Abstract
Zinc is crucial for neuron function, but whether and how labile zinc ion (Zn2+) acts as an intracellular signaling molecule remains unclear. In this work, we investigate the relationship between Ca2+ and Zn2+ dynamics using fluorescence imaging. Our findings reveal that manipulating Ca2+ influx through various pathways induces intracellular acidification, which subsequently elicits Zn2+ spikes that reflect transient increases in cytosolic Zn2+ levels. These Ca2+-dependent Zn2+ spikes have been recorded in both rat (Rattus norvegicus) primary neuron cultures and organotypic mouse (Mus musculus) hippocampal slice cultures prepared from both males and females. They are specific to neurons and astrocytes but are absent in other cell types we tested including HeLa cells, COS-7 cells, and fibroblasts. We further identify Metallothionein III (MT3), a Zn2+ buffering protein specifically expressed in brain cells, as the source of these Zn2+ spikes. Reduction in MT3 expression by knockdown with shRNAmiR techniques significantly decreases the amplitude of Zn2+ spikes, while overexpression of MT3 in HeLa and COS-7 cells is sufficient to induce Ca2+-dependent Zn2+ spikes, demonstrating the crucial roles of MT3 in Zn2+ release. Lastly, we explore the biological roles of MT3-mediated Zn2+ spikes in neurons. Suppressing Zn2+ spikes with either MT3 knockdown or mild Zn2+ chelation results in increased dendritic branching in primary rat hippocampal neurons. These results suggest that Zn2+ release from endogenous MT3 acts as a regulatory signal to inhibit dendrite branching and growth, establishing a critical role for Zn2+ spikes in neurite outgrowth and neuronal development.
Significance Statement
Zinc is essential for brain development, primarily known for its role in supporting protein structure and enzymatic activity. However, its function as an intracellular signaling molecule has been debated because labile zinc (Zn2+) concentrations inside cells are typically stable. In this study, we discovered a unique pathway where Ca2+ influx triggers cellular acidification, which subsequently releases Zn2+ from Metallothionein III (MT3), a Zn2+-binding protein highly expressed in the brain. More importantly, we found that depletion of these Zn2+ spikes via MT3 knockdown or chelation increases dendritic arborization, a critical step in forming neural connections. Our findings reveal that Ca2+ influx activates MT3-mediated Zn2+ signaling, which fine-tunes the neuronal network maturation, highlighting previously unrecognized signaling roles of Zn2+ in brain development.
Introduction
Both synchronous and asynchronous Ca2+ oscillations are distinctive patterns of activity in developing neurons, which have been observed in isolated neuron culture (Sun and Südhof, 2021), brain slices (Gust et al., 2003; Corlew et al., 2004), and in vivo (Adelsberger et al., 2005). These Ca2+ spikes depend on extracellular Ca2+ influx, but intracellular Ca2+ stores can also contribute to increased Ca2+ levels (Gu et al., 1994). Ca2+ transients occur in early neuronal development and contribute to brain development. Asynchronous Ca2+ spikes are involved in neuron differentiation, and their frequency can define neurotransmitter type (Gu and Spitzer, 1995; Borodinsky et al., 2004; Spitzer et al., 2004). On the other side, synchronized Ca2+ oscillations play a role in neurogenesis (Rash et al., 2016) and the maturation of neural networks (Opitz et al., 2002; Allène et al., 2008). Synchronous Ca2+ spikes have been reported in multiple brain areas, including cerebral cortex (Opitz et al., 2002; Allène et al., 2008), hippocampus (Mohajerani and Cherubini, 2006), cerebellum (Ramirez and Stell, 2016), thalamus (Pangratz-Fuehrer et al., 2007), embryonic spinal cord (Czarnecki et al., 2014), and retina (Protti et al., 2000).
Our previous work revealed that synchronous Ca2+ oscillations are accompanied by Zn2+ spikes in primary rat hippocampal and cortical neurons (Zhang et al., 2021). Intracellular Zn2+ concentrations are tightly regulated by Zn2+ transporters and Zn2+ binding proteins, which control its uptake, efflux, and storage. Two families of Zn2+ transporters, ZnTs and ZIPs, work together to maintain low cellular Zn2+ levels. ZnTs, consisting of 10 isoforms, are primarily responsible for Zn2+ efflux, while ZIPs, comprising of 14 isoforms, mediate Zn2+ influx into the cytosol (Baltaci and Yuce, 2018; Kambe et al., 2021). Additionally, metallothioneins (MTs), a class of cysteine-rich Zn2+ binding proteins, can bind up to seven Zn2+ ions within their two subdomains, serving as crucial buffers for cellular Zn2+ homeostasis (Babula et al., 2012; Baltaci et al., 2018). MT1 and MT2 isoforms are widely expressed throughout the body, while MT3 and MT4 isoforms have more specialized localization: MT3 is primarily localized to the central nervous system (CNS) and MT4 is found in stratified epithelial cells (Thirumoorthy et al., 2011). The brain-enriched isoform, MT3, was initially discovered due to its decreased expression in Alzheimer's disease (Uchida et al., 1991). MT3 is present in multiple brain regions (Masters et al., 1994) and, when applied exogenously to primary neuron cultures, has been found to inhibit neurite growth (Uchida et al., 1991). Additional studies show that MT3 protects neurons against excitotoxicity (Erickson et al., 1997; Koumura et al., 2009b), and changes in MT3 expression are associated with brain injury and neurodegeneration (Hozumi et al., 1998).
This study is a continuation of our previous discoveries to identify the source and function of Ca2+-dependent Zn2+ spikes in neurons. We showed that artificially inducing Ca2+ influx triggered Zn2+ spikes in primary neurons and astrocytes, but not in other cell types. In addition, stimulation-induced action potentials, which trigger robust Ca2+ influx, also induced Zn2+ transients in CA1 neurons within organotypic mouse hippocampal slice cultures. Furthermore, we provided evidence that the brain-enriched MT isoform MT3 is the primary source of such Zn2+ spikes. Knockdown of MT3 through shRNAmiR significantly reduced Zn2+ spike amplitude in primary rat hippocampal neurons, while overexpression of rat-MT3 in non-neuronal cell types was sufficient to initiate Ca2+-dependent Zn2+ spikes. Lastly, we found that both MT3 knockdown and Zn2+ chelation enhanced dendritic arborization and simultaneous MT3 knockdown and Zn2+ chelation did not further increase dendritic complexity, suggesting that MT3 might modulate neurite development via Zn2+ signals released from it. Overall, our results suggest that large, synchronized Ca2+ spikes induce neuronal acidification, which in turn promotes Zn2+ release from MT3. This process may contribute to the fine-tuning of neural circuit formation during brain development.
Materials and Methods
Animals
Pregnant Sprague Dawley rats were purchased from Charles River. All experiments were conducted in strict compliance with the Institutional Animal Care and Use Committee (IACUC) approved animal protocols from the University of Denver. Mice were acquired from the Jackson Laboratory (C57BL/6NJ), and experiments employing mouse brain slices were conducted in accordance with the IACUC of the University of Colorado on Anschutz Medical Campus and National Institutes of Health guidelines.
Primary rat neuron culture and transfection
Primary dissociated hippocampal and cortical neurons cultures were prepared from embryonic day 18 (E18) rats according to previous protocol (Kaech and Banker, 2006). The hippocampi or cortex were removed from the brains in dissection medium containing 1× HBSS, 10 mM HEPES buffer, pH 7.3, and 5 µg/ml gentamycin. The hippocampi or cortex were minced, treated with dissection solution containing 20 U/ml papain (Worthington), and dissociated by 50 µg/ml DNase I (Sigma). Neurons were plated on 1 mg/ml poly-d-lysine-coated (Sigma) round glass coverslip-covered dishes (MatTek). Approximately 1 × 105 cells/cm2 were plated in each dish in plating medium (MEM supplemented with d-glucose and 5% FBS). Cells were maintained in culture medium (Neurobasal medium supplemented with GlutaMAX-1 and B-27) after adhering. Neuron cultures were maintained at 37°C, 5% CO2 and fed every 3 d.
Hippocampal neurons were transfected from DIV 3 to DIV 6, using the Lipofectamine 3000 transfection kit (Thermo Fisher Scientific) in 250–500 µl Opti-MEM. The reagent-DNA mixture was incubated for 30 min at room temperature before directly adding to the neuron imaging dishes. Before reagent-DNA was added, 1 ml media was removed from each dish and filtered with an equal volume of fresh neuron culture media to make 50:50 media. After 4 h of 37°C incubation, the media was replaced with 2 ml 50:50 media. Neuron imaging dishes were incubated at 37°C, 5% CO2 until imaging.
Organotypic mouse hippocampal slice cultures
Organotypic slice cultures from the hippocampus (400 µm thick) were prepared from postnatal day 2 (P2)–P3 male and female pups, as described previously (Kleinjan et al., 2023). The age of slice culture is reported as equivalent postnatal (EP) day (i.e., postnatal day at slice culturing + days in vitro). For two-photon imaging and electrophysiology experiments, slices were transferred to a submersion-type, temperature-controlled recording chamber (TC-324C, Warner Instruments) and perfused with artificial CSF (ACSF; in mM: 127 NaCl, 25 NaHCO3, 1.25 NaH2PO4, 2.5 KCl, 25 d-glucose, aerated with 95% O2/5% CO2; Ogelman et al., 2024). Imaging and electrophysiological recordings were performed at 30°C.
