Restoring endogenous Dlg4/PSD95 expression by an artificial transcription factor ameliorates cognitive and motor learning deficits in the R6/2 mouse model of Huntington’s disease

Abstract

Background

Huntington’s disease (HD) is an incurable hereditary disorder caused by an expansion of CAG repeats in exon 1 of the Huntingtin gene (HTT). HD is characterized by motor dysfunction and cognitive decline. The pathophysiology of HD begins in cortico-striatal circuits and later spreads to other brain regions, notably the hippocampus. At the cellular level, structural changes in synapses have been observed prior to neuronal degeneration, significantly disrupting the formation and maintenance of neuronal circuits. The postsynaptic density protein 95 (PSD-95, hereafter Dlg4/PSD95) is a key synaptic plasticity protein reduced in HD and other neurodegenerative diseases such as Alzheimer’s disease (AD). Epigenetic silencing of plasticity and memory genes contributes to AD pathology and cognitive impairment. To restore endogenous Dlg4/PSD95 expression in AD, we previously developed an epigenetic editing strategy where a zinc finger DNA-binding domain targeting the Dlg4/PSD95 gene promoter was fused to the transactivation domain VP64 and driven under a CMV promoter. AAV-PhP.B-mediated delivery of this artificial transcription factor (ATF) CMV-PSD95-6ZF-VP64 improved cognition in an AD mouse model. Here, we assessed the therapeutic potential of AAV9-mediated delivery of the synapsin-driven ATF PSD95-6ZF-VP64 in the R6/2 HD mouse model.

Results

Consistent with the previous studies, R6/2 mice exhibited reduced hippocampal Dlg4/PSD95 mRNA and protein levels in young adulthood (7 weeks), which persisted into early adulthood (14 weeks). Starting at adolescents (4 weeks), the R6/2 mice also displayed motor (i.e., accelerated rotarod) and cognitive (i.e., Barnes maze and object location memory) impairments. In wild-type primary hippocampal cultures, AAV9-PSD95-6ZF-VP64 led to an increase in synaptic PSD-95 clusters and spine size. Intracerebroventricular injections of neonatal R6/2 mice with AAV9-PSD95-6ZF-VP64 elevated hippocampal Dlg4/PSD95 expression levels to those observed in control non-transgenic mice. Importantly, AAV9-PSD95-6ZF-VP64 effectively improved hippocampal-dependent deficits in spatial learning and memory in young adult HD mice, as well as impairments in motor coordination and motor skill learning, with these benefits persisting into adulthood.

Conclusion

This work validates Dlg4/PSD95 as a key player in the prodromal phase of HD pathology and establishes the ATF PSD95-6ZF-VP64 as an attractive therapeutic tool for treating the disease’s early phase.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-025-01903-2.

Introduction

The devastating neurodegenerative disorder Huntington’s disease (HD) is caused by an expanded CAG repeat in exon 1 of the gene encoding huntingtin (HTT) and is characterized by progressive motor, cognitive and psychiatric symptoms [100, 101]. Currently available treatments help manage some of the symptoms, however, no cure is available, and few advances have been made on reversing or slowing disease progression [103, 109]. Although the highly toxic HTT exon 1 protein containing the polyQ tract is a key focus of HTT-lowering therapies, recent studies -including those using antisense oligonucleotides (ASOs) such as Tominersen- have shown that selectively targeting the mutant allele without affecting the wild-type HTT allele remains a major challenge [111].

While cortico-striatal dysfunction is a core pathological feature of HD [20, 21], accumulating evidence supports a critical role for early cognitive impairments related to the hippocampus and adjacent structures [41]. Cognitive impairments -including deficits in attention, working memory, executive functions, and spatial navigation- are relatively subtle, but are among the earliest prodromal symptoms of HD. These alterations can emerge one to two decades before the onset of motor symptoms [12, 34, 41, 48, 95, 100]. While these cognitive impairments are commonly linked to deterioration of the cortico-striatal pathways [58, 60, 95, 113], growing evidence in humans indicates that hippocampal dysfunction may also contribute significantly [11, 26, 48, 98, 121]. Specifically, alterations in hippocampal volume [98, 99, 121] and impairments in functional neuronal plasticity [11, 48] have been reported in the early stages of the disease in human HD patients.

Studies utilizing transgenic HD mouse models have yielded significant insights into the neural mechanisms underlying cognitive deficits associated with HD. Consistent with clinical studies in HD patients, research using a variety of HD transgenic mouse models -including R6/2, R6/1, YAC128, Q175, and HdhQ111- has linked early hippocampal-associated learning and memory deficits [24, 67] to reduced spine density and impaired synaptic and extrasynaptic transmission, mediated by NMDA receptors (NMDARs) and AMPA receptors (AMPARs) in the hippocampus [21, 39, 78, 82, 84, 93, 94, 122, 126, 129].

PSD-95, encoded by Dlg4 (termed hereafter Dlg4/PSD95), is a principal synaptic scaffold protein that anchors NMDARs, AMPARs and dopamine receptors within the postsynaptic density (PSD) to other scaffold proteins (i.e., shank, homer), various cytoskeletal proteins, adhesion molecules, and intracellular signaling molecules (i.e., Pyk2, Karilin-7, nNOS), thereby playing a critical role in regulating synaptic transmission, synaptic plasticity, dendritic spine stability, as well as learning and memory [32, 66, 114–117]. Analysis of several HD mouse models during prodromal stages has revealed that the appearance of behavioral deficits and impaired glutamatergic activity in HD mice is accompanied by a loss of total and postsynaptic Dlg4/PSD95 protein levels. Specifically, biochemical and immunofluorescence studies investigating early stages of phenotypic progression have shown that both total and synaptic Dlg4/PSD95 protein levels are reduced in hippocampal, cortical and striatal neurons across various HD mouse models, including R6/2 [20, 81], R6/1 [39, 112], and Hdh^Q111 [89]. RNA profiling also revealed that Dlg4/PSD95 mRNA levels are reduced in pre-symptomatic HD knock-in mice, including Q175 [62] and Hdh^Q111 [4]. Consistently, Dlg4/PSD95 mRNA and/or PSD-95 protein levels are also diminished in striatal postmortem tissue from human HD patients [35, 53, 112, 125]. Strikingly, molecular studies provide direct evidence that Dlg4/PSD95 critically contributes to HD pathology and phenotype, with animals lacking Dlg4/PSD95 displaying typical HD phenotypes, including synaptic dysfunction, striatal neurodegeneration, and learning and motor deficits [127]. Considering that the loss of Dlg4/PSD95 could explain, at least in part, cognitive and motor impairments in HD pathogenesis, this protein is a promising molecular target for gene therapy. To restore Dlg4/PSD95 gene expression, here we employed an epigenetic editing strategy as a potential therapeutic approach.

