Artificial microRNAs targeting tau enable post-symptomatic functional recovery in aged tauopathy mice

Abstract

Tauopathies are a group of neurodegenerative disorders, including Alzheimer’s disease, frontotemporal dementia, and progressive supranuclear palsy, characterized by the pathological accumulation of tau protein. While tau reduction has emerged as a promising disease-modifying strategy, most preclinical studies have focused on preventive approaches, and the therapeutic potential after clinical onset remains largely unexplored. This limitation is critical, as patients are typically diagnosed after symptoms emerge. Furthermore, global tau suppression may disrupt physiological tau functions and lead to adverse effects, underscoring the need for targeted interventions. In this sense, RNA interference (RNAi)-mediated therapies using viral vectors offer high specificity, regional and cell-specific expression, and sustained target knockdown. We have engineered artificial microRNAs (Tau-miRNAs) to selectively reduce tau in vulnerable brain regions, minimizing off-target effects. Here, we tested the efficacy of these Tau-miRNAs in a tauopathy mouse model at advanced disease stages, delivering them into the prefrontal cortex after cognitive and electrophysiological deficit onset. This post-symptomatic intervention led to long-term improvements in memory, restoration of neuronal firing properties, and reduced pathological tau at synapses. Our findings highlight the potential of spatially targeted RNA-based tau-lowering strategies for late-stage intervention in tauopathies, addressing a critical unmet need in the treatment of these devastating disorders.

Graphical abstract

Facal and colleagues demonstrate that post-symptomatic delivery of tau-targeting artificial microRNAs (amiRNAs) rescues cognitive deficits, normalizes cortical neuronal firing, and reduces pathological tau accumulation in aged tauopathy mice. These findings provide proof of concept that regionally targeted RNA-based tau lowering can reverse established disease phenotypes and represents a promising therapeutic strategy for tau-driven neurodegenerative diseases.

Introduction

Tauopathies are a group of neurodegenerative diseases characterized by the pathological accumulation of tau protein in certain brain nuclei, leading to synaptic deficits, abnormal neuronal activity, and ultimately neurodegeneration.^1,2 These pathologies include Alzheimer’s disease (AD), progressive supranuclear palsy, and frontotemporal dementia, which have devastating cognitive and behavioural effects.

In the healthy brain, tau plays a myriad of physiological functions, maintaining microtubule dynamics and facilitating intracellular transport, both of which are essential for neuronal homeostasis.^3,4 However, several failures in genetic or metabolic pathways are known to alter tau physiology, leading to the accumulation of pathological tau variants, including hyperphosphorylated, misfolded, or truncated forms.^5,6 Under these pathological conditions, tau dissociates from the axon and relocates to the somatodendritic compartment where it exerts detrimental changes in neuronal function,^7,8,9 assembling into insoluble filaments and finally forming neurofibrillary tangles (NFTs).^6,7,10 Such mislocalization underlies severe neuronal impairments that could begin years before cell death.^10,11 NFTs correlate with the progression of the disease in a discrete pattern initiating in the primary affected areas to the spreading of pathological tau that leads to synapse loss and neurodegeneration.¹² However, evidence suggests that this “gain of toxic function” of tau is less attributed to the highly assembled structures like NFTs and more related to the previous oligomeric forms,^13,14 which are abundantly found in synaptic terminals of patients.⁷

In this context, disease-modifying therapies that lower the levels of intracellular pathological tau species upstream of protein aggregation may prove effective in reversing the mechanisms driving tauopathies.^5,15,16 Noteworthy, reducing or ablating tau can confer protective effects in preclinical models of neurodegeneration, including mitigating excitotoxicity and enhancing synaptic function.^17,18,19,20 Yet, most preclinical studies that validate such interventions so far have focused on preventive strategies, limiting the scope of the therapeutic window. Developing more realistic and efficient therapies would require targeting tau after the onset of clinical symptoms underlying disease.

As a proof of concept, we evaluated whether localized tau reduction, using an RNA interference-based strategy to suppress tau expression, could revert cognitive impairments in a mouse model of tauopathy after phenotypic onset. By targeting tau after symptoms are present, we sought to test the feasibility of rescuing established disease phenotypes in a clinically relevant context. Specifically, we developed artificial tau-targeting microRNAs (Tau-miRNAs), which have previously demonstrated efficient target engagement and tau reduction in both human neurons and mouse brain.²¹ These Tau-miRNAs were expressed using lentiviral vectors, ensuring stable, long-term expression after a single delivery. We performed administration of Tau-miRNAs into the prefrontal cortex (PFC) of aging htau mice after detecting cognitive decline and impairments in neuronal activity. Cognitive deficits were assessed in novel object recognition (NOR) tasks, followed by field recordings to determine neuronal firing patterns in prefrontal neurons. Postmortem analyses showed reduction of pathological tau accumulation, while array tomography showed a marked decrease of phospho-tau clusters at presynaptic terminals after Tau-miRNAs treatment. Together, our findings provide insights into the therapeutic potential of Tau-miRNAs and their ability to reduce pathological tau accumulation and rescue neuronal activity and cognitive function after phenotypic onset.

