Zn²⁺ entry through the mitochondrial calcium uniporter is a critical contributor to mitochondrial dysfunction and neurodegeneration

Abstract

Excitotoxic Ca²⁺ accumulation contributes to ischemic neurodegeneration, and Ca²⁺ can enter the mitochondria through the mitochondrial calcium uniporter (MCU) to promote mitochondrial dysfunction. Yet, Ca²⁺-targeted therapies have met limited success. A growing body of evidence has highlighted the underappreciated importance of Zn²⁺, which also accumulates in neurons after ischemia and can induce mitochondrial dysfunction and cell death. While studies have indicated that Zn²⁺ can also enter the mitochondria through the MCU, the specificity of the pore’s role in Zn²⁺-triggered injury is still debated. Present studies use recently available MCU knockout mice to examine how the deletion of this channel impacts deleterious effects of cytosolic Zn²⁺ loading. In cultured cortical neurons from MCU knockout mice, we find significantly reduced mitochondrial Zn²⁺ accumulation. Correspondingly, these neurons were protected from both acute and delayed Zn²⁺-triggered mitochondrial dysfunction, including mitochondrial reactive oxygen species generation, depolarization, swelling and inhibition of respiration. Furthermore, when toxic extramitochondrial effects of Ca²⁺ entry were moderated, both cultured neurons (exposed to Zn²⁺) and CA1 neurons of hippocampal slices (subjected to prolonged oxygen glucose deprivation to model ischemia) from MCU knockout mice displayed decreased neurodegeneration. Finally, to examine the therapeutic applicability of these findings, we added an MCU blocker after toxic Zn²⁺ exposure in wildtype neurons (to induce postinsult MCU blockade). This significantly attenuated the delayed evolution of both mitochondrial dysfunction and neurotoxicity. These data—combining both genetic and pharmacologic tools—support the hypothesis that Zn²⁺ entry through the MCU is a critical contributor to ischemic neurodegeneration that could be targeted for neuroprotection.

Introduction

Ischemic stroke is a leading cause of death and disabilities worldwide, with its incidence projected to increase significantly with the aging population (Benjamin et al., 2018). Yet, despite its toll, treatment in humans is limited to attempts at prompt restoration of blood flow, and there are no pharmacologic interventions with demonstrated neuroprotective efficacy. This limitation reflects in part the inability to predict who will suffer from an ischemic event, necessitating a primary focus on treatments delivered after an ischemic event has occurred, and highlights the importance of developing a better understanding of events after an ischemic episode that contribute to delayed neurodegeneration.

A large body of evidence has implicated “excitotoxicity”, caused by excessive release of the excitatory neurotransmitter glutamate, as an important contributor to neuronal injury, and most studies of excitotoxicity have focused on deleterious effects of rapid Ca²⁺ entry through glutamate activated channels (particularly the highly Ca²⁺ permeable N-methyl-D-aspartate [NMDA] receptors) (Choi et al., 1988; Rothman and Olney, 1986). Studies of such Ca²⁺-dependent toxicity indicate the occurrence of multiple injury promoting effects via actions on mitochondria as well as on cytosolic targets (Brennan et al., 2009; Nicholls and Budd, 2000; Siesjo, 1988). Mitochondrial effects of Ca²⁺ appear to result after entry through the mitochondrial Ca²⁺ uniporter (MCU). Yet, Ca²⁺ targeting therapies, primarily by blocking Ca²⁺ entry into the cell, have shown little efficacy (Hoyte et al., 2004; Ikonomidou and Turski, 2002). Furthermore, studies using MCU blockers (to block mitochondrial Ca²⁺ uptake) have not only shown mixed results, but also worsened injury in some studies, perhaps in part by preventing mitochondrial buffering of Ca²⁺ loads and resulting in more rapid cytosolic Ca²⁺ loading (Velasco and Tapia, 2000).

Another ion that has been implicated in ischemic neurodegeneration is Zn²⁺. Zn²⁺ accumulates in neurons after ischemia, primarily from two sources: 1) Zn²⁺ stored in presynaptic vesicles that can be released with activation (termed “synaptic Zn²⁺”) (Assaf and Chung, 1984; Howell et al., 1984), which can then enter postsynaptic neurons (through voltage-sensitive Ca²⁺ channels [VSCC], and with greater rapidity through a subset of AMPA-kainate receptors that are Ca²⁺-permeable [Ca-AMPAR]) (Jia et al., 2002; Kerchner et al., 2000; Weiss et al., 1993; Yin and Weiss, 1995), and 2) Zn²⁺ already present in postsynaptic neurons which is released from intracellular buffers (primarily metallothionein-III [MT-III]) in response to metabolic perturbations associated with ischemia (oxidative stress and acidosis) (Jiang et al., 2000; Maret, 1995). While studies have shown that Zn²⁺ stemming from either one of these sources alone is sufficient to induce injury (Aizenman et al., 2000; Sensi et al., 1999; Weiss et al., 1993), it is apparent that both can synergistically contribute to neuronal injury (Ji and Weiss, 2018), and in fact, the predominant source of Zn²⁺ accumulation contributing to epileptic or ischemic damage appears to differ between CA1 and CA3 hippocampal pyramidal neurons (Lee et al., 2003; Medvedeva et al., 2017). Prior studies have found Zn²⁺ chelation to be neuroprotective in ischemia (Calderone et al., 2004; Koh et al., 1996), and recent studies of hippocampal slices subjected to oxygen glucose deprivation (OGD) as a model of ischemia find that cytosolic Zn²⁺ rises occur early, preceding and contributing to the onset of lethal sharp Ca²⁺ rises (termed “Ca²⁺ deregulation”) (Medvedeva et al., 2009). Thus, Zn²⁺ may be an attractive target, as its rise occurs before injury becomes irreversible. Interestingly, a critical target of acute deleterious Zn²⁺ effects also appears to be the mitochondria—which Zn²⁺ also enters through the MCU—to potently trigger mitochondrial dysfunction (Clausen et al., 2013; Gazaryan et al., 2007; Ji et al., 2019; Jiang et al., 2001; Malaiyandi et al., 2005; Saris and Niva, 1994). However, past studies elucidating the role of the MCU in Zn²⁺ effects were limited to the use of pharmacologic MCU blockers which lack complete specificity (Tapia and Velasco, 1997). Consequently, one cannot be certain that beneficial effects are specifically due to MCU blockade.

To this end, we took advantage of the recent availability of MCU knockout (KO) mice (Pan et al., 2013) and combined this with use of pharmacologic MCU blockers to 1) assess the role of MCU specifically in Zn²⁺-induced mitochondrial dysfunction and neuronal injury, and 2) examine the ability of delayed MCU blockade (after Zn²⁺ exposures) to attenuate the downstream, deleterious effects of Zn²⁺ on mitochondria. Our findings in both cultured cortical neurons and hippocampal slices provide further support to the hypothesis that Zn²⁺ entry into the mitochondria through the MCU plays a critical role in promoting early and prolonged mitochondrial dysfunction—as well as the consequent neurodegeneration—and that delayed MCU blockade can specifically attenuate these contributions to injury.

Materials and methods

Ethics statement

All procedures for the current study were approved by the Institutional Animal Care and Use Committee of the University of California Irvine, and in accordance with the recommendations from the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Animals

Mice lacking Mitochondrial Calcium Uniporter (Ccdc109a [Mcu], IST11669F8; Texas A&M Institute for Genomic Medicine) were bred from homozygous breeding pairs. Genotype was validated with protocol as detailed previously (Pan et al., 2013), using the following primers: HP3’ (CCATCTGTTCCTGACCTTGA), IST11669F8-F (GGAGTTAAGTCATGAGCTGC), and IST11669F8-R (CTGGCTTAGTTGGCAGAGTT). CD-1 mice (used as control and wildtype) were ordered from Charles River (Crl:CD1[ICR]).

Reagents and indicators

Materials for culture maintenance (Minimum Essential Media [MEM], fetal bovine serum, glutamine, and horse serum) and for imaging (Newport Green, FluoZin-3 AM, MitoTracker Green, Pluronic F-127, Fura-FF) were purchased from Life Technologies (Grand Island, NY). Other reagents for experiments (2,2’-dithiodipyridine [DTDP], Carbonyl cyanide-p-trifluoromethoxyphenylhydrazone [FCCP], Rhodamine 123 [Rhod123], Ruthenium Red [RR], and N-methyl-D-aspartate [NMDA]) were purchased from Sigma-Aldrich (St. Louis, MO). Hydroethidine (HEt) was purchased from Assay Biotech (Sunnyvale, CA), MK-801 from Abcam, Apocynin from Acros Organics (Morris Plains, NJ), and XF Base Medium (minimal Dulbecco’s Modified Eagle’s Medium) from Agilent Technologies (Santa Clara, CA). All other chemicals and reagents were purchased from common commercial sources.

