Abstract
Action potential (AP)-independent (miniature) neurotransmission occurs at all chemical synapses but remains poorly understood, particularly in pathologic contexts. Axonal endoplasmic reticulum (ER) Ca2+ stores are thought to influence miniature neurotransmission, and aberrant ER Ca2+ handling is implicated in progression of Huntington disease (HD). Here, we report elevated mEPSC frequencies in recordings from YAC128 mouse (HD-model) neurons (from cortical cultures and striatum-containing brain slices, both from male and female animals). Pharmacological experiments suggest that this is mediated indirectly by enhanced tonic ER Ca2+ release. Calcium imaging, using an axon-localized sensor, revealed slow AP-independent ER Ca2+ release waves in both YAC128 and WT cultures. These Ca2+ waves occurred at similar frequencies in both genotypes but spread less extensively and were of lower amplitude in YAC128 axons, consistent with axonal ER Ca2+ store depletion. Surprisingly, basal cytosolic Ca2+ levels were lower in YAC128 boutons and YAC128 mEPSCs were less sensitive to intracellular Ca2+ chelation. Together, these data suggest that elevated miniature glutamate release in YAC128 cultures is associated with axonal ER Ca2+ depletion but not directly mediated by ER Ca2+ release into the cytoplasm. In contrast to increased mEPSC frequencies, cultured YAC128 cortical neurons showed less frequent AP-dependent (spontaneous) Ca2+ events in soma and axons, although evoked glutamate release detected by an intensity-based glutamate-sensing fluorescence reporter in brain slices was similar between genotypes. Our results indicate that axonal ER dysfunction selectively elevates miniature glutamate release from cortical terminals in HD. This, together with reduced spontaneous cortical neuron firing, may cause a shift from activity-dependent to -independent glutamate release in HD, with potential implications for fidelity and plasticity of cortical excitatory signaling.
SIGNIFICANCE STATEMENT Miniature neurotransmitter release persists at all chemical neuronal synapses in the absence of action potential firing but remains poorly understood, particularly in disease states. We show enhanced miniature glutamate release from cortical neurons in the YAC128 mouse Huntington disease model. This effect is mediated by axonal ER Ca2+ store depletion, but is not obviously due to elevated ER-to-cytosol Ca2+ release. Conversely, YAC128 cortical pyramidal neurons fired fewer action potentials and evoked cortical glutamate release was similar between WT an YAC128 preparations, indicating axonal ER depletion selectively enhances miniature glutamate release in YAC128 mice. These results extend our understanding of action potential independent neurotransmission and highlight a potential involvement of elevated miniature glutamate release in Huntington disease pathology.
Introduction
Neurotransmitter release occurs independently of Na+-driven action potentials (APs) at most, if not all, neuronal synapses. This activity-independent (or “miniature”) release has been shown to be physiologically relevant and can serve distinct biological roles. Dendritic protein synthesis, network excitability, and neurodevelopment are among the processes influenced by miniature transmission (Carter and Regehr, 2002; Sharma and Vijayaraghavan, 2003; Sutton et al., 2006; Andreae and Burrone, 2015). In addition, recent studies suggest that enhanced miniature glutamate release can drive pathologic conditions, contributing to major depressive disorder and neurodegeneration (Fishbein and Segal, 2007; Chanaday et al., 2021).
In contrast to activity-dependent release, several studies report that miniature synaptic events persist in the absence of extracellular Ca2+ (Yamasaki et al., 2006; Xu et al., 2009); however, such events still appear to be largely Ca2+-dependent, based on their near complete blockade by the fast, membrane-permeable Ca2+ chelator BAPTA-AM in cortical cultures (Xu et al., 2009). Rather, Ca2+ originating from axonal ER Ca2+ stores appears to largely mediate this type of miniature release (Emptage et al., 2001). However, other studies report a direct dependence of miniature release on extracellular Ca2+, mediated by stochastic openings of voltage-gated Ca2+ channels (VGCCs) (Ermolyuk et al., 2013), or ER Ca2+ depletion and consequent Ca2+ entry via the store-operated Ca2+ channel (SOC) response (Chanaday et al., 2021). One study also reported miniature release in cortical cultures was mediated by the calcium sensing receptor (CaSR), a GPCR expressed on the plasma membrane of cortical terminals (Vyleta and Smith, 2011), which when activated by extracellular Ca2+ (or other agonists), increased miniature event frequencies independently of intracellular Ca2+. Although these studies suggest a complex, diverse regulation of AP-independent neurotransmission, there is general consensus that most mini events require Ca2+ (extracellular or intracellular). Moreover, involvement of presynaptic ER Ca2+ is a recurrent theme, with ER stores either providing Ca2+ for vesicular release directly or recruiting SOC entry.
Abnormalities in neuronal Ca2+ handling, particularly in relation to ER function (Tang et al., 2009), are key features of multiple age-related neurodegenerative disorders, including Huntington disease (HD) (Raymond, 2017), a fatal, autosomal dominantly-inherited neurodegenerative disorder caused by a polyglutamine-encoding CAG repeat-expansion (>35 repeats) in exon 1 of the huntingtin gene (Huntington's Disease Collaborative Research Group, 1993). Abnormal glutamate signaling is also a key HD feature and likely contributes to synaptic dysfunction and the relatively selective degeneration of GABAergic striatal spiny projection neurons (SPNs) and cortical pyramidal neurons (CPNs) in HD (Graveland et al., 1985; Vonsattel et al., 1985; Beal et al., 1986; Hantraye et al., 1990; Raymond, 2017; Cepeda and Levine, 2022). Increased postsynaptic extrasynaptic NMDAR expression (Zeron et al., 2002; Fan et al., 2007; Milnerwood et al., 2010; Botelho et al., 2014; Plotkin et al., 2014; Kovalenko et al., 2018), which triggers cell death signaling (Hardingham and Bading, 2010), is thought to contribute to HD pathology. Altered glutamate release from cortical projections onto SPNs has also been reported in multiple HD mouse models (Cepeda et al., 2003; Joshi et al., 2009; Raymond et al., 2011), but the direction of this effect appears to be model and disease-stage dependent (Joshi et al., 2009).
The ER's apparent involvement in regulating miniature glutamate release suggests a link between ER dysfunction and altered glutamate signaling in HD. Mutant huntingtin protein directly interacts with ER Type 1 inositol (1,4,5)-triphosphate receptors (IP3Rs), sensitizing their Ca2+ release in response to IP3 (Tang et al., 2003). In addition, ryanodine receptors (RyRs: ER-localized Ca2+ channels that mediate Ca2+-induced Ca2+ release) are constitutively leaky in HD mouse models (Suzuki et al., 2012). Mutant huntingtin is expressed at presynaptic terminals (Li et al., 2003); however, ER Ca2+ handling in this subcellular compartment has not been studied in HD. Here, we investigate presynaptic Ca2+ signaling and miniature glutamate release in cortical pyramidal neurons from premanifest HD-model mice.
Materials and Methods
Culture preparation
All animal-related procedures were approved by and adhered to the guidelines of the University of British Columbia Committee on Animal Care and the Canadian Council on Animal Care (protocols A17-0295, A15-0069, and A19-0076). Cultures were prepared from both male and female embryonic day 17-18 pups from either WT FVB/N or transgenic yeast artificial chromosome-containing mice expressing the full-length human huntingtin genomic DNA with 128 CAG repeats (YAC128). YAC128 mice were maintained on the FVB/N background (homozygous line 55). WT and YAC128 mice used for ex vivo slice experiments (below) and bred for culture preparation (above) were group-housed under controlled conditions, free of know pathogens, at room temperature (22°C-24°C), under a 12 h light/dark cycle. Cortical cultures used in patch-clamp electrophysiology, and Ca2+-imaging experiments were prepared as previously described (Milnerwood et al., 2012; Smith-Dijak et al., 2019) and plated at a density of 225,000 neurons/ml. In a subsets of experiments, a portion of the total 2.7 million cortical neurons (plated per 24-well culture) were transfected with one or more transgenic reporters including the following: GFP (Addgene plasmid 37825); a synaptophysin-tagged GCaMP6-M construct (a generous gift from Anne Marie Craig, University of British Columbia); JGCaMP7-F (Addgene plasmid 104489); a postsynaptic density 95 (PSD95)-tagged M-cherry construct; or a GFP-tagged internally expressed anti-PSD95 antibody (a generous gift from D.B. Arnold, University of Southern California) (Gross et al., 2013).