Results

Local tau reduction in the prefrontal cortex rescues cognitive impairments and modulates neuronal activity in aged htau mice

Based on our previous observation that local injection of Tau-miRNAs effectively prevent pathological tau accumulation in htau mice,²¹ we investigated whether tau reduction shortly after phenotypic onset could rescue or reverse pathological phenotypes in this model. Htau mice accumulate hyperphosphorylated and insoluble tau predominantly in the PFC during aging.^22,23 In fact, Fluorodeoxyglucose-Positron Emission Tomography (FDG-PET) analyses have shown progressive metabolic decline in the prelimbic area of the PFC between 3 and 12 months of age.²⁴ From 6 months onward, htau mice exhibit cognitive decline as evidenced by deficits in the NOR task.²⁴ Additionally, in vivo electrophysiological field recordings reveal exacerbated neuronal firing rates in the PFC of aging htau mice.²¹ To evaluate the therapeutic potential of tau reduction in symptomatic mice, we performed a single administration of Tau-miRNAs into the PFC of htau and wild-type (WT) mice at 6 months of age, when deficits associated with prefrontal dysfunction are already established.

Six-month-old mice underwent baseline behavioural testing prior to injection to confirm cognitive decline in htau mice relative to WT controls. Subsequently, mice were randomly assigned to treatment groups to receive a single injection of either Tau-miRNAs or scrambled control (Scr)-miRNAs into the medial PFC (mPFC). Behavioural, electrophysiological, and molecular analyses were performed at 12 months of age (6 months post-injection) to evaluate phenotypic rescue (Figure 1A). At 12 months, htau mice treated with Scr-miRNA showed significant cognitive deficits in the NOR task, with an average discrimination index (DI) of approximately 50%, indicating impairment to distinguish the novel from the familiar object. In contrast, Tau-miRNAs-injected htau mice demonstrated a robust preference for the novel object, with a mean DI around 70%, comparable to WT littermates (Figures 1B and 1C). Noteworthy, no sex differences in DI were observed (Figure S1A), suggesting both males and females respond similarly to Tau-miRNAs treatment, and total exploration time did not differ between groups (Figure S1B). Within-group longitudinal analysis revealed that all htau mice receiving Tau-miRNA improved their NOR performance post-injection, whereas htau mice injected with Scr-miRNA did not change cognitive performance over time (Figure 1D). Additionally, total exploration time did not differ between pre- and post-injection analyses in either group (Figure S1C).

Figure 1.

Single post-symptomatic injection of Tau-miRNAs in the mPFC rescues cognitive decline in aged htau mice

(A) Timeline of lentiviral injection of Tau-miRNAs in the mPFC of htau mice. Six-month-old mice (htau and WT littermates) were injected after phenotypic onset related to pathological tau accumulation in the PFC of htau mice. Lentiviruses (LVs) contained either a scrambled control miRNA or an equimolar combination of two artificial microRNAs targeting MAPT mRNA. After injection, mice were maintained until 12 months of age, where behavioural, electrophysiological, and molecular analyses were performed. Time course of some pathological phenotypes related to tau accumulation in htau mice are described such as beginning of cognitive decline in the NOR task at age 6 months and the increase of firing rates of pyramidal neurons in the PFC described by electrophysiological field recordings in vivo. (B–D) NOR performance in miRNA-injected groups. (B) Representative traces of exploration, where circles describe the area for the novel (filled with red) and familiar (empty) objects. Discrimination index in the NOR task for each group (C) analyzed at 12 months of age and (D) comparing each animal before (Pre-inj 6m) and after (Post-inj 12m) the injection. For (C), WT Scr-miRNA = 18, WT Tau-miRNA = 13, htau Scr-miRNA = 14, and htau Tau-miRNA = 17; for (D), WT Scr-miRNA = 8, WT Tau-miRNA = 4, htau Scr-miRNA = 7, and htau Tau-miRNA = 9. (E–I) Open field task in miRNA-injected groups. (E) Representative traces of exploration (in red), with the center of the area marked by the black dotted lines. Total distance traveled for each group (F) assessed at 12 months and (G) before and after treatment. Time spent in the center relative to total time of exploration for each group (H) analyzed at 12 months and (I) before and after the injection. For (F and H), WT Scr-miRNA = 16, WT Tau-miRNA = 10, htau Scr-miRNA = 14, and htau Tau-miRNA = 16; for (G and I), WT Scr-miRNA = 4/5, WT Tau-miRNA = 5, htau Scr-miRNA = 4, and htau Tau-miRNA = 6. For (C, F, and H), data are shown as scatter dot plots with bars ±SEM, one-way ANOVA followed by Tukey’s post hoc test; for (D, G, and I), data are shown as scatter dot plots, paired t test. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

