YAC128 and wild-type (WT) littermate mice were maintained on the FVB/N background and experiments were performed according to protocols approved by the University of British Columbia Animal Care Committee (Protocol numbers A16-0130 and A16-0206). Heterozygous line 53 YAC128 mice or line W13 C6R mice (YAC128 mice bearing a D586A point mutation rendering mHTT resistant to caspase cleavage at this site) and their WT littermates were used for all experiments except neuronal culture transfections. For these experiments, homozygous line 55 YAC128 and FVB/N mice were used to remain consistent with previously published work (Milnerwood et al., 2012) as well as to improve culture health by avoiding an overnight genotyping step. For all cohorts, approximately equal numbers of male and female mice were used, and no differences in results were observed when assessed in each sex individually. All animals were sacrificed using CO₂.

Cortical or striatal total lysates were prepared by homogenizing snap-frozen tissue in stringent lysis buffer (20 mM HEPES pH 7.4, 150 mM NaCl, 40 mM β-Glycerophosphate, 10 mM NaF, 1% Triton-X, 1% SDS, 0.5% sodium deoxycholate) with protease and phosphatase inhibitors freshly added [1X Complete Protease Inhibitor Cocktail (Roche; Cat# 11697498001), 5 μM zVAD, 1 μg/ml Pepstatin A, 1 mM sodium orthovanadate]. Protein samples were sonicated, centrifuged to remove cellular debris, and quantified by DC assay (BioRad; Cat# 5000112). Equal amounts of protein from each sample were run by SDS-PAGE on 8 or 10% acrylamide gels and transferred to 0.45 μm nitrocellulose membranes. Membranes were blocked in 5% skim milk powder in TBST (0.1% Tween-20) for 1 h at room temperature (RT), blotted in primary antibody in 3% BSA/TBST for 2 h at RT, washed 4× in TBST, incubated with fluorescent secondary antibody for 1 h at RT, washed, and scanned with an LI-COR Odyssey imaging system. Band intensities were quantified in Image Studio Lite using top-bottom background subtraction. Measurements for proteins of interest were normalized to the intensity of a loading control which was probed on the same blot. All Western blot images within each figure panel were run on the same gel and cropped from a single scan image at identical brightness/contrast adjustments. Molecular weights (in kDa) are noted on the right side of each cropped blot image.

Subcellular fractionations from brain tissue were performed as previously described to isolate synaptic (post-synaptic density, PSD) and extrasynaptic (non-PSD) membranes, as well as nuclear fractions (Milnerwood et al., 2010). Striatal non-PSD preparations consistently produced low protein yield and unreliable detection of GluN2B phosphorylation which was not consistent between technical replicates from the same biological starting sample. Therefore, we chose an alternative method for evaluation of striatal “non-synaptic” pS1303 levels which has previously been described (Gladding et al., 2014), and provided much more reliable results due to decreased protocol time and lesser centrifugation and wash steps (and thus reduced protein/volume loss). We used the term “non-synaptic” throughout the “Results and Discussion” sections to distinguish where this modified method was used in place of standard fractionation (in which case the term “extrasynaptic fraction” is used). To generate non-synaptic fractions, tissue samples were homogenized in a gentle lysis buffer (20 mM HEPES pH 7.4, 150 mM NaCl, 40 mM β-Glycerophosphate, 10 mM NaF, 1% Triton-X) containing fresh protease and phosphatase inhibitors, but no SDS or sodium deoxycholate. This buffer is unable to lyse Triton-insoluble PSD membranes. Thus, after homogenization and high-speed centrifugation, the remaining supernatant was assumed to contain cytosol and non-synaptic membranes, including endosomes, while the insoluble pellet contained the synaptic PSD. Since GluN2B is a transmembrane protein, any GluN2B in the final non-synaptic protein sample was assumed to be present in membranes, rather than the cytosol. However, these non-synaptic membranes could represent either plasma membrane, internalized membrane vesicles, or both. To validate both our standard subcellular fractionation protocol as well as our modified lysis protocol, we confirmed that both methods resulted in the enrichment of the synaptic proteins PSD95 and GluN2B in the PSD fractions, as expected (Supplementary Figure 1). Synaptic, extrasynaptic, and “non-synaptic” protein fractions were quantified and used for Western blotting as described above. Antibodies and conditions used for Western blotting are listed in the following table:

Cortical tissue was homogenized in a gentle lysis buffer (50 mM Tris, 150 mM NaCl, 1% Igepal) with fresh protease inhibitors and pre-cleared with 10 μl of Protein G Dynabeads (Invitrogen; Cat# 10004D) for 1.5 h at 4°C. During the pre-clear step, 15 μl of Dynabeads per sample were washed once in lysis buffer, blocked with 3% BSA in lysis buffer, and conjugated to primary rabbit anti-DAPK1 antibody or control rabbit IgG (2.5 μg per sample) in blocking solution for 1 h at RT. Beads were washed twice in lysis buffer, added to pre-cleared lysates, and rotated overnight at 4°C. Beads were washed three times for 10 min in lysis buffer and protein was eluted in 1× NuPAGE LDS sample buffer (Invitrogen; Cat# NP0007) with 100 mM DTT at 70°C for 10 min. Samples were run by SDS–PAGE alongside previously set-aside input controls, transferred to nitrocellulose membranes, and blotted for DAPK1 and GluN2B using mouse primary antibodies. Co-immunoprecipitated GluN2B band intensity was normalized to the amount of immunoprecipitated DAPK1 for each sample.

Male WT and female YAC128 (line 53) mice were set up in mating pairs and provided either regular drinking water or water containing memantine. The concentration of memantine was calculated based on the weight of the female mouse and the estimated average consumption of 5 ml per day to administer a dose of 2 mg/kg/day. Final memantine concentrations were approximately 12 mg/l. Females were separated from males after 1 week, and pregnant females remained on either water or memantine treatment until pups were weaned at P20. At this time, male and female pups were separated and the average pup weight per cage was used to calculate a 1 mg/kg/day dose, assuming an average consumption of 4 ml per day per pup. Final memantine concentrations for pup treatment were approximately 2.5 mg/l. At 1 month of age, mice were sacrificed and brain tissue was collected for subcellular fractionation and Western blotting.

Autoclaved 12 mm No. 1 glass coverslips (Marienfeld Superior) were transferred to 24-well plates, and coated with 100 μg/ml high molecular weight poly-D-lysine (PDL) hydrobromide (Sigma–Aldrich; Cat# P6407) at RT overnight, washed 4x thoroughly with water, and air-dried before use. E16.5–17.5 embryos were removed from pregnant female mice, and brains were kept overnight at 4°C in Hibernate-E supplemented with L-glutamine (0.5 mM, Gibco; Cat# 25030) and B27 (Gibco; Cat# 17504) while excess embryonic tissue was used to perform genotyping. For dendritic spine experiments, cortical and striatal tissues were dissected separately in ice-cold Hank’s Balanced Salt Solution (Gibco; Cat# 14170). Dissected tissue was dissociated by gently pipetting once with a P1000 pipette, centrifuged, and enzymatically digested in 0.05% trypsin-EDTA (Gibco; Cat# 25300) for 8 min at 37°C. Trypsin was inactivated with 10% FBS in neurobasal medium (NBM; Gibco; Cat# 21103). Cells were then triturated gently 3–5× incomplete NBM (supplemented with L-glutamine, penicillin/streptomycin, and B27) containing DNase 1 (0.08 mg/ml) for further dissociation. Cells were centrifuged, resuspended in complete NBM, counted, and plated at a 1:3 cortical:striatal ratio with a final density of 160,000 cells per well. For evaluation of exogenous YFP-GluN2B surface expression, striatal neurons were nucleofected with a YFP-GluN2B construct (2 μg; a gift from Dr. Ann Marie Craig, University of British Columbia, Vancouver, BC, Canada) before being mixed with cortical neurons at a 1:1 ratio and plated in 500 μl 10% FBS/DMEM. The media was replaced with 500 μl of warm complete NBM after 3–4 h and topped up to 1 ml the following day. For biochemical experiments, cortical and striatal tissue were kept together during the culture process and seeded in 6-well PDL-coated plates at a density of 1 million cells per well. We estimate that the cortical:striatal ratio in these cultures was approximately 3:1–4:1. All cultures were fed with fresh complete NBM (25% well volume) every 6–7 days until DIV14 and every 3–4 days afterward unless undergoing drug treatments.