Electrophysiology
An Axopatch 200B amplifier and pClamp 9.2 software (Molecular Devices) were used to acquire whole-cell patch-clamp electrophysiology recordings. Data were digitized at 20 kHz and low-pass filtered at 1 kHz. For electrophysiology experiments, cultures were perfused with extracellular fluid (ECF) containing the following (in mm): 167 NaCl, 2.4 KCl, 10 glucose, 10 HEPES, 2 CaCl2, and 1 MgCl2; NaOH (1 mm) was used to adjust the pH to 7.30, and the osmolarity was adjusted to 305-310 mOsm. TTX (500 nm) and picrotoxin (PTX) (50 mm) were added to this ECF to block sodium channel-mediated APs and GABAA receptor-mediated currents, respectively. Neurons were patched with borosilicate glass pipettes pulled to a tip resistance of 3-6 mΩ when back-filled with intracellular solution containing the following (in mm): 130 Cs-methanesulfonate, 5 CsCl, 4 NaCl, 1 MgCl, 10 HEPES, 5 EGTA, 5 QX-314 Cl, 0.5 Na-GTP, 10 Na-phosphocreatine, and 5 Mg-ATP (∼286 mOsm). During experiments, neurons were held at –70 mV in voltage-clamp, with hyperpolarizing voltage steps (−10 mV) performed periodically to measure intrinsic membrane properties. Under these conditions, AMPAR-mediated mEPSCs appeared as transient inward current deflections. Recordings with a series resistance >25 mΩ were excluded from analysis; typical values were between 15 and 20 mΩ. A minimum of 2 min following establishment of the whole-cell configuration was allowed before experimental measurements, so neurons could fully dialyze with intracellular solution and membrane resistance and holding current stabilize. For experiments involving within-cell drug applications, a maximum 20% change in series resistance between control and drug measurements was tolerated. In these cases, drugs were applied locally to neurons with a fast perfusion system. Mini Analysis software (Synaptosoft) was used to detect mEPSCs and extract relevant parameters. A minimum of 100 and no more than 1000 mEPSCs were analyzed per neuron per experimental condition.
CPN morphology
CPNs in WT and YAC128 cultures were labeled by transfecting a subset of neurons (1 of 2.7 million) with a cytoplasmic GFP at the time of platting. Cultures were subsequently fixed at DIV17-19 and GFP-labeled CPN dendritic arbors imaged on a Zeiss Axiovert 200M fluorescence microscope (20× magnification, 0.8 NA), with a Zeiss 702 monochrome camera, using Zen software. Multiple image Z stacks were acquired and X,Y-tiling was used to ensure entire dendritic arbors were visualized. Images were exported to Fiji-ImageJ for analysis by a blinded observer. Images were flattened using the maximum Z-projection function. Background subtraction was performed, and neuronal processes were thresholded following adjustment of brightness and contrast. Automated Sholl analysis was performed using the ImageJ Sholl analysis plugin.
Excitatory cortical synapse staining
A subset of neurons (2 of 2.7 million) in WT and YAC128 cortical cultures were transfected with a GFP-tagged internally-expressed anti-PSD95 antibody (intrabody) (Gross et al., 2013) at the time of platting. At DIV17-DIV19, cells were fixed and stained for VGlut1 and the GluA2 AMPAR subunit as previously described (Buren et al., 2016). Briefly, cultures were first live-stained with a primary mouse anti-GluA2 antibody (Millipore), then fixed and stained with a secondary AlexaFluor-568-conjugated donkey anti-mouse antibody (Invitrogen). Subsequently, cultures were incubated with a primary guinea pig anti-VGlut1 antibody (Millipore), then stained with a secondary AMCA-conjugated donkey anti-guinea pig antibody (Jackson ImmunoResearch Laboratories). To amplify the GFP fluorescence of the anti-PSD95 intrabody, cultures were also incubated with a primary chicken anti-GFP antibody (1:1000) (Millipore), followed by a secondary AlexaFluor-488-conjugated antibody (1:500) (Invitrogen). Cultures were imaged on a Zeiss Axiovert 200M fluorescence microscope (63× magnification, 1.4 NA), using a Zeiss 702 monochrome camera and Zen software. CPNs expressing the anti-PSD95 intrabody were identified based on their diffuse cytoplasmic GFP fill, with bright GFP-labeled puncta expressed at dendritic spines. A portion of each CPN's arbor, containing multiple secondary and tertiary dendrites, was selected for imaging and sufficient image Z stacks were acquired to adequately capture all dendritic processes present in a given 63× field. Images were exported to Fiji-ImageJ for analysis by a blinded observer and flattened using the maximum Z-projection function. For each CPN image, the GFP channel was used to identify three secondary or tertiary dendritic segments, at least 40 µm away from the CPN soma, over which ROIs were drawn. Following background subtraction, fluorescent puncta in the green (GFP), red (AlexaFluor-568), and blue (AMCA) channels, visible within dendritic ROIs, were manually thresholded and detected with the analyze particles function. The ImageJ colocalization plugin was used to identify triple-colocalized puncta (PSD95, GluA2, and VGlut1), which we interpreted as functional CPN glutamatergic synapses. Synapse density was defined as the number of triple-colocalized puncta present within a dendritic segment divided by the area of the segment and averaged across all three dendrites analyzed in a given CPN.
GCaMP imaging
To directly image cytosolic Ca2+ in axonal boutons of neurons in our cortical monocultures, we transfected 1 million cells (of a total 2.7 million) at time of plating with a rat synaptophysin-tagged GCaMP6-M construct (rSyph-GCaMP6m). The rSyph-GCaMP6m construct was created by fusing GCaMP6-M (Chen et al., 2013) with the full-length rat synaptophysin protein (1-307 amino acids) via a small glycine-serine linker and inserting the fused rSyph-GCaMP6-M construct into a pLL3.7-hSyn vector to achieve neuron-selective expression. For some experiments, the same 1 million cells were also transfected at time of platting with an M-cherry-tagged PSD95 construct; in these cases, presumptive glutamatergic terminals targeting other CPNs were identified as rSyt-GCaMP6m-expressing boutons colocalized with M-Cherry-labeled postsynaptic spines. In other experiments, 1 million of the total 2.7 million cortical neurons were transfected at the time of plating with jGCaMP7-F to monitor AP-dependent somatic Ca2+ events.
For all Ca2+-imaging experiments, cultures were plated on 8-well cover glass chambers (Fisher Scientific, Nunc, Lab-Tek) and imaged at DIV17-DIV19 with a Zeiss Axiovert 200M fluorescence microscope (63× magnification, 1.4 NA), using a Zeiss 702 monochrome camera and Zen software. Movies were acquired using the Zen time-series mode. For experiments where only Ca2+-dependent GCaMP fluorescence was measured, data were acquired at 10 Hz (100 ms exposure per frame), with camera-steaming enabled. In some experiments, we simultaneously monitored Ca2+-dependent and independent (isosbestic) GCaMP fluorescence in boutons by rapidly switching between two filter cubes: one with a 488 nm excitation filter (Ca2+-dependent) and the other with a 405 nm (isosbestic) excitation filter (both filter cubes used 525 nm emission filters). In these experiments, a 100 ms exposure time was used for both channels, and data were acquired at 0.6 Hz per channel. These experiments were performed in standard ECF (as above) with or without TTX (500 nm) present, but in the absence of PTX.