In addition to HD, the expression of Dlg4/PSD95 is reduced in aging and in several neurodegenerative disorders, including in Alzheimer’s disease (AD) [7, 18, 25]. Although limited studies are available for HD, comprehensive analyses of global and locus-specific (epi)genomic and transcriptomic alterations in human and mouse brain samples at cell-type resolution have revealed key phenomena in AD. Accumulating evidence indicates that epigenetic silencing of genes involved in synaptic plasticity and memory contributes to AD pathology and cognitive impairment [116]. Several distinct mechanisms account for the epigenetic blockade, including: (i) reduced enrichment of acetylated histones H3/H4 (H4K12ac, H3K14ac) at gene promoters [45],(ii) decreased H3K4me3 levels at promoters [42],(iii) diminished H3K27ac at enhancer elements [42],and/or (iv) increased H3K9me2 enrichment at promoter regions [128]. Knock-down of HDAC2 expression or the treatment with diverse small molecules that inhibit HDAC2 activity (i.e., SAHA, sodium butyrate, TSA) was able to reinstate morphological alterations, synaptic plasticity, and rescue memory deficits observed in AD [45, 47, 116]

Epigenetic editing is a promising therapeutic approach to selectively modify chromatin for treating neurodegenerative disease such as HD and AD, and potentially other neurological diseases where Dlg4/PSD95 expression is aberrant [14, 114–116, 123]. To restore endogenous Dlg4/PSD95 expression in AD, we previously engineered an epigenetic editing strategy, utilizing an artificial transcription factor (ATF) [18, 114–116]. This ATF, termed PSD95-6ZF-VP64, contains 6 Cys2His2 zinc fingers (6ZF) to target an 18-bp sequence located in the proximal Dlg4/PSD95 gene promoter region upstream of the transcription start site (TSS). This DNA-binding domain was fused to the transactivation domain VP64, a tetramer of the viral protein VP16. This effector domain (ED) has shown to strongly promote transcriptional activation by recruiting proteins, including chromatin-modifying enzymes [40, 43]. Transduction of N2a cells and primary hippocampal neurons in vitro led to increased gene transcription via histone H3 acetylation (i.e., H3ac total, H3K9ac, H3K14ac) [18]. Transduction of the ATF PSD95-6ZF-VP64 in hippocampal neurons in vivo impacted several plasticity-associated processes (i.e., regulating synapse and spine maturation), and intriguingly, rescued memory deficits in aged and AD mice [18, 114–116].

Having confirmed the efficacy of PSD95-6ZF-VP64 in improving plasticity-associated processes and cognition in AD mice, we aimed to assess its therapeutic potential in the R6/2 HD mouse model, which exhibits early hippocampal-dependent synaptic dysfunction and cognitive deficits [24, 67, 82]. To this end, intracerebroventricular (ICV) injections were performed in postnatal (P) 0–1 R6/2 mouse brains using adeno-associated virus (AAV) serotype 9 (AAV9) to deliver a synapsin-driven ATF PSD95-6ZF-VP64. Subsequent analyses included hippocampal Dlg4/PSD95 mRNA and protein levels, as well as assessments of motor and cognitive impairments. AAV9 is widely regarded as the ‘gold standard’ serotype for gene delivery targeting the central nervous system (CNS) as its safety and efficacy has been extensively established in animal models and humans [54]. Importantly, a systematic evaluation of various AAV serotypes expressing a GFP reporter -including AAV2/1, AAVDJ8 and AAV9- showed that ICV injecting in P0 neonatal mice resulted in robust AAV9-mediated transgene delivery in the hippocampus, primary motor cortex (mainly layer 5) and substantia nigra, while expression in the striatum, cerebellum, and olfactory bulb was limited [49]. Thus, AAV9-mediated ICV delivery of the ATF PSD95-6ZF-VP64 offers insights into its potential therapeutic effects on motor learning and cognitive deficits by targeting mainly the hippocampus in R6/2 HD mice.

Results

The expression of PSD95-6ZF-VP64 in wild-type primary hippocampal cultures leads to an increase in synaptic PSD-95 clusters and spine size

In our previous study [18], we designed a specific and functional construct PSD95-6ZF-HA-NLS-GFP-VP64 where a zinc finger DNA-binding domain (6ZF) targeting the Dlg4/PSD95 gene promoter was fused to the transactivation domain VP64. This construct also contained a nuclear localization sequence (NLS), and a hemagglutinin (HA) tag and a green fluorescent protein (GFP) reporter to visualize transduced cells. For simplicity, this construct was termed PSD95-6ZF-VP64 (Fig. 1A). Transduction of this ATF increased Dlg4/PSD95 gene expression in N2a cells and in hippocampal neurons in vitro through histone H3 acetylation (see discussion) [18]. A 6ZF construct lacking an ED (or ´no ED´), and hence termed PSD95-6ZF-NoED, was also created as control (Fig. 1A). To transduce the ATF and its control in N2a cells and in the brain, we generated lentiviral as well as AAV with the novel capsid PHP.B bicistronic vectors, respectively [18]. The AAV vectors encoding the ATF (and its control) were driven by the chicken β-actin gene promoter, and GFP was driven by an internal ribosome entry site (IRES). For the study presented here, we retained the same core constructs (PSD95-6ZF-HA-NLS-GFP-VP64/NoED, (Fig. 1A), while modifying the promoters driving these designs as well as the reporter. Specifically, PSD95-6ZF-VP64/NoED was placed under the elongation factor-1 (EF1) gene promoter and the enhanced GFP (eGFP) sequence under the human synapsin 1 (hSyn) promoter to report expression specifically in neurons. To achieve stable and long-term transgene expression in vitro and in vivo (see below) with limited (or undetected) toxicity, AAVs were packaged using the AAV9 capsid. To verify the capacity of the ATF to increase Dlg4/PSD95 expression, primary hippocampal neurons from wild-type CBA mice were infected at 7 days in vitro (DIV) with AAV9-PSD95-6ZF-VP64 construct. CBA animals were chosen as these mice served as the genetic background for R6/2 animals (see below). As controls, neurons were infected with AAV9-PSD95-6ZF-NoED. Three days later (10 DIV), neurons transduced with either PSD95-6ZF-VP64 (Fig. 1B, E) or PSD95-6ZF-NoED (Fig. 1D) displayed a robust GFP signal in their soma, dendrites and smaller neuronal protrusions (i.e., spines), with no apparent signs of toxicity (i.e., blebbing).

Fig. 1.

AAV9-PSD95-6ZF-VP64 increases synaptic Dlg4/PSD95 clusters and spine size in cultured hippocampal neurons. A The basic PSD95-6ZF-HA-NLS-GFP construct contains a zinc finger DNA-binding domain (6ZF) targeting the Dlg4/PSD95 gene promoter fused to the nuclear localization sequence (NLS) and a hemagglutinin (HA) tag. The PSD95-6ZF-HA-NLS-GFP-VP64 construct, termed PSD95-6ZF-VP64, also contains the transactivation domain VP64 to induce histone H3 acetylation and hence transcription of Dlg4/PSD95. A control 6ZF construct lacking VP64 is termed PSD95-6ZF-NoED. The PSD95-6ZF-VP64/NoED constructs were placed under the EF1 gene promoter and the eGFP sequence under the human synapsin promoter (reporter) to enable structural details of transduced neurons. AAV9-PSD95-6ZF-VP64/NoED were infected in 7 DIV hippocampal neurons and fixed at 10 DIV for fluorescence analysis. B Representative confocal images show infected neuron (double arrow) distinguished by GFP fluorescence and HA expression. C Quantification of spine head areas shows a significant increase in larger spine heads in PSD95-6ZF-VP64-treated neurons compared to PSD95-6ZF-NoED (p < 0.001). D-E Immunolabeling reveals an increase in synaptic spine density and PSD95 clustering in PSD95-6ZF-VP64 neurons (D) compared to PSD95-6ZF-NoED-treated cells (E). Insets highlight dendritic details, showing increased PSD95 puncta density and spine size in PSD95-6ZF-VP64 neurons