For imaging experiments, neurons grown on coverslips were treated every day from either DIV14 or DIV18 until DIV20. Ifenprodil hemitartrate (Tocris Bioscience; Cat# 0545) and memantine hydrochloride (Tocris Bioscience; Cat# 0773) were dissolved in sterile H₂O and used at a final treatment concentration of 3 μM. The DAPK1 inhibitor TC-DAPK6 (DKI; Tocris Bioscience; Cat# 4301) was dissolved in DMSO and used at a final concentration of 1 μM (0.01% DMSO). Cultures were fixed and immunostained at DIV21.

For biochemistry, culture media was replaced with conditioned media containing 10 μM DKI or DMSO (0.1%) for 1 h. Cells were collected by scraping immediately following treatment, centrifuged, and frozen at −20°C for later processing. Pellets were lysed by pipetting in gentle lysis buffer to solubilize only non-synaptic proteins. Samples underwent SDS–PAGE and Western blotting as described above for the detection of protein and phosphorylation levels.

For spine experiments, neurons on coverslips were fixed at DIV21 for 15 min at RT in 4% paraformaldehyde/PBS. Coverslips were washed 3× in PBS after fixation and between all subsequent incubation steps. Samples were incubated at −20°C in methanol for 5 min, permeabilized in 0.03% Triton-X/PBS at RT for 5 min, blocked in 0.2% gelatin/PBS at RT for 30 min, incubated with primary antibody against DARPP32 in blocking solution overnight at 4°C and fluorescent secondary antibody for 1.5 h at RT. Coverslips were mounted on microscope slides using Prolong Gold Antifade Reagent with DAPI (Invitrogen; Cat# P36935). Z-stack images of step size 0.6 μm were acquired using a Leica SP8 confocal microscope with a 63× objective lens. Representative MSNs for imaging were selected based on a healthy nuclear appearance and high DARPP32 expression level. In ImageJ, files were converted to 2D images with the maximum intensity Z-projection option, background-subtracted with a rolling ball radius of 35 pixels, and smoothed using the de-speckle function. De-speckling was necessary, as it dramatically improved the efficiency and accuracy of automatic spine detection in NeuronStudio (Version 0.9.92). Images were imported into NeuronStudio for semi-automatic spine identification and classification using at least three different secondary or tertiary dendritic segments per neuron.

Exogenous YFP-GluN2B surface expression was assessed as previously described (Milnerwood et al., 2012). In transfected, unstained cultures, the basal fluorescence of the YFP (which is N-terminal and thus expressed on the extracellular domain of GluN2B) is extremely low and difficult to detect. Transfected cultures grown on coverslips were thus live-stained at DIV21 to detect surface GluN2B by incubating with primary chicken GFP antibody in conditioned media for 10 min in a 37°C CO₂ incubator. Cells were rinsed once with warm conditioned media, fixed, and incubated with Alexa Fluor 488-conjugated goat anti-chicken secondary antibody for 1.5 h at RT. Samples were subsequently permeabilized, blocked, and stained as described above for internal YFP and VGLUT1 using rabbit and guinea pig primary antibodies, respectively. This was followed by fluorescent secondary antibody staining (Alexa Fluor 568-conjugated goat anti-rabbit and Alexa Fluor 647-conjugated goat anti-guinea pig). In this way, surface YFP-GluN2B could be measured via the Alexa Fluor 488 (green) intensity, while internal YFP-GluN2B could be detected by measuring the intensity of Alexa Fluor 568 (red). Coverslips were mounted on slides and transfected cells were imaged at 63× objective magnification in three channels at constant laser intensity across samples. Three representative secondary or tertiary dendritic segments were selected and the mean intensity value from each GFP channel within each selection was measured and averaged to generate a surface-to-internal ratio value as a measure of surface expression. The numbers of total YFP-GluN2B punctae and punctae colocalized with VGLUT1 were counted manually and normalized to the dendritic area to yield GluN2B punctae and synapse densities. For all imaging experiments, a total of at least 24 neurons from three individual culture batches were imaged. Primary antibodies and dilutions used for immunocytochemistry experiments are listed in the table below. Alexa Fluor goat secondary antibodies directed to the appropriate primary antibodies were used at a 1:500 dilution. All imaging and analyses were performed with the researcher blinded to experimental conditions.