Additional cell culture and transfection
Astrocytes were prepared from cortical cells from embryonic day 18 (E18) rats following the same protocol stated above for primary hippocampal neurons. The cortexes were removed from the brains in dissection medium, minced, treated with dissection solution containing papain, and dissociated using DNase. Cells were plated on 1 mg/ml poly-ʟ-lysine-coated round glass coverslip-covered dishes. Approximately 1 × 105 cells/cm2 were plated in each dish in plating medium. Cells were maintained in culture media after adhering, then switched to astrocyte media (DMEM with 20% FBS) at DIV 2–3, and maintained at 37°C, 5% CO2. Cells were fed every 3 d. Astrocytes were transfected from DIV 7 to DIV 9, using the Lipofectamine 3000 transfection kit in 500 µl Opti-MEM. The reagent-DNA mixture was incubated for 30 min at room temperature before directly adding it to the imaging dishes. Before reagent-DNA was added, 1 ml media was removed from each dish and filtered with an equal volume of fresh astrocyte media to make 50:50 media. After 4 h of 37°C incubation, media was replaced with 2 ml 50:50 media. Astrocyte imaging dishes were incubated at 37°C, 5% CO2 until imaging.
HeLa cells were maintained in DMEM with 10% FBS at 37°C, 5% CO2. HeLa cells were transfected at ∼40–50% confluency using 3 µl homemade polyethylenimine (PEI) transfection reagent and 1.5 µg DNA in 250 µl Opti-MEM. The mixture was incubated for 30 min at room temperature before directly adding the mixture to the imaging dishes. Imaging dishes were incubated at 37°C, 5% CO2 for 48 h and then imaged.
COS7 cells were maintained in DMEM with 10% FBS and 1× pen-strep at 37°C, 5% CO2. COS7 cells were transfected at ∼40–50% confluency using 3 µl PEI transfection reagent and 1.5 µg DNA in 250 µl Opti-MEM. The mixture was incubated for 30 min at room temperature before directly adding the mixture to the imaging dishes. Imaging dishes were incubated at 37°C, 5% CO2 for 48 h and then imaged.
Wild-type patient fibroblast cell line GM03340F was maintained in DMEM with 10% FBS and 1× GlutaMAX-1 at 37°C, 5% CO2 until imaging.
Recording of intracellular Ca2+, Zn2+, and pH
FluoZin-3 AM (Thermo Fisher Scientific), Fura Red AM (Thermo Fisher Scientific), Fluo-4 AM (Thermo Fisher Scientific), pHrodo Red AM (Thermo Fisher Scientific), and pHrodo Green AM (Thermo Fisher Scientific) were loaded following the manufacturer's instructions. For simultaneous recording of Ca2+ and Zn2+ dynamics in neurons, cells were stained with 2 µM Fura Red AM and 1 µM FluoZin-3 AM in 1 ml of neuron culture medium for 15 min at 37°C, 5% CO2. For simultaneous recording of Ca2+ and pH dynamics in neurons, cells were stained with 1 µM Fluo-4 AM for 10 min at 37°C, 5% CO2, followed by 1.25 µM pHrodo Red AM for another 5 min. For simultaneous recording of Zn2+ and pH dynamics in neurons, cells were stained with 1 µM FluoZin-3 AM for 10 min at 37°C, 5% CO2, followed by 1.25 µM pHrodo Red AM for another 5 min. After staining, dye-containing medium was replaced with 1 ml of fresh, prewarmed neuron culture media, and neurons were incubated for another 15 min at 37°C, 5% CO2. Neurons were then washed with the indicated buffer before imaging.
Astrocytes, HeLa cells, COS7 cells, and fibroblasts used to measure Ca2+ dynamics were stained with 2 µM Fluo-4 AM in homemade phosphate-free HHBSS buffer for 1 h at 37°C, 5% CO2. To measure Zn2+ dynamics in these cells, they were stained with 2 µM FluoZin-3 AM in phosphate-free HHBSS buffer for 1 h at 37°C, 5% CO2. To measure pH dynamics, cells were stained with 1 µM pHrodo Green AM in phosphate-free HHBSS buffer for 15 min at 37°C, 5% CO2. Fibroblast dishes used to measure Ca2+ and pH dynamics were stained with 2 µM Fluo-4 AM in phosphate-free HHBSS buffer for 45 min at 37°C, 5% CO2, then with 1.25 µM pHrodo Red AM for another 15 min at 37°C, 5% CO2. HeLa cell dishes used to record intracellular Zn2+ and pH dynamics in experiments over 1 h were stained with 5 µM FluoZin-3 AM in phosphate-free HHBSS buffer for 20 min and then stained with 1.25 µM pHrodo Red AM for another 10 min at 37°C, 5% CO2. After staining, dye-containing buffer was replaced with fresh, prewarmed buffer and cells were incubated for another 30 min at 37°C, 5% CO2.
All imaging was performed on a Nikon/Solamere CSUX1 spinning disk microscope. Images were collected with a 40× 1.4 NA oil immersion objective. The recording of intracellular Ca2+ dynamics by Fluo-4 AM was acquired every 5 s (488 nm excitation, 200 ms exposure, 10 mW power). The ratiometric measurements of intracellular Ca2+ dynamics by Fura Red AM were made using dual-excitation at 435 nm (Ca2+ bound state) and 488 nm (Ca2+ free state) with 100 ms exposure at 10 mW power. The acquisition intervals were 5 or 20 s. The recording of intracellular Zn2+ dynamics by FluoZin-3 AM was acquired every 5, 10, or 20 s (488 nm excitation, 200 ms exposure, 10 mW power). The recording of intracellular pH dynamics by pHrodo Red AM was acquired every 20 s (560 nm excitation, 200 ms exposure, 10 mW power). The recording of intracellular pH dynamics by pHrodo Green AM was acquired every 20 s (488 nm excitation, 200 ms exposure, 10 mW power).
Two-photon imaging of Zn2+ dynamics in CA1 hippocampal neurons
FluoZin-3 (Thermo Fisher Scientific) was introduced into CA1 pyramidal neurons via the recording pipette during whole-cell recordings (see below, Electrophysiology). Imaging was performed on CA1 pyramidal neurons at depths of 20–50 µm of hippocampal slice cultures at EP16–18 using a two-photon microscope (Investigator, Bruker) with a pulsed Ti:sapphire laser (MaiTai HP, Spectra-Physics) tuned to 920 nm (3–5 mW at the sample). The microscope and data acquisition were controlled with Prairie View (Bruker) and 302RM (Conoptics). Neurons were imaged at 10 min intervals in recirculating ACSF at 30°C aerated with 95% O2/5% CO2 (∼310 mOsm), pH 7.2, with 2 mM CaCl2 and 1 mM MgCl2, and 0.001 mM tetrodotoxin (TTX) was added for controls. For each neuron, image stacks (512 × 512 pixels, 0.048 μm/pixel) were acquired with 1 μm z-steps, targeting the soma of CA1 pyramidal neurons under conditions with and without action potential induction. All images shown are maximum projections of 3D image stacks after applying a median filter (2 × 2) to the raw image data.
Electrophysiology
Whole-cell recordings (electrode resistance, 6–9 MΩ; series resistance, 20–50 MΩ) were performed at 30°C on visually identified CA1 pyramidal neurons of hippocampal slice cultures (EP16–18) using a MultiClamp 700B amplifier (Molecular Devices). To evoke action potentials, whole-cell recordings were performed in current-clamp mode using potassium-based internal solution containing FluoZin-3 (in mM: 136 K-gluconate, 10 HEPES, 17.5 KCl, 9 NaCl, 1 MgCl2, 4 Na2-ATP, 0.4 Na-GTP, 0.06 FluoZin-3, and ∼300 mOsm, ∼pH 7.26). To examine Zn2+ dynamics at the soma, a train of action potentials was evoked by 5 step current injections (300 pA, 300 ms each) delivered at 0.2 Hz, 10 min after whole-cell break-in. Signals were filtered at 2 kHz and digitized at 10 kHz and responses were analyzed using Clampfit 10.3 (Molecular Devices) and OriginPro 8.5 software (OriginLab).
Drugs and buffers
Cells were imaged using phosphate-free HHBSS buffer, which contains the following (in mM): 1.26 CaCl2, 5.4 KCl, 1.1 MgCl2.6H2O, 137 NaCl, 16.8 d-glucose, and 30 HEPES. Cells were also imaged using calcium-free phosphate-free HHBSS buffer, which contains the following (in mM): 5.4 KCl, 1.1 MgCl2.6H2O, 137 NaCl, 16.8 d-glucose, and 30 HEPES. The high KCl HHBSS used to depolarize neurons contains the following (in mM): 1.26 CaCl2, 100 KCl, 1.1 MgCl2.6H2O, 41.14 NaCl, 16.8 d-glucose, and 30 HEPES. The calcium-free high KCl HHBSS buffer was made with the following (in mM): 41.14 NaCl, 100 KCl, 16.8 d-glucose, 30 HEPES, and 1.1 MgCl2.6H2O. All buffers were adjusted to pH 7.2.
TPA was prepared as a 25 mM stock solution in DMSO. AITC was prepared as a 100 mM stock solution in DMSO. Glutamate was prepared as a 10 mM stock solution in DMSO. Glycine was prepared as a 10 mM stock solution in DMSO. BAPTA-AM was prepared as a 4 mM stock solution in DMSO. 2-Amino-5-phosphonopentanoate (APV) was prepared as a 50 mM stock solution in water. All solutions were diluted in the phosphate-free HHBSS buffer indicated to their respective working concentrations during imaging experiments.
Optical excitation in hippocampal neurons
Hippocampal neurons were transfected with pCAG-Chrimson-tdTomato (Klapoetke et al., 2014), then stained with Fluo-4 AM or FluoZin-3 AM as previously described, and imaged from DIV 7 to DIV 14. These dishes were protected from light during the staining process. ChR (Ex 561 nm, 2 mW, 4,000 ms) and GFP (Ex 488 nm, 10 mW, 200 ms) channels were set up, and images were acquired every 5 s for Ca2+ imaging and 10 s for Zn2+ imaging.