Next, we analyzed 10 DIV hippocampal neurons as at these early/intermediate stages of in vitro development the PSD95 immunoreactivity (IR) is low and diffuse, with only few PSD-95-IR clusters detected at synapses [19, 118]. We first verified the subcellular localization of the PSD95-6ZF constructs by performing immunostaining with specific antibodies against the HA tag that is fused to 6ZF construct and against Dlg4/PSD95. NucBlue was used as nuclear counterstain. Transduction of neurons with PSD95-6ZF-VP64 resulted in GFP-positive neurons displaying a nuclear HA signal and a global increase in the Dlg4/PSD95 signal (Fig. 1B). Next, we tested the capacity of ATF PSD95-6ZF-VP64 to increase spine density in GFP-positive hippocampal neurons in vitro by performing reconstructions and measurements of morphological spine parameters using Spine-j software [64]. After studying 135 spines from three different cell cultures treated PSD95-6ZF-VP64, and 85 spines from an equal number of neurons infected with PSD95-6ZF-NoED, it was observed that the distribution of spine head area was significant different between the two conditions, with neurons expressing PSD95-6ZF-VP64 displaying larger spines (Fig. 1C). These results align with previous in vitro studies demonstrating the critical role of Dlg4/PSD95 in synapse and spine maturation in glutamatergic neurons, as shown through conventional knockdown (shRNA-PSD95) and/or overexpression (cDNA-PSD95) approaches in neuronal cultures [19, 30, 31, 68]. Furthermore, these findings support our in vivo study, where GFP-positive dentate gyrus hippocampal neurons expressing ATF PSD95-6ZF-VP64, but not ATF PSD95-6ZF-NoED, exhibited a higher density of mushroom spines [18]. Important, and consistent with the finding that neurons expressing PSD95-6ZF-VP64 display larger spines, we also found that transduction of this ATF led to an increased expression of Dlg4/PSD95 signal in spines (Fig. 1D). Similar experiments were performed with PSD95-6ZF-NoED, which also resulted in a nuclear HA signal in GFP-positive neurons, without leading to an increase in global (data not shown) or synaptic Dlg4/PSD95 protein levels (Fig. 1E).

ICV injections of AAV9 viral particles carrying PSD95-6ZF lead to widespread and long-term expression of the PSD95-6ZF constructs in the mouse brain

In our previously study, AAV vectors were produced with the PHP.B capsid to enable widespread and long-term transduction of the constructs in the mouse brain [18]. The PHP.B capsid (and variants such as PHP.eB) has shown a ~ 60-fold higher efficiency than AAV9 in transducing the adult mouse brain of several mouse strains (e.g., C57BL/6 J, FVB/NJ, DBA/2 J) [10, 22, 28]. However, the PHP.B capsid requires the LY6A receptor to cross the blood–brain barrier (BBB), and a number of inbred strains (e.g., BALB/cJ, BALB/cByJ, A/J, NOD/ShiLtJ, CBA/J) express a specific genetic variant of the Ly6 genes (Ly6a), making AAV-PHP.B ineffective for systemic CNS gene transfer. Given these limitations, and the potential to deliver the ATF to the CNS through non-invasive method (i.e., intravenous delivery) in future experiments, we opted to use the AAV9 capsid to generate viral particles. We tested whether widespread and long-term transduction of the PSD95-6ZF constructs in the mouse brain could be achieved when AAV9 viral particles (5 × 10¹¹ IU/µl) were delivered via ICV injections in P0-1 mice (Fig. 2A). A non-invasive in vivo imaging system (Bruker´s In-Vivo FX PRO, see methods) was employed to determine fluorescence signals in control (non-infected) and AAV9-PSD95-6ZF-VP64 and AAV9-PSD95-6ZF-NoED treated animals. During the first 48 h post-injection, when the cranium of neonatal mice remains transparent, transcranial fluorescence imaging can be conducted. During this timeframe, AAV9 infected mice exhibited a strong fluorescence signal compared to control littermates (non-infected or injected with trypan blue) (Fig. 2B, C). At 72 h post-injection, transcranial fluorescence signals were no longer detectable, requiring the dissection of the brains for further analysis (Fig. 2C). To assess long-term expression of the constructs, mice were ICV injected with AAV9-PSD95-6ZF-NoED at P0-1, and the brain was extracted after one year. These animals showed a robust GFP signal in the hippocampus near the lateral ventricle, a phenomenon not observed in animals injected with trypan blue neonatally (Fig. 2C).

Fig. 2.

ICV injections of AAV9-PSD95-6ZF ensures widespread and long-term transduction in the mouse brain. A Upper model shows ICV injections sites and lower model illustrates striatum, lateral ventricles, and hippocampus. B GFP fluorescence was monitored 48- and 72-h post-injection in control mice treated with PSD95-6ZF-NoED, PSD95-6ZF-VP64, or trypan blue (control) using the In Vivo FX PRO fluorescent imaging system. Robust GFP signals were observed in PSD95-6ZF-treated groups, while no fluorescence was detected in trypan blue controls. C Heatmaps display persistent GFP fluorescence in hippocampal regions one-year post-infection with PSD95-6ZF-NoED construct, confirming long-term viral expression. Fluorescence intensity is scaled relative to background. Representative result of 3 independent experiments

Hippocampal Dlg4/PSD95 mRNA and protein levels are reduced in R6/2 mice

Biochemical analysis of hippocampal and cortical extracts from R6/2 mice have revealed that PSD95 mRNA and protein levels were significantly reduced at 6 weeks of age and continued to decline as the disease advanced (8 and 12 weeks tested), compared to wild-type non-transgenic (NTg) littermates [69, 81]. Here, we examined whether total Dlg4/PSD95 protein and mRNA levels are also reduced in R6/2 mice hippocampi at 7 and 14 weeks of age, compared to their NTg littermates. The analysis revealed a ~ 50% reduction in hippocampal Dlg4/PSD95 mRNA levels in 7-week-old R6/2 mice (Fig. 3A), a decrease that persisted through week 14 (Fig. 3C). To determine whether the changes in mRNA levels correlated with changes in protein levels, Dlg4/PSD95 protein levels were assessed by western blot with a specific and well-characterized antibody [18, 19, 50, 61, 80, 124]. Consistent with the RT-qPCR assays, western blot assays showed a ~ 50% reduction in hippocampal Dlg4/PSD95 protein levels in 7-week-old R6/2 mice (Fig. 3B), a decrease that was maintained 14-week-old R6/2 mice (Fig. 3D, Suppl. Figures 1, 2).

Fig. 3.

Restoration of Dlg4/PSD95 mRNA and protein expression in R6/2 mice treated with AAV9-PSD95-6ZFP-VP64. A, C Dlg4/PSD95 mRNA were measured in hippocampal whole lysates obtained from 7-week-old (A) and 14-week-old (C) mice in four conditions: control non-transgenic (NTg) mice, untreated R6/2 mice, R6/2 mice treated with PSD95-6ZF-VP64 or PSD95-6ZF-NoED via AAV9-mediated delivery of the constructs through ICV injections at neonatal stages (P0-1) B, D Dlg4/PSD95 and protein levels were measured in the hippocampus of 7-week-old (B) and 14-week-old (D) under the same four conditions. All data are reported as mean ± S.E.M. Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001. n = 4–6 independent animals as indicated by each circle. The determination of the linear range for Dlg4/PSD95 and actin quantification by western blot, along with representative blots, is presented in Suppl. Figures1-2

PSD95-6ZF-VP64 increases hippocampal Dlg4/PSD95 mRNA and protein levels in R6/2 mice

Next, we investigated whether transduction of the ATF PSD95-6ZF-VP64 in hippocampal neurons could increase Dlg4/PSD95 mRNA and protein levels in R6/2 mice. To ensure stable and widespread transgene expression in the hippocampus and other brain regions from early developmental stages, AAV9-PSD95-6ZF-VP64 viral particles were ICV administered to P0-1 R6/2 animals. As control, HD mice were treated with AAV9-PSD95-6ZF-NoED. We found that 7-week-old R6/2 mice transduced with PSD95-6ZF-VP64 -but not PSD95-6ZF-NoED- exhibited significantly increased Dlg4/PSD95 mRNA (Fig. 3A) and protein (Fig. 3B) levels, which were comparable to those observed in NTg mice. Analysis of 14-week-old R6/2 mice revealed that transduction with PSD95-6ZF-VP64 resulted in sustained elevation of Dlg4/PSD95 protein levels (Fig. 3D; Suppl. Figures 1, 2), while mRNA levels remained unchanged (Fig. 3C). These data demonstrate that the engineered ATF targeting the Dlg4/PSD95 locus is effective in enhancing its expression in hippocampal neurons in the R6/2 mouse model. Analysis of 14-week-old R6/2 mice revealed that Dlg4/PSD95 protein levels (Fig. 3D; Suppl. Figures 1, 2), but not mRNA levels (Fig. 3C), remained elevated following transduction with PSD95-6ZF-VP64.