To assess hyperlocomotion, another phenotype previously described in the htau model,^22,24,25 mice were also tested in the open field arena (Figures 1E–1G). At 12 months of age, htau mice injected with Scr-miRNA displayed significantly increased spontaneous locomotor activity compared to WT controls (Figure 1F), consistent with previous reports.^22,24,25 In contrast, Tau-miRNA-injected htau mice exhibited locomotor activity levels comparable to those of WT mice (Figure 1F), although longitudinal analyses revealed no changes over time (Figure 1G). Notably, no differences were observed between groups in the time spent in the center of the open field arena (Figure 1H) or in the time spent in the open arms of the elevated plus maze (Figures S1D and S1E) at 12 months of age. However, both parameters were elevated in htau mice at 6 months of age (pre-injection) compared to WT littermates (Figures 1I and S1F), consistent with our previous findings indicating that behavioural disinhibition is a prominent trait of young htau mice that diminishes with age.²⁴

We next assessed whether Tau-miRNA treatment could also normalize dysregulated neuronal activity in htau mice²¹ by performing in vivo extracellular field recordings in the mPFC of 12-month-old animals. Recorded neurons were classified as putative pyramidal cells or interneurons based on their mean spike waveforms (Figures 2A and S2A), and their firing activity was analyzed. Moreover, all groups exhibited burst firing, as represented by the interspike interval (ISI) histograms of representative neurons per group, showing a bimodal distribution with a peak around 10 ms (Figures S2B–S2D). To quantitatively evaluate this bursting behaviour, the burstiness index was calculated for each treatment, discriminating between pyramidal and interneurons as well (see materials and methods). In line with previous findings showing a correlation between increased cortical pyramidal firing rates and cognitive impairment in aging htau mice, Scr-miRNA controls exhibited elevated firing rates (Figures 2B, 2C, and S2E) and increased burstiness (Figure 2D) in pyramidal neurons compared to WT littermates. In contrast, Tau-miRNAs-injected htau mice displayed significantly lower pyramidal firing rates and bursting activity, although firing levels were still higher compared to age-matched WT controls (Figures 2B–2D and S2E). Distribution of firing rates of pyramidal neurons per animal shows that WT mice injected with Tau-miRNA presented a similar pattern compared to WT scrambled controls (Figure S2E).

Figure 2.

Tau-miRNAs-directed expression modulates dysregulated firing and bursting activity of cortical neurons in the mPFC of htau mice

(A) (Middle) Representative traces of an electrophysiological signal band-pass filtered between 300 and 6,000Hz for miRNAs-injected groups. Putative pyramidal neurons (top, red) or putative interneurons (bottom, blue) were classified upon their average spike waveforms. Mice: WT Scr-miRNA = 7, htau Scr-miRNA = 8, and htau Tau-miRNA = 10. (B–D) Firing rate (B), cumulative frequencies (C), and burstiness index (D) of putative pyramidal neurons (pPYR) registered in the mPFC of injected groups. WT Scr-miRNA = 106, htau Scr-miRNA = 100, and htau Tau-miRNA = 123. (E–G) Firing rate (E), cumulative frequencies (F), and burstiness index (G) of putative interneurons (pINT) registered in the mPFC of injected groups. WT Scr-miRNA = 34, htau Scr-miRNA = 42, and htau Tau-miRNA = 37. Each dot represents the mean firing rate (B and E) or the burstiness index (D and G) of each recorded neuron along the session. Black lines indicate the median value per group, and statistical analyses were performed by Kruskall-Wallis followed by Dunn’s post hoc test. For (C and F), statistical analyses were performed by Kolmogorov-Smirnov test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

Interestingly, a similar recovery was observed in interneurons. Scr-injected htau mice showed elevated interneuron firing rates compared to WT controls, while Tau-miRNA treatment reduced interneuron activity to WT-like levels (Figures 2E and 2F). Likewise, interneurons from htau Scr-miRNA controls exhibited an increased burstiness index, suggesting a greater propensity for bursting events, which was normalized by Tau-miRNAs treatment (Figure 2G). Together, these findings demonstrate that post-onset tau silencing via Tau-miRNAs effectively normalizes dysregulated firing patterns in both pyramidal neurons and interneurons in the mPFC of htau mice.