Characterizing the effects of pH on cellular Zn2+
To assess the changes in Zn2+ concentrations at different pH levels, pHrodo Red AM and FluoZin-3 AM were used to measure pH and Zn2+ dynamics simultaneously. Rat hippocampal neurons and HeLa cells were stained as previously described. The pH base buffer contains the following: 1.3 mM CaCl2, 25.4 mM KCl, 1 mM MgCl2, 2 µM carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP), 134.2 mM K-gluconate, 3.6 mM KHCO3, and 10 mM 2-[N-morpholino]ethanesulfonic acid (MES), or piperazine-1,4-bis(2-ethanesulfonic acid) (PIPES), or HEPES following a published protocol (Kiedrowski, 2011). The pH buffers were adjusted to different pHs (MES, pH 5.5–6.5; PIPES, pH 6.6–7; HEPES, pH 7.2–8) and treated with 5 µM nigericin and 5 µM monensin before acquiring images.
Immunofluorescence staining
Following a published method (Glynn and McAllister, 2006), imaging dishes were washed three times with 1 ml 1× PBS, pH 7.4, containing the following (in mM): 137 NaCl, 2.7 KCl, 10 Na2HPO4, and 1.8 Na2HPO4. Then, 2 ml of fixing buffer, pH7.4, containing 4% paraformaldehyde, 0.05 M MgCl2, 0.01 M EGTA, and 0.04 g/ml sucrose was added to each dish for 10 min at 4°C. Dishes were washed three times with 1 ml of 1× PBS, and then 2 ml of 0.1 M glycine was added to each dish for 5 min to quench fixation at 4°C. Dishes were then washed three times with 1 ml of 1× PBS, and 2 ml of cold 0.2% Triton X-100 was added to each dish for 5 min at 4°C to permeabilize the cells. Dishes were then washed three times with 1 ml of 1× PBS. Each wash was incubated for 5 min at 4°C. Dishes were then incubated with 10% BSA for blocking for 30 min at 4°C and then washed three times with 1 ml of 1× PBS. Each wash was incubated for 5 min at room temperature. Then, 0.22 µM syringe filtered primary antibody in 3% BSA was added. Anti-MT3 (Thermo Fisher Scientific, 12179-1-AP) was diluted 1:200, then added to the dish, and incubated for 1 h at room temperature. After 1 h, dishes were washed three times with 1 ml of 1× PBS. Each wash was incubated for 5 min at room temperature. Then, 0.22 µM syringe filtered secondary antibody in 3% BSA was added to each dish. Goat anti-rabbit IgG cross-absorbed Alexa Fluor 488 secondary antibody (Thermo Fisher Scientific, A-11008) was diluted to 1:500, added to the dish, and incubated for 45 min at room temperature. The dish was then washed three times with 1 ml of 1× PBS. Each wash was incubated for 5 min at room temperature. Finally, 4 µl of 300 µM DAPI in 1 ml of 1× PBS was added and incubated for 10 min at room temperature before imaging.
shRNAmiR-mediated knockdown of MT3
shRNAmiR-mediated knockdown of MT3 was performed using shRNAmiR techniques from a published protocol (Ritter et al., 2017). miRNAs were designed based off sequences generated by Horizon siDESIGN Center (https://horizondiscovery.com/en/ordering-and-calculation-tools/sidesign-center) or using Thermo Fisher Scientific BLOCK-iT RNAi Designer (https://rnaidesigner.thermofisher.com/rnaiexpress/). Five different nonoverlapping knockdown sequences were tested. The 64-nucleotide shRNAmiR oligos were cloned into the pcDNA6.2/GW-mRFP-miR (kind gift from Dr. Peter S. McPherson). The shRNAmiR sequences used are as follows: Control 5′-TGC TGA CGT GAC ACG TTC GGA GAA TTG TTT TGG CCA CTG ACT GAC AAT TCT CCA CGT GTC ACG T-3′; shRNA #1 5′-TGC TGT GGC ACA CTT CTC ACA TCC TGG TTT TGG CCA CTG ACT GAC CAG GAT GTG AAG TGT GCC A-3; shRNA #2 5′-TGC TGT ATT CAC ATA GGC TGT GTG GGG TTT TGG CCA CTG ACT GAC CCC ACA CAC TAT GTG AAT A-3′; shRNA #3 5′-TGC TGT GCT GTG CAT GGG ATT TAT TCG TTT TGG CCA CTG ACT GAC GAA TAA ATC ATG CAC AGC A-3′; shRNA #4 5′-TGC TGT GCC CTG GTA CAA AGA CTC GTG TTT TGG CCA CTG ACT GAC ACG AGT CTG TAC CAG GGC A-3′; shRNA #5 5′-TGC TGT GGA TGG GCA GGT AGG GAC AGG TTT TGG CCA CTG ACT GAC CTG TCC CTC TGC CCA TCC A-3′.
RNA extraction
RNA extraction was performed using RNAqueous-Micro Total RNA Isolation Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. At the indicated DIV, neuron culture media was removed, and 2 ml of DPBS was added to wash neurons. Then, 700 µl of lysis/binding solution was added to the well, and the lysis buffer was pipetted up and down several times. Next 700 µl of 64% ethanol was added to the lysis buffer and mixed gently by carefully pipetting. Then, 700 µl of the lysate/ethanol mixture was added to a filter cartridge and centrifuged at 10,000 rpm for 1 min. The waste collected in the collection tube was discarded. This step was repeated to transfer all the lysate/ethanol mixture to the filter cartridge. After all the lysate/ethanol mixture had been passed through the filter, 700 µl of Wash Solution #1 was added to the filter cartridge and centrifuged at 10,000 rpm for 1 min. Waste was discarded, and 500 µl of Wash Solution #2/3 was added and centrifuged at same speed for 1 min. This step was repeated. The empty cartridge was centrifuged at 10,000 rpm for 1 min. Finally, 50 µl of preheated elution solution was added to the filter cartridge and collected into a new collection tube. RNA yield was measured by nanodrop.
cDNA synthesis
cDNA synthesis was performed using iScript Select cDNA Synthesis Kit according to manufacturer's instructions (Bio-Rad). First, 4 µl of 5× iScript select reaction mix, 2 µl of Oligo(dT)20 primer, 13 µl of RNA sample, and 1 µl of iScript reverse transcriptase were combined in a PCR tube. If 13 µl of RNA exceeded a total RNA concentration of 1 µg, then 1 µg of RNA was used with water to account for the remaining volume. This reaction was incubated for 90 min at 42°C and then heat-inactivated at 85°C for 5 min. cDNA yield was measured by NanoDrop.
RT-qPCR
RT-qPCR was performed using the QuantStudio 3 instrument. First, 2 µl of 5 ng/µl cDNA, 10 µl of 2× PowerTrack SYBR Green Master Mix for qPCR (Thermo Fisher Scientific), 1 µl of forward and reverse primers (8 µM MT3, 16 µM GAPDH), and 17 µl of nuclease-free water were mixed in a clear-bottom 96-well plate. Results were analyzed using the ΔΔCT method (Livak and Schmittgen, 2001). For experiments measuring MT3 mRNA levels at different DIVs, ΔΔCT levels were calculated by first calculating the ΔCT for MT3-GAPDH at each DIV, then ΔΔCT was calculated by subtracting the ΔCT value of DIV 5 from all other DIVs and then normalizing that value to DIV 5. Values were normalized to the average ΔΔCT value for DIV 5 in each independent experiment when multiple replicates of DIV 5 were conducted in the same RT-qPCR experiment. The primer sequences used for RT-qPCR are as follows: Rat MT3 qPCR Fwd, 5′-TGG TTC CTG CAC CTG CTC GG-3′; Rat MT3 qPCR Rev, 5′-TCA CTG GCA GCA GCT GCA TTT C-3′; Rat GAPDH Fwd, 5′-GAC ATG CCG CCT GGA GAA A-3′; Rat GAPDH Rev, 5′-AGC CCA GGA TGC CCT TTA GT-3′.
Imaging data analysis
Imaging data was collected using MicroManager software and analyzed using Fiji (ImageJ). Raw data output from Fiji was analyzed using Excel, KaleidaGraph, and JMP Pro.
For single wavelength sensors (Fluo-4 AM, FluoZin-3 AM, pHrodo Red AM, and pHrodo Green AM), changes in fluorescent intensity (ΔF = F − F0) were normalized to the baseline at 0 s (F0) indicated as ΔF/F0. For ratiometric sensors (Fura Red AM), the indicator fluorescence ratio signal, R, was calculated to indicate calcium levels (equal to the emission intensity measured with 435 nm excitation divided by that measured with 488 nm excitation). The changes in excitation ratio (ΔR = R − R0) were normalized to the baseline at 0 s (R0) indicated as ΔR / R0.
In immunostaining experiments used to confirm MT3 expression, the fluorescence for the A488 secondary antibody was reported (a.u). In immunostaining experiments used to quantify MT3 knockdown efficiency, the fluorescence of the A488 secondary antibody recorded was normalized to the average value of the control shRNA for the respective DIV.
The Sholl analysis plugin on Fiji (Ferreira et al., 2014) was used to measure the number of intersections marked by EGFP-MAP2 up to 40 µm away from the soma. An ROI was selected in the center of the soma for each cell. The data was adjusted to account for the size of the soma, so only values marking dendrites up to 30 µm away from the soma were reported.
FluoZin-3 two-photon imaging data was analyzed using ImageJ (NIH). The fluorescence changes (ΔF / F0) were calculated using the formula (F − F0) / F0, in which F0 represents the baseline fluorescence signal, measured 10 min after whole-cell break-in. After measuring baseline fluorescence (F0) from the cell body of CA1 pyramidal neurons, fluorescence changes were measured (F) immediately and 10 min after action potential induction by injecting current (see above, Electrophysiology). For controls, fluorescence changes (F) were measured only at 10 min.