Untreated R6/2 mice exhibit motor and cognitive impairments

The progression of symptoms in the R6/2 model has been extensively characterized through a series of behavioral tests conducted at various life stages [23, 74, 83, 110, 122]. Building on this knowledge, weight measurements were conducted in conjunction with a series of behavioral assays designed to assess key cognitive and motor symptoms during the early stages of the disease (Fig. 4A). In agreement with the previous studies [51, 74], and as discussed in more detail below, untreated R6/2 mice exhibited weight loss (Fig. 4B), as well as the characteristic cognitive (Fig. 4C-F) and motor (Fig. 5) impairments associated with HD. To determine whether weight, cognitive, and motor performances could be improved in R6/2 mice, AAV9-PSD95-6ZF-VP64 was administrated by ICV injections to neonatal stages (P0-1). Unlike weight (Fig. 4B), AAV9-PSD95-6ZF-VP64 ameliorated several cognitive impairments in treated R6/2 mice, as discussed in the following sections. From week ten on, pronounced motor symptoms of the disease begin to appear, such as coordination failure, seizures, tremors, and severe dehydration, and by week fourteen, these symptoms are deemed to represent a humanitarian endpoint under our experimental conditions.

Fig. 4.

AAV9-PSD95-6ZF-VP64 improves cognitive functions in R6/2 mice. A Timeline of the performed physiological and behavioral test battery. Weight and cognitive tests including object location memory (OLM), novel object recognition (NOR) and Barnes maze (Fig. 4), and motor tests including accelerated rotarod test and grip test (Fig. 5). Mouse examinations were conducted under four conditions: control NTg mice, untreated R6/2 mice, and R6/2 mice treated with PSD95-6ZF-VP64 or PSD95-6ZF-NoED. B Weight measurements of mice under the four experimental conditions. C Schematic representation of OLM and NOR tests to measure spatial and object recognition. D Discrimination index represents the proportion of time spent exploring the displaced object (B) or novel object (C) relative to the total time spent exploring both objects (B/C + A). An index greater than 0.5 indicates that mice successfully discriminate between the displaced or novel object and the non-displaced object. The data show that untreated R6/2 mice display an impairment in OLM, but not in NOR. Treatment of R6/2 mice with PSD95-6ZF-VP64, but not PSD95-6ZF-NoED, significantly improves OLM. E Schematic of the Barnes maze test setup with spatial cues. F Primary latency in the Barnes maze demonstrates that untreated R6/2 mice display an impairment in the Barnes maze. Treatment of R6/2 mice with PSD95-6ZF-VP64, but not PSD95-6ZF-NoED, significantly improves Barnes maze, reaching spatial memory levels comparable to NTg mice. All data are reported as mean ± S.E.M. Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001. (n = 4–6 independent animals as indicated by each circle, except in weight measurements). For graph F, statistical significance was set at ###p < 0.001 NTg and R6/2 + PSD95-6ZF-VP64 versus R6/2 and R6/2 + PSD95-6ZF-NoED at day 1, and $$$p < 0.001 NTg and R6/2 + PSD95-6ZF-VP64 versus R6/2 and R6/2 + PSD95-6ZF-NoED at day 7

Fig. 5.

AAV9-PSD95-6ZF-VP64 improves motor skill learning and coordination in R6/2 mice. A Timeline and number of trials (left) and schematic (right) of accelerating rotarod test for motor skill learning and coordination, as well as the grip strength test for general muscle strength. Mouse behavioral experiments were conducted under four conditions: control NTg mice, untreated R6/2 mice, and R6/2 mice treated with PSD95-6ZF-VP64 or PSD95-6ZF-NoED. B, C The graph shows that latency to fall in the rotarod test is significantly increased in untreated R6/2 mice at 7 and 9 weeks of age, compared to NTg mice. Treatment of R6/2 mice with PSD95-6ZF-VP64, but not PSD95-6ZF-NoED, significantly improves the rotarod test. D, E Grip strength was unaffected in any condition tested by treatment. All data are reported as mean ± S.E.M. Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001. (n = 4–7 independent animals as indicated by each circle)

PSD95-6ZF-VP64 improves cognitive functions in R6/2 mice

To assess the function and relative health of specific brain regions associated with learning and memory, the following behavioral testing were conducted on R6/2 mice, both in the absence and presence of AAV9-PSD95-6ZF-VP64 treatment: object location (OLM) and novel object recognition (NOR) tasks at week 4, when mice are adolescents (Fig. 4C, D), and the Barnes maze test at week 7, when mice reach young adulthood (Fig. 4E, F). In these tasks, the animals freely explore, hence avoiding stress. The OLM task is a simple, single-trial, hippocampal-dependent behavioral memory test that assesses spatial learning and relies on the rodent’s innate preference for novelty [18, 63, 90, 119]. Briefly, the OLM test consists of three steps: habituation (3 days), a training session (on day 4) followed by the OLM session (on day 4) (Fig. 4C). During the training session the animals were presented with two identical non-displaced two objects (i.e., Objects A and B) and, as predictable, control NTg mice showed no preference to any of the objects (data not shown). Next, mice were taken to their original cage, while changing the location of one of the objects (i.e., Object B). On return, the exploration time was recorded and defined as the time that the animal spent sniffing or touching the objects with its nose and/or forepaws. The ‘discrimination index’ of OLM calculates the time spent in exploring the displaced object (i.e., Object B) over the total time exploring both objects (i.e., Objects A + B) (Fig. 4C). As expected, 4-week-old control NTg mice exhibited an increased exploration index for the displaced object, indicating a preference for exploring the object in the new location (Fig. 4D). In contrast, and consistent with previous studies [24], untreated 4-week-old R6/2 mice exhibited impaired performance in the OLM test, failing to demonstrate an increase in the exploration index for the displaced object. Importantly, treatment of R6/2 mice with AAV9-PSD95-6ZF-VP64 -but not AAV9-PSD95-6ZF-NoED- slightly, but significantly, reduced the OLM impairment (Fig. 4D).

After conducting the OLM task, animals were taken again to their original cage, while replacing the initially non-displaced object (i.e., Object A) to a new type of object (i.e., Object C) to test NOR (Fig. 4C). NOR evaluates non-spatial learning of object identity and relies on multiple brain regions, including hippocampus, entorhinal, perirhinal, and parahippocampal cortex [6, 8, 17, 119, 120]. We observed a high exploration index for the new object in the entire experimental group, even in the untreated R6/2 mouse group (Fig. 4D), in agreement with previous studies [15, 39]. The results from the OLM and NOR tasks indicate that 4-week-old R6/2 HD mice exhibit impaired cognitive functions associated mainly with hippocampal-dependent spatial learning and memory, rather than in broader cognitive processes involving other brain regions.

To get further support this finding, untreated and treated R6/2 and NTg mice were subjected to another task that heavily relies on hippocampal activity, the Barnes maze test (Fig. 4E). This test aims to evaluate short and long them spatial working memory [9, 38]. The parameter measured in this test is primary latency, defined as the time it takes for an animal to identify a shelter among 19 decoys, consisting of false-bottomed holes. The progression of primary latency was assessed over 6 days, with a rest day interspersed between sessions five and seven (day 5 and day 7 in Fig. 4F). A decrease in primary latency is interpreted as an indication that the animal is capable to recognize the shelter across sessions. Consistent with the previous studies [15, 23, 79], untreated 7-week-old R6/2 mice exhibited long escape latencies at day 1 of the test and showed poor performance in reliably identifying the shelter across sessions. This suggests a hippocampal-dependent spatial learning and memory deficits in R6/2 mice. While untreated R6/2 and HD mice treated with AAV9-PSD95-6ZF-NoED showed overall poorer performance, their reduced latencies over time suggest a residual capacity for learning. In contrast, NTg mice as well as R6/2 mice treated with AAV9-PSD95-6ZF-VP64 performed well from the start, with little improvement across sessions (Fig. 4F). This performance ceiling may reflect rapid task acquisition or indicate that the Barnes maze design lacks sensitivity to detect gradual learning in high-performing animals. Together, the findings from the OLM and Barnes maze tests indicate that neonatal ICV treatment of R6/2 mice with AAV9-PSD95-6ZF-VP64 effectively mitigates hippocampal-dependent spatial learning deficits in young mice (4 weeks old) and that these benefits persist into adulthood (7 weeks old) in R6/2 mice.