Spontaneous Ca2+ waves were commonly evident in rSyph-GCaMP6m-labeled axons, particularly when AP-dependent Ca2+ events were blocked with TTX. These signals were quantified either with the Astrocyte Quantitative Analysis (AQuA) software (running in MATLAB) (Wang et al., 2019) or an in house software developed to export and quantify GCaMP time courses from boutons colocalized with MCherry-labeled dendritic spines (also in MATLAB). Rather than relying on neuronal morphology-based segmentation, AQua defines events as spatially and temporally connected signals surpassing user-defined thresholds. This allowed the spatial extent of Ca2+ waves to be quantified, but also meant that multiple boutons were often classified as part of a single event when contacted by a given Ca2+ wave. Frequencies of AQuA-detected spontaneous axonal events were quantified as numbers of events occurring within a standardized area [178.6 µm × 113.1 µm (a maximal 63× FOV)], during a standardized (3 min) consecutive imaging interval. AQuA detection parameters were empirically determined to best match a small number of manually analyzed experiments and applied across all cultures and conditions analyzed to facilitate meaningful comparisons.
Our in-house software adapted an algorithm from Nelson et al. (2015) to detect centroids of fluorescent puncta and to identify colocalizations between puncta in separate imaged channels (here GCaMP and MCherry-PSD95) based on a user-defined minimum distance between centroids (Nelson et al., 2015). GCaMP puncta (presumptive boutons) were classified as colocalized with CPN dendritic spines when their centroids fell within 4 pixels of an MCherry-labeled particle centroid. Both Ca2+-dependent and isosbestic GCaMP fluorescence was measured in these experiments (as above). As background autofluorescence was often high in the isosbestic channel, we restricted analysis of GCaMP puncta to those in which signal/noise was sufficient for our algorithm to detect corresponding puncta in the isosbestic channel. Each point of a bouton's Ca2+-dependent GCaMP time course was divided by the corresponding Ca2+-independent isosbestic channel signal to account for differences in GCaMP expression between individual boutons and cultures and facilitate quantitative Ca2+ comparisons. To compare relative resting bouton cytosolic Ca2+ levels between genotypes and various pharmacological conditions, we quantified both mean and minimum (488/405) fluorescence values of individual bouton traces (following low pass filtering to remove miniature Ca2+ events). The MATLAB FindPeaks function was used to detect miniature events in these (488/405) fluorescence traces and extract relevant parameters, including peak event amplitudes and event half-amplitude widths. Event frequencies were calculated by dividing numbers of detected miniature events in a trace by the trace interval in minutes.
In a subset of experiments conducted in the presence of TTX, responses of individual WT or YAC128 rSyph-GCaMP6m-labeled synaptic boutons to caffeine or ionomycin were measured. In these experiments, culture fields were imaged for 3 min (as above), after which caffeine (1 mm) or ionomycin (10 μm) was applied and imaging continued for an additional 1-2 min. For these experiments, an analyzer (blinded to culture genotype) used Fiji-ImageJ (National Institutes of Health) to manually assign elliptical ROIs to 10 boutons per movie and exported each ROI's fluorescence-intensity time course. Time courses were subsequently imported to MATLAB, where the curve fitting tool was used to model each time course's photobleaching profile based on the initial 3 min recording; this curve was then extrapolated to the entire time course including caffeine treatment. The resultant “bleaching curve” was subtracted from the raw rSyph-GCaMP6m fluorescence curve and this time course subsequently divided by the “bleaching curve” to yield a final curve reflecting the caffeine response in DF/F units. Ionomycin experiments were analyzed similarly, except in this case, the blinded analyzer selected 10 active boutons (showing at least 1 clear Ca2+ event during the initial 3 min recording) and 10 inactive boutons (with no Ca2+ events present during the initial 3 min recording). Time courses derived from inactive boutons were exported to MATLAB, where the ionomycin-mediated DF/F responses were calculated as above. In the case of active boutons, spontaneous events occurring during the first 3 min were detected with MATLAB's FindPeaks function and excised from the raw fluorescence time courses before curve fitting, but otherwise processed as above.
As expected, spontaneous events were far more frequent when rSyph-GCaMP6m-expressing cultures were imaged in the absence of TTX. Most boutons showed many events during a 3 min imaging session. These events, which were presumably driven by the AP-dependent opening of VGCCs, were far shorter in duration than typical TTX-resistant events. To assess for genotype differences in these signals, a blinded analyzer used Fiji-ImageJ to randomly assign elliptical ROIs to 20 boutons per movie and exported each ROI's fluorescence-intensity time course to MATLAB. The high frequency of AP-dependent events often precluded extracting the photobleaching profile of a bouton's time course, necessitating an alternative means of calculating the DF/F values of spontaneous events. We therefore averaged the gray-value fluorescence intensity across an entire time course, subtracted this average fluorescence pointwise from the time course, then divided pointwise by the average fluorescence to convert to DF/F units. Any nonstationary present in these DF/F time-series because of photobleaching was removed with the MATLAB detrend function. The MATLAB FindPeaks function was subsequently used to detect events in these DF/F traces as above. Bouton event frequencies and event parameters in the absence of TTX were compared between WT and YAC128 cultures under baseline conditions and following application of ryanodine (5 μm). A similar algorithm run entirely in MATLAB was used to monitor activity-dependent somatic Ca2+ events in JGCaMP7f expressing cortical neurons.
Intensity-based glutamate-sensing fluorescence reporter (iGluSnFR) imaging in acute brain slices
Expression of the genetically encoded iGluSnFR (Marvin et al., 2013) in WT and YAC128 mice was achieved with stereotaxic injection of a viral construct as described previously (Parsons et al., 2016). Briefly, under isoflurane anesthesia, 1-1.4 µl of the AAV1.hSyn.iGluSnFr.WPRE.SV40 construct (Penn Vector Core; Loren Looger, Janelia Farm Research Campus of the Howard Hughes Medical Institute) was directly injected into the dorsal striatum of 4- to 6-week-old mice. Following surgery, mice were closely monitored for a week to ensure adequate recovery.
After waiting 3-6 weeks, to ensure optimal iGluSnFR expression, acute brain slices from 2- to 4-month-old YAC128 mice and age-matched WT controls were prepared as described previously (Parsons et al., 2016; Koch et al., 2018). Briefly, mice were decapitated following deep isoflurane anesthesia, and their brains rapidly removed and placed in an ice-cold slicing solution, bubbled with carbogen (95% O2, 5% CO2) gas, containing the following (in mm): 125 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 2.5 MgCl2, 0.5 CaCl2, and 10 glucose; 300-µm-thick striatum-containing sagittal brain slices were cut with a Leica VT1200S vibratome. Slices were subsequently incubated for 30 min in warmed ACSF containing 2 mm CaCl2 and 1 mm MgCl2; ACSF constituents and concentrations were otherwise identical to the slicing solution (above).
Slices were transferred to a submerged recording chamber for experiments and perfused with carbogen-bubbled ACSF at a rate of 2-3 ml/min at room temperature. Cortical release of glutamate into the striatum was evoked by delivering paired 0.1 ms electrical pulses at 100 Hz with an A-M Systems isolated pulse stimulator (model 2100) and tungsten monopolar stimulating electrode (tip resistance 0.1 mΩ). The electrode was placed into a corpus callosum segment adjacent to the dorsal striatum at an ∼50-100 µm depth. During and immediately before the electrical stimulation, iGluSnFR fluorescence was excited with a 470 nm LED; slices were not illuminated between experimental measurements to minimize phototoxicity and bleaching. Stimulation and LED activation were triggered by Clampex software (Molecular Devices). iGluSnFR fluorescence was isolated with a 530 nm bandpass filter and imaged with a CCD camera (1 M60, Pantera, Dalsa) and XCAP software (Epix) at 150 Hz with 8 × 8 pixel binning. Experimental measurements encompassing four stimulation trials and two blank trials were performed at 3 min intervals. The four stimulation trials were averaged and the blank trials, in which slice fluorescence was imaged without electrical stimulation, were averaged and used to account for photobleaching and to calculate the stimulation mediated changes in iGluSnFR fluorescence over basal fluorescence (ΔF/F) as described previously (Parsons et al., 2016). Videos were analyzed offline with ImageJ software. The average iGluSnFR signal was measured over 10 × 10 pixel (93.8 × 93.8 μm) ROI placed over the maximal area of evoked iGluSnFR activity within the striatum adjacent to the stimulating electrode. To normalize for potential differences in iGluSnFR expression between slices, in subsets of experiments we applied a near saturating dose of glutamate (10 mm) in the presence of TTX (500 nm), kyneurenate (100 μm), and DL-TBOA (10 μm) to slices following evoked iGluSnFR experiments. Evoked iGluSnFR measurements and the subsequent exogenous glutamate calibrations were performed under identical excitation LED intensity. Peak evoked DF/F responses were normalized to the mean iGluSnFR fluorescence grayscale intensity over the analysis ROI following the plateau of this signal in the continued presence of the glutamate calibration solution.