COS-7 cells were grown at 37°C and 5% CO₂ in high glucose DMEM (Gibco; Cat# 11965) supplemented with 10% FBS (Gibco; Cat# 26140). Cells were transiently co-transfected with constructs encoding GluN1a (A gift from Dr. Lynn Raymond, University of British Columbia), GFP-tagged GluN2B (a gift from Dr. Katherine Roche, National Institutes of Health; Sanz-Clemente et al., 2013), and C-terminal FLAG-tagged DAPK1 (a gift from Dr. Ruey-Hwa Chen, Academia Sinica; You et al., 2017) at a 2:2:1 ratio using X-tremeGENE 9 transfection reagent (Roche; Cat# 6365809001). For relevant experiments, DMSO or DKI (10 μM) were added to cell culture media at the time of transfection. Live GFP-GluN2B surface staining was performed 24 h post-transfection, followed by immunocytochemistry for internal GFP-GluN2B and DAPK1-FLAG, as described in the previous section. Microscopy and image analysis were performed as done for neuronal culture experiments, except surface expression was quantified within three representative surfaces/cytosolic non-nuclear regions within each imaged cell identified as positively co-expressing surface GFP-GluN2B and DAPK1-FLAG. Primary antibodies and dilutions used for immunocytochemistry experiments are listed in the table below. Alexa Fluor goat secondary antibodies directed to the appropriate primary antibodies were used at a 1:500 dilution (488-conjugated goat anti-chicken, 568-conjugated goat anti-mouse, 647-conjugated goat anti-rabbit).

Pulse-chase assays were performed using bio-orthogonal labeling with the methionine analog azido-homoalanine (AHA), as previously described (Dieterich et al., 2007; Ullrich et al., 2014; Wang et al., 2017). COS-7 cells were grown at 37°C in DMEM supplemented with 10% FBS. Cells were transiently co-transfected with GluN1a, GFP-tagged GluN2B, and FLAG-tagged DAPK1 DNA constructs as described above. 24 h post-transfection, cells were deprived of methionine and cysteine and incubated with 50 μM AHA and non-labeled cysteine for 1 h (pulse). After the pulse, cells were washed with PBS and incubated with regular media (chase). Cells were harvested after 0, 0.5, 1, 2, 3, 5, 8, 10, or 24 h of chase time and lysed in RIPA buffer containing protease and phosphatase inhibitors. GFP-tagged GluN2B and FLAG-tagged DAPK1 were immunoprecipitated from cell lysates by overnight incubation with Protein G Dynabeads and goat anti-GFP polyclonal (Eusera; Cat# EU4) or mouse anti-FLAG monoclonal (Millipore Sigma–Aldrich; Cat# F1804) antibodies, respectively. Bio-orthogonal click chemistry of AHA-labeled proteins with biotin-PEG4-alkyne (Sigma–Aldrich; Cat# 764213) was subsequently performed. GluN2B and DAPK1 immunoprecipitates were adjusted to 1% SDS and incubated with 100 mM Tris (benzyl-triazolyl methyl)amine (TBTA), 1 mM CuSO₄, 1 mM Tris-carboxyethyl phosphine (TCEP), and 100 mM azido-biotin at 37°C in darkness for 30 min. Samples were washed and boiled to remove protein from the beads. Total GluN2B and DAPK1 were detected by Western Blot analysis with mouse anti-GluN2B (Thermo Fisher Scientific; Cat# MA1–2014; 1:800) or rabbit anti-DAPK1 (Millipore Sigma; Cat# D1319; 1:1,000), respectively. GluN2B or DAPK1 bio-orthogonally labeled with biotin was detected with Alexa Fluor 680 conjugated to streptavidin (Life Technologies; Cat# S-32358). Protein turnover was determined by calculating the ratio of biotin-labeled protein to total protein signal and normalizing to t = 0 h.