PSD95-6ZF-VP64 improves motor skill learning and coordination in R6/2 mice

The accelerated rotarod test (Fig. 5A) is a sensitive assay to detect variations in motor skill learning and coordination. The test involves measuring the time that the animal can remain on a rotating rod (termed latency), which progressively increases its speed from 4 to 20 RPM within a 5-min period [83]. To ensure familiarity with the setup, mice underwent a pre-test session at a constant speed before three consecutive testing trials. To track motor decline as the disease progressed, performance was evaluated at two time points: at week 6, when mice are adolescents (Fig. 5B) and week 9, when mice reach adulthood (Fig. 5C). The accelerated rotarod test showed a decrease close to 80% in latency in 6-week-old and 9-week-old R6/2 mice, compared to NTg mice (Fig. 5B, C). In contrast, R6/2 mice treated with AAV9-PSD95-6ZF-VP64 showed a significant improvement, reaching latencies comparable to NTg controls. This effect was absent in R6/2 mice treated with AAV9-PSD95-6ZF-NoED, whose latencies remained similar to those of untreated R6/2 mice. The findings from the accelerated rotarod test indicate that neonatal ICV treatment of R6/2 mice with AAV9-PSD95-6ZF-VP64 effectively enhances motor skill learning and coordination in young adult R6/2 mice, with these benefits persisting into full adulthood. These findings are consistent with those reported by other authors, who observed a similar impairment in accelerated rotarod test in R6/2 mice from the sixth week onwards [51, 52, 74, 83].

To assess motor strength and rule out confounding factors related to body weight or fatigue from rotarod latency, we performed a grip strength test, in parallel to rotarod test at weeks 6 and 9. The test was conducted using a standardized grip strength meter, where mice were allowed to grasp a metal grid while being gently pulled back until they released their grip (Fig. 5A). The peak force (in Newtons) was recorded over three consecutive trials, and the best performance was considered for analysis. In agreement with the previous studies showing that grip strength was changed starting in R6/2 mice at 12 weeks of age [74], no significant differences were observed in grip strength between untreated R6/2 and NTg mice at week 6 (Fig. 5D) and week 9 (Fig. 5E), indicating that young adult and adult R6/2 mice preserved motor strength. The findings further suggest that the increased latency observed in R6/2 mice in the rotarod test following AAV9-PSD95-6ZF-VP64 treatment is attributable to improved motor learning and coordination rather than changes in muscle strength.

Discussion

Here, we studied R6/2 mice and found that this HD mouse model exhibited reduced hippocampal Dlg4/PSD95 mRNA and protein levels, as well as cognitive (i.e., Barnes maze and OLM) and motor (i.e., accelerated rotarod) impairments. To demonstrate a causal relationship between low Dlg4/PSD95 expression and behavioral deficits, we treated neonatal R6/2 mice with AAV9 carrying the PSD95-6ZF-VP64 sequence. We found that the ATF restored hippocampal Dlg4/PSD95 mRNA and protein levels comparable to those observed in NTg mice. Notably, AAV9-PSD95-6ZF-VP64 also effectively alleviated hippocampal-dependent deficits in spatial learning and memory in young adult R6/2 mice, as well as impairments in motor coordination and motor skill learning. These behavioral benefits persisted into adulthood. Our study confirms Dlg4/PSD95 as a crucial factor in the prodromal phase of HD pathology and highlights ATF PSD95-6ZF-VP64 as a promising therapeutic approach for treating the disease’s early stage.

R6/2 mice recapitulate key aspects of the human disease within a short time frame, making them a relevant HD model for evaluating therapeutic interventions and gaining insights into the underlying disease mechanisms [51, 77]. Starting in adolescence and continuing into adulthood, R6/2 mice exhibit progressive motor impairments, including deficits in coordination, balance, and motor skill learning; these impairments have been consistently observed using the accelerated rotarod performance test [51, 52, 74, 83]. In agreement with these previous reports, we also observed that untreated adolescent R6/2 mice exhibit a progressive decline in accelerated rotarod performance. In addition to motor impairments, spatial learning and memory deficits have been reported in R6/2 mice as early as four weeks of age, manifesting as reduced performance in hippocampal-dependent tasks such as the Barnes Maze and OLM tests [23, 24, 41]. In agreement with these studies, and using the same hippocampal-dependent tasks, we also found that untreated 4-week-old adolescent R6/2 mice exhibit the characteristic progressive spatial cognitive impairments associated with HD. Biochemical assays conducted on whole lysates extracted from the hippocampi further revealed a significant reduction in total Dlg4/PSD95 protein levels in R6/2 mice at both 7 and 14 weeks of age. It is very likely that synaptic loss of Dlg4/PSD95 proteins starts much earlier in hippocampal neurons of R6/2 mice. To gain direct insights in this, synaptic membranes (synaptosomes) can be analyzed. However, in our experience pooling hippocampi of at least three adult mice (12-month-old) is required to achieve a single data point [18]. Due to the large number of animals required to assess whether synaptic Dlg4/PSD95 protein levels are reduced in 4-week-old adolescent R6/2 mice in at least three independent experiments, this objective exceeds the scope of this study. Our data from R6/2 hippocampi aligns with a previous study on this synaptic scaffold molecule, which reported that membrane-associated Dlg4/PSD95 protein levels (i.e., P2 fraction or synaptosome fraction that is enriched in PSD proteins) are significant reduced in hippocampus of 4-week-old R6/2 mice [69]. Significant reductions in the total and PSD-enriched Dlg4/PSD95 protein levels have also been detected in the cortex of 4-week-old [69] and 6, 8, and 12-week-old R6/2 mice [81].

Biochemical studies in other HD mouse models, including in R6/1, Q175 and Hdh^Q111, have also revealed that Dlg4/PSD95 mRNA and/or protein levels are significantly decreased in the hippocampus, cortex and striatum [4, 39, 62, 81, 89, 112]. Furthermore, functional, structural and behavioral studies in R6/2 and other HD transgenic mouse models have found that cognitive decline is associated with reduced spine density and impaired synaptic and extrasynaptic transmission, mediated by NMDARs and AMPARs in the hippocampus [21, 39, 78, 82, 84, 93, 94, 126, 129]. Given that Dlg4/PSD95 is a critical synaptic protein that scaffolds hippocampal AMPARs and NMDARs to signaling molecules in the PSD, thereby regulating synaptic plasticity and dendritic spine stability, and hence learning and memory [32, 114–117], we hypothesized that the cognitive decline and motor skill learning deficits observed in R6/2 mice is mediated, at least in part, by the loss of Dlg4/PSD95 levels. Indeed, treating neonatal R6/2 mice with AAV9-PSD95-6ZF-VP64 restored hippocampal Dlg4/PSD95 protein levels comparable to those observed in wild-type mice, improving spatial learning and memory, motor coordination and motor skill learning, effects that persisted into adulthood.