Experimental design and statistical analyses
All neuron experiments were performed at specific DIVs, noted in the manuscript. All neuron experiments include replicates from multiple neuron preparations.
Statistical analysis was performed using Student's t test, ANOVA, or Kruskal–Wallis tests followed by the indicated post hoc analysis. The specific statistical tests used are indicated in the figure caption, and exact p values are reported in the figure or figure caption. Error bars indicate standard error of the mean (SEM). Significance levels are defined as p < 0.05. Not statistically significant data is noted as n.s. Group data are presented as mean ± SEM unless otherwise noted. All statistical tests were two-tailed.
Results
Induced Ca2+ influx evokes endogenous Zn2+ spikes in neurons
We previously found that synchronous Ca2+ spikes in primary neuron cultures are followed by Zn2+ spikes, suggesting that neurons can produce Ca2+-dependent Zn2+ spikes (Devinney et al., 2005; Zhang et al., 2021). To further explore the Ca2+ dependence of these Zn2+ spikes, we artificially induced Ca2+ influx to generate the synchronous Ca2+ spikes by three different methods. First, we depolarized neurons with 50 mM KCl, activating endogenous voltage-gated calcium channels (VGCCs) and resulting in Ca2+ influx. The Zn2+ spikes were monitored using FluoZin-3 AM because it is insensitive to Ca2+ and pH from 5.2 to 9 (Zhao et al., 2008), while Ca2+ spikes were monitored using Fura Red AM, allowing simultaneous recording of Ca2+ and Zn2+ dynamics. Depolarization with high KCl resulted in significant cellular Zn2+ elevation following Ca2+ influx (Fig. 1A,B), and such Zn2+ spikes cannot be detected in Ca2+-free buffer (Fig. 1B), confirming that Zn2+ spikes depend on Ca2+ influx. Next, we induced Ca2+ influx through activation of endogenous NMDA receptors by 100 μM glutamate and 10 μM glycine. This also resulted in Ca2+ spikes followed by increased cellular Zn2+ signals (Fig. 1C,D). When NMDA receptors were stimulated in calcium-free buffer, there was no change in Ca2+ and Zn2+ levels (Fig. 1D). We confirmed Zn2+ spikes can be generated by Ca2+ influx through NDMA receptors using NMDA receptor inhibitor, 2-amino-5-phosphonopentanoate (APV), which significantly decreased both Ca2+ and Zn2+ spikes under glutamate stimulation (Fig. 1E). Lastly, we activated neurons by optogenetic activation (Nagel et al., 2002). Hippocampal neurons were transfected with light-sensitive ion channel ChrimsonR-tdTomato (ChrR-tdT). Because the channelrhodopsin is tagged with a red fluorescent protein, Ca2+ and Zn2+ dynamics were measured in separate experiments using green, fluorescent dyes. Neurons stained with either Ca2+ dye, Fluo-4 AM, or Zn2+ dye, FluoZin-3 AM. Ca2+ and Zn2+ spikes were detected after 4 s light stimulation was applied to cells (Fig. 1F,G). Again, no Ca2+ and Zn2+ spikes could be detected via optogenetic activation in Ca2+-free buffer (Fig. 1G). These results using diverse stimulation methods demonstrated that large Ca2+ influx result in cytosolic Zn2+ elevations, or Zn2+ spikes, inside primary hippocampal neurons.
Induced Ca2+ influx is followed by increased Zn2+ concentrations in neurons. A, Representative traces and images of Ca2+ (Fura Red ΔR/R0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in neurons depolarized with 50 mM KCl in calcium-containing buffer. Scale bar, 10 µm. B, Quantification of amplitude of Ca2+ (Fura Red ΔR / R0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) spikes after treatment with 50 mM KCl in calcium-containing buffer or calcium-free buffer (Ca2+, n = 20 cells; 0 Ca2+, n = 24 cells; t test; ****p ≤ 0.0001). C, Representative traces and images of Ca2+ (Fura Red ΔR / R0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in neurons activated with 100 µM glutamate and 10 µM glycine in calcium-containing buffer. Scale bar, 10 µm. D, Quantification of amplitude of Ca2+ (Fura Red ΔR / R0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) sikes after treatment with 100 µM glutamate and 10 µM glycine in calcium-containing buffer or calcium-free buffer (Ca2+, n = 19 cells; 0 Ca2+, n = 38 cells; t test; ****p ≤ 0.0001). E, Quantification of amplitude of Ca2+ (Fura Red ΔR / R0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) spikes after 5 min pretreatment treatment with 0 µM (Control) or 50 µM APV (APV), followed by activation with 100 µM glutamate and 10 µM glycine in calcium-containing buffer (Control, n = 16 cells; APV, n = 17 cells; t test; ****p ≤ 0.0001). F, Representative traces of Ca2+ (Fluo-4 AM ΔF / F0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in neurons overexpressing pCAG-ChrimsonR-tdT and activated using 4 s light stimulation in calcium-containing buffer. Ca2+ and Zn2+ traces were taken from different experiments. G, Quantification of amplitude of Ca2+ (Fluo-4 AM ΔF / F0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) spikes in neurons overexpressing pCAG-ChrimsonR-tdT after activation with 4 s light stimulation in calcium-containing buffer or calcium-free buffer (Ca2+ dynamics with Ca2+, n = 5 cells; Ca2+ dynamics 0 Ca2+, n = 5 cells; Zn2+ dynamics with Ca2+, n = 7 cells; Zn2+ dynamics 0 Ca2+, n = 5 cells; t test; ****p ≤ 0.0001). H, Representative trace and quantification of peak Ca2+ (Fura Red ΔR / R0 max) dynamics after 5 min pretreatment with 100 µM BAPTA-AM, followed by depolarization with 50 mM KCl in buffer containing varying concentrations of extracellular Ca2+ (0, 1, or 2 mM; 0 mM Ca2+, n = 20 cells; 1 mM Ca2+, n = 23 cells; 2 mM Ca2+, n = 19 cells; ANOVA Tukey post hoc; 0 vs 1 mM *p = 0.0253, 0 vs 2 mM ****p ≤ 0.0001, 1 vs 2 mM ****p ≤ 0.0001). I, Quantification of amplitude of Zn2+ (FluoZin-3 ΔF / F0 max) spikes after 5 min pretreatment with 100 µM BAPTA-AM, followed by depolarization with 50 mM KCl in buffer containing varying concentrations of extracellular Ca2+ (0, 1, or 2 mM; 0 mM Ca2+, n = 20 cells; 1 mM Ca2+, n = 23 cells; 2 mM Ca2+, n = 19 cells; ANOVA Tukey post hoc; 0 vs 1 mM *p = 0.0456, 0 vs 2 mM ****p ≤ 0.0001, 1 vs 2 mM *p = 0.0332).
To further evaluate the effects of Ca2+ influx on Zn2+ spikes, we used the membrane permeable Ca2+ chelator, BAPTA-AM (Kd = 110 nM) which can reduce intracellular Ca2+ concentrations. After treatment with 100 μM BAPTA-AM, neurons were depolarized with high KCl in buffers containing varying amounts of extracellular Ca2+ (0, 1, and 2 mM). BAPTA-AM treatment led to a reduction in Ca2+ influx and a corresponding decrease in Zn2+ spike amplitudes (Fig. 1H,I). In addition, we observed a stepwise decrease in both Ca2+ and Zn2+ spikes as extracellular Ca2+ was reduced from 2 to 1 to 0 mM. These results indicate a correlation between the intracellular Ca2+ levels and Zn2+ spike amplitude, supporting the conclusion that Ca2+ influx is necessary for the generation of Zn2+ spikes.
Neuron stimulation evokes endogenous Zn2+ spikes in hippocampal slice cultures
Next, we investigated whether Ca2+-dependent Zn2+ spikes also occur under more physiologically relevant conditions in which neuronal connectivity is preserved. To this end, we used organotypic mouse hippocampal slice cultures and focused on CA1 pyramidal neurons. FluoZin-3 dye was injected into soma, and somatic Zn2+ dynamics were monitored using two-photon imaging following the induction of action potential spikes (Fig. 2A). Action potentials resulted in a significant increase in Zn2+ transients, an effect that was not observed under control conditions lacking neuronal stimulation in slices (Fig. 2B,C). These results demonstrate that Ca2+-dependent Zn2+ spikes are not limited to dissociated cultures but also occur in an ex vivo system that better retain native neuronal architecture and synaptic networks, indicating the physiological relevance of these Zn2+ spikes.
Zn2+ spikes are present in hippocampal brain slice culture. A, Schematic of two-photon imaging combined with whole-cell electrophysiology. B, Representative images of Zn2+ dynamics (FluoZin-3) in CA1 pyramidal neurons with (left) and without (right) action potential spikes induced by current injections. C, Quantification of Zn2+ transients (FluoZin-3, ΔF / F0) following action potential induction or under control conditions (AP stim, n = 10 cells; No stim, n = 6 cells; t test; error bars represent SEM. n.s., not significant; **p values: AP stim, 0.0004, 2 min; 0.0015, 10 min; No stim, 0.2, 10 min; AP stim vs No stim, 0.009).