Media

For culture growth and/or maintenance, the following media were used: “Maintenance media” (MEM supplemented with 25 mM glucose), “growth media” (maintenance media supplemented with 10% heatinactivated horse serum 2 mM glutamine), and “plating media” (growth media supplemented with 10% fetal bovine serum). For imaging experiments and/or Zn²⁺ exposures, HEPES-buffered media (consisting of the following [in mM]: 120 NaCl, 5.4 KCl, 0.8 MgCl₂, 20 Hepes, 15 glucose, 10 NaOH, in pH 7.4; with either 0 mM CaCl₂ [0 mM Ca²⁺ HSS] or 1.8 mM CaCl₂ [1.8 mM Ca²⁺ HSS]) were used, and kept at room temperature unless specified otherwise. To prepare hippocampal slices, isolated brains were maintained in the “preparation solution” (consisting of the following [in mM]: 220 sucrose, 3 KCl, 1.25 NaH₂PO₄, 6 MgSO₄, 26 NaHCO₃, 0.2 CaCl₂, 10 glucose, and 0.42 ketamine, kept in pH 7.35, maintained at 310 mOsm, and equilibrated with 95% O₂/5% CO₂). The hippocampal slices were maintained in artificial cerebral spinal fluid (ACSF; consisting of the following [in mM]: 126 NaCl, 3 KCl, 1.25 NaH₂PO₄, 1 MgSO₄, 26 NaHCO₃, 2 CaCl₂, 10 glucose, kept in pH 7.35, maintained at 310 mOsm with sucrose, and equilibrated with 95% O₂/5% CO₂).

Neuronal culture preparation

Mixed cortical neuronal cultures were prepared as described previously (Yin et al., 1994). Briefly, neocortical regions from CD-1 or MCU KO mouse embryos (15 – 16 gestational day; mixed gender) were isolated and cells suspended in plating media. The neuronal suspension was then plated on previously established astrocytic monolayers (of the same genotype) on glass-bottomed dishes, Seahorse XF24 cell culture microplates, or culture-treated 24 well microplates. 2 – 3 days after plating, culture was treated with 10 μM cytosine arabinoside for 24 hrs (to stop non-neuronal cell division), and media was switched to growth media. Cells were kept in 5% CO₂/37°C incubator for at least 10 days prior to experiment. To prepare astrocytic monolayers, the same protocol was used with the following changes: 1) neocortical tissues from postnatal mice (1 – 3 days old; mixed gender) from CD-1 or MCU KO mice were used, 2) plating media was supplemented with epidermal growth factor (10 ng/ml), and 3) suspended cells were plated on poly-D-lysine and laminin-coated glass-bottomed dishes, culture treated 24 well microplates, and Seahorse XF24 cell culture microplates.

Zn²⁺ exposure

Cultured neurons were placed in either 0 mM Ca²⁺ or 1.8 mM Ca²⁺ HSS prior to starting all experiments. After 10 min baseline and pre-treatment with 60 μM DTDP ± 500 μM Apocynin (if indicated), Zn²⁺ and/or Ca²⁺ loading was induced. To permit ion influx, neurons were exposed to indicated levels of Zn²⁺ (0 – 300 μM) and/or 1.8 mM Ca²⁺ in 90 mM K⁺ HSS (“high K⁺”; HSS modified with 90 mM K⁺ and Na⁺ adjusted to 35 mM to maintain osmolarity) for 5 min to depolarize neurons, inducing Zn²⁺ and/or Ca²⁺ entry through VSCC. NMDA antagonist MK-801 (10 μM)—used to inhibit Ca²⁺ entry through NMDA receptors—was added during all Zn²⁺ and/or Ca²⁺ exposures (even when Ca²⁺ was not present during the experiment) to maintain consistency. After 5 min Zn²⁺ and/or Ca²⁺ exposure in high K⁺, neurons were washed 3 times in HSS (60 μM DTDP ± 500 μM apocynin or 10 μM RR, as indicated). If the experiment involved incubation, cultured neurons were washed 3 times into maintenance media and kept in the 5% CO₂/37°C incubator until needed. When attempting to best model the events that may occur during ischemic conditions, the following standardized paradigm was used when indicated (10 min pre-exposure to 60 μM DTDP, followed by 5 min 100 μM Zn²⁺ + high K⁺ + MK-801 + DTDP and 3x wash into DTDP, all in 1.8 mM Ca²⁺ HSS; termed “ischemic Zn²⁺ exposure”).

Quantitative imaging studies

Neuronal cultures aged 10 – 16 days in vitro (DIV) were placed on the stage of a Nikon Diaphot inverted microscope equipped with 75 Watts Xenon-lamp, a filter wheel controlled by a computer, a 40x (1.3 numerical aperture) epifluorescence oil-immersion objective along with the following fluorescence cubes: a green FITC (Ex: 480 nm, dichroic: 505 nm, Em: 535 nm) and a red TRITC fluorescence cube (Ex: 540 nm, dichroic: 565 nm, Em: 605 nm). Emitted signals were captured every min with a Sensys Photometrics intensified charge-coupled device camera, and converted digitally by the MetaFluor software (Version 7.0, Molecular Devices LLC, Sunnyvale, CA). Gain and exposure of the camera were adjusted to obtain baseline fluorescence level of 200 – 300 arbitrary units of a 12-bit signal output of 4,096 for all fluorophores. For imaging, only fields with at least 20 healthy looking cells (non-clustered neurons with robust processes) were used. For analysis, background fluorescence (the lowest value in a neuron-free region of the field) was subtracted from images, and fluorescence measurement for each neuron (F_x) was normalized to the average fluorescence intensity of the 10 min baseline measure (F₀) to trend normalized fluorescence over time (F_x/F₀). The F_x/F₀ of each cell in the field was averaged to produce a value constituting one experiment repetition.

Cytosolic Zn²⁺

Either low affinity cytosolic Zn²⁺ indicator Newport Green diacetate (K_d ~ 1 μM, Ex: 490 nm, Em: 530 nm; using a green FITC fluorescence cube) or the high affinity cytosolic Zn²⁺ indicator FluoZin-3 AM (K_d ~ 15 nM, Ex: 494 nm, Em: 516 nm; using a green FITC fluorescence cube) was used to quantify neuronal Zn²⁺ levels (Canzoniero et al., 1999; Gee et al., 2002; Sensi et al., 1999). Cultured neurons were loaded with 5 μM Newport Green or FluoZin-3 in 0 mM Ca²⁺ HSS containing 0.2% Pluronic F-127 and 1.5% dimethlyl sulfoxide (in the dark room, at room temperature) prior to use for experiments. To assess mitochondrial Zn²⁺ levels, FCCP (1 μM) was added at the end of each experiment to release mitochondrially sequestered Zn²⁺, with the resulting increase in cytosolic Zn²⁺ level permitting estimate of degree of mitochondrial Zn²⁺ accumulation.

Mitochondrial membrane potential

Mitochondrial membrane potential sensitive indicator Rhod123 (Ex: 507 nm, Em: 529 nm; using a green FITC fluorescence cube) was used to quantify mitochondrial depolarization. Neurons were used for imaging after loading the cultures with 2 μM Rhod123 in 1.8 mM Ca²⁺ HSS for 30 min (in the dark room, at room temperature). Rhod123 is a positively charged dye which accumulates in mitochondria, where its fluorescence is quenched. With mitochondrial depolarization, the Rhod123 sequestered in the mitochondria gets released, resulting in increase in Rhod123 fluorescence intensity (Duchen et al., 2003). FCCP (1 μM; 5 min) was added at the end of each experiment to induce maximal mitochondrial depolarization, permitting measure of maximal Rhod123 ΔF.

Reactive oxygen species

Superoxide sensitive dye HEt (Ex: 510–560 nm; Em: > 590 nm; using a red TRITC fluorescence cube) was used to quantify ROS generation. Neurons were used for imaging after loading the cultures with 5 μM HEt in 0 mM or 1.8 mM Ca²⁺ HSS for 30 min (in the dark room, at room temperature). As HEt is oxidized into highly fluorescent ethidium, ROS generation is quantified as the HEt fluorescence increase.