A limitation of the current study is the exclusive use of the R6/2 transgenic mouse model, which, although widely used and well-characterized, exhibits a rapidly progressing and severe phenotype that may not fully capture the slower, more insidious course of HD in human patients. The R6/2 model expresses a truncated form of the human HTT gene with ~ 150 CAG repeats and develops early-onset behavioral and molecular deficits. While this allows for efficient therapeutic screening, it may overlook more subtle and progressive features of HD. Slower-progressing knock-in models, such as the Q175 or Hdh^Q111 mice, express full-length mutant HTT under endogenous regulatory elements and recapitulate key aspects of human pathology, including gradual synaptic decline, transcriptional dysregulation, and delayed behavioral phenotypes [4, 44, 75]. It would be interesting in future experiments to test whether the ATF PSD95-6ZF-VP64 also benefits these relevant HD knock-in mouse models.

Increasing hippocampal Dlg4/PSD95 protein levels in R6/2 mice treated with AAV9-PSD95-6ZF-VP64 likely improved cognition and motor learning by modulating hippocampal neuron plasticity processes. This notion is supported by data obtained in our previous study, demonstrating that transduction of PSD95-6ZF-VP64 in adult wild-type hippocampal dentate gyrus neurons significantly enhanced synaptic transmission mediated by NMDARs and AMPARs and increased the number of mature mushroom spines [18]. Regarding NMDARs, Dlg4/PSD95 preferentially anchors NR2A-NMDARs but can also indirectly scaffold NR2B subunits as NMDARs can be formed as tri-heteromeric NR1NR2ANR2B NMDAR complexes, which are highly abundant in mature striatum, hippocampus, and cortex [3, 29, 65, 69, 105, 108, 117]. Numerous studies performing electrophysiological recordings in HD mouse models, along with findings from treatments with NMDAR antagonists, indicate that enhanced glutamatergic activity -particularly mediated by extrasynaptic NR2B-containing NMDARs- is a major cause of excitotoxicity and neuronal dysfunction and thereby critically contribute to phenotype onset (for reviews see [21, 85, 93, 94, 129]). Hence, we expect that restoring synaptic Dlg4/PSD95 levels would lead to the reinsertion of NR1NR2A-NMDARs as well as NR1NR2ANR2B-NMDAR complexes within the PSD along with scaffolding proteins such as homer and shank. We also anticipate that restoring Dlg4/PSD95 expression could rescue PSD levels of intracellular signaling molecules Pyk2 and Kalirin-7. In line with this idea, overexpression of Pyk2 and Kalirin-7 rescued synaptic Dlg4/PSD95 levels, as well as synaptic transmission, spine densities and cognitive impairments in HD mouse models [39, 57, 70, 89]. Together, we expect that restoring Dlg4/PSD95 expression in R6/2 mice could rescue (sub)synaptic levels of glutamatergic receptors (NMDARs, AMPARs), scaffolding proteins (shank, homer), as well as intracellular signaling molecules (Pyk2, Kalirin-7), thereby enhancing synaptic transmission, plasticity, and dendritic spine stability. Future experiments employing diverse approaches -including biochemical, molecular, neuroanatomical, and electrophysiological methods- will be essential to test this hypothesis.

Our study using mouse models provides important biological insights by not only validating hippocampal Dlg4/PSD95 as a key player in cognition -previously demonstrated in aging and AD [18]- but also by establishing its critical role in motor skill learning. While this study was performed in the R6/2 mouse model, our findings may have relevant implications for understanding and potentially treating HD patients as Dlg4/PSD95 mRNA and protein levels are diminished in human patients [35, 53, 112, 125]. Moreover, spatial memory and learning emerge early in HD patients, often preceding motor impairments by a decade or more [11, 48]. Consistent with this, studies report hippocampal atrophy and impaired synaptic plasticity in postmortem studies and imaging, respectively [98, 121], underscoring the hippocampus as a relevant therapeutic target. Our data indicate that the targeted expression of the ATF PSD95-6ZF-VP64 improves cognition and motor skill learning in R6/2 mice by restoring hippocampal Dlg4/PSD95 levels to those observed in control NTg mice.

Given that treatment was initiated at neonatal stages, the epigenetic editing strategy most likely prevented the emergence of cognitive and motor skill learning deficits in these HD mice. Hence, preventing the reduction of Dlg4/PSD95 levels in pre-symptomatic human patients through epigenetic editing approaches may offer a promising avenue for early intervention and enhancement of quality of life of the patient. In line with this, current therapeutic strategies for HD are increasingly shifting toward targeting early neuronal dysfunction and impaired synaptic plasticity, rather than focusing solely on later-stage neurodegeneration [104].

It remains also crucial to develop strategies capable of restoring hippocampal deficits in symptomatic HD patients. In our previous study, we demonstrated that stereotaxic administration of AAV9-PSD95-6ZF-VP64 in the hippocampi of 6.5-month-old symptomatic AD mice successfully rescued NOR impairment, restoring performance to levels comparable to wild-type controls [18]. It would be important to perform similar studies in symptomatic HD mice.

In our previous study [18], we showed by chromatin immunoprecipitation (ChIP) assays that N2a cells and primary hippocampal neurons transducing the ATF PSD95-6ZF-VP64 led to increased histone H3 acetylation (i.e., H3ac total, H3K9ac, H3K14ac) and hence enhanced Dlg4/PSD95 gene expression. In vivo studies further demonstrated that the enrichment of histone H3 acetylation during hippocampal development (E18 to > P90) paralleled with the recruitment of histone acetyltransferases (HATs) -including CBP/p300 and GCN5- to the Dlg4/PSD95 gene promoter sequence and increased Dlg4/PSD95 gene expression [18]. It is likely that transduction of the ATF in hippocampal neurons of R6/2 mice increased Dlg4/PSD95 mRNA and protein levels through epigenetic mechanisms involving the recruitment of histone acetyltransferases (HATs), which promote histone H3 acetylation at the Dlg4/PSD95 gene promoter, similar to previous findings using VP64 as an effector domain [18].

We also have previously shown that the ATF PSD95-6ZF-VP64 specifically and efficiently binds to the rat Dlg4/PSD95 locus in hippocampal neurons at a target sequence that is highly conserved across vertebrate species, with 100% identity among mice, rats, and importantly humans, thereby supporting its potential for therapeutic applications in patients [18]. With the initiation of the first human clinical trial involving epigenetic editors earlier this year, and several others expected to begin soon (Chemistry World, 2025: https://www.chemistryworld.com/news/epigenetic-editors-enter-clinical-trials/4021267.article), there is growing optimism that gene therapy and epigenetic editing could offer a novel strategy to combat disease [96]. Beyond the use of AAV9, newly developed viral delivery systems -such as STAC-BBB capsid from Sangamo Therapeutics- enable more efficient delivery of epigenetic editors to the CNS in non-human primates. Therefore, in combination with these emerging delivery technologies, there is increasing hope that epigenome-editing tools, such as the ATF PSD95-6ZF-VP64, could be utilized in future therapeutic interventions to delay the onset and/or progression of HD.

Methods

Animals

All protocols involving mice were carried out according to NIH and ARRIVE guidelines and were approved by the Ethical and Bio-security Committees of Universidad Andrés Bello. The original R6/2 transgenic male mouse, which expresses the 5' end of the human HTT gene [77], was kindly gifted by Dr. Rene Vidal Gomez (Universidad Mayor, Chile). The colony was maintained by backcrossing the R6/2 congenic male offspring with CBA/C57BL/6 J female mice in successive generations. Genotyping was performed at birth (P0) by PCR analysis of tail snips using the following primer set: F: 5 ‘-CCG CTC AGG TTC TGC TTT TA 3’ and R: 5 ‘-TGG AAG GAC TTG AGG GAC TC 3’ [76]. Animals were labeled with an intradermal p-Chip (p-Chip Corporation) and kept with their mothers and littermates until 3 weeks of age. At this point, they were separated by sex and the experimental group. Each group included a non-transgenic animal (NTg), transgenic animal (R6/2), transgenic animal injected with AAV-PSD95-6ZF-VP64 (PSD95-6ZF-VP64), and transgenic control animal injected with AAV-PSD95-6ZF-NoED (PSD95-6ZF-NoED).