Ca2+-dependent Zn2+ spikes are specific to brain cells
As Ca2+ oscillations are ubiquitous signals that involved diverse cell types, we wondered whether we could observe Zn2+ spikes in other cell types in addition to hippocampal neurons. To test this, we induced synchronized Ca2+ spikes by triggering Ca2+ influx in four different cell types: astrocytes, HeLa cells, COS-7 cells, and primary fibroblasts. Astrocytes are glial cells mainly found in the CNS and are critical to healthy brain function, playing many roles including synapse development and function (Nedergaard et al., 2003; Halassa et al., 2007; Sofroniew and Vinters, 2010). We first tested if astrocytes could elicit Ca2+-dependent Zn2+ spikes like those observed in hippocampal neurons. To control the Ca2+ influx into the astrocytes, we transfected the astrocytes with TRPA1-mCherry, a plasma membrane channel permeable to cations. We stained transfected astrocytes with either Fluo-4 AM or FluoZin-3 AM and then activated the TRPA1 channel using Allyl isothiocyanate (AITC), where we did observe Ca2+-dependent Zn2+ spikes (Fig. 3A). Next, we induced Ca2+ influx using the same strategy in HeLa cells. When HeLa cells overexpressing TRPA1 were stimulated by AITC, we only detected cytosolic Ca2+ increase, but no Zn2+ elevation (Fig. 3B). Similar results were found in COS7 cells that Ca2+ elevations were not followed by cytosolic Zn2+ spikes (Fig. 3C). Lastly, we looked at Ca2+-dependent Zn2+ spikes in primary fibroblasts, which express endogenous TRPA1 channels. We stained GM03440F WT patient fibroblasts with Fluo-4 AM or FluoZin-3 AM and activated the TRPA1 channels with AITC. We were again only able to detect Ca2+ spikes, but no changes in cytosolic Zn2+ in fibroblasts (Fig. 3D). Overall, these data demonstrate that only astrocytes exhibited the Ca2+-dependent Zn2+ spikes, which were absent in HeLa cells, COS-7 cells, and primary fibroblasts. This suggests that the Ca2+-induced Zn2+ spikes might be specifically associated with brain cells.
Ca2+-dependent Zn2+ spikes are brain specific. A, Representative traces and images of Ca2+ (Fluo-4 AM ΔF / F0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in astrocytes overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer. Ca2+ and Zn2+ traces were taken from different experiments. Scale bar, 10 µm. B, Quantification of amplitude of Ca2+ (Fluo-4 AM ΔF / F0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) spikes in astrocytes overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer or calcium-free buffer (Ca2+ dynamics with Ca2+, n = 5 cells; Ca2+ dynamics 0 Ca2+, n = 5 cells; Zn2+ dynamics with Ca2+, n = 4 cells; Zn2+ dynamics 0 Ca2+, n = 5 cells; t test; *p = 0.0325, ***p = 0.0008). C, Representative traces and images of Ca2+ (Fluo-4 AM ΔF / F0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in HeLa cells overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer. Ca2+ and Zn2+ traces were taken from different experiments. Scale bar, 10 µm. D, Quantification of amplitude of Ca2+ (Fluo-4 AM ΔF / F0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) sikes in HeLa cells overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer or calcium-free buffer (Ca2+ dynamics with Ca2+, n = 9 cells; Ca2+ dynamics 0 Ca2+, n = 7 cells; Zn2+ dynamics with Ca2+, n = 6 cells; Zn2+ dynamics 0 Ca2+, n = 4 cells; t test; ***p = 0.0003). E, Representative traces and images of Ca2+ (Fluo-4 AM ΔF / F0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in COS7 overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer. Ca2+ and Zn2+ traces were taken from different experiments. Scale bar, 10 µm. F, Quantification of amplitude of Ca2+ (Fluo-4 AM ΔF / F0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) spikes in COS7 cells overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer or calcium-free buffer (Ca2+ dynamics with Ca2+, n = 6 cells; Ca2+ dynamics 0 Ca2+, n = 3 cells; Zn2+ dynamics with Ca2+, n = 8 cells; Zn2+ dynamics 0 Ca2+, n = 6 cells; t test; **p = 0.0018). G, Representative traces and images of Ca2+ (Fluo-4 AM ΔF / F0) and Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in fibroblasts, activated with 100 µM AITC in calcium-containing buffer. Ca2+ and Zn2+ traces were taken from different experiments. Scale bar, 10 µm. H, Quantification of amplitude of Ca2+ (Fluo-4 AM ΔF / F0 max) and Zn2+ (FluoZin-3 AM ΔF / F0 max) spikes in fibroblasts, activated with 100 µM AITC in calcium-containing buffer or calcium-free buffer (Ca2+ dynamics with Ca2+, n = 7 cells; Ca2+ dynamics 0 Ca2+, n = 3 cells; Zn2+ dynamics with Ca2+, n = 8 cells; Zn2+ dynamics 0 Ca2+, n = 5 cells; t test; **p = 0.0026).
Ca2+ influx causes cytoplasmic acidification in hippocampal neurons, astrocytes, HeLa cells, COS7 cells, and fibroblasts
Next, we sought to determine mechanistically how Ca2+ influx can lead to intracellular Zn2+ release in brain cells. Because these Zn2+ spikes are detected in Zn2+-free buffer, they are not caused by influx from extracellular buffer but rather by Zn2+ released from an intracellular source. Literature shows that Ca2+ influx can cause cellular acidification (OuYang et al., 1995; Wu et al., 1999; Hwang et al., 2011), and it has been established that cellular acidification liberates Zn2+ from intracellular stores (Kiedrowski, 2014). We sought to determine if Ca2+ influx causes cellular acidification in the five cell types tested here. First, we stained primary hippocampal neurons with Fluo-4 AM and pHrodo red AM to simultaneously monitor Ca2+ and pH dynamics. For the pHrodo AM sensors, increased fluorescence indicates a decrease in pH. We then induced Ca2+ influx by depolarizing neurons with high KCl as previously mentioned. The Ca2+ spike observed after depolarization was accompanied by a pH decrease, signifying cellular acidification (Fig. 4A). Similarly, Ca2+ influx induced by glutamate and glycine treatment also induced cellular acidification (Fig. 4B). Next, we examined if the similar Ca2+ influx-induced acidification occurs in astrocytes. Again, astrocytes were transfected with TRPA1-mCherry, then stained with pHrodo green AM, and activated with AITC. This resulted in a small but detectable pH reduction in astrocytes (Fig. 4C). The same experiment was conducted in HeLa cells and COS7 cells transfected with TRPA1-mCherry, both of which also displayed a pH drop upon AITC-induced Ca2+ influx (Fig. 4D,E). Finally, we examined GM03440F fibroblasts that express endogenous TRPA1 channels. Cells were stained with pHrodo red AM to monitor pH dynamics, as well as Fluo-4 to monitor Ca2+ dynamics. We observed a small but measurable pH decrease in fibroblasts following AITC-activated Ca2+ influx (Fig. 4F). Overall, these data show that Ca2+ influx is consistently associated with cellular acidification across multiple cell types, including hippocampal neurons, astrocytes, HeLa cells, COS7 cells, and fibroblasts.
Ca2+ influx caused acidification of the cytoplasm in hippocampal neurons, astrocytes, HeLa cells, COS7 cells, and fibroblasts. A, Representative traces of Ca2+ (Fluo-4 AM ΔF / F0) and pH (pHrodo Red ΔF / F0) dynamics in hippocampal neurons depolarized with 50 mM KCl in calcium-containing buffer and quantification of pH (pHrodo Red ΔF / F0) dynamics in calcium-containing buffer or calcium-free buffer (Ca2+, n = 13 cells; 0 Ca2+, n = 9 cells; t test; **p = 0.0050). B, Representative traces of Ca2+ (Fluo-4 AM ΔF / F0) and pH (pHrodo Red AM ΔF/F0) dynamics in hippocampal neurons activated with 100 µM glutamate and 10 µM glycine in calcium-containing buffer and quantification of pH (pHrodo Red ΔF / F0) dynamics in calcium-containing buffer or calcium-free buffer (Ca2+, n = 6 cells; 0 Ca2+, n = 14 cells; t test; **p = 0.0077). C, Representative traces of pH (pHrodo Green AM ΔF / F0) dynamics in astrocytes overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer and quantification of pH (pHrodo Green ΔF / F0) dynamics in calcium-containing buffer or calcium-free buffer (Ca2+, n = 8 cells; 0 Ca2+, n = 5 cells; t test; **p = 0.0089). D, Representative traces of pH (pHrodo Red AM ΔF/F0) dynamics in HeLa cells overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer and quantification of pH (pHrodo Green ΔF/F0) dynamics in calcium-containing buffer or calcium-free buffer (Ca2+, n = 6 cells; 0 Ca2+, n = 7 cells; t test; **p = 0.0071). E, Representative traces of pH (pHrodo Red AM ΔF/F0) dynamics in COS7 cells overexpressing TRPA1-mCherry, activated with 100 µM AITC in calcium-containing buffer and quantification of pH (pHrodo Green ΔF/F0) dynamics in calcium-containing buffer or calcium-free buffer (Ca2+, n = 10 cells; 0 Ca2+, n = 5 cells; t test; **p = 0.0033). F, Representative traces of Ca2+ (Fluo-4 AM ΔF / F0) and pH (pHrodo Red AM ΔF / F0) dynamics in fibroblasts, activated with 100 µM AITC in calcium-containing buffer and quantification of pH (pHrodo Red ΔF / F0) dynamics in calcium-containing buffer or calcium-free buffer (Ca2+, n = 4 cells; 0 Ca2+, n = 6 cells; t test; *p = 0.0323).