Tau-miRNAs post-symptomatic injection reduces insoluble tau and reverts pathological accumulation in synaptic terminals in the mPFC of aged htau mice

We next aimed to determine whether the phenotypic recovery observed after Tau-miRNAs treatment correlates with reduction in pathological tau accumulation. Western blot analyses performed with 12-month-old mouse brain extracts revealed that Tau-miRNAs postsymptomatic administration reduced total tau protein levels by approximately 30% in the injected mPFC of htau mice (Figures 3A, 3B, and S3A), consistent with our previous findings in presymptomatic—3-month-old—animals.²¹ Moreover, hyperphosphorylated tau (p-tau), detected with the PHF1 antibody (Figures 3C, 3D, and S3B), and insoluble tau species, assessed by sarkosyl-insolubility assay (Figures 3E, 3F, and S3C), were both significantly decreased in Tau-miRNAs-injected htau mice compared to Scr-miRNA controls. These findings indicate that Tau-miRNAs delivery after phenotypic onset not only reduces total tau levels but also diminishes tau pathological burden in its aggregated and hyperphosphorylated forms.

Figure 3.

Tau-miRNAs reduce insoluble and p-tau levels in the mPFC of htau mice

(A) Representative immunoblots and (B) quantification of total tau protein contents in the mPFC of miRNAs-injected mice. WT = 6, htau Scr-miRNA = 4, and htau Tau-miRNA = 4. WT group represents pooled samples from Scr- and Tau-miRNAs injected groups, with similar values obtained for total tau levels. (C) Representative immunoblots and (D) quantification of p-tau (PHF1) contents in the mPFC of miRNAs-injected mice. WT = 4, htau Scr-miRNA = 6, and htau Tau-miRNA = 5. (E) Representative immunoblots and (F) quantification of insoluble tau contents in the mPFC of miRNAs-injected mice. WT Scr-miRNA = 2, htau Scr-miRNA = 5, and htau Tau-miRNA = 4. (G) Representative images obtained by array tomography, showing p-tau (AT180, green) in the mPFC synaptic neuropil of miRNAs-injected mice. Yellow arrows indicate AT180 with the presynaptic marker synapsin 1a (magenta). (H–J) Quantitative analyses of (H) density of AT180/synapsin colocalized puncta, (I) density of AT180 clusters, and (J) density of synaptic boutons in the mPFC. WT Scr-miRNA = 4, htau Scr-miRNA = 3, and htau Tau-miRNA = 5. Data are shown as scatter dot plots with bars ±SEM, and statistical analyses were performed by one-way ANOVA followed by Tukey’s post hoc test. ∗p < 0.05 and ∗∗p < 0.01.

To further investigate the subcellular distribution of hyperphosphorylated tau in htau mice, we performed high-resolution array tomography immunofluorescence to assess the colocalization of p-tau (Thr231) with presynaptic terminals in the mPFC (Figure 3G). As previously described, Scr-injected htau mice exhibited a higher number of p-tau-positive presynaptic terminals compared to age-matched WT controls, along with an increased number of p-tau clusters in the mPFC (Figures 3G–3I). Remarkably, six months after Tau-miRNAs administration, p-tau accumulation at synaptic terminals was significantly reduced in htau mice (Figures 3G, right panel and 3H–3I, red bars). These results indicate that pathological tau accumulation at the synapse can be effectively diminished following sustained suppression of tau synthesis via Tau-miRNAs. Importantly, the number of synaptic boutons remained unchanged across all groups (Figure 3J), suggesting that the reduction in synaptic p-tau is not secondary to synapse loss.

Taken together, these biochemical and histological analyses demonstrate that a single local administration of Tau-miRNAs in the mPFC after phenotypic onset achieves efficient target engagement, leading to a significant reduction in pathological tau accumulation.

Discussion

This study shows that local tau reduction using Tau-miRNAs can effectively rescue cognitive impairments, modulate neuronal activity, and reduce pathological tau accumulation in htau mice after phenotypic onset, providing a strong foundation for the development of targeted tau-lowering therapies for post-onset interventions in the treatment of tauopathies. Artificial Tau-miRNAs administered into the mPFC of aging htau mice reduced total tau protein levels by approximately 30% in the injected area, consistent with previous study.²¹ A significant decrease in insoluble tau species and hyperphosphorylated tau (p-tau) clusters following Tau-miRNAs administration further supports its efficacy in mitigating tau pathology even after phenotypic onset. These effects were accompanied by a rescue of cognitive deficits and hyperactivity, aligning with prior studies showing protective effects of tau downregulation in models of excitotoxicity and behavioural dysfunction.^19,26,27,28 Unlike preventive approaches, our study demonstrates the capacity of tau knockdown to reverse already-established phenotypes. Yet, further studies to evaluate additional mPFC-related behavioural phenotypes in htau mice—such as social interaction tests and working-memory-dependent paradigms—would contribute to determine the impact and translational relevance of tau reduction in rescuing other behavioural deficits.