Although we did not directly quantify the number of CAG repeats in our animals, all R6/2 breeders were obtained from a well-established colony maintained by paternal transmission, in which male founders carry approximately 150 CAG repeats -a range widely used in studies evaluating behavioral and molecular deficits [74, 77]. While repeat instability may occur under paternal transmission (Mangiarini et al., 1996), controlled breeding conditions help maintain consistent phenotypes in R6/2 mice with approximately 150 CAG repeats (Menalled et al., 2009). Although most experiments were conducted in male R6/2 mice to minimize sex-related variability, behavioral and molecular data did not show evidence of variability that could be attributed to sex. Individual-level data, including sex and treatment allocation, are available in the public repository associated with this study.

The health status of the animals was monitored periodically, and individuals exhibiting significant deterioration or aggressive behavior were excluded from subsequent analyses. Behavioral experiments were conducted between weeks 4 and 14. Hippocampi from experimental groups, aged 7 and 14 weeks were collected for quantitative PCR and Western blot analyses. Mice were anesthetized with isoflurane, and their brains were rapidly extracted. Hippocampi were dissected on ice and quickly frozen in liquid nitrogen, and stored at − 80 °C until analysis.

ICV injections

Recombinant AAV9 were produced at the Virus Unit of the Institute of Biomedical Sciences at Universidad Andrés Bello, following established protocols [18, 50]. The PSD95-6ZF-VP64 and PSD95-6ZF-NoED constructs were driven by the EF1 promoter, and each vector included a separate eGFP reporter gene under the control of the hSyn1 promoter. Viral vectors were titered and diluted to a concentration of approximately 5 × 10¹¹infectious units (IU/µl), and the same viral dose was used for both AAV-PSD95-6ZF-VP64 and AAV-PSD95-6ZF-NoED constructs. Viral administration was performed via ICV injection, following established procedures [2, 16, 56]. Briefly, P0-P) animals were anesthetized by cold exposure, placed on an aluminum plate within an ice bucket, lined with absorbent paper to prevent direct contact with the metal surface and avoid skin injury. A 2 µL Hamilton Neuros syringe (Hamilton Company, Cat. No. 65459–02) was used to bilaterally administer 1 µL of viral solution into the lateral ventricles. The target site for injection was located 0.8–1 mm lateral from the sagittal suture, halfway between lambda and bregma; these landmarks are visible through the skin at P0-2 [56]. To confirm correct injection, some control animals were injected with trypan blue, a non-fluorescent dye, to ensure there was no non-specific fluorescent signal. The expression of ATF was monitored using the eGFP reporter 48 and 72 h after injection, using an In-Vivo FX PRO fluorescent imaging system (Bruker).

Cell cultures

Primary cultures of mouse hippocampal neurons were prepared from P0-P1 CBA/C57BL/6 J animals, as previously described [55]. Briefly, the hippocampal tissue was excised, minced, and enzymatically treated in ice-cold PBS containing 2 mg/ml papain (Sigma-Aldrich, Cat. No. P4762) and 20 mg/ml DNase I (Roche, Cat. No. 10104159001). To facilitate tissue dissociation, samples were incubated at 37 °C for 30 min in a fresh aliquot of the enzymatic solution. The cells were plated on poly L-lysine-coated coverslips in 24-well culture plates and maintained in neurobasal medium (Gibco, Cat. No. 21103–049) supplemented with B-27 (Gibco, Cat. No. 810–1048), GlutaMAX (Gibco, Cat. No. 810–1048), and 1% penicillin–streptomycin (Gibco, Cat. No. 15140–122) at 37 °C in a 5% CO₂ atmosphere. Neurons were maintained under culture conditions for seven days with periodic medium changes. On day 7, the cells were subjected to infection with AAV-PSD95-6ZF-VP64 or AAV-PSD95-6ZF-NoED. Expression of the constructs was confirmed by the observation of eGFP fluorescence, which became detectable within the initial 48 h post-infection.

Immunostaining

Immunostaining of hippocampal neurons for PSD95 were performed as previously described [5, 19, 97]. Briefly, primary cultures of mouse hippocampal neurons were fixed 72 h post infection with fresh 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton X-100 in PBS for 20 min, and blocked with 10% goat serum (Gibco, Cat. No. 50062Z) for 30 min. Cells were then incubated overnight at 4 °C with primary antibodies: PSD95 (Neuromab, Cat. No. 75–028, 1:500, mouse), and HA-Tag (Santa Cruz, Cat. No. sc-271111, 1:800, rabbit), which is located at the N-terminal of the peptide. After three washes with 0.5% Triton X-100 in PBS, the samples were incubated with Alexa Fluor-conjugated secondary antibodies (Alexa Fluor 568 [Invitrogen, Cat. No. A11001], or Alexa Fluor 635 [Invitrogen, Cat. No. A11010]) for 1 h at room temperature. The nuclei were stained with NucBlue (Invitrogen, Cat. No. R37605), and samples were mounted in Fluoromount-G mounting medium (Invitrogen, Cat. No. 4958–02).

Images of hippocampal neurons were taken on a Leica TCS SP8 confocal microscope with a × 63 oil objective (NA = 1.4; HC PL CS2 APO) and digital zooms of 2X and 5X, with a z-step of 0.5 μm optical sections (velocity scan 600 Hz; resolution 1024 × 1024 pixels, equivalent to 87.3 μm × 87.3 μm). The following laser wavelengths were used to detect NucBlue (Ex 405 nm), GFP (Ex 488 nm), Alexa568-HA (Ex 552 nm), and Alexa635-PSD95 (Ex638).

Image analysis and spine morphology assessment

Images were processed and analyzed using ImageJ software [102, 106]. Cells expressing the viral construct were identified by eGFP fluorescence, and the maximum projection of the z-stack was used to determine the localization of the HA-tag. For each condition, three to five full-cell images were captured during the same imaging session, for a total of six sessions. To analyze spine morphology, approximately 60 μm segments of secondary apical dendrites from eGFP/HA-positive cells were manually selected. The ImageJ plug-in SpineJ [64] was used to quantify spine head area.

RT-qPCR

Levels of Dlg4/PSD95 transcripts were quantified as previously described [18]. Briefly, total RNA was isolated from frozen tissues using TRIzol reagent (Invitrogen, Cat. No. 15596026) according to the manufacturer's instructions. The RNA concentration was measured using a NanoDrop One spectrophotometer (Thermo Fisher Scientific). For cDNA synthesis, 2 µg of RNA was mixed with 0.5 µL oligo dT primers (Invitrogen) and adjusted to a final volume of 10 µL with nuclease-free water. The mixture was heated at 72 °C for 5 min to anneal the oligo-dT primers to mRNA. Reverse transcription was performed in a reaction mixture containing 2 µL RT buffer, 0.5 µL reverse transcriptase (NEB, Cat. No. M0253S), 0.5 µL RNase inhibitor (New England biolabs, Cat. No. M0307S), 0.5 µL dNTP mix, and 6.5 µL nuclease-free water. The reaction mixture was incubated at 42 °C for 59 min, followed by enzyme inactivation at 90 °C for 10 min. qPCR was performed using Brilliant II SYBR Green Master Mix (Agilent, Cat. No. 600828) using the following primer sets: Dlg4/PSD95 (F: 5 ‘-AGA CGG TGA CGC AGA TGG AA-3,’ R: 5 ‘-TCG GGG AAC TCG GAG AGA AG-3’) and GAPDH (F: 5 ‘-CAG TCA AGG CCG AGA ATG GGA-3,’ R: 5 ‘-CCT TCT CCA TGG TGG TGA AGA CA-3’) as the reference gene. Melting curve analysis was performed to verify the primer specificity.