Experimental design and statistical analysis
Statistical analysis and creation of figures were performed using GraphPad Prism (version 7). All data distributions were tested for normality with the D'Agostino-Pearson omibus normality test.
The Student's unpaired t test was used for unpaired comparisons between two data groups, such as when mean mEPSC frequencies were compared between WT and YAC128 neurons, as long both data groups were normally distributed. When one or both groups failed, the D'Agostino-Pearson omibus normality test, the nonparametric Mann–Whitney test was used instead.
When parameters of the same group of neurons or axonal boutons were compared before and after a drug treatment, statistical significance was assessed with the Student's paired t test, unless data points in the control or drug-treatment group failed the D'Agostino-Pearson omibus normality test, in which case, the nonparametric Wilcoxon matched-pairs signed rank test was used instead.
A two-way ANOVA with the Bonferroni post test (as appropriate) was used when testing for genotype differences in a dependent variable measured at different time points, as was the case for our brain-slice iGluSnFR experiments.
Depending on the experimental design, “n” numbers in figures refer to number of neurons, numbers of imaged culture fields, numbers of individual axonal boutons, or numbers of brain slices. Clarifying details are present within individual figure legends, as are the numbers of culture batches or mice used.
Differences in mean values were considered significant at p < 0.05. Comprehensive descriptions of statistical analysis are included in the figure legends.
Results
mEPSC frequencies are elevated in YAC128 striatal and cortical neurons at early time points
Our group previously found that mEPSCs were significantly more frequent, compared with WT, in striatal SPNs from yeast artificial chromosome (YAC128) mouse-derived cortical-striatal cocultures at DIV14. This genotype difference was not seen at DIV 21, at which point higher presynaptic miniature glutamate release rates may have been masked by a significant reduction in YAC128 SPN total dendritic length relative to WT (and therefore reduced total excitatory synapse numbers) (Buren et al., 2016).
The above data indicated a higher rate of AP-independent glutamate release from YAC128 presynaptic CPN terminals at early time points. To follow-up on these findings, we capitalized on a relatively simpler culture preparation containing only cortex-derived neurons, along with YAC128 mouse-derived cortical-striatal brain slices, to mechanistically dissect how mutant huntingtin protein expression affects AP-independent and -dependent glutamate release from CPN terminals.
To determine whether miniature glutamate release is enhanced at YAC128 CPN terminals targeting other CPNs, we measured CPN mEPSCs in WT and YAC128 cortical cultures with patch-clamp recordings at a series of DIV time points: 7, 14, 18, and 21. Both genotypes showed increased mEPSC frequencies as neurons matured and formed synaptic connections with time in culture. However, YAC128 CPN mEPSC frequencies were consistently higher than age-matched WT controls after DIV7 and up until DIV21, at which point mEPSC frequencies became comparable between genotypes (Fig. 1A–L), as was also the case in the cortico-striatal coculture model (Buren et al., 2016). The greatest genotype mEPSC difference occurred at DIV14, when mean YAC128 CPN mEPSC frequencies were more than double that of WT: 10.83 ± 1.80 Hz versus 4.52 ± 0.60 Hz, respectively (Fig. 1D,E). YAC128 mEPSC frequencies remained significantly higher than WT at DIV18: 14.07 ± 1.54 Hz versus 9.92 ± 1.54 Hz, respectively (Fig. 1G,H). Age-matched YAC128 and WT CPNs showed similar mEPSC amplitudes across all DIV time points examined (Fig. 1C,F,I,L). We chose to perform subsequent experiments probing the mechanistic details underlying increased cortical mini glutamate release in DIV18 cortical monocultures, since this was the most mature culture stage at which mEPSC frequencies were elevated in YAC128 CPNs.
To test whether mHTT expression similarly affects mini glutamate release in a more physiologically representative preparation, we measured SPN mEPSCs in ex vivo cortical-striatal brain slices prepared from 2- to 4-month-old WT and YAC128 mice, time points when behaviorally evident disease is modest, if not absent (Slow et al., 2003). In these slices, YAC128 SPN mEPSC frequencies were approximately triple that of WT cells: 13.09 ± 3.06 Hz versus 4.33 ± 0.45 Hz respectively, while event amplitudes did not differ significantly between genotypes (Fig. 1M–O).
Synapse numbers and dendritic complexity are similar in WT and YAC128 cultured cortical pyramidal neurons at DIV18
The higher mEPSC frequencies seen in DIV18 YAC128 CPNs suggest an increased presynaptic glutamate release probability. However, relative differences in synapse numbers could also account for this finding. To estimate numbers of synapses, we first expressed cytosolic GFP in a small proportion of neurons in WT and YAC128 cortical cultures and imaged full CPN dendritic arbors at DIV18. Sholl analysis revealed similar arborization patterns and total dendritic length in WT and YAC128 CPNs (Fig. 2A–D). In separate cultures, we expressed an internal GFP-tagged anti-PSD95 antibody (Gross et al., 2013) in a subset of neurons and immunostained for VGlut1 to identify glutamatergic synapses, and the GluA2 AMPAR subunit to identify functional synapses. Functional excitatory synapse numbers, defined here as GFP-labeled PSD95 puncta colocalized with VGlut1 and GluA2 immunofluorescent-labeled puncta, were not significantly different between DIV18 WT and YAC128 CPNs, although there was a trend toward lower synapse density in YAC128 CPNs (Fig. 2E,F). These results, together with the above data showing increased YAC128 CPN mEPSC frequencies at DIV18 (Fig. 1), point to increased miniature vesicular glutamate release from cortical terminals in YAC128 cultures.
Releasing ER calcium with low-dose ryanodine or caffeine increases the mEPSC frequency in WT, but not YAC128 cultures
Studies using mouse models suggest that Ca2+ release from ER stores is aberrant in HD because of increased IP3 and ryanodine receptor activity (Tang et al., 2003; Suzuki et al., 2012). Although effects of mHTT on the presynaptic ER have not been specifically studied, we hypothesized that a presynaptic ER Ca2+ leak contributes to the elevated miniature glutamate release seen in our YAC128 cultures. To test this hypothesis, we first recorded mEPSCs before and during local application of low-dose (5 μm) ryanodine to cultured CPNs; 5 μm ryanodine releases ER Ca2+ by opening ryanodine receptors (Meissner, 2017). In WT cultures, 5 μm ryanodine nearly doubled the CPN mEPSC frequency: from 6.74 ± 1.17 Hz to 11.91 ± 2.36 Hz [Students' paired t test; p = 0.0280; n = 13 cells] (Fig. 3A–C). However, in YAC128 cultures, ryanodine (5 μm) did not significantly alter the CPN mEPSC frequency: 11.98 ± 1.66 Hz versus 13.89 ± 2.01 Hz, under control conditions and in ryanodine (5 μm), respectively (Fig. 3D–F), suggesting that the effect of ryanodine was occluded in YAC128 cultures. A comparison of the percent change in mEPSC frequency following ryanodine (5 μm) revealed a significantly greater response in WT than in YAC128 CPNs [31.7 ± 7.5% (n = 13) versus 5.2 ± 9.5% (n = 15) in WT and YAC128 CPNs, respectively (Students' unpaired t test; p = 0.0418)]. We next used caffeine (1 mm) as an alternative means of agonizing ryanodine receptors; this more prominently increased the mEPSC frequency in WT cultures but otherwise produced similar results [WT: 99.2 ± 34.1% increase in mEPSC frequency (n = 16; 8 cultures), versus YAC128: 1.2 ± 7.6% increase in mEPSC frequency (n = 13; 5 cultures)] [p = 0.0025 (exact); Mann–Whitney test] (Fig. 3G–L). Neither ryanodine nor caffeine altered mEPSC amplitude in either genotype (Fig. 3C,F,I,L).