Electrophysiological recordings in the medial PFC of htau mice further support these findings, showing that Tau-miRNAs treatment normalized the increased firing rate and bursting activity in pyramidal neurons and interneurons. Noteworthy, in contrast to our previous results in which no statistically significant differences were detected in PFC interneurons between htau and WT mice, the present findings indicate that interneuron dysfunction emerges alongside pyramidal neuron hyperactivity as tau pathology progresses, pointing to a broader disruption of cortical network activity in the htau model. The significant differences now observed between htau and WT interneurons are likely attributable to the increased number of recorded animals and neurons in the present study, compared with our earlier work, where only nonsignificant trends were detected and electrophysiological alterations were conservatively attributed only to pyramidal neurons in 12-month-old injected mice.²¹ This modulation is consistent with the proposed relationship between tau and neuronal excitability^29,30 and expands on previous observations by demonstrating that post-onset tau knockdown can mitigate dysregulated cortical activity.³¹

At the molecular level, the observed reduction of insoluble tau and presynaptic p-tau clusters highlights the relevance of tau accumulation at synaptic sites in driving both cognitive and electrophysiological impairments. Importantly, array tomography revealed that Tau-miRNAs significantly reduced p-tau accumulation at presynaptic terminals—compartments known to be affected early in tauopathies⁷—without changes in synaptic bouton density, suggesting that tau-targeting strategies can restore synaptic integrity without compromising its structure. Together, these findings provide mechanistic support for targeting tau to alleviate tauopathy-related neuronal and synaptic dysfunction.

This study underscores the therapeutic potential of targeted, localized tau-lowering strategies to reverse established phenotypes. Since clinical diagnosis often coincides with advanced tau pathology, the ability to rescue function after symptom onset is critical for translational applications. Moreover, the absence of effects on anxiety, general locomotion, and electrophysiological recordings in treated animals, including WT mice receiving Tau-miRNAs injection, highlights the specificity of local Tau-miRNAs intervention and supports its potential safety and tolerability.

Tau reduction targeting MAPT mRNA has emerged as one of the most promising therapeutic strategies under clinical evaluation. Several studies have shown that moderate reductions of up to 50% are well tolerated and do not lead to structural or functional impairments in mouse models,^{19,30,31,32,33,34} although other authors reported that global tau reduction might have detrimental effects in brain function.^35,36,37 A recent phase 1b clinical trial using antisense oligonucleotides targeting tau, has demonstrated the feasibility and safety of MAPT suppression in humans.³⁸ However, challenges remain, including the need for repeated intrathecal administrations and uncertainty about the long-term effects of sustained global tau reduction in the human brain.

In this context, endogenous RNA-targeting offers several advantages: they can be designed using endogenous miRNA backbones to achieve high specificity for MAPT mRNA and their integration into cellular machinery allows for long-lasting and spatially restricted expression in targeted regions, potentially reducing the need for repeated dosing and minimizing off-target effects.^39,40,41,42

While this study demonstrates the efficacy of Tau-miRNAs treatment, it also has inherent limitations. The intervention was confined to a single brain region (the mPFC) and evaluated over a 6-month period, leaving open questions regarding the durability and generalizability of the therapeutic effects. Future studies should address long-term behavioural and circuit-level outcomes, explore systemic or multi-site delivery strategies, and incorporate single-cell multiomic approaches—particularly transcriptomic and proteomic profiling—to define the broader molecular consequences of tau knockdown. In addition, investigating interactions between tau reduction and other pathological pathways, including inflammation, autophagy, and mitochondrial dysfunction, as well as regional and cell-specific spatiotemporal dynamics over the course of treatment, may reveal new combinatorial strategies to halt or reverse neurodegeneration.

Materials and methods

Mice and experimental design

All animal procedures were designed in accordance with the NIH Guidelines for the Care and Use of Laboratory Animals. Protocols were approved by Institutional Animal Care and Use Committee of INGEBI-CONICET. Mice were housed in standard conditions under 12 h dark/light cycle with ad libitum access to food and water. Htau transgenic mice⁴³ in a C57BL/6J background, were obtained from Jackson Laboratories (Bar Harbor, Maine, USA; B6.Cg-Mapt^tm1(EGFP)KltTg(MAPT)8cPdav/J. Strain number: 005491) and bred in-house. To confirm the presence of the human MAPT transgene and the mouse Mapt^−/− background, all mice used in this study were genotyped by PCR as previously described.²² Htau mice were in-house backcrossed to C5BL/6J mice every 12 generations to refresh breeders.