Cellular acidification causes Zn2+ elevation in neurons
We showed that while all tested cell types exhibited Ca2+-induced cellular acidifications, Ca2+-dependent Zn2+ spikes were observed exclusively in neurons and astrocytes. Previous work showed that a pH drop to 6.0 was sufficient to increase intracellular Zn2+ concentrations in hippocampal neurons (Kiedrowski, 2012). As such, we investigated and compared the impact of cellular acidification on Zn2+ release in HeLa cells and neurons. To quantify the pH-dependent changes in intracellular Zn2+ concentrations, we adjusted cellular pH from 5.2 to 8.9 using the ionophores nigericin and monensin. Intracellular Zn2+ and pH levels were simultaneously recorded using FluoZin-3 AM and pHrodo Red AM dyes, respectively. To confirm the specificity of Zn2+ signals, TPA was used to chelate Zn2+ without affecting cellular pH. Using this in situ pH calibration assay, we observed distinct differences between neurons and HeLa cells. In neurons, the intracellular Zn2+ concentrations were unchanged between pH 7.0 and 8.9 but displayed a significant increase when the pH drooped below 6.5 (Fig. 5A). In contrast, HeLa cells only displayed a slight increase in intracellular Zn2+ concentrations when pH was reduced to below 6.0 (Fig. 5B). This response in HeLa cells was minimal and not comparable with the robust Zn2+ increase observed in neurons. By comparing the pH-Zn2+ response curves between neurons and HeLa cells (Fig. 5C), we found that Zn2+ concentrations in HeLa cells remained relatively stable across a large range of cellular pH levels, while intracellular Zn2+ levels sharply increased in response to pH drops in neurons. These findings suggest that cellular acidification induced more pronounced Zn2+ signals in neurons compared with HeLa cells, indicating neurons have a unique sensitivity to intracellular pH changes. This differential sensitivity in neurons is likely due to the presence of pH-sensitive intracellular stores that can readily release Zn2+, a feature not found in other cell types.
Acidification induced a more pronounced Zn2+ increase in neurons than in HeLa cells. A, Recording of Zn2+ (FluoZin-3 AM F) and pH (pHrodo Red AM ΔF / F0) dynamics in neurons calibrated at different pHs. The trace represents the average ± SEM for 23 cells. B, Recording of Zn2+ (FluoZin-3 AM F) and pH (pHrodo Red AM ΔF / F0) dynamics in HeLa cells calibrated at different pHs. The trace represents the average ± SEM for 32 cells. C, Plot of Zn2+ (FluoZin-3 AM F) concentrations in neurons (blue) and HeLa cells (pink) against different cellular pHs.
Rat hippocampal neurons express MT3
To discover a potential intracellular source of Zn2+ spikes, we investigated the metallothionein (MT) family of Zn2+ binding proteins, which are known for their multiple Zn2+ binding sites and Zn2+ buffering capacity (Krężel and Maret, 2006). Among the MT isoforms, we focused on the brain-enriched isoform, MT3. MT3 was reported to be predominantly expressed in the CNS (Shuttleworth and Weiss, 2011) and has been detected in the rat brain (Hozumi et al., 2008); however, reports differ as to whether MT3 is expressed in neurons (Velázquez et al., 1999; Yanagitani et al., 1999), astrocytes (Uchida et al., 1991; Kobayashi et al., 1993), or both (Yamada et al., 1996). Due to conflicting reports, we first examined whether the primary rat hippocampal neurons express MT3. We measured endogenous MT3 mRNA expression levels by RT-qPCR (Fig. 6A). The expression level of MT3 mRNA significantly varied from DIV 5 to DIV 21 but was consistently present in neurons through this period, nonetheless.
Primary hippocampal neurons express MT3. A, RT-qPCR results of relative mRNA levels (ΔΔCT values) of MT3 in primary hippocampal neurons at indicated DIVs (n = 7–8 replicates per DIV; ANOVA Tukey post hoc; groups with the same letter are not significantly different from each other. DIV 5 vs DIV 14 p = 0.0292, DIV 5 vs DIV 21 p = 0.0446, DIV 8 vs DIV 11 p = 0.0087, DIV 8 vs DIV 14 p = 0.0259, DIV 8 vs DIV 21 p = 0.0139). B, Representative images of HeLa cells transfected with tdTomato alone (control) or tdTomato and kozak-rat-MT3 plasmids (MT3 OE), then immunostained with anti-MT3 primary antibody and Alexa-488 secondary antibody, followed by DAPI stain. Scale bar, 10 µm. C, Representative image of a hippocampal neuron on DIV 5 transfected with Tau-mCherry and then immunostained with anti-MT3 primary antibody and Alexa-488 secondary antibody, followed by DAPI stain. Scale bar, 10 µm. D, Quantification of MT3 fluorescence in HeLa cells transfected with tdTomato (control) or tdTomato and kozak-rat-MT3 plasmids (MT3 OE) and immunostained with anti-MT3 primary antibody and Alexa-488 secondary antibody, followed by DAPI stain (Control, n = 54 cells; MT3 OE, n = 54 cells; t test; ****p ≤ 0.0001). MT3 fluorescence is reported as the fluorescence intensity from the A-488 secondary antibody (a.u.). E, Quantification of MT3 fluorescence in hippocampal neurons after immunostaining with anti-MT3 primary antibody and Alexa-488 secondary antibody, followed by DAPI stain (DIV 5, n = 90 cells; DIV 14, n = 105 cells; DIV 21, n = 67 cells; ANOVA Tukey post hoc; DIV 5 vs DIV 14 ****p ≤ 0.0001, DIV 5 vs DIV 21 ****p ≤ 0.0001, DIV 14 vs DIV 21 ****p ≤ 0.0001). MT3 fluorescence is reported as the fluorescence intensity from the A-488 secondary antibody (a.u.).
To confirm protein expression, we used immunofluorescence staining with the MT3 antibody. To evaluate the specificity of the MT3 antibody and test whether it reacts to other MT isoforms, we transfected HeLa cells with tdTomato plasmid alone (control) or cotransfected HeLa cells with tdTomato and kozak-rat-MT3 plasmids to induce MT3 overexpression, along with DAPI for nuclear labeling (Fig. 6B). Since HeLa cells do not endogenously express MT3, almost no fluorescence was observed in control cells (Fig. 6B,D). In contrast, cells overexpressing MT3 exhibited strong antibody labeling (Fig. 6B,D). These results demonstrate that MT3 antibody has low reactivity with other MT isoforms and is specific for MT3. We then examined the MT3 expression in neurons. Neurons were transfected with Tau-mCherry to indicate neuron morphology and identity, followed by staining with an MT3 primary antibody and Alexa-488-conjugated secondary antibody, along with DAPI for nuclear labeling (Fig. 6C). MT3 immunofluorescence staining results in neurons indicated that MT3 protein was expressed in the hippocampal neuron cultures from DIV 5 to DIV 21 (Fig. 6E). Overall, these results demonstrate that primary rat hippocampal neurons express MT3 at both the mRNA and protein levels.
Reduction in MT3 expression decreases Ca2+-dependent Zn2+ spikes in neurons
Next, we reduced MT3 expression using shRNAmiRs in primary hippocampal neurons (Ritter et al., 2017). We designed five nonoverlapping knockdown sequences targeting MT3 and created corresponding shRNAmiR plasmids. Knockdown efficiency of each construct was assessed by monitoring MT3 expression levels with immunofluorescence staining. Among the five constructs, shRNA #3 and shRNA #4 demonstrated the most significant reduction in MT3 immunofluorescence compared with the control construct (Fig. 7A). We next transfected neurons with the shRNA constructs and stained them with FluoZin-3 AM to monitor intracellular Zn2+ dynamics. We stimulated endogenous NMDA receptors with glutamate and glycine to induced Ca2+-dependent Zn2+ spikes. The amplitude (ΔF / F0 max) of Zn2+ spikes were significantly lower in neurons transfected with shRNA constructs compared with control neurons transfected with the nontargeting construct (Fig. 7B,C). The shRNA #3 and shRNA #4 constructs demonstrated the greatest reduction in Zn2+ spike amplitude. These data show that MT3 knockdown decreases Zn2+ spike amplitude, which suggests that MT3 acts as a source for Zn2+ liberation in neurons following Ca2+ influx and cellular acidification.
Knockdown of MT3 by shRNA in primary hippocampal neurons decreases Ca2+-induced Zn2+ spikes. A, Relative MT3 fluorescence in hippocampal neurons DIV 8–12 after transfection with shRNA constructs. Data values are normalized to average nontargeting control values for each set of experiments (n = 38–40 cells for each shRNA construct; ANOVA Dunnett's post hoc; Control vs #1 n.s., Control vs #2 ***p = 0.0004, Control vs #3 ****p < 0.0001, Control vs #4 ****p < 0.0001, Control vs #5 ****p < 0.0001). B, Trace of average Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in hippocampal neurons DIV 7–15 transfected with shRNA constructs, then activated with 100 µM glutamate and 10 µM glycine in calcium-containing buffer. The traces represent the average ± SEM for 23 cells from control, 18 cells from shRNA #3, and 16 cells from shRNA #4 transfected cells. C, Fluorescence maximum values of FluoZin-3 AM ΔF / F0 in hippocampal neurons DIV 7–15 transfected with shRNA constructs after 100 µM glutamate and 10 µM glycine treatment (n = 12–23 cells for each shRNA construct; ANOVA Dunnett's post hoc; Control vs #1 *p = 0.0170, Control vs #2 *p = 0.0197, Control vs #3 ****p < 0.0001, Control vs #4 ****p < 0.0001, Control vs #5 **p = 0.0088).