YAC128 mEPSCs are more sensitive to removal of extracellular Ca2+ but resistant to chelation of intracellular Ca2+
We next reasoned that, if an ongoing release of presynaptic ER Ca2+ in YAC128 cultures occludes potentiation of miniature glutamate release by low-dose ryanodine and caffeine, inhibiting ER Ca2+ release should lower YAC128 miniature event frequencies to WT levels. To test this, we pre-incubated cultures with ECF containing the sarco/endoplasmic reticulum Ca2+-ATPase (SERCA) pump inhibitor cyclopiazonic acid (CPA) (30 μm) to deplete ER Ca2+ stores. Since ER Ca2+ depletion can increase cytosolic Ca2+ by engaging the SOC response, and this effect can increase mini neurotransmitter release (Emptage et al., 2001; Chanaday et al., 2021), we performed these experiments in the absence of extracellular Ca2+. Surprisingly, under these conditions, mean mEPSC frequencies were significantly lower in YAC128 CPNs compared with WT [3.33 ± 0.57 Hz (n = 14) vs 8.41 ± 1.50 Hz (n = 17), respectively (Student's unpaired t test; p = 0.0062)] (Fig. 4A–C), suggesting that YAC128 mini glutamate release is far more dependent on extracellular Ca2+ and/or release from ER stores. If the above conditions fully deprived neurons of presynaptic Ca2+, this result might indicate a lower intrinsic (Ca2+-independent) synaptic vesicle release probability in YAC128, versus WT, cultures. However, remaining synaptic events can still be Ca2+-dependent, as pharmacological SERCA pump inhibition has been reported to only partially deplete the axonal ER Ca2+ (de Juan-Sanz et al., 2017). Alternatively, mini release in YAC128 cultures could be more dependent on extracellular Ca2+ because of a greater role of CaSR signaling (Vyleta and Smith, 2011) or SOC-mediated Ca2+ entry (Chanaday et al., 2021). To address these possibilities, we loaded WT and YAC128 cortical cultures with the cell membrane-permeant Ca2+ chelator BAPTA-AM and measured CPN mEPSCs in standard ECF (after extracellular BAPTA-AM washout). WT CPN mEPSC frequencies were significantly more sensitive to reduced intracellular Ca2+ than YAC128 CPNs (16.9 ± 4.6% that of vehicle controls in WT (n = 10) vs 50.2 ± 11.8% in YAC128 (n = 9); Students' unpaired t test; p = 0.0142) (Fig. 4D,E). Their near complete block by BAPTA-AM indicates that WT mEPSCs are highly dependent on intracellular Ca2+, whereas YAC128 mEPSCs are far less so. Presumably, this reduced intracellular Ca2+ dependency also means that the substantial reduction of YAC128 mEPCS frequencies in CPA and 0 Ca2+ was due primarily to the absence of extracellular Ca2+ rather than pharmacological SERCA pump inhibition.
Presynaptic Ca2+ sparks and waves are altered in YAC128 cultures
To directly monitor presynaptic Ca2+ dynamics, we next exploited a genetically encoded Ca2+ sensor that preferentially localizes to presynaptic terminals, created by fusing GCaMP6-M with the rat synaptophysin protein via a small glycine-serine linker (rSyph-GCaMP6m) (Chen et al., 2013). We expressed rSyph-GCaMP6m in cultured cortical neurons and performed Ca2+ imaging experiments, first in the presence of TTX (500 nm) to relate presynaptic Ca2+ signaling to our mEPSC findings (above). For clarity, we will refer to axonal Ca2+ events in TTX (AP-independent) as miniature, and events present in the absence of TTX (both activity-dependent and -independent) as spontaneous. Both YAC128 and WT cultures showed miniature axonal Ca2+ events, as well as slow Ca2+ waves traversing multiple neighboring boutons (Fig. 5A,B). In some cases, these axonal waves traveled extensively, crossing entire 63× (178.6 µm × 113.1 µm) imaging fields over multiple seconds; in contrast, spontaneous Ca2+ events seen in the absence of TTX (i.e., AP-dependent) could not be temporally resolved owing to relatively slow GCaMP kinetics and appeared simultaneously across all imaged boutons of a given axon. Miniature presynaptic Ca2+ events imaged in TTX were also strikingly long lasting, on average 5-10 times longer than typical events seen in the absence of TTX (Fig. 5C,D). We first quantified these AP-independent miniature axonal Ca2+ events with the AQuA, an ROI-agnostic algorithm that detects waves as suprathreshold signals connected in time and space (Wang et al., 2019). Such waves varied considerably between culture batches; and although more were detected on average in YAC128 cultures, this effect did not reach statistical significance (Fig. 5E). Interestingly, Ca2+ waves tended to travel further in WT than YAC128 axons, reflected in a significantly larger mean event area (Fig. 5F).
In subsets of WT and YAC128 cultures, we identified boutons from presumptive glutamatergic synapses, by coexpressing an mCherry-tagged PSD95 construct and restricting GCaMP analysis to rSyph-GCaMP6m-labeled boutons clearly colocalized with mCherry-labeled puncta (Fig. 6A). As in the larger rSyph-GCaMP6m-labeled bouton population (including those of GABA interneurons), numerous PSD95 colocalized boutons were active in the presence of TTX in cultures of both genotypes (Fig. 6B,C). To systematically compare miniature event amplitudes between genotypes in these experiments, we capitalized on the ability to measure GCaMP fluorescence independent of its Ca2+-bound state by exciting it with light at its isosbestic frequency (∼405 nm) (Barnett et al., 2017). By normalizing GCaMP traces to isosbestic fluorescence intensity, we thereby accounted for any differences in GCaMP expression between cultures and individual boutons. These experiments revealed the amplitudes of miniature Ca2+ events were, on average, 32% smaller in YAC128 boutons (Fig. 6D). YAC128 miniature events were also significantly shorter in duration when measured at half amplitude (Fig. 6E). In contrast to analysis using the AQuA algorithm (above), miniature event frequencies were slightly, but significantly, lower in mCherry-colocalized YAC128 boutons (Fig. 6F). However, inspection of the Ca2+ event amplitude distributions from both genotypes revealed highly skewed distributions with modes (or distribution peaks) very near the event-amplitude detection threshold (Fig. 6G). These results suggest a significant number of smaller amplitude events may not have been detected amid trace noise and that YAC128 cultures would be more susceptible to undercounting of such events, given the smaller average YAC128 event amplitude.
Responses to the Ca2+ ionophore ionomycin are reduced in YAC128 presynaptic boutons
The brightness of GCaMP fluorescence at a given excitation light intensity is determined both by its expression level and the degree of its Ca2+ binding. Quantitative Ca2+ concentration comparisons are therefore best made when an additional calibration step is performed to account for differential GCaMP expression at individual ROIs. To do so, in subsets of experiments we permeabilized neuronal membranes to Ca2+ with the Ca2+ ionophore ionomycin following GCaMP imaging. This theoretically allowed GCaMP fluorescence to be measured in the presence of the known extracellular Ca2+ concentration (2 mm for these experiments) and for differences in GCaMP expression to be inferred based on the differential fluorescence intensity of ROIs under these conditions.