Mitochondrial respiration assay

Mitochondrial respiration was assessed by measuring changes in O₂ consumption rate (OCR) using the Seahorse XF24 Extracellular Flux Analyzer, as described previously with adjustments (Yao et al., 2009). Briefly, neurons plated on top of astrocytic monolayers on Seahorse XF24 Cell Culture Microplates were exposed to the ischemic Zn²⁺ exposure, washed into DTDP for 20 min, and incubated for 2 hrs (in maintenance media and % CO₂/37°C incubator). After completing the incubation, cultures were washed into XF Base Medium (supplemented with 2 mM glutamine, 15 mM glucose, and 1 mM sodium pyruvate), and maintained at 37°C for 1 hr prior to starting the Seahorse XF24 assay. The machine measures the OCR of cells during baseline and after the sequential exposure to 1 μM oligomycin, 2 μM FCCP, and antimycin A/rotenone (both at 1 μM). The concentrations of mitochondrial inhibitors were determined empirically, with FCCP specifically adjusted to induce approximately 1.5x the baseline. Prior to each experiment, all culture wells were visually inspected (to ensure equal distribution of cells) and randomly assigned to treatment groups. All calibration instructions were followed according to the manufacturer’s protocols, and Seahorse XF24 program and Wave software were used to run the assay and analyze the data respectively. We carried out prior validation studies confirming astrocytes make minimal contributions to observed OCR and its changes, as our cultured neurons are plated on top of astrocyte monolayers (as discussed above).

Confocal imaging of mitochondrial morphology

Mitochondrial morphology was assessed as described previously (Ji and Weiss, 2018). Briefly, confocal microscopy was performed using an inverted stage Nikon Eclipse Ti chassis microscope with a Yokogawa CSUX spinning disk head. Images were observed at 1000x oil-immersion objective (1.49 numerical aperture) using a Hamamatsu electromultiplying CCD camera. Field was scanned sequentially with excitation (488 nm) via a Coherent sapphire laser source synchronized with the camera, and emission monitored with a 525 (50) nm filter. Images were acquired using the MicroManager Image Acquisition software (ver 1.52). After cultured neurons were exposed to the ischemic Zn²⁺ exposure paradigm (discussed above) + 20 min wash + 12 hr incubation (in maintenance media at 37°C/5% CO₂), cultures were loaded with 200 nM MitoTracker Green (Ex: 490 nm, Em: 516 nm) in 1.8 mM Ca²⁺ HSS for 30 min at room temperature in the dark, and then switched into 1.8 mM Ca²⁺ HSS for the imaging. Only healthy appearing neurons with intact outer membranes (as could be visualized via brightfield imaging) were chosen for imaging. Camera gain, laser intensity, and exposure times were adjusted to make MitoTracker Green fluorescence intensity 1.5 – 2 times that of the background fluorescence value. Heat fan was used to maintain imaging rig at 37°C. Images acquired were blinded, and lengths and width of distinct mitochondria measured manually in the ImageJ software. The values were used to calculate length/width (L/W) ratio of individual mitochondria, which were then averaged to produce one average L/W ratio representing one repetition of an experiment.

Neurotoxicity Assessment

Neurotoxicity was examined in cultured neurons plated on 24 well microplates at 14 – 16 DIV. Briefly, neurons were switched into HSS (0 or 1.8 mM Ca²⁺), and exposed to 60 μM DTDP (10 min), Zn²⁺ ± Ca²⁺ (in high K⁺ and 10 μM MK-801, as discussed above; 5 min) and 10 μM RR (20 min), with concentration as detailed above. After the exposures and treatment, cultures were washed in maintenance media and kept in 37°C/5% CO₂ for 24 hrs, at which cell death was quantified with by lactate dehydrogenase (LDH) efflux assay as described (Koh and Choi, 1987). In all experiments, the background LDH present in the media was subtracted by measuring the level in the sham wash protocol, used as negative control, and scaled to levels in positive controls (obtained in cultures exposed to 300 μM NMDA for 24 hrs, a paradigm that reliable destroys >90% of neurons while sparing glia). All wells were visually inspected to estimate degree of cell death, which was compared to quantification from LDH assay, for validation.

Hippocampal slice oxygen-glucose deprivation

Acute hippocampal slices were prepared as described previously (Medvedeva et al., 2009). Briefly, brains from 25 ± 3 days old mice were rapid removed after anesthesia with isoflurane and kept in the ice-cold preparation solution. Hippocampal slices were cut with Leica VT1200 vibratome at 300 μm thickness and placed in ACSF. After slices were equilibrated in ACSF for 1 hr at 34°C, slices were switched into oxygenated ACSF in room temperature for ≥1 hr prior to use.

Slices were mounted on the stage of an upright microscope (BX51WI; Olympus), placed in a flow-through chamber (RC-27L chamber with plastic slice anchor; Warner Instruments), and perfused with the oxygenated ACSF (2 ml/min; 32°C). To measure intracellular Ca²⁺ dynamics, Fura FF (K_d ~ 6 μM) was loaded into individual hippocampal pyramidal neurons using a patch pipette held in the whole cell configuration at – 60 mV for 5 min. After pipette withdrawal, the injected cells were left to recover for 20 min prior to experiment. Fluorescence was alternately excited at 340(20)/380(20) nm via 40× water-immersion objective using a xenon light source (Sutter Instruments) and emitted fluorescence was collected at 532(40) nm using a cooled CCD camera (Hamamatsu). All filters are band pass with bandwidths indicated in parentheses. Images were taken every 15 or 30 s, background subtracted, and analyzed using the MetaFluor software. Changes in Ca²⁺ levels are displayed as the ratio of background subtracted Fura FF emission intensities upon excitation at 340 and 380 nm (“340/380 ratio”).

To model ischemic injury, slices were placed in hypoxic-hypoglycemic condition (via oxygen-glucose deprivation [OGD]). ACSF was changed to the same solution lacking glucose (and replaced with sucrose to maintain osmolarity) and bubbled with 95% N₂/5% CO₂. OGD was continued for ≥ 15 min and maintained until Ca²⁺ deregulation occurred.

Statistical analysis

All values on traces are displayed as mean ± standard error of the mean (SEM), and on bars displayed as mean + SEM. All experiments were interleaved and have been repeated ≥ 3 times. All comparisons reflect matching sets of cell cultures or slices from the same set of dissections or slice preparation, and all efforts were made to match biologic variables between comparisons. Two-tailed student’s t-test was used to assess significance.

Results

Genetic deletion of MCU attenuates mitochondrial Zn²⁺ accumulation in cortical neurons

As prior studies of mitochondrial Zn²⁺ entry through the MCU have only used pharmacologic agents (as discussed above), using the MCU KO mice, we first sought to validate that Zn²⁺ can indeed enter the mitochondria through the MCU, with greater specificity than previously possible. Wildtype (WT) and MCU KO cultured neurons were loaded with the low affinity cytosolic Zn²⁺ indicator Newport Green (K_d ~ 1 μM) in 0 mM Ca²⁺ HSS, and exposed to a sequence of 25, 75, and 300 μM exogenous Zn²⁺ (5 min each). Because VSCC are routes of exogenous Zn²⁺ entry through the plasma membrane that are more uniformly expressed across cortical neurons (in contrast to Ca-AMPAR, which are more selectively expressed in specific subsets of neurons), all Zn²⁺ exposures were performed in the presence of 90 mM K⁺ (“high K⁺”; to depolarize neurons, permitting Zn²⁺ influx through VSCC). MK-801 (10 μM), an NMDA receptor blocker used to prevent Ca²⁺ entry, was also added during Zn²⁺ exposures (even in Ca²⁺-free conditions) in all experiments to eliminate any confounders caused by the channel activation. The dose-response curve consistently displayed greater Newport Green fluorescence rises (ΔF)—indicating higher cytosolic Zn²⁺ levels—in MCU KO neurons (Fig 1 top). After the Zn²⁺ exposures, the mitochondrial protonophore, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (“FCCP”; 1 μM, 5 min) was added. This dissipates the electrochemical gradient necessary for Zn²⁺ to be retained in mitochondria, leading to release of mitochondrial Zn²⁺ to the cytosol, and the consequent Newport Green ΔF reflects the degree of mitochondrial Zn²⁺ accumulation (Clausen et al., 2013; Ji and Weiss, 2018; Medvedeva et al., 2017; Sensi et al., 2003; Sensi et al., 2002). Note that FCCP increased Newport Green ΔF to similar peak levels in both genotypes, despite MCU KO displaying greater cytosolic Zn²⁺ levels during Zn²⁺ exposures, indicating attenuation of mitochondrial Zn²⁺ accumulation in the MCU KO neurons (Fig 1 top).