Experimental groups (htau or WT littermates, Table S1) at age 6 months old were randomly allocated to receive either Tau-miRNA (166 + 724; 1:1) or Scr-miRNA as previously described.²¹ At 12 months of age, behavioural tests and electrophysiological in vivo recordings were performed as previously described.^21,24 For protein extraction, mice were sacrificed by cervical dislocation and the injected area was dissected and stored at −80°C until use. For immunofluorescence array tomography analyses, mice were perfused transcardially with 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4).

Design of amiRNAs and subcloning in viral vectors

amiRNAs were designed following described rules for small interfering RNA (siRNA)^44,45 following an in-house method described.^21,46 Target regions were chosen at exon 2/3 junction (alternative exons) and at exon 11 (constitutive exon) to maximize silencing. Thermodynamic values of the duplex were calculated according to the energy at the 5′ end of the sense strand (Es) and the energy of the 5′ end on the strand considered antisense (Eas). Only Eas < Es were selected with ΔT values between −5 and −3. Five initial sequences were obtained targeting the human MAPT transcript, of which 2 were selected after BLAST against human and mouse (sequences with >17 nts off-target match were discarded). This conservative selection reduces the off-target effect of miRNAS.^39,42 The Scr sequence was obtained with the same GC content with less than 15-nt match obtained after BLAST. Selected siRNA sequences were embedded into a microRNA backbone containing the arms of miR-155 followed by the antisense sequence, the loop, and the sense sequence. From the siRNA sense sequence, nts 10 and 11 were removed to allow the formation of a 3D structure that optimizes the binding with the RNA-induced silencing complex system.⁴⁴ Full sequences of the artificial miRNAs used in this study are as follows: [miRNA5′arm -Antisense(21nt) - Loop-sense (19nt)- miRNA 3′arm].

Tau-miRNA 166 (target E2/3):

ACCGGTGTCGACTTTAAAGGGAGGTAGTGAGTGGACCAGTGGATCCTGGAGGCTTGCTGAAGGCTGTATGCTGAATGCCTGCTTCTTCAGCTTTGTTTTGGCCACTGACTGACAAAGCTGAAAGCAGGCATTCAGGACACAAGGCCTGTTACTAGCACTCACATGGAACAAATGGCCCAGATCTGGCCGCACTCGAGATATCTAGAATTCACTAGTGAGCTC.

Tau-miRNA 724 (target E11):

ACCGGTGTCGACTTTAAAGGGAGGTAGTGAGTGGACCAGTGGATCCTGGAGGCTTGCTGAAGGCTGTATGCTGTAATGAGCCACACTTGGAGGTGTTTTGGCCACTGACTGACACCTCCAAGTGGCTCATTACAGGACACAAGGCCTGTTACTAGCACTCACATGGAACAAATGGCCCAGATCTGGCCGCACTCGAGATATCTAGAATTCACTAGTGAGCTC.

Scr-miRNA:

accggtGTCGACTTTAAAGGGAGGTAGTGAGTGGACCAGTGGATCCTGGAGGCTTGCTGAAGGCTGTATGCTGAAATGTACTGCGCGTGGAGACGTTTTGGCCACTGACTGACGTCTCCACGCAGTACATTTCAGGACACAAGGCCTGTTACTAGCACTCACATGGAACAAATGGCCCAGATCTGGCCGCACTCGAGATATCTAGAATTCACTAGTGAGCTC.

amiRNAs were subcloned under the human synapsin promoter between AgeI and EcoRI sites into a lentiviral vector backbone previously described.^{21,22,24,25,46} Lentiviral particles were generated as previously described.²¹ Briefly, HEK293T cells were grown on DMEM, supplemented with 10% (v/v) fetal bovine serum (NATOCOR, Argentina), 0.5 mM L-glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin (ThermoFisher). Cells at 80%–85% confluence were co-transfected with a lentiviral shuttle vector (Tau-miRNA 166, Tau-miRNA 724, or Scr-miRNA) together with helper vectors encoding packaging and envelope proteins (CMVΔ8.9 and CMV-VSVg, respectively). Viral particles were harvested from the culture medium 36 h after transfection and treated with RNase-free DNase I (Thermo Fisher). Viral vectors were purified by centrifugation and filtering (45-μm pore), concentrated by ultracentrifugation at 100,000 × g (Ti 90 rotor, Beckman), and resuspended in sterile PBS. After performing titration, 10 μL aliquots of viral particles were stored at −80°C.