Boutons in YAC128 cultures responded to ionomycin (10 μm) application with significantly smaller fluorescence intensity increases (DF/F) compared with WT (Fig. 7A–C), a finding that would be consistent with higher resting cytosolic Ca2+ in YAC128 boutons (Lindhout et al., 2019). Furthermore, in both genotypes, ionomycin-mediated changes in presynaptic GCaMP fluorescence were significantly smaller in boutons showing at least one miniature Ca2+ event in the previous 3 min recording (Fig. 7D–F), suggesting that higher basal cytosolic Ca2+ concentrations occur in a population of boutons undergoing TTX-resistant Ca2+ events. Interpretations were confounded, however, by the observation that cytosolic Ca2+ responses typically peaked following ionomycin application then rapidly declined (Fig. 7A,B), inconsistent with the plateau expected if cytosolic bouton Ca2+ levels truly equilibrated with the 2 mm extracellular Ca2+ concentration. Several studies suggest a more complex mechanism of ionomycin's actions on cytosolic Ca2+ concentrations, with a critical involvement of ER Ca2+ release and subsequent SOC entry (Morgan and Jacob, 1994; Müller et al., 2013). Moreover, some evidence suggests that ionomycin's mechanism of action varies considerably with its concentration. We therefore repeated the above experiments with a higher (60 μm) ionomycin dose, and in subsets of these experiments, recorded for a longer duration following ionomycin treatment (6-8 vs 2 min) to see if Ca2+ responses eventually plateaued. Despite a 6 × increase in dose, we observed similar results to those seen with the 10 μm ionomycin (Fig. 7G,H).
GCaMP isosbestic point measurements suggest that resting cytosolic Ca2+ levels are not elevated in YAC128 presynaptic boutons
We next used isosbestic measurements as an alternative means to calibrate GCaMP expression at individual boutons (as above) and to test whether reduced GCaMP Ca2+ responses to 10 and 60 μm ionomycin were best explained by depleted bouton ER Ca2+ stores or indeed reflect higher resting cytosolic Ca2+ concentrations in YAC128 boutons. We first verified that GCaMP isosbestic point fluorescence intensity measurements were independent of the cytosolic Ca2+ concentrations under our imaging conditions. In recordings where we rapidly switched between isosbestic (405 nm excitation) and conventional GCaMP (480 nm excitation) measurements, isosbestic time courses showed similar bleaching dynamics to that of the corresponding GCaMP trace, but were of a lower fluorescence intensity and importantly lacked the Ca2+ transients seen in the GCaMP traces (Fig. 7I). Transient artifacts were commonly seen in the isosbestic channel following ionomycin (60 μm) application. These manifested as either small increases in signal (as in Fig. 7I) that occurred throughout the entire imaging field, or a small reduction in signal because of bouton swelling or movement causing some GCaMP signal to move out of the analysis ROI. In either case, normalizing GCaMP traces to the isosbestic channel helped to compensate for these artifacts; and subsequently, ionomycin responses typically did plateau (Fig. 7J,K). However, this ionomycin response plateau was still typically preceded by a substantially larger rise in Ca2+. Furthermore, if ionomycin application freely permeabilized the cell membrane to Ca2+ (thus equilibrating intracellular and extracellular compartments), GCaMP/isosbestic measurements in ionomycin should depend only the extracellular Ca2+ concentration and not differ between genotypes. However, GCaMP/isosbestic values following ionomycin (60 μm) remained significantly larger in WT than YAC128 boutons (Fig. 7L).
Surprisingly, resting GCaMP/isosbestic measurements were significantly lower in YAC128 boutons (Fig. 7J,K,M), a result suggesting lower basal cytosolic Ca2+, as opposed to the higher resting cytosolic Ca2+ concentrations expected. Together, these results suggest that reduced ionomycin responses in YAC128 boutons are because of decreased ER Ca2+ content and not differences in the cytosolic Ca2+ concentration. Reduced ionomycin responses in boutons showing miniature Ca2+ events in both genotypes may therefore reflect ER Ca2+ depletion because of ongoing ER release.
Caffeine increases basal Ca2+ and miniature events in WT, but not YAC128 cortical boutons
The slow kinetics of the above miniature presynaptic Ca2+ events are consistent with ER-mediated Ca2+ waves reported in postsynaptic neuronal compartments (Ross, 2012). We next tested whether these presynaptic Ca2+ events were affected by caffeine (1 mm), an ER ryanodine receptor agonist that substantially increased mEPSC frequencies in WT, but not YAC128 cultures (above). In the presence of TTX, caffeine (1 mm) significantly increased the presynaptic Ca2+ event frequency in WT cultures (Fig. 8A) but did not significantly alter event frequencies in YAC128 cultures (Fig. 8B).
When exploring the effects of caffeine (and other pharmacological agents; see below) on miniature axonal Ca2+ events, we compared separate imaging fields to minimize impacts of photobleaching and toxicity. However, we observed relatively rapid changes in basal rSyph-GCaMP6m fluorescence in response to caffeine (1 mm) application, conducive to within-bouton measurements. A clear increase in basal rSyph-GCaMP6m fluorescence was seen in most WT boutons following caffeine (1 mm) application, which was attenuated in YAC128 boutons (Fig. 8C). This caffeine (1 mm)-mediated increase in rSyph-GCaMP6m fluorescence was significantly greater in WT than in YAC128 boutons (Fig. 8D), consistent with depleted ER Ca2+ stores in YAC128 CPNs, as suggested by the results of the ionomycin experiments.
CPA increases miniature axonal Ca2+ event frequencies and reduces ionomycin responses in WT, but not YAC128 cortical boutons
Pharmacological inhibition of the ER SERCA pump has been shown to increase presynaptic cytosolic Ca2+ concentrations and potentiate miniature glutamate release. We next tested whether SERCA pump inhibition with CPA differentially impacts axonal Ca2+ in WT versus YAC128 cultures. WT cultures incubated with CPA (30 μm) in the presence of TTX (500 nm) showed significantly more miniature axonal Ca2+ events (vs experiments in TTX alone) (Fig. 8E), whereas axonal Ca2+ event frequencies did not significantly differ between YAC128 cultures in CPA and TTX (vs TTX alone) (Fig. 8F). In WT cultures pre-incubated with CPA (30 μm) and TTX, responses to ionomycin (60 μm) were significantly reduced by 43% (vs responses in TTX alone) (Fig. 8G). Conversely, pre-incubation with CPA did not significantly alter ionomycin responses in YAC128 boutons (Fig. 8H), which showed similar amplitude responses to WT after CPA treatment. These results further support that ER Ca2+ is depleted in YAC128 cortical axons. As expected, CPA also increased cytosolic Ca2+ in WT boutons, as reflected by increased resting GCaMP/isosbestic values (data not shown). However, this effect was more complex in 2 of 3 cultures tested, in which the GCaMP/isosbestic decreased substantially below the baseline value following prolonged (>60 min) CPA incubation.
