changes in the duration and number zinc spikes were coordinated with the changes in pH spikes (Fig. 4b & Fig. 4e). When we applied the metal ion chelator TPEN which abolishes both calcium and zinc spikes, pH spikes were depleted as well (Fig. 4c & Fig. 4e). However, with the treatment of TPA to selectively abolish zinc spikes (Fig. 2b), cellular pH spikes were unaffected (Fig. 4d & Fig. 4e). The data indicates that the generation of zinc spikes depends on pH spikes, while pH spikes are not affected by zinc spikes.

3.5. pH spikes depend on calcium influx

After discovering that the zinc spikes were coordinated with pH spikes, we then examined the causative relationship between calcium spikes and pH dynamics. We had shown that the blockade of L-type VGCCs, NMDA receptors, and AMPA receptors all abolished both calcium spikes and zinc spikes (Fig. 3b–d). Similarly, inhibiting calcium influx by blocking VGCC and NMDA receptors also depleted the pH spikes (Fig. 5a & 5b). In addition, inhibition of AMPA receptors by NBQX also impeded the generation of pH spikes (Fig. 5c). These data show that blocking calcium influx can suppress spontaneous pH spikes, suggesting that pH spikes are generated by the introduction of calcium during neuronal excitation.

Figure 5. Calcium influx elevated cytosolic zinc and acidified cytoplasm.

(a) Representative traces of zinc and pH spikes in neurons treated with 10 μM nifedipine. Nifedipine abolished both zinc and pH spikes. (b) Representative traces of zinc and pH spikes in neurons treated with 20 μM APV. APV depleted both zinc spikes and pH spikes. (c) Representative traces of zinc and pH spikes in neurons treated with 10 μM NBQX. NBQX eliminated both zinc spikes and pH spikes. (d) The effects of nifedipine, APV, and AMPA treatment on quantity of zinc and pH spikes within 30 minutes (nifedipine treated: n = 26 cells from at least 3 independent imaging dishes; APV treated: n = 14 cells from at least 3 independent imaging dishes; NBQX treated: n = 11 cells from at least 3 independent imaging dishes. ****p<0.0001, ***p<0.001 paired t-test).

4. Discussion

This work made a serendipitous discovery that some neurons (~48%) in primary hippocampal cultures can fire spontaneous synchronized zinc spikes, usually between DIV 15 and DIV 21 (Fig. S4). In addition to primary hippocampal neurons, such zinc spikes were also detected in primary cortical neurons (Fig. S8). Previous studies have found that spontaneous neural activity and calcium oscillations can also be identified during this critical early neuronal developmental stage (Luhmann et al. 2016; Kirischuk et al. 2017; Bacci et al. 1999; Penn et al. 2016). Based on our results, we propose a model that zinc spikes are initiated by large intrinsic neural excitability, hence synchronized calcium spikes, in the young developing neurons (Fig. 6). At the early developmental stages, the neuronal population can generate spontaneous large burst of neural activity mediated by the developmental shift from electrical synaptic transmission to chemical transmission (Luhmann et al. 2016). This chemical transmission includes activation of both glutamate receptors and GABA_A receptors (Opitz et al. 2002). The majority (85–90%) of neurons in primary hippocampal cultures are excitatory glutamatergic neurons (Benson et al. 1994). When we blocked the NMDA receptor and AMPA receptor, spontaneous zinc spikes were depleted (Fig. 3c & 3d), suggesting that zinc spikes were driven by the large spontaneous neural excitability in developing glutamatergic neurons. Such neural activity is synchronized to generate giant depolarizing potentials, which can evoke substantial calcium influx, generating spontaneous and synchronous calcium spikes. During simultaneous recordings of intracellular calcium and zinc in hippocampal neurons, each zinc spike was always preceded by a calcium spike (Fig.1b & 1c). The zinc spikes were depleted in the absence of extracellular calcium (Fig. 3a) or by blocking of calcium influx (Fig. 3b – d), whereas abolishing zinc spikes with the zinc-specific chelator TPA did not reduce the calcium spikes (Fig. 2b). All these lines of evidence demonstrate that the generation of zinc spikes depends on calcium spikes, while generation of calcium spikes does not require zinc spikes.