To assess gene expression, a small piece of anterior cortical tissue was cut and RNA was extracted using an RNeasy Mini Kit (Qiagen; Cat# 74106) and RNase-free DNase set (Cat# 79254). cDNA was synthesized with the Superscript III First-Strand system according to manufacturer instructions (Life Technologies; Cat# 18080–051). Quantitative RT-PCR reactions were prepared in MicroAmp Fast optical 96-well reaction plates (Applied Biosystems; Cat# 4346907) with Power SYBR™ Green PCR Master Mix (Thermo Fisher Scientific; Cat# 4368706), and run on the Applied Biosystems 7500 Fast Real-Time system. Dapk1 expression was quantified based on ΔΔCt values using the housekeeping gene Rpl13a. Dapk1 primer sequences: 5′- GTCCGCTACCTCTGTTTGATG-3′ and 5′-GTGTCCTTCCGCAGTCTTG-3′. Rpl13a primer sequences: 5′-GGAGGAGAAACGGAAGGAAAAG-3′ and 5′-CCGTAACCTCAAGATCTGCTTCTT-3′.

Mice (4 weeks old) were anesthetized with ketamine and xylazine (100/10 mg/kg i.p.) and then placed in a supine position with their noses upright and heads flat on the bench surface. The DAPK1 inhibitor TC-DAPK6 (DKI) was dissolved in 50% PEG, 10% DMSO, 40% NaCl (0.9% w/v), and used at a final concentration of 1.67 μM. A 10 μl Hamilton syringe with a blunted needle tip was used to administer either vehicle or DKI in 2 μl increments per nare every 4 min, alternating sides until the total dose (30 μl; 50 nmol) was given. Mice remained supine for 30 min after intranasal administration. After 6 h, mice were sacrificed and samples were collected for subcellular fractionation experiments.

GraphPad Prism 5 was used for all statistical analysis and graph preparation. Data are presented as mean ± SEM. Two-tailed Student’s t-test, one-way ANOVA, or two-way ANOVA statistical tests with post hoc analysis (Bonferroni or Dunnett) were used for all experiments with an alpha level set to 0.05. In some cases, two-tailed t-tests were performed between specific groups after an ANOVA test when a planned comparison was established during the experimental design process. Significance for these tests is noted with the “#” symbol instead of “*”. For all experiments, data were tested to ensure normality and equal variance between groups. Figures were generated in Adobe Photoshop CS5.

Given the established role of exNMDAR excitotoxicity in HD, we sought to evaluate whether DAPK1 is dysregulated in YAC128 mice. One-month-old animals were used because YAC128 mice at this age demonstrate enhanced sensitivity to excitotoxins and elevated striatal exGluN2B, before behavioral or neuropathological phenotypes (Graham et al., 2009; Milnerwood et al., 2010). We observed significantly increased DAPK1 protein levels in total lysates from both the cortex and striatum of YAC128 mice, as well as reduced site-specific phosphorylation at the autoinhibitory S308 residue (normalized to protein level), indicating greater kinase activation (Figures 1A,B). These changes were not observed in the cerebellum, which is largely spared in HD (Vonsattel et al., 2008; Figure 1C). We also did not observe any differences in DAPK1 protein level or phosphorylation in the cortex of transgenic human wtHTT-expressing YAC18 control mice (Supplementary Figure 2), signifying that these changes are not due to general overexpression of human HTT. When we evaluated Dapk1 mRNA levels, we found no differences between genotypes (Figure 1D), suggesting altered DAPK1 protein stability or turnover in YAC128 brains, as opposed to elevated transcription. Interestingly, in 1-month-old WT FVB animals, DAPK1 expression was significantly higher in cortical and striatal tissues when compared directly to the cerebellum (Figure 1E). Furthermore, phosphorylation of DAPK1 at S308 was reduced in the striatum compared to the cortex and was dramatically increased in the cerebellum (Figure 1E). Thus, in the unaffected brain, DAPK1 expression and activation are highest in the regions most affected by HD.