Stereotaxic injections

Lentiviruses were delivered into the mPFC as previously described.^21,24 Briefly, mice (males and females) aged 22–26 weeks (weight 25–32 g) were anesthetized with isoflurane 0.5%–2.5% (2.5% for induction/0.5%–1% for maintenance, Baxter) in medical-grade oxygen with an air flow at 2.5 L min⁻¹ and placed into a stereotactic frame (Stoelting CO.). A 10-μL Hamilton syringe coupled to a 36G stainless steel tube (Coopers Needle Works Ltd, UK) was used to inject 1.5 μL of lentiviral suspension (0.5 × 10⁷ transducing units (TU)/mL; 0.2 μL/min) per site of injection, bilaterally, at four sites into the mPFC, following coordinates of mouse atlas (Paxinos and Franklin, 2013) (in mm): AP = +2.3, LM = ±0.5, DV = −1.8 and −2.2. One hour before surgery, mice received subcutaneous injection of systemic analgesic (meloxicam, 5 mg/kg), which was repeated 24 and 48 h after surgery. Local analgesic was injected before surgery (lidocaine 0.5%, 5 mg/kg). Any animal showing signs of pain or discomfort after surgery was sacrificed following the endpoint protocol.

Protein extraction and western blotting

mPFC was dissected and homogenized with a buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 2 mM EGTA, and protease and phosphatase inhibitor cocktail (Thermo Fisher). Tissue disruption was performed with a motorized tissue grinder followed by sonication, and protein extracts were centrifuged at 13,500 rpm for 15 min at 4°C. Western blotting was performed as previously described.²¹ Equal amounts of total protein (determined with Pierce BCA Protein Assay Kit, Thermo Fisher) were separated on 10%–12% SDS-polyacrylamide gels (prepared with acrylamide and N,N′-methylenebisacrylamide 30%) and transferred using a semi-dry transfer system to nitrocellulose membranes (Bio-Rad). SeeBlue Plus 2 (Thermo Fisher) was used as a molecular weight marker. Membranes were blocked in 5% (w/v) non-fat dry milk and 0.05% v/v Tween 20 in TBS for 1 h at room temperature. Primary antibodies were used in diluted blocking solution to incubate blots overnight at 4°C: PHF1 (1:1,000, mouse monoclonal, provided by Dr. Peter Davies), anti-total tau (1:10,000, rabbit polyclonal, Dako, Denmark), and anti-β-actin-HRP conjugated (1:10,000, mouse monoclonal, Sigma-Aldrich). After washing 3 times in TBS containing 0.05% v/v Tween 20, blots were incubated with secondary antibody goat anti-mouse or rabbit-HRP conjugated (1:2,000; Thermo Fisher) for 2 h at room temperature. Proteins were visualized using ECL reagent (Thermo Fisher) exposing membranes on the GeneGnomeXRQ (Syngene). Optical density was quantified using FluorChem software (Alpha Innotech), and p-tau or total tau contents were normalized to actin, used as a loading control.

Sarkosyl insolubility assay

Fractionation of insoluble proteins was performed using 250 μg of mPFC protein extract, which was incubated with 1% sarkosyl reagent (Sigma-Aldrich) for 1 h in minimal agitation at room temperature, following the protocol previously reported.²¹ Protein extract with 1% of sarkosyl reagent was ultracentrifuged at 39,000 rpm (1 h) at 20°C to obtain the pellet and then washed with sarkosyl 1% by ultracentrifugation at the same conditions for 15 min. The pellet was resuspended for 1 h at room temperature for western blotting.

In vivo electrophysiological recordings

Data acquisition

Electrodes for extracellular recording were made as previously described.²¹ Mice were deeply anesthetized with isoflurane (2% for induction, and 0.5%–1% for maintenance, Baxter) and placed into a stereotaxic frame. The skull was exposed to clearly locate bregma. A craniotomy was performed over the mPFC coordinates (AP = +2.1 mm, LM = ±0.5 mm, bregma as reference). The tetrodes were lowered inside the brain at a speed rate of 10–20 μm/s. Stable spontaneous action potentials were sought up between −1 and −2.5 mm from the surface. Electrophysiological data were recorded at different positions (up to 7 recordings were acquired per animal), with durations ranging from 15 to 30 min each.