Figure 6. Model of calcium influx-induced zinc spikes in neurons.

(1) Intrinsic neural excitability evokes large calcium influx in neurons at the early developing stages. (2) Calcium buffering systems on the plasma membrane or organellar membranes pump calcium out of the cytosol and protons into the cytosol, thereby reducing intracellular calcium concentration, (3) resulting in intracellular accumulation of protons and downward pH spikes. (4) This cellular acidification liberates zinc from zinc-binding ligands such as metallothionein and glutathione, (5) generating periodic zinc spikes synchronized with neural excitability and calcium spikes.

Further studies in this work reveal that cellular acidification mediates the signal transduction process from calcium spikes to zinc spikes. We found that the neurons can generate downward pH spikes that were synchronized with zinc spikes with similar timing of both initiation and duration (Fig. 4a). These pH spikes were diminished when calcium influx was blocked (Fig. 5a – c), but were not affected by zinc chelation (Fig. 4d), suggesting that pH spikes occur after calcium spikes, but before zinc spikes in the signal transduction sequence. In our proposed model, the pH downward spikes are generated by calcium spikes, which is supported by previous findings that calcium influx can induce cellular acidification. For example, NMDA can induce calcium-dependent cellular acidification in rat hippocampal neurons (Irwin et al. 1994). The neurotransmitter glutamate and pungent compound capsaicin can evoke cellular acidification through an increase in cellular calcium concentration in sensory neurons (Takahashi & Copenhagen 1996; Hwang et al. 2011). Activation of hippocampal CA1 neurons can also induce extracellular alkalization through calcium influx (Makani & Chesler 2010). Calcium-induced intracellular acidification might be mediated by two different pathways (Fig. 6). First, high concentration of cytosolic calcium can be quickly pumped out of cells by the plasma membrane calcium-ATPase, which exchanges cytosolic calcium for extracellular protons (Daugirdas et al. 1995), thereby reducing intracellular pH in neurons. In addition, calcium/proton exchangers on the organellar membranes could also contribute to the intracellular acidification (Nishi & Forgac 2002). Intracellular pH homeostasis is maintained by the v-ATPase proton pump (Pamarthy et al. 2018). When the v-ATPase was inhibited, both zinc spikes and pH spikes were reduced (Fig. 4b). Previous studies, especially Kiedrowski’s work, support our model that zinc spikes are produced through cellular pH dynamics. Acidification has been shown to enhance zinc release from zinc binding residues both in hippocampal neurons (Kiedrowski 2012) and in vitro (Kiedrowski 2014).

It is estimated that total zinc concentration in mammalian cells ranges from 100 to 500 μM, with the majority of cellular zinc bound to proteins and small ligands, while the free, labile zinc in the cytosol and intracellular organelles such as ER, Golgi, and mitochondria are in the picomolar range at a resting state (Qin et al. 2011; Park et al. 2012). In glutamatergic neurons, which are the major neuronal type in the primary hippocampal cultures (85 – 90%) (Benson et al. 1994), there are three intracellular zinc stores that might contribute to the observed spontaneous zinc spikes. First, it has long been known that high levels of labile zinc are present in the synaptic vesicles of glutamatergic neurons and the concentration may reach hundreds of micromolar to millimolar range (Frederickson et al. 2000). Although it is unclear if such synaptic zinc can be delivered from the lumen of synaptic vesicles into the cytosol, the zinc transporters such as ZnT3 localized on synaptic vesicles might be able to utilize the zinc/proton mechanism to mediate zinc release. Previous studies showed that a zinc transporter (ZnT) homologue in E. coli, ZitB, utilizes the zinc/proton exchange mechanism to regulate zinc homeostasis (Chao & Fu 2004). Second, endolysosomal vesicles store a pool of zinc in hippocampal neurons. Activation of the lysosomal cation channel TRPML1 can release zinc into the cytosol which can be detected by a sensitive zinc sensor, GZnP3, which we recently developed (Minckley et al. 2019). Lastly, the zinc buffering protein metallothionein (MT) can act as a source for zinc due to its high expression levels and multiple zinc binding sites (Krężel & Maret 2006). MTs are a family of cysteine-rich proteins (~30% cysteine composition). The brain-specific MT gene, MT3 is predominantly expressed in hippocampal neurons (Aschner 1996; Juárez-Rebollar et al. 2017). Studies have suggested that zinc can be liberated from MTs by oxidants or acidosis (Shuttleworth & Weiss 2011). Small ligands such as glutathione and free cysteine amino acids can also buffer a pool of zinc that can be released by protons (Kiedrowski 2014). Because large zinc spikes were observed in the soma, much higher than previously observed lysosomal zinc release (Minckley et al. 2019), we speculate that the zinc buffering proteins and ligands are the major source from where zinc spikes are generated. Further studies will be needed to clarify the roles of these low pH-sensitive intracellular stores in the generations of spontaneous zinc spikes in neurons.