Activity-dependent Ca2+ transients in presynaptic cortical terminals are less frequent in YAC128 cultures because of reduced AP firing rates
We next used the rSyph-GCaMP6m construct in the absence of TTX to examine presynaptic AP-dependent Ca2+ transients in WT and YAC128 cortical cultures. When neuronal AP firing was intact, rSyph-GCaMP6m-expressing boutons in both WT and YAC128 cultures were dominated by presumably VGCC-mediated signals (Fig. 9A,B). These spontaneous signals were more frequent than those seen in the presence of TTX, but of far shorter duration. Interestingly, these spontaneous Ca2+ events were nearly twice as frequent in WT axonal boutons, compared with those in YAC128 cultures (Fig. 9A,B,E). Ryanodine (5 μm) modestly but significantly reduced the frequency of these events in WT cultures (by 17%) (Fig. 9A,C,E). Conversely, 5 μm ryanodine elicited a small, but significant increase in the frequency of these events in YAC128 cultures (Fig. 9B,D,E). Although it has been reported that axonal ER Ca2+ can be released following AP-mediated presynaptic Ca2+ influx (Emptage et al., 2001), amplitudes of AP-dependent Ca2+ events were not significantly lower in YAC128 boutons, as might be expected if ER Ca2+ was a major contributor to these signals in WT, but not YAC128, boutons. Alternatively, reduced AP-dependent axonal Ca2+ event frequencies could be explained by lower YAC128 CPN AP firing rates. To test this possibility, we expressed cytosolic GCaMP 7f in cultured cortical neurons and recorded AP-dependent Ca2+ events from the soma of transfected CPNs. Frequencies of such activity-dependent somatic signals were significantly lower in YAC128 versus WT CPNs (Fig. 9F,G), suggesting that lower activity-dependent axonal Ca2+ event frequencies in YAC128 cultures are mediated by decreased CPN AP firing rates rather than a specifically axon-localized mechanism. Activation of a Ca2+-dependent K+ conductance following release of ER Ca2+ stores is a plausible, although highly speculative, explanation for reduced YAC128 CPN AP firing and ryanodine's effects on WT spontaneous axonal Ca2+ events. The small but significant reduction in spontaneous Ca2+ events in YAC128 boutons following ryanodine treatment is more puzzling, but perhaps explained by ryanodine-mediated RyR closure predominating in YAC128 CPN, a plausible scenario if the YAC128 RyR population normally favors the open conformation.
Together, experiments up to this point suggest that ER Ca2+ depletion in YAC128 cultures elevates miniature vesicular glutamate release but may also contribute to reduced AP firing rates, which should ultimately reduce activity-dependent glutamate release.
Evoked glutamate release does not differ between YAC128 and WT cortical-striatal brain slices
We next performed experiments in cortical-striatal ex vivo brain slices prepared from 2- to 4-month-old WT and YAC128 mice expressing the fluorescent glutamate sensor iGluSnFR in striatal neurons (Parsons et al., 2016; Koch et al., 2018). This preparation allowed direct optical measurement of glutamate release in the striatum, independent of postsynaptic neuronal properties. Striatal iGluSnFR signals evoked by stimulating cortical axons of the corpus callosum (Fig. 10A,B) were significantly, albeit modestly, decreased by ryanodine (5 μm) in WT, but not YAC128 slices (Fig. 10C,D). These results suggested that evoked glutamate release may be reduced in YAC128 mice by an ongoing release of axonal ER Ca2+ stores, which also occludes YAC128 responses to low-dose ryanodine. To directly compare evoked cortical glutamate release between genotypes, we next measured striatal-evoked iGluSnFR responses, at a variety of stimulation intensities, then normalized these responses to a near saturating dose of exogenously applied glutamate at the end of the experiment (under like illumination), to account for any differences in iGluSnFR expression between brain slices, animals, and genotypes. However, these experiments revealed similar evoked glutamate responses between genotypes (Fig. 10E,F). de Juan-Sanz et al. (2017) found that axonal ER depletion attenuated evoked glutamate release from hippocampal neurons, but that this effect was only apparent at physiological temperature. As experiments up to this point were performed at room temperature, we also examined striatal-evoked iGluSnFR responses at 34°C. However, we continued to observe no significant difference between WT and YAC128 slices at near physiological temperature (Fig. 10G), suggesting that equivalent cortical AP firing evokes similar glutamate release in the YAC128 and WT striatum at this young age.
Discussion
Presynaptic neurotransmitter release, in concert with postsynaptic signaling, underlies most between-neuron communication and can be divided into AP-dependent and -independent forms. The former requires sodium AP-mediated presynaptic VGCC activation, while the latter persists without neuronal activity. Although miniature neurotransmission is relatively poorly understood, its physiological relevance is increasingly accepted (McKinney et al., 1999; Frank et al., 2006; Sutton et al., 2006). Moreover, miniature release is regulated relatively independently of its AP-dependent counterpart and may use distinct vesicular pools (Sara et al., 2005; Fredj and Burrone, 2009), targeting different postsynaptic receptors (Atasoy et al., 2008).
Altered synaptic signaling, particularly at glutamatergic excitatory synapses, has been reported in HD models (Raymond et al., 2011; Sepers et al., 2018; Tyebji and Hannan, 2017; Cepeda and Levine, 2022) and other neurodegenerative diseases (Wang and Reddy, 2017). Increased SPN extrasynaptic NMDAR expression favors excitotoxic postsynaptic glutamate signaling in HD models (Milnerwood et al., 2010; Dau et al., 2014). Mounting evidence also suggests aberrant glutamate release from cortical afferents in HD; however, the direction of this alteration is disease stage-dependent (Cepeda et al., 2003; Joshi et al., 2009).
The ER, a continuous intracellular membrane-system involved in Ca2+ storage and protein synthesis, is found in all neuronal processes, including axons and presynaptic boutons. Increased ER-to-cytosol Ca2+ release has been shown in HD models because of enhanced IP3R responsiveness (Tang et al., 2003) and a constituent RyR Ca2+ leak (Suzuki et al., 2012). Although previous HD studies have focused on the postsynaptic ER, presynaptic ER Ca2+ release is reported to modulate neurotransmission (Llano et al., 2000; Emptage et al., 2001). Our results suggest that mHTT depletes ER Ca2+ stores proximal to cortical presynaptic terminals, occluding pharmacological facilitation (by RyR activation) of miniature postsynaptic (glutamate) and presynaptic (Ca2+) event frequencies; this underlying ER leak presumably mediates increased basal YAC128 mini glutamate release. However, our results paradoxically suggest that resting cytosolic Ca2+ levels are lower in YAC128 than WT boutons, while YAC128 mini axonal Ca2+ events were smaller in amplitude and similar in frequency to WT. Furthermore, YAC128 mEPSCs were less sensitive to chelation of intracellular Ca2+. The ultimate mechanism by which ER Ca2+ dysfunction elevates miniature glutamate release from YAC128 boutons is therefore more complex than simply providing cytosolic Ca2+ for vesicular release.
AP-independent cortical glutamate release is selectively elevated in the YAC128 model
No genotype differences were seen in evoked striatal glutamate release in 2- to 4-month-old mice brain slices, despite substantially higher YAC128 SPN mEPSC frequencies, suggesting selective potentiation of miniature glutamate release at this early stage. Others reported altered evoked cortical glutamate release in YAC128 brain slices; however, genotype differences were complex and disease-stage dependent, with early increases followed by reduced release in aged mice (Joshi et al., 2009). Perhaps our experiments were performed during the transition between juvenile and aged phenotypes, obscuring genotype differences. In any case, our results support a dissociation between mHTT's modulation of miniature and evoked glutamate release, consistent with differential regulation of these two forms of neurotransmission. As we have focused exclusively on glutamatergic synapses, future experiments are required to determine whether other neurotransmitter systems are similarly affected.
Elevated YAC128 mEPSC frequencies are mediated by increased glutamate release and not differences in synapse numbers
YAC128 CPNs showed higher mEPSC frequencies at DIV14 and DIV18 (vs WT cultures). Striatal SPNs (which receive substantial excitatory cortical innervation) likewise showed higher mEPSC frequencies in acute brain slices from young YAC128 mice, indicating this phenotype is not an artifact of culture conditions. However, CPN mEPSC frequencies were similar between genotypes by DIV21, consistent with findings of Buren et al. (2016) in a cortical-striatal coculture model, who attributed this to reduced YAC128 SPN total dendritic length (thus, reduced overall synapse numbers). Notably our DIV18-aged cortical cultures, used for mechanistic experiments, showed no genotype differences in CPN synapse density or dendritic morphology, suggesting that differential mEPSC frequencies reflect altered presynaptic release rates. We suspect that increased miniature cortical glutamate release is an early disease feature, most evident in the postsynaptic mEPSC frequency before loss of synaptic connections.