The C6R mouse model expresses the YAC128 transgene bearing point mutations at residues D586 and D589, rendering mHTT resistant to caspase cleavage at D586 (Graham et al., 2006). Unlike YAC128 mice, C6R mice are resistant to excitotoxicity, do not demonstrate enhanced exNMDAR activity, and display no motor deficits or neuropathology, indicating that mHTT cleavage is a critical contributor to exNMDAR dysfunction in HD mice (Graham et al., 2006; Milnerwood et al., 2010). We found that DAPK1 expression is reduced in the cortex and unaltered in the striatum of 1-month-old C6R mice, and pS308 levels are comparable to WT in both brain regions (Figures 1A,B), suggesting that D586 cleavage is required for mHTT-induced dysregulation of DAPK1.

DAPK1 interacts with and phosphorylates exGluN2B at S1303, which contributes to the amplification of receptor function and surface expression (Tu et al., 2010; Sanz-Clemente et al., 2013). We hypothesized that an increase in DAPK1 expression and activity in the YAC128 brain would correlate with enhanced phosphorylation at this GluN2B residue. We evaluated GluN2B and pS1303 levels in synaptic (post-synaptic density, PSD) and extrasynaptic (non-PSD) membrane compartments from 1-month-old WT and YAC128 tissues. We observed an increase in pS1303 in YAC128 cortices, which was specific to the extrasynaptic compartment, and not observed in synaptic fractions (Figures 2A,B). We did not detect altered GluN2B or pS1303 levels in YAC128 striatal synaptic fractions (Figure 2C). However, using a previously-described modified method to evaluate striatal “non-synaptic” protein (Gladding et al., 2014), we observed a significant increase in non-synaptic pS1303 in the YAC128 striatum (Figure 2D).

DAPK1 has been shown to interact more strongly with GluN2B receptors in vivo under pathological conditions and when S308 is dephosphorylated (Tu et al., 2010; Goodell et al., 2017). Furthermore, phosphorylation of GluN2B at S1303 enhances the DAPK1-GluN2B interaction (Goodell et al., 2017). We hypothesized that a chronic elevation in DAPK1 expression, activation, and exGluN2B pS1303 levels would be associated with an enhanced basal DAPK1-GluN2B interaction in YAC128 brains. In non-synaptic compartments, which were isolated using a gentle lysis buffer, we observed an increased interaction between GluN2B and DAPK1 in the YAC128 cortex (Figure 2E).

Even though GluN2B pS1303 was elevated specifically in the extrasynaptic compartment in YAC128 brains, we found that DAPK1 expression itself was increased in both synaptic and extrasynaptic membrane fractions from the YAC128 cortex and striatum (Supplementary Figure 3). Thus, although additional consequences of increased DAPK1 at the synapse may exist, the effect on GluN2B phosphorylation at S1303 appears specific to the extrasynaptic receptor population.

Calcium/calmodulin-dependent protein kinase II (CaMKII) is a DAPK1-related kinase with established roles in synaptic signaling, neuronal plasticity, and receptor regulation (Shonesy et al., 2014). Similar to DAPK1, CaMKII has been implicated in ischemic excitotoxicity related to NMDAR activation, and also phosphorylates GluN2B at S1303 (Omkumar et al., 1996; Coultrap et al., 2011), influencing receptor surface expression (Sanz-Clemente et al., 2013). However, we found that CaMKII expression and autonomous activation measured by phosphorylation at T286 were unaltered in total, synaptic and extrasynaptic compartments of the YAC128 cortex and striatum (Supplementary Figure 4).