Given the integral roles of zinc in brain development, the zinc spikes might be involved in modulating neuronal development. Zinc deficiency during brain development can result in learning and memory abnormalities in rat pups (Halas et al. 1986). At the cellular level, zinc deficiency reduces neuronal precursor proliferation, impairs neuronal differentiation, and hinders neurogenesis (Levenson & Morris 2011). In addition, evidence has suggested that zinc plays paramount roles in synaptogenesis and synapse maturation by targeting potential zinc effectors. Biochemical evidence suggests that changes in the physiological levels of zinc (picomolar concentration) can modulate the activities of tyrosine phosphatase (Wilson et al. 2012), which is an important enzyme for hippocampal synapse formation and learning (Fuentes et al. 2012). In addition, zinc can modulate the brain-derived neurotrophic factor (BDNF) signaling which is involved in neuronal differentiation, synaptogenesis, and synaptic plasticity (Zagrebelsky & Korte 2014; Scharfman & MacLusky 2014; Leal et al. 2015). Zinc regulates BDNF expression levels, by activating matrix metallopeptidases (MMPs) that cleave pre-BDNF into BDNF and transactivate tyrosine receptor kinase B (TrkB) through BDNF-independent mechanisms (Hwang et al. 2005). In dendritic postsynaptic spines, zinc can be recruited by the excitatory postsynaptic scaffolding protein Shank3 and Shank2 to regulate the development and maturation of synapses in the hippocampus and cortex (Grabrucker et al. 2011; Roussignol et al. 2005). The thickness of the postsynaptic density (PSD) and the synaptic localization of Shank3 could be influenced by local zinc concentration (Ha et al. 2018). Therefore, these zinc spikes might exert vast downstream effects in neuronal development by regulating the function and localization of a cohort of proteins.

In summary, we report a novel finding that primary hippocampal neurons can generate spontaneous and synchronous zinc spikes that are initiated from spontaneous neural activity (Fig. 6). The dynamic changes in zinc concentration might alter the activity and localization of proteins related to neural growth, synapse maturation, and synaptogenesis, and hence play a significant role in the regulation of neuronal development.

Abbreviations:

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid

APV

DL-2-amino-5-phosphonovaleric acid

BDNF

Brain-derived neurotrophic factor

DIV

Days in vitro

DMEM

Dulbecco’s Modified Eagle Medium

E18

embryonic day 18

FBS

Fetal bovine serum

HHBSS

HEPES-Buffered Hanks Balanced Salt Solution

IACUC

Institutional Animal Care and Use Committee

MMPs

Matrix metallopeptidases

MT

Metallothionein

NBQX

1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide

NMDA

N-methyl-d-aspartate

PSD

Postsynaptic density

SEM

Standard error of the mean

TPA

Tris(2-pyridylmethyl)amine

TPEN

N,N,N’,N’-tetrakis(2-pyridylmethyl)ethylenediamine

TrkB

Tyrosine receptor kinase B

TRPA1

Transient Receptor Potential Ankyrin 1

VGCCs

Voltage-gated calcium channels

ZnT

Zinc transporter