Altered axonal Ca2+ waves in YAC128 cultures cannot fully explain differences in miniature glutamate release
RyR agonism, with caffeine or low-dose ryanodine, failed to increase the mEPSC frequency in YAC128 CPNs, despite robustly increasing WT frequencies. This result suggests tonic presynaptic ER Ca2+ release chronically elevates mini release in YAC128 cultures, occluding further facilitation. Consistent with this mechanism, caffeine significantly increased the frequency of axonal Ca2+ waves in WT, but not YAC128 cultures. Similarly, pharmacological inhibition of the ER SERCA pump, known to increase miniature glutamate release in cultured hippocampal neurons (Emptage et al., 2001; Chanaday et al., 2021), increased axonal Ca2+ wave frequency and transiently increased cytosolic Ca2+ levels in WT, but not YAC128 boutons. Since ionomycin's mechanism of action involves ER Ca2+ release and subsequent SOC influx (Morgan and Jacob, 1994; Müller et al., 2013), reduced ionomycin responses in YAC128 boutons also suggest decreased YAC128 axonal ER Ca2+ content, as previously reported in YAC128 SPNs (Tang et al., 2003). Interestingly, in 0 extracellular Ca2+, WT and YAC128 boutons showed similarly large ionomycin response magnitudes (data not shown), perhaps indicating ER Ca2+ depletion in YAC128 boutons requires sufficient extracellular Ca2+ to initiate IP3 and or ryanodine receptor-mediated Ca2+-induced ER Ca2+ release with effects of mHTT otherwise masked. Alternatively, CaSRs, which are implicated in mini release, are often Gq-type GPCR coupled, mediating their signaling in part via cytosolic IP3 release; it is possible this mechanism is upregulated in YAC128 axonal boutons (discussed below).
Amplitudes of axonal Ca2+ signals were significantly lower in YAC128 boutons. Likewise, Ca2+ waves were more spatially restricted, consistent with reduced YAC128 ER Ca2+ content decreasing Ca2+ release per individual axonal event. Unexpectedly, miniature axonal wave frequencies were not significantly different between genotypes when measured with the AQuA algorithm, while analysis of mCherry-PSD95 colocalized boutons with our in-house software found significantly lower YAC128 Ca2+ event frequencies. However, lower average Ca2+ event amplitudes may have caused more events to evade detection in YAC128 cultures, obscuring the expected frequency difference.
Neuronal ER Ca2+ waves are predominantly mediated by regenerative IP3R activity (Nakamura et al., 1999; Larkum et al., 2003; Hagenston et al., 2008); Ca2+ released from RyR may modulate these processes (Miyazaki and Ross, 2013), consistent with our data. Spatially restricted, low-amplitude Ca2+ puffs from individual IP3R clusters can induce propagating ER waves but can also remain localized, depending on IP3 concentrations, cytosolic Ca2+ concentrations, and activity of neighboring channel clusters. Increased IP3R and/or RyR responsiveness in YAC128 cultures might increase axonal wave generation. Conversely, ER Ca2+ depletion (because of ongoing release) could also reduce the amplitude of individual Ca2+ puffs, such that neighboring receptor cluster activation is impeded and wave generation reduced. Reduced axonal Ca2+ event amplitude and area are more consistent with this latter possibility. An increase in the frequency of below-detection-threshold ER Ca2+ release events (which would cause ER Ca2+ depletion) in concert with reduced complex wave generation, potentially explains our findings. However, based on the reduced sensitivity of YAC128 CPN mEPSCs to BAPTA-AM, any such subdetection-threshold ER release events are unlikely to account for higher YAC128 mEPSC frequencies.
Resting cytosolic Ca2+ concentrations are not higher in YAC128 boutons
Surprisingly, basal GCaMP intensity normalized to isosbestic fluorescence was lower in YAC128 boutons, suggesting reduced cytosolic Ca2+ concentrations compared with WT. Future experiments using ratiometric Ca2+ sensors are needed to definitively test this result. Nonetheless, this refutes the possibility that cytosolic Ca2+ concentrations are increased in YAC128 boutons.
It is generally assumed that mHTT elevates cytosolic Ca2+ concentrations in vulnerable neuron types. However, conclusions to this effect have typically been derived from differential responses to pharmacological manipulations, rather than direct cytosolic Ca2+ concentration comparisons. One study reported fivefold higher cytosolic Ca2+ concentrations in R6/2 mouse SPNs compared with WT, based on ratiometric fura-2 measurements (Hansson et al., 2001); however, neurons were dialyzed with fura-2-containing intracellular solution, presumably altering intracellular signaling and neuronal Ca2+ buffering. Additionally, most studies of neuronal Ca2+ handling in HD models have focused on striatal SPNs, raising the possibility of differential Ca2+ handling in CPNs or specifically in axonal boutons. However, given the exquisite spatial/temporal regulation of Ca2+ signaling, it is not surprising that compensatory processes would be engaged to prevent increased spontaneous ER Ca2+ release from chronically elevating cytosolic Ca2+. We speculate that increased mitochondrial Ca2+ uptake or enhanced extrusion via the plasma membrane Ca2+ ATPase or Na+/Ca2+ exchanger underlies such compensation and merits future study.
We found substantial evidence that axonal ER stores are depleted in YAC128 cortical neurons, and that this is associated with increased miniature glutamate release. However, cytosolic Ca2+ concentrations were not elevated in YAC128 boutons and YAC128 mini glutamate release was less dependent on intracellular Ca2+, suggesting that increased mini glutamate release is not directly mediated by increased ER-to-cytosol Ca2+ release or an enhanced SOC response. mEPSC frequencies in YAC128 CPNs were, however, more sensitive to extracellular Ca2+, consistent with CaSR upregulation in YAC128 cortical terminals mediating enhanced mini glutamate release. This possibility can be tested in future experiments.
Implications for postsynaptic signaling
Miniature glutamate release can mediate differential postsynaptic signaling (Sutton et al., 2007), and elicits postsynaptic NMDAR-mediated Ca2+ influx under physiologically relevant conditions (Espinosa and Kavalali, 2009; Beaulieu-Laroche and Harnett, 2018). In cultures, miniature glutamate-mediated events can become toxic to CPNs following prolonged silencing of neuronal activity with TTX (Fishbein and Segal, 2007). Future studies will be necessary to determine if and how the increased mini glutamate release shown here interacts with the well-described alterations in postsynaptic NMDAR expression in the YAC128 and other HD mouse models, and whether enhanced activity-independent glutamate release contributes to neurodegeneration in HD.
Footnotes
This work was supported by resources made available through the Dynamic Brain Circuits cluster, the DataBinge forum, and the NeuroImaging and NeuroComputation Center at the University of British Columbia (UBC) Djavad Mowafaghian Center for Brain Health (RRID:SCR_019086); and Canadian Institutes of Health Research (CIHR) Foundation Grants FDN-143210 to L.A.R. and FDN-154278 to M.R.H and Project Grant PJT-178043 to L.A.R. J.P.M. was supported by a Michael Smith Foundation for Health Research Trainee Award RT-2020-0614 and a Hereditary Disease Foundation Fellowship. C.B. was supported by a University of British Columbia 4-year Graduate Fellowship (UBC 4-YF). A.I.S.-D. was supported by a CIHR Canada Graduate Scholarship Doctoral award and a UBC 4-YF. E.T.K. was supported by a CIHR Canada Graduate Scholarship Master's award, Canadian Open Neuroscience Platform Scholar award, and UBC 4-YF. M.S. was supported by a Vanier Canada Graduate Scholarship and a UBC 4-YF. W.B.N. was supported by a UBC-CIHR-MD/PhD studentship and a Vanier Canada Graduate Scholarship. L.A.R. holds the UBC Department of Psychiatry Louise A. Brown Chair in Neuroscience. M.R.H. holds a Canada Research Chair. We thank Dr. Anne Marie Craig (UBC) for expert advice on experiments using rSyph-GCaMP6m; Pankaj Kumar Gupta for providing data analysis code; and Dr. Lily Zhang and Dr. Rujun Kang for technical support and assistance.
The authors declare no competing financial interests.
- Correspondence should be addressed to Lynn A. Raymond at Lynn.raymond{at}ubc.ca