Figure 1. Mitochondrial Zn²⁺ accumulation is attenuated in mice with genetic deletion of MCU.

WT (blue) and MCU KO (red) cultured cortical neurons were loaded with either low affinity cytosolic Zn²⁺ indicator Newport Green (top and bottom) or high affinity cytosolic Zn²⁺ indicator FluoZin-3 (middle), and exposed to indicated doses of exogenous Zn²⁺ with 90 mM K⁺ (“high K⁺”, along with 10 μM MK-801 to inhibit Ca²⁺-entry through the NMDA receptor). DTDP (60 μM; to disrupt cytosolic Zn²⁺ buffering) and FCCP (1 μM; to release mitochondrially sequestered Zn²⁺) were added as indicated. Studies were carried out in HSS with either 0 mM Ca²⁺ (top and middle; to ensure observations of Zn²⁺ specifically) or 1.8 mM Ca²⁺ (bottom; to better model physiological condition). Traces show time course of indicator ΔF (background subtracted and normalized to baseline, [F_x/F₀]), and represent means ± standard error of the mean (SEM) of at least 6 experiments, consisting of ≥ 120 neurons. Grey bars indicate time points of comparison (** indicates p < 0.01, *** indicates p < 0.001, by two-tailed t-test). Note that MCU KO neurons consistently displayed greater ΔF in response to exogenous Zn²⁺ exposures and/or disrupted buffering, as well as reduced FCCP responses, both of which indicate the critical role that MCU plays in facilitating Zn²⁺ entry into mitochondria.

After validating Zn²⁺ uptake through the MCU with a spectrum of Zn²⁺ exposures alone, we next examined the role of the pore in permitting Zn²⁺ entry under more physiologic conditions, as may occur during homeostatic Zn²⁺ movement, by inducing partial disruption of endogenous Zn²⁺ buffering and far lower extracellular Zn²⁺ entry. To do so, WT and MCU KO neurons were loaded with high affinity cytosolic Zn²⁺ indicator FluoZin-3 (K_d ~ 15 nM) in 0 mM Ca²⁺ HSS, and exposed to 2,2’-dithiodipyridine (DTDP), a thiol oxidant that disrupts Zn²⁺ binding to intracellular buffering proteins, at 60 μM, a concentration which induces partial intracellular Zn²⁺ mobilization (Ji and Weiss, 2018). Neurons were subsequently exposed to 1 μM Zn²⁺ + high K⁺ with continued DTDP, followed by wash, and FCCP exposure as indicated (Fig 1 middle). As above, MCU KO neurons displayed significantly greater cytosolic Zn²⁺ levels in response to both intracellular Zn²⁺ mobilization alone as well as the subsequent low Zn²⁺ exposure. Furthermore, the lack of FluoZin-3 ΔF increase in response to FCCP in MCU KO indicates minimal mitochondrial Zn²⁺ accumulation, in contrast to the robust mitochondrial uptake occurring in WT.

While findings thus far support the notion that MCU facilitates mitochondrial Zn²⁺ entry, as Zn²⁺ exposures were carried out in the absence of Ca²⁺, it is possible that the role MCU plays in Zn²⁺-triggered mitochondrial dysfunction in pathophysiologic conditions (where Ca²⁺ is present) may be different. To address this, WT and MCU KO neurons were loaded with Newport Green and exposed to DTDP, 100 μM Zn²⁺ + high K⁺, wash, and FCCP, all in 1.8 mM Ca²⁺ HSS, as indicated (Fig 1 bottom). This paradigm—combining both disruption of intracellular Zn²⁺ buffering and moderate exogenous Zn²⁺ exposure—was chosen, as it likely best reflects what may occur under pathophysiological conditions like ischemia. Similarly to findings in Ca²⁺-free conditions, MCU KO neurons still displayed both higher cytosolic Zn²⁺ levels during Zn²⁺ exposure and reduced FCCP-induced mitochondrial Zn²⁺ release, further validating the critical role played by the MCU in permitting mitochondrial Zn²⁺ accumulation, whether from extracellular or intracellular sources.

Zn²⁺-triggered mitochondrial ROS production is specifically attenuated in MCU KO neurons

We next examined whether Zn²⁺-triggered ROS production, long considered to be a critical contributor to downstream injury cascades (Ji et al., 2019), is also attenuated in MCU KO neurons. Indeed, as we have previously found that acute Zn²⁺-induced ROS generation was almost entirely of mitochondrial origin (unlike Ca²⁺-induced ROS, which had important contributions from both mitochondria and the cytosolic enzyme NADPH oxidase [NOX]) (Clausen et al., 2013), we expected minimal Zn²⁺-triggered ROS generation in MCU KO, corresponding to the attenuated mitochondrial Zn²⁺ accumulation (as shown in Fig 1). Thus, WT and MCU KO neurons were loaded with Hydroethidine (HEt; a superoxide sensitive indicator whose fluorescence increase, ΔF, indicates total ROS production) (Bindokas et al., 1996; Clausen et al., 2013; Ji and Weiss, 2018), and exposed to DTDP (60 μM, 10 min), 100 μM Zn²⁺ + high K⁺ (5 min), and wash, all in 0 mM Ca²⁺ HSS, as indicated (Fig 2, Top). In contrast to our expectation, there was no difference in HEt ΔF.

Figure 2. Zn²⁺-triggered mitochondrial—but not cytosolic—ROS production is attenuated in MCU KO neurons.

WT (blue) and MCU KO (red) cultured neurons loaded with superoxide preferring ROS indicator Hydroethidine (HEt) in either 0 mM Ca²⁺ (top and middle) or 1.8 mM Ca²⁺ (bottom) HSS were exposed to 100 μM Zn²⁺ + high K⁺, with DTDP (60 μM) and apocynin (500 μM) added as indicated. Traces display mean ± SEM F_x/F₀ values for HEt ΔF and represent ≥ 8 experiments consisting of ≥ 160 neurons. Grey bars indicate time points of comparison (* indicates p < 0.05, *** indicates p < 0.001, by two-tailed t-test). Despite the similar total ROS generation (top), note that Zn²⁺-triggered ROS rise specifically from mitochondria (assessed via addition of NOX inhibitor apocynin to attenuate cytosolic ROS production) was significantly attenuated in MCU KO neurons (middle and bottom).

This appeared to contradict our prior finding that acute Zn²⁺-triggered ROS is primarily of mitochondrial origin (Clausen et al., 2013). However, in light of previous studies which found that prolonged low level Zn²⁺ exposures could induce NOX activity (Noh and Koh, 2000), we hypothesized that the ROS generation in MCU KO could result from NOX upregulation secondary to the impaired mitochondrial Zn²⁺ buffering in the MCU KO, resulting in persistently higher cytosolic Zn²⁺ levels. To test this idea, the experiment was repeated in the presence of the NOX inhibitor, apocynin (500 μM), which almost completely eliminated Zn²⁺-triggered ROS production in MCU KO neurons while having little effect in WT neurons (Fig 2 Middle). Repeating the experiment in physiological (1.8 mM) Ca²⁺ HSS again revealed near elimination of Zn²⁺-triggered ROS rise in MCU KO neurons with NOX inhibition (Fig 2 Bottom), further supporting the hypothesis that Zn²⁺-triggered mitochondrial ROS generation is critically dependent upon entry through the MCU.

Critical role of the MCU in acute and delayed Zn²⁺-triggered mitochondrial dysfunction

We next examined the MCU dependence of other measures of Zn²⁺-triggered mitochondrial dysfunction. As our goal was to better understand the pore’s role in pathophysiologic conditions, we decided to use an exposure paradigm that models events likely to occur in pathological events like ischemia, incorporating both disruption of buffering and uptake of synaptic Zn²⁺ (henceforth termed the “ischemic Zn²⁺ exposure”; that parallels the paradigms used in Fig 1 and 2 bottom panels, and comprises 10 min pre-exposure to 60 μM DTDP, followed by 5 min 100 μM Zn²⁺ + high K⁺ + DTDP and 3x wash into DTDP, all in 1.8 mM Ca²⁺ HSS; see Materials and methods).