It was previously shown that the treatment of cortical neurons with NMDA caused activation of DAPK1, phosphorylation of GluN2B at S1303, and increased exNMDAR-mediated calcium transients (Tu et al., 2010; Fan et al., 2014). These effects were abolished by the NMDAR antagonist AP5 and by genetic depletion of DAPK1, indicating that DAPK1 is activated downstream of NMDARs and subsequently augments exGluN2B phosphorylation and conductance (Tu et al., 2010). In YAC128 mice, low-dose treatment with memantine, which preferentially blocks exNMDAR channels, normalizes striatal cell survival (pCREB) and death (p-p38 MAPK) signaling, returns exGluN2B expression to WT levels, and prevents motor function decline and neuropathology (Okamoto et al., 2009; Milnerwood et al., 2010; Dau et al., 2014). We thus hypothesized that a positive feedback loop may be pathologically amplified in HD mice, whereby elevated DAPK1 activation in YAC128 brains may both result from and contribute to increased exNMDAR function. To address the former, we treated WT and YAC128 mice with low-dose memantine from conception by administering memantine in the drinking water to breeders and subsequently to pups once weaned. Tissues were collected for biochemistry at 1 month of age. We found that memantine treatment normalized DAPK1 S308 phosphorylation in YAC128 total cortical lysate to WT levels (Figure 3A). Treatment also normalized elevated YAC128 exGluN2B S1303 phosphorylation (Figure 3B). These findings indicate that DAPK1 dephosphorylation and enhanced pS1303 levels may occur downstream of exNMDAR activity in the YAC128 cortex.

Our next experiments aimed to address whether DAPK1 contributes to increased exGluN2B phosphorylation or surface expression in YAC128 neurons. Consistent with the role of DAPK1 in amplifying exGluN2B function in ischemia, genetic manipulations eliminating DAPK1 expression or kinase activity promote neuroprotection and synaptic preservation in models of stroke, excitotoxicity, and Alzheimer disease. Furthermore, a selective, small-molecule DAPK1 inhibitor (TC-DAPK6, referred to here as “DKI”) protects against glutamate toxicity in SH-SY5Y cells, enhances the maturation of primary cultured neurons, and prevents chronic stress-induced GluN2B S1303 phosphorylation, CREB dephosphorylation, and loss of brain-derived neurotrophic factor (BDNF) levels (Kim et al., 2014; Tian et al., 2014; Li et al., 2018). We used this same compound to evaluate the effect of targeting DAPK1 kinase activity on GluN2B receptors in YAC128 neurons. Primary YAC128 corticostriatal cultures exhibited increased endogenous pS1303, GluN2B, and DAPK1 levels at DIV21 compared to WT (Figure 4A). These experiments were performed using a gentle lysis buffer to only solubilize non-synaptic proteins. Treatment with DKI normalized all of these measures to WT levels (Figure 4A). We validated this finding in vivo by delivering DKI via intranasal administration to YAC128 mice at 4 weeks of age. Mice were sacrificed 6 h post-treatment and examination of extrasynaptic cortical membrane fractions indicated a trend to a reduction in GluN2B levels (Figure 4B). There was also a slight trend to decreased phosphorylation of exGluN2B at S1303 when normalized to GluN2B protein levels. Interestingly, when normalized to the β-actin loading control, pS1303 levels were significantly reduced with DKI treatment (Figure 4B). Furthermore, DKI treatment significantly reduced extrasynaptic DAPK1 protein levels (Figure 4B), thus confirming in vivo what we observed in DKI-treated YAC128 primary neuronal cultures. Together, this suggests that that DAPK1 activity contributes to pathologically elevated non-synaptic GluN2B and pS1303 in the presence of mHTT.

We further explored our observation that DKI normalized DAPK1 protein levels in YAC128 cultures compared to WT by assessing whether this effect was due to a treatment-induced alteration in the stability of the DAPK1 protein. We performed pulse-chase experiments with the traceable amino acid azidohomoalanine (AHA) in COS-7 cells triple-transfected with DAPK1, GluN1, and GluN2B constructs and found that DKI significantly increased DAPK1 protein turnover compared to DMSO control conditions (Figure 4C), thus supporting our hypothesis that activity of DAPK1 promotes stability of the protein.