Quantitative western blot analysis

Protein levels of Dlg4/PSD95 were quantified as previously described [18], but with minor modifications. Briefly, proteins were extracted from frozen tissues using RIPA buffer (Pierce, Cat. no. 89900), supplemented with protease inhibitors (Roche, Cat. No. 4693132001) and homogenized on ice. Quantification was performed using the Bradford method (Bio-Rad) by measuring the absorbance at 590 and 450 nm [33]. The dynamic linear range for Dlg4/PSD95 and β-actin quantification by western blot analysis was determined using 1, 10, 20, and 30 µg of protein obtained from hippocampal lysates of control NTg and transgenic animals. The results showed that the signal intensity for both Dlg4/PSD95 and actin at 5 µg remained within the unsaturated dynamic range of the acquisition system (Suppl. Figure 1). Next, for quantitative western blot of analysis of all sample conditions, protein samples (5 µg) were resolved on 12% acrylamide gel at 100 V for 90 min. A PageRuler (ThermoScientific, Cat. No. 26616) was used as the molecular weight marker. The proteins were transferred onto methanol-activated PVDF membranes at 250 mA for 2 h at 4 °C. Transfer efficiency was confirmed by Ponceau Red staining, followed by blocking with 5% skim milk in PBS/Tween 20 (0.05%) for 1 h. Immunodetection involved overnight incubation with primary antibodies diluted in PBS/Tween 20 (0.05%) containing 5% skim milk at 4 °C. The primary antibodies used were anti-actin (Sigma, Cat. No. A5441, 1:5000) and anti-PSD95 (Neuromab, Cat. No. 75–028, 1:1000). Membranes were then incubated for 1 h at room temperature with secondary antibodies diluted in the same solution: anti-mouse IgG-HRP (GeneTex, Cat. No. GTX213111-01, 1:10,000), and anti-rabbit IgG-HRP (GeneTex, Cat. No. GTX213110-01, 1:10,000). Detection was performed using SuperSignal West Dura (Thermo Scientific, Cat. No. 34076), and the membranes were imaged with a multispectral detector (Bruker Co.) at 5, 10, and 15-min exposures.

Behavioral test battery

A test battery was designed based on the progression of symptoms in R6/2 animals and similar protocols described in the literature [51]. Specifically, the battery includes the following tests:

Accelerated rotarod test [83]: This test was conducted at two time points, week 6 and week 9, using a 5-animal rotarod apparatus (Panlab, LE8305) to enable parallel measurements across the experimental group. Each session lasted for four days. On day 0, animals were habituated to the rotarod at a constant speed of 4 RPM, with two adaptation trials separated by a 20 min interval. On days 1–3, animals were initially adapted to 4 RPM, after which the rotation speed progressively increased from 4 to 40 RPM over a 5 min period. Latency was defined as the time an animal remained on the rod before falling, with the test automatically stopped via an optical switch. The best latency time from three consecutive trials, each separated by 20 min, was used for the analysis.

Combined NOR and OLM tests [27]: These tests were performed at week 4 to evaluate object recognition and spatial memory. The animals were first habituated to an open-field arena (34 × 34 × 34 cm) with plexiglass walls located inside a soundproof box equipped with a camera (Isolation Cubicle, Ugo Basile, 46,590-PB). Over three consecutive days, the animals spent 5 min per session inside the field. On the fourth day, the animals were exposed to an arena containing two objects placed equidistantly at the center. They were allowed to explore freely until the visit frequency for both objects were similar, typically within 5 min. After a 20 min rest in their home cage, the OLM test was conducted by moving one object to a new location, while the other remained stationary. Animals explored the field for 10 min and the recognition index was calculated as the percentage of time spent exploring the displaced object. Following another 20 min rest, the NOR test was performed by replacing the stationary object with a novel object. Animals explored the field for 5 min, and the discrimination index was calculated as the percentage of time spent exploring a novel object. To ensure consistency and minimize olfactory cues, the arena and objects were cleaned with 70% ethanol between trials, and the animals always entered the arena from the same point. Video analysis for the NOR and OLM tests was conducted using Anymaze (Stoelting Co.).

Barnes maze test: This test was applied during week 7 as a widely used behavioral assay for assessing spatial navigation and short- to medium-term memory in rodents, serving as an alternative to the morris water maze [88]. On day 0, animals were acclimated to the behavioral testing room for 30 min. Visual markers were placed on the walls surrounding the maze to provide spatial cues, and the maze surface was cleaned with 70% ethanol after each trial to eliminate olfactory cues. Testing began the following day and continued for five days, with three trials per day. During each trial, animals were placed under a small box at the center of the maze and marked for identification in the video recordings. After 10 s, the box was lifted, and the animals were allowed to explore freely. If an animal located the escape hole within the 5-min trial period, it was allowed to remain in the refuge briefly before being returned to its home cage. If an animal failed to locate the escape hole, it was gently guided to it and allowed to remain there for a few seconds before being removed. The trial with the shortest latency was chosen for analysis. On day 7, a single-probe trial was conducted, during which the escape hole was blocked, requiring animals to rely solely on spatial memory. Video recordings were analyzed using DeepLabCut [72] to quantify the primary latency, defined as the time taken for the animal to locate the escape hole. This measure assessed spatial learning and memory performance across sessions and conditions.

Statistical methods

The statistical analyses employed in this study were designed to evaluate the hypotheses based on data distribution and variance characteristics. Behavioral data (Barnes Maze, Rotarod, and Grip Test) were collected using a repeated-measures design, as the same cohort of animals was tested at multiple time points. In contrast, biochemical analyses, including Western blot and RT-qPCR, were based on independent measurements, as each subject provided a single tissue sample for analysis. RT-qPCR data were normalized against the reference gene (Gapdh) and compared to the control NTg average, necessitating a distinct statistical approach compared to repeated-measures analyses [36].

Normality was assessed using the Shapiro–Wilk test [107], and variance homogeneity was evaluated using Levene’s test [37]. Parametric tests were applied when assumptions of normality and homoscedasticity were met, whereas nonparametric alternatives were used otherwise.

For pairwise comparisons, Student’s t-test or Welch’s t-test was used depending on variance homogeneity, while the Mann–Whitney U test was used for nonparametric data. Multiple group comparisons were conducted using one-way ANOVA or Welch’s F-test for parametric data, and the Kruskal–Wallis test for nonparametric data [59, 73, 92].

For behavioral experiments involving repeated measures, a repeated-measures ANOVA was applied to assess the effects of time, treatment, and their interaction. When significant effects were found, post hoc pairwise comparisons were performed using Tukey’s or Bonferroni’s correction, as appropriate. The statistical test applied to each experiment and the corresponding sample size are detailed in the respective figure captions.

All data are reported as mean ± S.E.M. Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001 relative to controls (wild-type or NTg mice, as specified in each figure).

All statistical analyses and visualizations were performed in R [91], using custom scripts developed by the author. Key packages include the datawizard [87], ggstatsplot [86], ggsignif [1], bayestestR [71], and effect size [13]. All scripts are available at GitHub (https://github.com/Gfernandezv).

Author contributionS

Conceptualization: BvZ and GF. Methodology design for ATF: FJB and BvZ. Methodology -design for behavioral studies: GF and BvZ. Methodology—WB and RT-qPCR assays: GF and KL. Writing original draft: GF and BvZ. Review and editing: FJB and KL. Figure preparations: GF. Funding acquisition: BvZ.

Funding

This research was funded by ANID-FONDECYT, grant number 1221745 (BvZ); ANID-MILENIO (NCN2023_32, BvZ, FJB); ANID-EXPLORADOR (13220203, BvZ). UNAB DI-06-24/REG (FJB), ANID-FONDECYT 1250955 (FJB).

Availability of data and materials

No datasets were generated or analyzed during the current study.

Declarations

Ethics approval

All protocols involving rodents were carried out according to the NIH, ARRIVE and ANID/CONICYT guidelines and were approved by the Ethical and Bio-security Committees of Universidad Andres Bello.

Competing interests

The authors declare no competing interests.