We first examined the role of MCU in Zn²⁺-triggered loss of mitochondrial membrane potential (ΔΨ_m) using Rhodamine 123 (Rhod123; 2 μM), a cationic indicator for which a cytosolic fluorescence increase (ΔF) indicates loss of ΔΨ_m (see Materials and Methods). WT and MCU KO neurons loaded with Rhod123 were treated to the ischemic Zn²⁺ exposure, washed, and exposed to FCCP as indicated (Fig 3A). The lack of Rhod123 ΔF in MCU KO in response to the Zn²⁺ exposure (compared to the acute ΔF seen after addition of Zn²⁺ in WT neurons), indicates that the Zn²⁺ entry through the MCU is necessary to induce ΔΨ_m loss (Fig 3A). FCCP, which causes maximal mitochondrial depolarization, increased Rhod123 ΔF to similar levels in both WT and MCU KO, indicating that the observed difference was not due to differences in the basal ΔΨ_m between the two genotypes.

Figure 3. Genetic deletion of MCU confers prolonged protection from Zn²⁺-triggered mitochondrial dysfunction.

WT (blue) and MCU KO (red) cultured neurons were exposed to the ischemic Zn²⁺ exposure (DTDP [60 μM; 10 min pre-exposure, followed by 5 min during Zn²⁺ exposure and 20 min during wash], 100 μM Zn²⁺ + high K⁺ [5 min], followed by wash [20 min]). Grey bars indicate time points of comparison (** indicates p < 0.01, *** indicates p < 0.001, by two-tailed t-test).

A. Zn²⁺-induced mitochondrial depolarization is attenuated in MCU KO neurons. WT and MCU KO neurons loaded with ΔΨ_m-sensitive indicator Rhodamine 123 (Rhod123) were treated to the ischemic Zn²⁺ exposure as indicated, followed by FCCP (1 μM; 5 min). Traces represent mean ± SEM F_x/F₀ values for Rhod123 ΔF and represent 8 experiments consisting of ≥ 160 neurons. Note the absence of Rhod123 ΔF in response to Zn²⁺ in MCU KO neurons, indicating protection from Zn²⁺-induced mitochondrial depolarization.

B. Mitochondrial respiration is preserved after Zn²⁺ exposure in MCU KO neurons. WT and MCU KO neurons were treated to either sham (dashed lines) or the ischemic Zn²⁺ exposure (solid lines) as described above. After 2 hr incubation, cultures were placed in the Seahorse device, which measures O₂ consumption rate (OCR) during baseline and after sequential application of oligomycin (Oligo; 1 μM), FCCP (2 μM), and antimycin A/rotenone (AA/Rot; both 1 μM). Traces show time course of OCR and represent mean ± SEM of 3 separate experiments, each consisting of 3 – 4 wells of cultured neurons, with arrows indicating time point at which mitochondrial inhibitors were added. Note that mitochondrial respiration was preserved in MCU KO neurons in response to Zn²⁺, compared to its dramatic inhibition WT.

C. MCU KO neurons are protected from Zn²⁺-triggered disruption of mitochondrial morphology. 12 hrs after WT and MCU KO neurons were treated to ischemic Zn²⁺ exposure, mitochondria were labeled with MitoTracker Green and observed under confocal microscopy (1000x). Length/width ratio (“L/W ratio”) was calculated based on blinded measurement. Representative images (left) and average L/W ratios (right) for WT and MCU KO neurons are displayed. Bar graphs show mean + SEM of L/W ratio, each representing ≥ 5 experiments comprising ≥ 50 mitochondria. Note the significant mitochondrial fragmentation and swelling in WT neurons, compared to the elongated morphology maintained by MCU KO neuronal mitochondria, both of which correspond to their respective L/W ratios.

Next, we examined Zn²⁺ effects on mitochondrial energy production using the Seahorse assay, which measures oxygen consumption rate (OCR) at baseline and in response to sequential application of mitochondrial inhibitors (oligomycin, FCCP, and antimycin A/rotenone), respectively indicating baseline respiration, ATP-linked respiration, maximum respiratory capacity, and extramitochondrial respiration. WT and MCU KO neurons were treated with either sham or the ischemic Zn²⁺ exposure, washed for 20 min, and incubated for 2 hrs prior to running the assay. Note the substantial preservation of mitochondrial respiration in MCU KO, in contrast to its significant inhibition in WT (Fig 3B).

Finally, we examined the long term effect of Zn²⁺ entry through the MCU on mitochondrial morphology. WT and MCU KO neurons were treated to the ischemic Zn²⁺ exposure, washed for 20 min, and after 12 hr incubation, cultures were loaded with the mitochondrial morphology marker, MitoTracker Green (200 nM) and examined under confocal microscopy (1000x; Fig 3C left). The length and the width of individual mitochondria were then measured blindly to determine their length/width (L/W) ratios, which were then collected to calculate the average L/W ratio of the mitochondria in a single neuron (Fig 3C right). Note that WT neurons showed multiple fragmented, swollen mitochondria (and a corresponding low L/W ratio), while MCU KO mitochondria preserved their morphology, maintaining their elongated structures with high L/W ratio (Fig 3C). These findings, showing significant protection from Zn²⁺-effects in MCU KO over a spectrum of time, further validate the crucial role of the MCU in the induction of Zn²⁺ triggered mitochondrial dysfunction.

Zn²⁺-specific contributions to neurotoxicity and ischemic neurodegeneration are attenuated in MCU KO neurons

Next, we sought to determine whether the beneficial effects observed in MCU KO neuronal mitochondria also extend to Zn²⁺-triggered neurodegeneration. To this aim, WT and MCU KO neurons were subjected to the ischemic Zn²⁺ exposure (labeled as Ca²⁺ & Zn²⁺ exposure in Fig 4A), washed for 20 min, and returned to the incubator for 24 hrs prior to assessing cell death. Degree of neurodegeneration was quantified by lactate dehydrogenase (LDH) efflux assay and validated by morphological examination (see Materials and methods). To our surprise, in contrast to our findings on mitochondrial dysfunction, both WT and MCU KO neurons displayed similar degrees of neurodegeneration (Fig 4A). In light of prior reports that cells from MCU KO mice were not protected from Ca²⁺-induced cell death (Nichols et al., 2017; Pan et al., 2013), we next sought to separately assess the Ca²⁺ and Zn²⁺ contributions to neurotoxicity in MCU KO neurons. When neurons were exposed to 1.8 mM Ca²⁺ + high K⁺ alone (Ca²⁺ only exposure), MCU KO was not protective, and in fact markedly potentiated the Ca²⁺ dependent injury compared to WT neurons. However, with Zn²⁺ + high K⁺ (with DTDP) and no Ca²⁺ (Zn²⁺ only exposure), MCU KO was protective compared to WT. To determine if MCU KO neurons would display neuroprotection if Ca²⁺ contributions were attenuated, cell death in WT and MCU KO neurons was quantified after the ischemic Zn²⁺ exposure (which includes both Zn²⁺ and Ca²⁺) was repeated in the presence of the NOX inhibitor apocynin (as in Fig. 2, above), aiming to reduce the injury burden caused by cytosolic Ca²⁺ actions (including Ca²⁺-induced NOX activation) (Brennan et al., 2009). Indeed, with NOX inhibition, we found significant neuroprotection from Zn²⁺-induced injury in MCU KO neurons (Fig 4A), supporting the hypothesis that targeting Zn²⁺ entry through the MCU specifically—unlike indiscriminate MCU blockade against both Ca²⁺ and Zn²⁺—could be neuroprotective.

Figure 4. Zn²⁺-specific contributions to neurotoxicity and ischemic neurodegeneration are attenuated in MCU KO cultured neurons and hippocampal slices.

A. Zn²⁺-triggered neurotoxicity is attenuated in cultured MCU KO neurons. WT (blue) and MCU KO (red) neurons were treated to the following exposures: “Ca²⁺ & Zn²⁺” (10 min pre-exposure to 60 μM DTDP, followed by 5 min 100 μM Zn²⁺ + high K⁺ + MK-801 + DTDP and 3x wash into DTDP for 20 min, all in 1.8 mM Ca²⁺ HSS), “Ca²⁺” (5 min exposure to high K⁺ + MK-801, followed by 3x wash for 20 min, all in 1.8 mM Ca²⁺ HSS), “Zn²⁺” (same as “Ca²⁺ & Zn²⁺” except in 0 mM Ca²⁺ HSS), or “Ca²⁺ & Zn²⁺” with Apocynin (same as “Ca²⁺ & Zn²⁺” with 500 μM Apocynin co-present with DTDP; indicated by diagonal striped bar). After 24 hr incubation, cell death was quantified with LDH efflux assay and validated with direct morphological examination. Concentrations used are indicated. Bars represent mean % cell death ± SEM of 6 independent experiments, each consisting of 4 wells of cultured neurons. Note that MCU KO neurons showed heightened sensitivity to Ca²⁺ triggered injury, but with attenuation of the Ca²⁺-contributions (either via Ca²⁺-free conditions or NOX inhibition with apocynin), MCU KO neurons were significantly protected from Zn²⁺-triggered neurotoxicity.