Overexpression of MT3 induces Ca2+-dependent Zn2+ spikes in HeLa and COS7 cells
HeLa and COS7 cells did not produce Ca2+-dependent Zn2+ spikes (Fig. 3). We hypothesized that this was due to the absence of endogenous MT3 expression in these cell types. We then examined if MT3 overexpression could regenerate Ca2+-dependent Zn2+ spikes in HeLa and COS7 cells. We cotransfected HeLa cells with TRPA1-mCherry and kozak-rat-MT3 plasmids, stained them with FluoZin-3 AM to monitor Zn2+ dynamics, and then activated TRPA1 channels using AITC. We were able to detect a Ca2+-dependent Zn2+ increase in the cytosol of HeLa cells overexpressing MT3 (Fig. 8A). The amplitude of the Zn2+ spike was significantly higher in HeLa cells overexpressing MT3 than control cells without MT3 expression (Fig. 8B). Similarly, we cotransfected COS7 cells with TRPA1-mCherry and kozak-rat-MT3 plasmids, stained them with FluoZin-3 AM, and activated TRPA1 channels using AITC. COS7 cells overexpressing MT3 also exhibited a Ca2+-dependent Zn2+ increase in the cytosol (Fig. 8C). The amplitude of the Zn2+ spike was significantly higher in COS7 cells overexpressing MT3 than control cells (Fig. 8D). These results show that overexpressing MT3 can produce Ca2+-dependent Zn2+ spikes in nonbrain cells, further supporting the role of MT3 as a source for Zn2+ liberation in neurons.
Overexpression of MT3 can induce Ca2+-dependent Zn2+ spikes. A, Average trace of Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in HeLa cells overexpressing TRPA1-mCherry only (Control) or TRPA1-mCherry and rat-MT3 plasmids (MT3 OE), activated with 100 µM AITC in calcium-containing buffer. Trace represents the average ± SEM for 21 cells from 12 independent experiments for control and 21 cells from 12 independent experiments for MT3 OE. B, Fluorescence maximum values of FluoZin-3 AM ΔF / F0 for HeLa overexpressing TRPA1-mCherry only (Control) or TRPA1-mCherry and rat-MT3 (MT3 OE), activated with 100 µM AITC (Control, n = 21 cells; MT3 OE, n = 21 cells; t test; **p = 0.0067). C, Average trace of Zn2+ (FluoZin-3 AM ΔF / F0) dynamics in COS7 cells overexpressing TRPA1-mCherry only (Control) or TRPA1-mCherry and rat-MT3 plasmids (MT3 OE), activated with 100 µM AITC in calcium-containing buffer. Trace represents the average ± SEM for 15 cells from 10 independent experiments for control and 13 cells from 10 independent experiments for MT3 OE. D, Fluorescence maximum values of FluoZin-3 AM ΔF / F0 for COS7 cells overexpressing TRPA1-mCherry only (Control) or TRPA1-mCherry and rat-MT3 (MT3 OE), activated with 100 µM AITC (Control, n = 15 cells; MT3 OE, n = 13 cells; t test; **p = 0.0097).
MT3-mediated Zn2+ release inhibits dendritic arborization in primary hippocampal neurons
After confirming that MT3 is responsible for the generation of Ca2+-dependent Zn2+ spikes in neurons, we next sought to determine the function of MT3-mediated Zn2+ spikes in neuronal development. We noticed that neurons transfected with shRNA #3 or shRNA #4 constructs appear to exhibit increased dendritic branching compared with neurons transfected with the control construct. This observation suggested that MT3 may be an important regulator of the dendritic arbor of developing neurons. To test this, we cotransfected hippocampal neurons with either control, shRNA #3, or shRNA #4 constructs along with EGFP-MAP2, followed by Hoechst staining directly before imaging (Fig. 9A). MAP2 expression allowed us to visualize and quantify dendritic branching in these neurons (Dehmelt and Halpain, 2005). Using Sholl analysis, we quantified dendritic branching by counting the number of intersections at different distances from the soma. We found that neurons transfected with shRNA #3 or shRNA #4 had significantly more intersections at 10, 20, and 30 µm radii away from the soma compared with control neurons (Fig. 9B), supporting the conclusion endogenous MT3 suppresses dendritic complexity.
Knockdown of MT3 and TPA treatment can increase dendritic complexity in primary hippocampal neurons. A, Representative image of primary hippocampal neurons transfected with either nontargeting control, shRNA #3, or shRNA #4 constructs and EGFP-MAP2. Scale bar, 10 µm. B, The number of intersections determined from the Sholl analysis of EGFP-MAP2 from the soma up at 10, 20, or 30 µm radius away from the soma in primary hippocampal neurons transfected with either nontargeting control, shRNA #3, or shRNA #4 constructs (Control, n = 167 cells; shRNA #3, n = 198; shRNA #4, n = 88 cells; Wilcoxon/Kruskal–Wallis was done at each distance; Control vs #3 at 10 µm ****p < 0.0001, Control vs #4 at 10 µm ***p = 0.0006, #3 vs #4 at 10 µm n.s., Control vs #3 at 20 µm ****p< 0.0001, Control vs #4 at 20 µm ****p < 0.0001, #3 vs #4 at 20 µm *p = 0.0199, Control vs #3 at 30 µm ****p < 0.0001, Control vs #4 at 30 µm ****p < 0.0001, #3 vs #4 at 30 µm **p = 0.0048). C, Average traces of Zn2+ (GZnP3 ΔF / F0) and pH (pHrodo red AM ΔF / F0) dynamics in hippocampal neurons DIV 7–13, treated with indicated TPA concentrations. Trace represents the average ± SEM for 7 cells from 7 independent experiments. D, Comparison of peak Ca2+ spikes after pretreatment with the indicated TPA concentration followed by 100 µM glutamate and 10 µM glycine activation in calcium-containing buffer (0 µM TPA, n = 31 cells; 0.5 µM TPA, n = 21 cells; 1 µM TPA, n = 21 cells; ANOVA Tukey post hoc; 0 vs 0.5 µM n.s., 0 vs 1 µM n.s., 0.5 vs 1 µM n.s.). E, Comparison of peak Zn2+ spikes pretreated with the indicated TPA concentration followed by 100 µM glutamate and 10 µM glycine activation in calcium-containing buffer (0 µM TPA, n = 31 cells; 0.5 µM TPA, n = 21 cells; 1 µM TPA, n = 21 cells; ANOVA Tukey post hoc; 0 vs 0.5 µM ****p < 0.0001, 0 vs 1 µM ****p < 0.0001, 0.5 vs 1 µM n.s.). F, Representative images of primary hippocampal neurons transfected with EGFP-MAP2 treated with treated overnight (16–24 h) with 0.5 µM TPA or no treatment (control). Scale bar, 10 µm. G, The number of intersections determined from Sholl analysis of EGFP-MAP2 at 10, 20, or 30 µm radius from the soma in primary hippocampal neurons treated overnight (16–24 h) with 0.5 µM TPA or no treatment (control; Control, n = 93 cells; TPA, n = 118 cells; Kruskal–Wallis was done at each distance; Control vs TPA at 10 µm n.s., Control vs TPA at 20 µm ***p = 0.0004, Control vs TPA at 30 µm ****p < 0.0001). H, The number of intersections determined from Sholl analysis of EGFP-MAP2 at 10, 20, or 30 µm radius from the soma in primary hippocampal neurons transfected with either nontargeting control, shRNA #3, or shRNA #4 constructs with or without overnight (16–24 h) treatment of 0.5 µM TPA (Control, n = 25 cells; Control/TPA, n = 35 cells; shRNA #3/TPA, n = 60 cells; shRNA #4/TPA, n = 51 cells; Wilcoxon/Kruskal–Wallis was done at each distance; Control vs #3/TPA at 10 µm ***p = 0.0007, Control vs #4/TPA at 10 µm **p = 0.0093, Control/TPA vs #3/TPA at 10 µm *p = 0.0402, Control/TPA vs #4/TPA at 10 µm n.s., Control vs #3/TPA at 20 µm **p = 0.0028, Control vs #4/TPA at 20 µm *p = 0.0319, Control/TPA vs #3/TPA at 20 µm n.s., Control/TPA vs #4/TPA at 20 µm n.s., Control vs Control/TPA at 30 µm *p = 0.0344, Control vs #3/TPA at 30 µm **p = 0.0032, Control vs #4/TPA at 30 µm **p = 0.0058, Control/TPA vs #3/TPA at 30 µm n.s., Control/TPA vs #4/TPA at 30 µm n.s).
We next sought to determine if the effect of MT3 on dendritic arborization is mediated by Zn2+. To this end, we first investigated the impact of Zn2+-specific chelator TPA (0.5–100 μM) on steady-state Zn2+ and pH concentrations. Baseline cytosolic Zn2+ concentrations were measured using the Zn2+ sensor GZnP3 (Kd = 1.3 nM; Minckley et al., 2019), and pH levels were measured with pHrodo Red AM. We identified 0.5–1 µM TPA as an optimal concentration that selectively reduced Zn2+ spikes without affecting steady-state Zn2+ or pH levels (Fig. 9C). To further investigate the effect of Zn2+ chelation, neurons stained with Fura Red AM and FluoZin-3 AM were treated with 0.5 or 1 µM TPA and then activated using 100 µM glutamate and 10 µM glycine. TPA treatment showed no significant effect on the amplitude of Ca2+ spikes (Fig. 9D), but it significantly decreased the amplitude of Zn2+ spikes at both 0.5 and 1 µM concentration (Fig. 9E). Based on these results, we chose 0.5 µM for subsequent experiments. Hippocampal neurons were transfected with EGFP-MAP2 and subjected to an overnight treatment (16–22 h) with 0.5 µM TPA. The cells were then stained with Hoechst followed by microscopy imaging (Fig. 9F). Sholl analysis revealed that TPA-treated hippocampal neurons showed an increased number of intersections at 20 and 30 µm from the soma compared with untreated control neurons (Fig. 9G). TPA treatment had no effects on dendritic arborization at 10 µm because it was only applied overnight. In addition, neurons treated simultaneously with MT3 knockdown and TPA did not show a further increase in dendrite branching compared with treatment with TPA alone (Fig. 9H), indicating that both MT3 knockdown and Zn2+ chelation converge on the same signaling pathway that involves Zn2+ release from MT3. These results suggest that MT3-mediated Zn2+ signaling might play an inhibitory role in dendritic development.