Data processing and analysis

Analysis and statistical tests were implemented in MATLAB (The MathWorks Inc., USA). Raw signals were band filtered between 300 and 6,000 Hz. Spikes sorting was performed as follows. An automatic threshold was set at five times the standard deviation above the mean to detect the spike events. Detected spikes were partitioned into many clusters with a k-means method and then were aggregated according to their interface energy for each pair. Clusters were manually split and merged according to their principal components for the subsequent analysis. Single units were classified into two groups based on their mean spike-width or waveform, measured as the time from the trough to the next peak of the mean action potential. Units with a “valley-to-peak” (y axis in the figure) greater than 440 μs were considered as putative pyramidal neurons (Figure S2A, red dots) and otherwise as putative interneurons (Figure S2A, blue dots). In this way, we distinguished neurons based on their average spike waveform, regardless of their firing rate. Rasters at 1-ms resolution containing a sequence of 0’s (no spike event) and 1’s (spike event) were constructed for each isolated unit. On these rasters, the firing rate (spikes per second) and the ISIs were computed for each neuron. To measure the degree of burstiness in the firing of a given neuron, an autocorrelogram with time shifts ranging from 0 to +500 ms was computed from the rasters. Next, the time shift (Δt) that reached half of the accumulated autocorrelogram value was computed. Burstiness index was defined as $BI=\left(500ms-\Delta t\right)/500ms$.⁴⁷

High-resolution immunofluorescent array tomography

After transcardiac perfusion of mice, brains were post-fixed in 4% paraformaldehyde overnight at 4°C and incubated in 15% sucrose for 24 h and then in 30% sucrose for another 24 h. Tissue blocks containing the mPFC were cut to 300-μm-thick sections using a vibratome and then processed for array tomography as previously described,⁴⁸ using LR White resin (medium grade, Ted Pella, USA). Embedded tissue was cut with a Jumbo Histo Diamond Knife (DiATOME) in an ultramicrotome (Reichert-Jung, Germany). A series of 20–30 sections 200 nm in thickness were collected in ribbons onto glass coverslips and processed for immunofluorescence. Antibodies anti-p-tau (Thr231) AT180 (1:100; mouse; Thermo Fisher) and anti-synapsin-1a (1:200; rabbit; Cell Signaling Technology) were used. Fluorescent-conjugated secondary antibodies raised in donkey (Alexa 488 and Alexa 647, 1:100; Jackson ImmunoResearch, UK) were used. Sections were mounted on glass slides with SlowFade Gold Antifade (Life Technologies) and then imaged in a Leica DMR fluorescence microscope using a PL APO 63X NA = 1.32 oil objective and a Retiga R1 camera (Q-Imaging, UK). Serial images were aligned and converted into stacks using Fiji. A sampling mask of 120 μm × 120 μm in the mPFC was used for quantitative analysis using the Analyze Particles function, yielding values of puncta (for Synapsin) or clusters (for AT180) per μm³.

Behavioural tests

Mice tested were sibling cohorts of age 6 or 12 months depending on the experimental design. NOR, open field, and elevated plus maze tasks were performed as described previously.^21,22,24,25 Experiments were conducted between 13:00 h and 17:00 h under dim illumination, in a separated behavioural room, where mice were transferred in advance. Recordings were analyzed by ANY-maze (Stoelting Co.). All arenas and devices were cleaned between subjects to minimize odor cues.

Statistical analyses

Data were analyzed with GraphPad Prism software. Datasets of each experiment were classified according to the p value obtained for the Shapiro-Wilk test, which determines normality. If the dataset passed the test (p value > 0.05), the data structure was classified as normal distribution, and if the dataset did not pass the test (p value < 0.05), the data structure was classified as non-normal distribution. For datasets classified as normal distribution, statistical tests used for comparing groups depended on the number of groups and independent or dependent variables used in the experiment: unpaired t test (two groups, one independent variable), paired t test (two groups, one dependent variable), and one-way ANOVA test (three or more groups, one independent variable). For one-way ANOVA, Tukey’s post hoc test was used. When datasets had non-normal distribution, statistical tests used for comparing groups were the non-parametric tests: Kruskall-Wallis test (followed by Dunn’s post hoc test) and two-sample Kolmogorov-Smirnov test (for comparisons of cumulative distribution of the datasets).

Data and code availability

All raw datasets and videos are available upon request.

Acknowledgments

This work was supported by the National Scientific and Technical Research Council of Argentina (CONICET), Argentina Ministry of Science (MINCYT-RIP) program Argentina Redes de Alto Impacto, the National Research Agency (ANPCyT), and CurePSP. We thank Caro Lell for technical assistance, Mariano DiGuilmi for insightful discussion, and E. F. Arce for permanent support.

Author contributions

Conceptualization, M.E.A.; funding acquisition, M.E.A.; investigation, formal analysis, and resources, C.L.F., I.P.-P., A.E.P., R.C.-D., C.G., R.F., and M.S.-R.; project administration, C.L.F. and M.E.A.; supervision, M.S.-R. and M.E.A.; visualization, C.L.F.; writing – original draft, C.L.F. and M.E.A.; writing – review & editing, all authors.

Declaration of interests

Tau miRNAs are under provisional patent WO/2025/027577 (CONICET; inventor M.E.A).