B. Ischemic neurodegeneration is delayed in MCU KO hippocampal CA1 neurons when Ca²⁺-contributions are moderated. Individual CA1 pyramidal neurons in hippocampal slices from WT (black; “control”) or MCU KO (blue; “MCU”) were loaded with the cytosolic Ca²⁺ indicator Fura FF and subjected to prolonged OGD in ACSF containing either 2 mM (left) or 200 μM (right) Ca²⁺. Traces show time course of ratio of Fura FF emissions upon excitation at 340 and 380 nm and represent mean ± SEM of ≥ 5 separate repetitions. Note that while MCU KO CA1 neurons showed more rapid Ca²⁺ deregulation (as indicated by irreversible Fura FF signal rise) with physiologic (2 mM) Ca²⁺ (left; likely due to rapid, unbuffered cytosolic Ca²⁺ accumulation leading to more rapid catastrophic injury), when effects of Ca²⁺ were attenuated, ischemic neurodegeneration was significantly delayed (right).

To test these findings in a model closer to in vivo injury, we next examined response to prolonged oxygen-glucose deprivation (OGD), as a model of ischemia, in acute hippocampal slices from MCU KO and WT mice. CA1 pyramidal neurons were loaded with low affinity, ratiometric cytosolic Ca²⁺ indicator Fura FF (K_d ~ 6 μM) (Medvedeva et al., 2009; Medvedeva and Weiss, 2014) and subjected to OGD (via perfusion with artificial cerebral spinal fluid [ACSF] bubbled with nitrogen gas to remove oxygen, and with glucose replaced by sucrose) with either physiologic (2 mM; Fig 4B left) or low Ca²⁺ levels (200 μM, to minimize Ca²⁺-contributions to injury; Fig 4B right). Consistent with prior studies using pharmacologic MCU blockade (Medvedeva and Weiss, 2014), MCU KO slices displayed an accelerated onset of the lethal Ca²⁺ deregulation event in response to OGD with physiologic Ca²⁺ (Fig 4B left), possibly reflecting loss the of mitochondrial Ca²⁺ buffering capacity, and consequent increased cytosolic Ca²⁺ loading and activation of cytosolic Ca²⁺ dependent cell death pathways. However, when Ca²⁺ contributions to injury were attenuated (via lower Ca²⁺ levels), onset of Ca²⁺ deregulation was significantly delayed in MCU KO neurons (Fig 4B right). These data from hippocampal slices show that when extramitochondrial effects of Ca²⁺ are moderated, MCU deletion can confer significant protection against ischemic neurodegeneration. As the degree of this protection is similar to that conferred by Zn²⁺ chelation observed in prior studies (Medvedeva and Weiss, 2014), taken together with our current findings from cultured neurons, these observations support the hypothesis that attenuating mitochondrial Zn²⁺ accumulation through the MCU is neuroprotective.

Delayed MCU blockade attenuates Zn²⁺-triggered mitochondrial dysfunction and cell death

Finally, to address whether present findings of attenuated Zn²⁺-triggered injury in MCU KO neurons might be applicable after cellular Zn²⁺ loading has already occurred, we examined the effects of delayed MCU blockade using the pharmacologic agent Ruthenium Red (RR). We first assessed changes in rapidly triggered mitochondrial dysfunction. WT neurons, loaded with either Rhod123 (Fig 5A left) or HEt (Fig 5A right), were treated with the ischemic Zn²⁺ exposure and washed into DTDP alone or DTDP + RR (10 μM), followed by FCCP in Rhod123-loaded cultures, as indicated. We found that delayed addition of RR attenuated both loss of ΔΨ_m and ROS production even after initial mitochondrial Zn²⁺ loading had already occurred (Fig 5A).

Figure 5. Delayed MCU blockade attenuates the Zn²⁺-triggered mitochondrial dysfunction and neurodegeneration.

WT neurons were exposed to the ischemic Zn²⁺ exposure, followed by 20 min wash into DTDP alone (blue, “CTRL”) or DTDP + 10 μM RR (red, “RR”). Traces represent mean ± SEM F_x/F₀ values for indicator ΔF (A) and represent 8 experiments consisting of ≥ 160 neurons. Bar graphs show mean + SEM of either L/W ratio (B) or % cell death (C), each representing ≥ 5 independent experiments. Grey bars indicate time points of comparison (** indicates p < 0.01, *** indicates p < 0.001, by two-tailed t-test).

A. Delayed MCU blockade attenuates acute Zn²⁺-triggered mitochondrial dysfunction. Neurons loaded with either Rhod123 (left) or HEt (right) were treated to the ischemic Zn²⁺ exposure as described above, with FCCP (1 μM) added in Rhod123 loaded neurons as indicated. Note that delayed RR attenuated both loss of ΔΨ_m (left) and rise in ROS production (right) in response to Zn²⁺ exposure.

B. MCU blockade after Zn²⁺ exposure preserves mitochondrial morphology. WT neurons were treated to the ischemic Zn²⁺ exposure as described above. After 12 hr incubation, neuronal mitochondria were labeled with MitoTracker Green and observed under confocal microscopy (1000x), after which L/W ratio of mitochondria were calculated blindly. Note the preserved mitochondrial structure (indicated by their elongated morphology [left] and corroborated with high L/W ratio [right]) in RR-treated neurons.

C. Delayed MCU blockade attenuates neurodegeneration. WT neurons were treated to the ischemic Zn²⁺ exposure as described above. After 24 hr incubation, cell death was assessed via LDH efflux assay. Note that delayed MCU blockade (via transient RR treatment) significantly attenuated Zn²⁺-triggered neurotoxicity.

We next set out to determine whether delayed MCU blockade by RR would provide prolonged preservation of mitochondrial morphology. WT neurons were exposed to the same paradigm above (the ischemic Zn²⁺ exposure followed by wash into DTDP ± RR for 20 min) and incubated for 12 hrs, after which mitochondria were labeled with MitoTracker Green and observed under confocal microscopy (as in Fig 3C). Note once again the preservation of elongated mitochondrial structure (corroborated by greater L/W ratio) in neurons treated with RR (Fig 5B). Finally, to test whether delayed MCU blockade could also have neuroprotective effects, the exposure was repeated and cell death assessed after 24 hr incubation (as above). The delayed but transient (20 min) MCU blockade produced a remarkable decrease in neurotoxicity (Fig 5C). These findings show that relatively brief MCU blockade after the insult can dramatically attenuate the downstream events that contribute to neuronal injury.

Discussion

Summary of findings

In the current study, we used the recently available MCU KO mice to evaluate the role of MCU in Zn²⁺-induced neuronal injury with greater specificity than was previously possible with pharmacological blockers. In cultured cortical neurons, genetic deletion of MCU attenuated mitochondrial Zn²⁺ accumulation, and consequently, diminished its deleterious effects on mitochondria, including loss of ΔΨ_m, inhibition of respiration, and swelling. While MCU KO surprisingly had little effect on the total cellular Zn²⁺-triggered ROS generation, this appeared to be due to an increased cytosolic NOX activity in the MCU KO neurons. Indeed, using the NOX antagonist apocynin, we found that mitochondrial ROS generation was nearly eliminated in the MCU KO neurons.

In assessing the role of MCU on neurodegeneration 24 hrs after exposures to both Ca²⁺ and Zn²⁺, we found genetic deletion of MCU was not protective. This death appeared to reflect primarily Ca²⁺ contribution, as further studies revealed accentuation of Ca²⁺-induced cell death and attenuation of Zn²⁺-triggered neurodegeneration in the MCU KO neurons. When a NOX antagonist was present during the combined Ca²⁺ and Zn²⁺ exposure, injury in the MCU KO was again significantly attenuated, suggesting that Ca²⁺-induced activation of upregulated NOX (as mentioned above) was a major contributor to Ca²⁺ triggered neurotoxicity in MCU KO neurons. These findings from cultured neurons were also reflected in the hippocampal slice OGD ischemia model. Whereas MCU KO accelerated OGD-induced cell death of CA1 pyramidal neurons when physiological levels of Ca²⁺ were present, when Ca²⁺ was lowered to abrogate its cytosolic effects, the neurodegeneration in the MCU KO pyramidal neurons was markedly delayed. These findings from the MCU KO cultured neurons and hippocampal slices further strengthen the notion that Zn²⁺ entry through the MCU is a critical contributor to mitochondrial dysfunction and the consequent neurodegeneration.