Previously, an increase in surface expression (relative to internal expression) of transfected YFP-GluN2B was observed in YAC128 MSNs co-cultured with cortical neurons at a 1:1 ratio, compared to WT (Milnerwood et al., 2012). As the basal fluorescence of YFP conjugated to GluN2B is extremely low, this experiment requires amplification of the YFP by immunocytochemistry using primary antibodies and fluorophore-conjugated secondary antibodies. We confirmed the phenotype reported in Milnerwood et al., 2012 in the present study and found that inhibition of DAPK1 reduced YFP-GluN2B surface expression in MSNs of both genotypes without affecting the number of receptor clusters, percent punctae colocalization with the presynaptic marker vesicular glutamate transporter 1 (VGLUT1), or the number of synapses (YFP-GluN2B/VGLUT1 colocalized points; Figure 5A). To assess whether the effect of blocking DAPK1 activity on surface expression may be mediated via altered S1303 phosphorylation, we performed GluN2B surface expression analysis on COS-7 cells triple-transfected with DAPK1, GluN1, and GluN2B constructs. We found that co-expression of a kinase-dead DAPK1 mutant (K42A; KA) resulted in reduced surface expression of GluN2B compared to co-expression with WT DAPK1 (Figure 5B). As described previously (Sanz-Clemente et al., 2013), rendering S1303 phospho-deficient by mutating to alanine (S1303A; SA), also significantly decreased GluN2B surface expression (Figure 5B). Notably, the S1303A mutation occluded the effect of the DAPK1 K42A mutation. Altogether, these results demonstrate that active DAPK1-induced phosphorylation of S1303 increases exGluN2B phosphorylation and surface expression, and suggest that pathological amplification of this process may contribute to exGluN2B dysfunction in the presence of mHTT.

MSN spine loss occurs with disease progression in HD patients and animal models (Ferrante et al., 1991; Klapstein et al., 2001; Laforet et al., 2001; Spires et al., 2004; Lerner et al., 2012; Marco et al., 2013; Rocher et al., 2016; Wu et al., 2016), as well as in DIV21 YAC128 MSNs co-cultured with cortical neurons at a 1:3 cortico:striatal (CS) ratio (Wu et al., 2016; Schmidt et al., 2018). Recently, DAPK1 was found to contribute to cortical spine and synapse degeneration after ischemia in vivo, and genetic deletion of the DAPK1 catalytic domain reversed hippocampal spine and synapse loss in Tg2576-APP Alzheimer disease mice (Pei et al., 2015; Shu et al., 2016). Therefore, we sought to determine if inhibiting DAPK1 could prevent or reverse striatal spine loss in YAC128 1:3 CS co-cultures. Consistent with previous reports, YAC128 MSNs co-cultured with cortical neurons at a 1:3 ratio exhibited significantly reduced numbers of total spines at DIV21, and this was primarily due to a loss of mature mushroom-shaped spines with a small contribution of immature (stubby, thin, and filopodia) spine types (Figure 6A). Treatment with DKI from DIV14-DIV21 completely rescued total, mature, and immature spine numbers in YAC128 MSNs to WT control levels (Figure 6A).

Our previous work showed that YAC128 MSN spine instability in co-culture is apparent by DIV18, but not DIV14, meaning DKI treatment from DIV14 to 21 prevented spine instability from occurring. Next, we sought to determine if DKI treatment could reverse the YAC128 spine loss after it had already occurred. To assess this, we performed a late-intervention treatment from DIV18 to 21 and observed complete reversal of total spine numbers, including the near-complete rescue of mushroom spine density with DKI treatment (Supplementary Figure 5). We did not observe a basal genotypic difference in immature spine numbers in this batch of control-treated neurons (Supplementary Figure 5), indicating that this phenotype may show more culture-to-culture variation compared to the reliable and robust loss of mushroom spines observed in all experiments.

Next, we reasoned that if the beneficial effect of inhibiting DAPK1 on spine stability is mediated through its downregulation of exGluN2B function, then blockade of exNMDARs or GluN2B should have the same effect. Similar to DAPK1 inhibition, blockade of exNMDARs with low-dose memantine for 7 days completely rescued total and mature spine numbers in co-cultured YAC128 MSNs (Figure 6B). Surprisingly, GluN2B-selective antagonism with ifenprodil provided no benefit in YAC128 MSNs and significantly reduced mushroom spine density in WT MSNs (Figure 6C), suggesting that ifenprodil may inhibit the beneficial activity of GluN2B-containing synaptic NMDARs in this in vitro system. Ultimately, these findings support our hypothesis that DAPK1 activity and consequent upregulation of exNMDAR function contribute to mHTT-induced instability of MSN dendritic spines.