Discussion
Zinc plays a critical role in maintaining normal brain development and function, as its deficiency is linked to a range of neurological conditions. Studies in rats demonstrated that zinc deficiency leads to impairments in learning and memory (Halas et al., 1986). In humans, zinc deficiency has been associated with neurodevelopmental and psychiatric disorders, including autism spectrum disorder (Alsufiani et al., 2022), depression (Petrilli et al., 2017), and schizophrenia (Joe et al., 2018). The importance of zinc in the brain is attributed to its roles in maintaining protein structural stability, supporting enzymatic activity, and facilitating signaling functions (Zhang et al., 2022).
Zinc signaling via its free, labile ionic form is very important to neuron function, serving roles within and around the synapse. Extracellular Zn2+ at the synapse can inhibit NMDA (Westbrook and Mayer, 1987; Vogt et al., 2000; Amico-Ruvio et al., 2011; Krall et al., 2020), AMPA (Kalappa et al., 2015; Carrillo et al., 2020), and GABA receptors (Westbrook and Mayer, 1987; Gingrich and Burkat, 1998; Hosie et al., 2003), thereby modulating excitatory and inhibitory synaptic activity. While much is known about the extracellular roles of Zn2+ in modulating neuronal communication, the function of Zn2+ as an intracellular signaling molecule is less defined. This work reports Ca2+-dependent Zn2+ spikes in both dissociated neuronal cultures and organotypic hippocampal slices and identifies a novel pathway where intracellular Zn2+ is released from MT3 in response to Ca2+ influx, producing Zn2+ spikes, which act as secondary messengers to regulate dendritic arborization.
Cross talk between Ca2+ and Zn2+ has been reported in diverse biological processes. In this work, we demonstrated a new cross talk mechanism in neurons and astrocytes where they can establish a Ca2+-proton-Zn2+ axis. Specifically, Ca2+ influx is associated with cellular acidification, which liberates intracellular Zn2+ release. Previous studies showed that Ca2+ influx in rat neurons induced by glutamate, KCl, and ionomycin resulted in a pH decrease (OuYang et al., 1995). Consistent with this, we observed cellular acidification in our rat hippocampal neurons after KCl and glutamate treatments. Similar cellular acidification was observed in astrocytes, HeLa cells, COS7 cells, and fibroblasts following Ca2+ influx. The mechanism underlying this phenomenon is suggested to involve the Ca2+/H+ exchange mechanism, which extrudes intracellular Ca2+ in exchange for an extracellular H+ ion. In snail neurons, low cytosolic Ca2+ concentrations are restored after Ca2+ influx by Ca2+/H+ exchange driven by ATPase activity (Schwiening et al., 1993). Similarly, Ca2+ induced cellular acidification was caused by activation of the plasma membrane Ca2+ ATPase (PMCA) in rat cerebellar granule cells (Wu et al., 1999), rat trigeminal ganglion cells (Hwang et al., 2011), and rat pancreatic acini cells (González et al., 1997). Based on these findings, the PMCA may be responsible for cellular acidification observed in our study; however, this determination would require further investigation.
Our findings suggest that the acidification triggered by Ca2+ spikes could act as a bridge to link Ca2+ influx to intracellular Zn2+ release. Zn2+ release has been shown to occur in response to a pH decrease in the cytosol. For example, glutamate and glycine activation was shown to increase both cytosolic Ca2+ and Zn2+ levels, associated with cellular acidification (Kiedrowski, 2011, 2012). A pH drop alone was enough to increase cytosolic Zn2+ levels in hippocampal neurons at pH 6. It was hypothesized that the pH-mediated Zn2+ increase may result from Zn2+ release from intracellular ligands like cysteine residues at pH 6.6–6.1 (Kiedrowski, 2014). We observe Zn2+ release in hippocampal neurons at acidic pH lower than 6.5, which aligns with previous observations.
MTs are Zn2+ binding proteins that play a large role in the regulation of cytosolic Zn2+ concentrations. In general, MTs regulate metal ion homeostasis, heavy metal detoxification, and oxidative stress (Babula et al., 2012). MT1 and MT2 are ubiquitously expressed, but more MT3 mRNA can be found in areas of the CNS compared with MT1 and MT2 isoforms (Scudiero et al., 2017). Different from MT1 and MT2, MT3 does not respond to changes in metal ion concentration (Palmiter et al., 1992; Faller, 2010). For example, MT2 mRNA levels were shown to increase after Zn2+ treatment, while MT3 mRNA levels remained unchanged (Bousleiman et al., 2017). This suggests roles for MT1/2, but not MT3, in heavy metal detoxification.
MT3 was originally identified and named as growth inhibitory factor (GIF) after its discovery in Alzheimer's disease (AD) pathology. Loss of inhibitory factors in AD brain extracts resulted in increased neurotrophic activity and abnormal neurite sprouting (Uchida and Tomonaga, 1989). This inhibitory factor was discovered to be MT3, which was shown to be significantly under expressed in AD brains (Uchida et al., 1991), although other studies suggest MT3 expression can be increased (Carrasco et al., 1999) or unchanged (Carrasco et al., 2006) in AD. MT3 has been further demonstrated to prevent neurite formation and growth when applied exogenously in rat cortical neurons (Chung et al., 2002; Uchida et al., 2002). MT3 KO mice exhibit abnormal psychological disorders but retain normal memory and locomotor activity (Koumura et al., 2009a). Our results revealed that, in addition to exogenously applied MT3, endogenous MT3 inhibits dendrite arborization via Zn2+ signals released from MT3 in rat hippocampal neurons.
In addition to its inhibitory function in neurite growth, MT3 can also uniquely regulate lysosome function in astrocytes (Lee et al., 2010) and protect against Aβ toxicity (Irie and Keung, 2001). Despite functional differences, MT1, MT2, and MT3 have been shown to release Zn2+ in response to redox signals (Jiang et al., 1998; Khatai et al., 2004; Maret, 2008). In addition, MT3 has been shown to release Zn2+ during ischemia (Shuttleworth and Weiss, 2011), a condition associated with a pH drop in solid tissues (Lemasters, 1999). We hypothesize that, in our hippocampal neuron cultures, MT3 can respond to pH changes caused by Ca2+ influx and release Zn2+. A previous report demonstrated that Zn2+ release from MTs occurs only at pH 5 or lower (Jiang et al., 2000). However, these experiments used the MT2 isoform. Our results showed that acidification leads to a greater increase in intracellular Zn2+ release in neurons compared with HeLa cells (Fig. 4), suggesting that brain-enriched MT3 may be more pH sensitive than other MT isoforms. MT3 has been shown to have unique characteristics from MT1 and MT2 (Vašák and Meloni, 2017) and has additional glutamate residues compared with MT1 and MT2 making it more acidic overall (Uchida et al., 1991). This may influence MT3's increased sensitivity to pH decreases in the cytosol. Compared with other isoforms, MT3 is considered more unstable and contains an additional threonine residue (position 5) and a CPCP (position 6–9) motif that are correlated to its unique inhibitory effect on neurite growth (Ding et al., 2010). The CPCP motif was also shown to influence the binding affinity of MT3, decreasing MT3's affinity for Zn2+ (Palumaa et al., 2005). A weaker binding affinity could facilitate a better Zn2+ release in response to cellular signals. It was also shown that MT3 can be more reactive and release more Zn2+ in response to NO compared with MT1 and MT2 (Chen et al., 2002). These structural differences between MT3 and other MT isoforms may contribute to MT3's unique functions, including the possibility of a more pH-sensitive domain that enables increased Zn2+ release compared with other isoforms. We demonstrated that manipulation of MT3 did alter the Zn2+ spike dynamics, alluding to the involvement of MT3 in Zn2+ spikes and another unique function of MT3.
Dendrite growth can be regulated by several factors including synaptic activity and microtubule associated proteins (McAllister, 2000). A possible explanation for the inhibitory function of MT3-mediated Zn2+ signals may be via its interactions with cytoskeletal proteins. Zn2+ has been shown to bind to microtubules in vitro (Hesketh, 1982). In addition, x-ray fluorescence imaging demonstrated the colocalization of Zn2+ and microtubules in dendrites of hippocampal neurons (Domart et al., 2020). Furthermore, we had previously showed that Zn2+ can displace MAP2C, doublecortin, and tau from microtubules in situ (Minckley et al., 2023). It is possible that Zn2+ released from MT3 can interact directly with the microtubules to inhibit dendrite growth. MT3 was also shown to interact with β-actin (Lahti et al., 2005), another cytoskeleton component. In astrocytes, MT3 was shown to be associated with F-actin (Lee et al., 2011). Zn2+ was implicated in actin remodeling and cell migration (Li et al., 2016). In addition, Zn2+ was found to contribute to synaptic strength through its influence on postsynaptic scaffolding proteins SHANK2 and SHANK3 (Ha et al., 2018; Vyas et al., 2020, 2021). All this evidence suggests that Zn2+ released from MT3 might inhibit dendritic branching and synaptic function via targeting on cytoskeleton proteins.
In this study, we identified a new signaling pathway involving the Ca2+-H+-MT3-Zn2+ axis and uncovered its critical roles in in neurite development. Our findings provide valuable insights into the physiological functions of endogenous MT3 and Zn2+ spikes during neuronal development, offering a foundation for future research on exploring intracellular Zn2+ signaling and the consequences that arise from Zn2+ dysregulation.
Footnotes
We thank the following sources for general financial support: NIH/NINDS Grant R01NS110590 (to Y.Q.) and NIH R01MH124778, R21MH137409, and R21NS133681 (to W.C.O.).
The authors declare no competing financial interests.
- Correspondence should be addressed to Yan Qin at Yan.Qin{at}du.edu.