Finally, as therapeutic interventions in humans will generally entail application of drugs after an ischemic event has occurred, we tested effects of delayed administration of the MCU antagonist, RR, after the Zn²⁺ exposure (again in presence of physiological Ca²⁺). We found the delayed RR to diminish the induction of acute mitochondrial dysfunction (ΔΨ_m and ROS), late mitochondrial swelling, and subsequent cell death, supporting the possible therapeutic utility of delayed targeting of Zn²⁺ entry through the MCU after an ischemic episode has occurred.

Zn²⁺ and ischemic neurodegeneration

As noted above, there is a strong need for greater understanding of injury pathways activated after brief ischemia in order to develop effective interventions. Excitotoxic injury has been strongly implicated in ischemia, and most studies of relevant mechanisms have focused on the critical role of rapid Ca²⁺ entry through NMDA receptors. One target of interest has been mitochondria, which Ca²⁺ enters through the MCU, and with sufficient accumulation, causes a myriad of deleterious effects, including disrupted function, ROS generation, swelling due to opening of mitochondrial permeability transition pore (mPTP), and release of Cytochrome C (CytC) (Choi et al., 1988; Nicholls and Budd, 2000; Rothman and Olney, 1986). But it has become clear that Ca²⁺ can also induce injury via activation of cytosolic targets, including NOX and catabolic enzymes, with some suggesting that cytosolic Ca²⁺ accumulation may be the key determinant of cell death (Brennan et al., 2009; Murphy et al., 2014; Siesjo, 1988). However, NMDA receptor targeted therapies have shown limited clinical efficacy (Hoyte et al., 2004; Ikonomidou and Turski, 2002). Furthermore, a number of studies finding Zn²⁺ accumulation in injured neurons after ischemia and neuroprotection with Zn²⁺ chelation strongly implicated Zn²⁺ as a contributor to the injury cascade (Calderone et al., 2004; Koh et al., 1996; Tonder et al., 1990; Yin et al., 2002).

An early question concerned the source of Zn²⁺ accumulation. As Zn²⁺ is found in presynaptic vesicles of certain excitatory pathways, from which it can be released with activation (Assaf and Chung, 1984; Frederickson, 1989; Howell et al., 1984; Sloviter, 1985), it was thought that uptake of this synaptic Zn²⁺—through routes including VSCC and Ca-AMPAR—would account for the accumulation. Yet, studies in ZnT3 KO mice (in which synaptic Zn²⁺ is eliminated due to deletion of the presynaptic vesicular Zn²⁺ transporter, ZnT3) (Cole et al., 1999) found increased Zn²⁺ accumulation in CA1 neurons after prolonged seizures (in contrast to CA3, in which decreased accumulation was noted in the ZnT3 KOs) (Lee et al., 2000; Lee et al., 2003). These findings clearly indicated another source for the Zn²⁺ accumulation, which turned out to be that released from buffering proteins (in particular MT-III) in response to metabolic changes (oxidative stress / acidosis) associated with pathological conditions including ischemia and prolonged seizures (Erickson et al., 1997; Maret, 1995). Indeed, intracellular Zn²⁺ mobilization from this source alone (via the oxidizing agent DTDP) was sufficient to induce both mitochondrial dysfunction and neurotoxicity (Aizenman et al., 2000; Sensi et al., 2003), and it is highly likely that during ischemia, chronic disruption of the intracellular buffering promotes continuous mitochondrial Zn²⁺ influx even after the acute insult (Ji et al., 2019; Medvedeva et al., 2017).

Studies on CA1 and CA3 hippocampal pyramidal neurons (HPNs) from transgenic (ZnT3 and MT-III KO) mouse brain slices subjected to OGD provide further support to the idea that the two sources may contribute to their differential vulnerabilities in disease, with CA3 neurons preferentially degenerating after prolonged seizures (Ben-Ari et al., 1980a; Ben-Ari et al., 1980b; Tanaka et al., 1988), and CA1 neurons after transient ischemia (Kirino, 1982; Ordy et al., 1993; Sugawara et al., 1999). In CA3, uptake of synaptic Zn²⁺, as may occur with repeated neuronal firing during seizures, was the predominant contributor to cytosolic Zn²⁺ accumulation, whereas in CA1, Zn²⁺ release from MT-III predominated, possibly consistent with prolonged metabolic disruption during ischemia causing protracted disruption of intracellular buffering (Medvedeva et al., 2017). Although most studies have examined the 2 sources separately, it is highly likely that they play synergistic roles, and more recent studies on cultured neurons using both exogenous Zn²⁺ exposures and DTDP have found even partial disruption of endogenous buffering (via DTDP) to dramatically increase toxic effects of relatively low exogenous Zn²⁺ entry (Ji and Weiss, 2018). Importantly for our current study, this combined exogenous Zn²⁺ and DTDP exposure permits us to more comprehensively model contributions from both Zn²⁺ sources, as likely occurs in ischemia, and when applied to MCU KO neurons, provides important insight into the role of the MCU in Zn²⁺-contributions to neuronal injury.

Mitochondria: critical, early targets of Zn²⁺ effects

Many studies have identified various targets of Zn²⁺ accumulation that lead to neuronal injury (Kim and Koh, 2002; McLaughlin et al., 2001; Noh and Koh, 2000). However, paralleling Ca²⁺, early studies suggested Zn²⁺ also has potent effects on mitochondria. Indeed, studies on both isolated mitochondria and cultured neurons have revealed that MCU blockers can also block Zn²⁺ entry into mitochondria, where it appears to induce mitochondrial dysfunction (including loss of ΔΨ_m, ROS production, mitochondrial swelling, inhibition of respiration) with far greater potency than Ca²⁺ (Clausen et al., 2013; Gazaryan et al., 2007; Jiang et al., 2001; Saris and Niva, 1994; Sensi et al., 2003), although others have reported a relative paucity of Zn²⁺ effects on mitochondria under conditions where little Ca²⁺ is present (Devinney et al., 2009; Pivovarova et al., 2014). Yet, despite the ability to block both Ca²⁺ and Zn²⁺ uptake into mitochondria, therapeutic attempts using MCU blockers have been disappointing, showing evidence of neuroprotection in some cases, while worsening outcomes in others (Velasco and Tapia, 2000). The reasons for this are not completely clear, but it is possible that by preventing mitochondrial Ca²⁺ uptake, MCU blockade promoted excess cytosolic Ca²⁺ accumulation, leading to loss of cytosolic Ca²⁺ homeostasis that ultimately induced further neuronal injury (Murphy et al., 2014). Furthermore, as the pharmacologic MCU blockers have been characterized to lack full specificity for the MCU and exert numerous nonspecific effects (including VSCC blockade and antioxidant) (Meinicke et al., 1998; Tapia and Velasco, 1997), it was not possible to ascertain whether the effects—protective or not—were indeed specifically due to MCU blockade. In that regard, a key aim of the current study was to model ischemia in in vitro culture and slice experiments (via exogenous Zn²⁺ exposure + DTDP and OGD respectively, as discussed above) and use genetic tools to validate and better define the role of the MCU in Zn²⁺ contributions to ischemic neurodegeneration. Our current findings using MCU KO mice not only provide validation that MCU truly is an important route of mitochondrial Zn²⁺ entry that promotes mitochondrial dysfunction, but also reveals that targeting Zn²⁺ specifically through the MCU—rather than indiscriminate MCU blockade, which can actually exacerbate Ca²⁺-triggered injury—can provide significant neuroprotection. Indeed, the recent identification of the MCU gene and its associated regulatory peptides (De Stefani et al., 2015; Kamer and Mootha, 2015; Marchi and Pinton, 2014) provided a major opportunity for breakthrough in the study of these channels. Interestingly, the regulatory subunits MICU1 and 2 appear to respond to Ca²⁺ to enable pore opening, preventing pore opening when Ca²⁺ is low and opening it when Ca²⁺ rises – conferring a sigmoid shaped Ca²⁺ concentration, channel activation relationship. These findings may have implications for Zn²⁺ effects on mitochondria as well, especially in light of prior evidence that Ca²⁺ not only is necessary for Zn²⁺ effects in certain conditions (as discussed above), but also enhances Zn²⁺ entry through the MCU (Saris and Niva, 1994). Furthermore, we and others have observed evidence for synergism between Zn²⁺ and Ca²⁺ on mitochondria, with far greater effects of Zn²⁺ in presence of Ca²⁺ (Gazaryan et al., 2007; Ji and Weiss, 2018; Jiang et al., 2001; Sensi et al., 2000), which could be explained in part by Ca²⁺ dependence of the MCU pore opening.

Delayed targeting of Zn²⁺ for protection against ischemia

Studies in hippocampal slice OGD models of ischemia have yielded important clues to how Zn²⁺ may contribute to ischemic neurodegeneration. Early studies clearly showed acute extracellular and intracellular Zn²⁺ rises (Li et al., 2001; Wei et al., 2004), but interactions between Zn²⁺ and Ca²⁺ effects were unexplored. Thus, we carried out the first study monitoring Ca²⁺ and Zn²⁺ simultaneously in single CA1 HPNs, and found Zn²⁺ rises to precede and contribute to lethal Ca²⁺ deregulation. Furthermore, this seemed to depend upon Zn²⁺ uptake into mitochondria (Medvedeva et al., 2009). To test the relevance of these findings, we carried out in vivo global ischemia studies in rats, where after inducing transient cardiac arrest, we used the Timm’s sulfide silver stain to label reactive Zn²⁺, and observed neurons under electron microscopy to examine both subcellular localization of Zn²⁺ and mitochondrial morphology (Yin et al., 2019). We found evidence of delayed and progressive mitochondrial Zn²⁺ accumulation in CA1 neurons, with the degree of accumulation correlating with the extent of mitochondrial structural disruption. These findings in hippocampal slices and whole animal models further support a critical role of mitochondrial Zn²⁺ accumulation as an early factor in ischemic neurodegeneration the targeting of which may yield protection.

To test the role of MCU in this injury cascade, pharmacologic blockers (RR or the more specific analogue Ru-360) were added prior to prolonged OGD in hippocampal slices, which surprisingly accelerated the lethal Ca²⁺ deregulation. As this may have been due to MCU blockers hastening cytosolic Ca²⁺ accumulation by inhibiting mitochondrial Ca²⁺ uptake, the experiment was repeated with ACSF Ca²⁺ lowered from 2 mM to 200 μM; under these circumstances, the blocker became highly protective. Additional findings that (in low Ca²⁺ conditions) the degree of protection from MCU blockade was similar to that from Zn²⁺ chelation supported the idea that the protection from MCU blockers was due to antagonism of Zn²⁺ entry into mitochondria via the MCU (Medvedeva and Weiss, 2014). Present studies using the MCU KO slice model subjected to prolonged OGD further validate these prior results, supporting the idea that attenuating Zn²⁺ entry specifically through MCU can provide neuroprotection. Yet, as these studies entailed either pretreatment with MCU blocker or genetic knockout, they are limited in utility for translation. Indeed, attenuating MCU activity prior to insult appears to exacerbate peak Ca²⁺ loading and worsens Ca²⁺ dependent injury, and as ischemic events cannot be predicted in patients, most opportunities for intervention are after the acute ischemic insult. Furthermore, given the profound protective effects of attenuating Zn²⁺ entry found in the MCU KO mice, delayed interventions targeting the MCU may be particularly beneficial, as it would reduce Zn²⁺ influx into mitochondria that likely continues to occur after ischemic insult.

To address mechanisms of such delayed injury after transient ischemia, we subjected slices to sublethal episodes of OGD (in which OGD was aborted shortly prior to the expected time of Ca²⁺ deregulation). We found that mitochondrial Zn²⁺ entry was an early event in response to sublethal OGD in both CA1 and CA3 regions. However, CA1 neurons had prolonged (> 1hr) Zn²⁺ accumulation in mitochondria that progressed during the period after OGD, and which appeared to contribute to delayed mitochondrial swelling. In contrast, far less late mitochondrial Zn²⁺ accumulation was observed in the CA3 neurons. Given the particular vulnerability of CA1 neurons to delayed ischemic degeneration, it is possible that this early mitochondrial Zn²⁺ accumulation and prolonged sequestration is a key trigger of the persistent mitochondrial dysfunction, including mitochondrial swelling, CytC release and mitochondrial multi-conductance channel activation observed in some animal studies (Bonanni et al., 2006; Calderone et al., 2004; Sugawara et al., 1999). Late administration of the MCU blocker RR diminished both the delayed Zn²⁺ accumulation within the mitochondria and the consequent mitochondrial swelling 1 hr after sublethal OGD in CA1 HPNs (Medvedeva et al., 2017).

Present findings of protective effects of late RR after Zn²⁺ exposures to cultured WT cortical neurons are consistent with prior studies, providing further support to the idea that delayed mitochondrial Zn²⁺ accumulation contributes to mitochondrial dysfunction and delayed neurodegeneration and supports the utility of late interventions. Indeed, by permitting buffering of acute cytosolic Ca²⁺ loads by the mitochondria prior to MCU blockade, late administration of these antagonists after the insult may abrogate the problem of enhanced cytosolic Ca²⁺ loading seen with pretreatment. Although the available antagonists lack full specificity for this channel and their other actions (including antioxidant effects) could theoretically contribute to their benefit, we found that application of RR did not significantly attenuate Zn²⁺-triggered ROS production in MCU KO neurons (data not shown), suggesting the benefits in RR treated WT neurons were indeed due to MCU blockade rather than nonspecific effects. Combined with present findings using the MCU KO, our data on delayed MCU blockade provide compelling support to the notion that post-injury antagonism of mitochondrial Zn²⁺ accumulation could play a crucial role as a future neuroprotective strategy.

Conclusions

In summary, present findings lend new support to the hypothesis that the MCU plays a critical role in Zn²⁺-triggered mitochondrial dysfunction and the subsequent neurotoxicity. Furthermore, the strong protection provided by delayed MCU blockade may lend further new support to the potential utility of delayed MCU blockade to attenuate mitochondrial Zn²⁺ accumulation even after ischemia.

Based on these observations, we propose the following model during ischemia: 1) Zn²⁺ accumulates in the cytosol due to combination of synaptic Zn²⁺ uptake and release from buffers, 2) accumulated Zn²⁺ enters the mitochondria via the MCU, with the degree of mitochondrial dysfunction reflecting combination of both the amount and the duration of mitochondrial Zn²⁺ loading. 3) In lethal ischemia, overwhelming mitochondrial Zn²⁺ accumulation occurs during ischemic insult to cause mitochondrial failure and promote acute lethal cytosolic Ca²⁺ overload, ultimately leading to rapid cell death. 4) If ischemic episode is transient, the mitochondria may have already depolarized sufficiently to release much of their accumulated Zn²⁺, but during recovery, they begin to repolarize and Zn²⁺ re-accumulates into the mitochondria. 5) This re-accumulation promotes more mitochondrial ROS generation, and aided by the post-injury oxidative environment (which can continue disrupting intracellular Zn²⁺ buffering to create a feed forward cascade of mitochondrial Zn²⁺ loading), lead to more mitochondrial dysfunction. 6) The accompanying consequences, including mPTP opening, CytC release, and irreversible oxidative damage, may be critical upstream events in triggering delayed degeneration pathways. Thus, targeting of either the early mitochondrial Zn²⁺ accumulation and/or the delayed Zn²⁺ re-accumulation into the mitochondria—both through the MCU—may yield significant benefit by impeding activation of these delayed injury pathways.

Highlights.

• Zn²⁺ thought to enter the mitochondria via the mitochondrial Ca²⁺ uniporter (MCU)

• Mitochondrial Zn²⁺ loading and its consequences are attenuated in MCU knockout mice

• Zn²⁺ contributions to neurodegeneration are attenuated in MCU knockout

• Delayed MCU blockade attenuates mitochondrial dysfunction and neurotoxicity

• Targeting mitochondrial Zn²⁺ through the MCU may offer significant neuroprotection

Abbreviations

Ca-AMPAR

Ca²⁺ permeable AMPA receptors

FCCP

carbonyl cyanide-p-trifluoromethoxyphenylhydrazone

DTDP

2,2’-dithiodipyridine

HEt

hydroethidine

KO

knockout

MT

metallothionein

MCU

mitochondrial calcium uniporter

NMDA

N-methyl-D-aspartate

OGD

oxygen glucose deprivation

ROS

reactive oxygen species

Rhod123

rhodamine 123

VSCC

voltage sensitive Ca²⁺ channels