AAV5-miHTT gene therapy demonstrates suppression of mutant huntingtin aggregation and neuronal dysfunction in a rat model of Huntington’s disease

Huntington’s disease (HD) is a fatal progressive neurodegenerative disorder caused by a mutation in the huntingtin (HTT) gene. To date, there is no treatment to halt or reverse the course of HD. Lowering of either total or only the mutant HTT expression is expected to have therapeutic benefit. This can be achieved by engineered micro (mi)RNAs targeting HTT transcripts and delivered by an adeno-associated viral (AAV) vector. We have previously showed a miHTT construct to induce total HTT knock-down in Hu128/21 HD mice, while miSNP50T and miSNP67T constructs induced allele-selective HTT knock-down in vitro. In the current preclinical study, the mechanistic efficacy and gene specificity of these selected constructs delivered by an AAV serotype 5 (AAV5) vector was addressed using an acute HD rat model. Our data demonstrated suppression of mutant HTT messenger RNA, which almost completely prevented mutant HTT aggregate formation, and ultimately resulted in suppression of DARPP-32-associated neuronal dysfunction. The AAV5-miHTT construct was found to be the most efficient, although AAV5-miSNP50T demonstrated the anticipated mutant HTT allele selectivity and no passenger strand expression. Ultimately, AAV5-delivered-miRNA-mediated HTT lowering did not cause activation of microglia or astrocytes suggesting no immune response to the AAV5 vector or therapeutic precursor sequences. These preclinical results suggest that using gene therapy to knock-down HTT may provide important therapeutic benefit for HD patients and raised no safety concerns, which supports our ongoing efforts for the development of an RNA interference-based gene therapy product for HD.


INTRODUCTION
Huntington's disease (HD) is a fatal, currently untreatable, neurodegenerative disorder with a prevalence rate of 1-10 in 100 000 individuals worldwide. 1 The presence of 40 and more CAG triplets in exon 1 of the huntingtin (HTT) gene has been identified as a fully penetrant trigger for the neuropathological process, which usually stretches over decades. 2,3 The resultant neuronal death affects primarily GABAergic medium spiny neurons in the early stage of HD, as well as neurons in other brain regions in the later stages. 4,5 The CAG expansion is translated into a polyglutamine (polyQ) tract in the N-terminus of the HTT protein, causing the mutant HTT to misfold and aggregate. 6 Growing evidence suggests that the mutant HTT disturbs multiple critical cellular pathways and that its aggregation is a prerequisite to neurodegeneration. Therefore, clearance of mutant HTT is currently accepted as being key for HD treatment. 7 Notably, Yamamoto et al. 8 showed that conditional blockage of HTT expression after symptom onset results in clearance of HTT aggregates and behavioral improvements, suggesting that HD may be partially reversible.
In contrast to other progressive neurodegenerative disorders, such as Alzheimer's or Parkinson's disease, the monogenic nature of HD allows for the development of disease-modifying therapies that aim to halt or suppress production of the aberrant HTT. The discovery of RNA interference (RNAi) in 1998 revolutionized the progress of therapeutic interventions focusing on gene silencing at the post-transcriptional level. 9 Since then, several groups, including ours, applied RNAi principles to design artificial small interfering (si)RNAs, short hairpin (sh)RNAs or micro (mi)RNAs that bind to the HTT transcript and reduce its translation. 10,11 To circumvent several challenges regarding the delivery and stable expression of RNAi products in the central nervous system, many studies have used viral vectors as delivery vehicles, which can be injected directly into the brain. reviewed in Kantor et al., 12 Adeno-associated viral (AAV) vectors are the most common vehicles of choice and a large number of AAV capsid serotypes provide cell-and tissue-specific tropism. reviewed in Srivastava 13 For the central nervous system, studies in rodents and non human primates have shown AAV serotype 5 (AAV5) to be strong and effective in the brain, making it an attractive candidate for the RNAi-based gene transfer. [14][15][16][17] Importantly, the AAV-delivered-miRNA-based gene therapy approach comprises continual expression of artificial miRNAs following a single administration of an AAV vector, resulting in long-term HTT lowering.
To develop a disease-modifying gene therapy for HD, we have previously designed several therapeutic miRNAs targeting either both 'total' or preferentially the mutant HTT transcripts 'allele-selective' and their HTT knock-down efficiency was addressed in vitro and in the humanized transgenic Hu128/21 HD mouse model. 18 Based on this study, we selected a miHTT construct that showed the strongest efficacy in vitro and induces a potent total HTT knock-down in the striatum and cortex of the Hu128/21 HD mice. 18 For the allele-selective HTT knock-down, we selected the miSNP50T and miSNP67T constructs that showed strongest efficacy in vitro.
Although several transgenic or knock-in HD animal models have been established and used for preclinical testing, none completely recapitulates the neuropathology that occurs in HD patients. [19][20][21] Therefore, a combination of several in vivo preclinical studies is required to address the necessary treatment outcome and safety measures for HD before entering the clinic. In respect to the latter, the present study was designed to evaluate the on-target efficacy of continuously expressed miHTT, miSNP50T, and miSNP67T constructs in suppressing the neuropathology associated with HD using an acute lentiviral (LV) HD rat model. To generate the HD rat model, wild-type rats were injected intrastriatally with a LV expressing a chimeric mutant HTT fragment, which is shown to induce local formation of mutant HTT aggregates followed by severe neuronal dysfunction at two months post-infection. 22 Therefore, this model allowed us to address the HD treatment response downstream of the mutant HTT protein in a larger rodent brain. Moreover, the chimeric LV sequence enabled us to study both total and allele-selective approaches in the context of mechanistic efficacy and gene specificity. 23 RESULTS miHTT-155 delivered by an AAV5 vector suppressed mutant HTT aggregate formation and DARPP-32-associated neuronal dysfunction in HD rats To assess distribution, mechanistic efficacy and allele selectivity of the therapeutic miRNA sequences, we initiated a pilot study in the acute LV HD rat model using the miHTT construct for the total HTT knock-down and the miSNP67T construct for the allele-selective knock-down. Both miHTT and miSNP67T constructs were previously designed in the engineered mmu-miR-155 precursor, named miHTT-155 and miSNP67T-155 ( Figure 1a). The miHTT and miSNP67T expression cassettes also contained a sequence encoding green fluorescent protein (GFP) to assess transduction efficiency ( Figure 1b). 18 We generated AAV5 viruses carrying the miHTT and miSNP67T expression cassettes and co-injected each virus bilaterally in the striatum of rats with a LV vector encoding a chimeric human mutant HTT sequence (Figures 1b and c). The LV vector consisted of an 82-long glutamine (82Q) chain fused with the target regions for the miHTT and miSNP67T constructs, named LV-mtHTT-67T. 23 To address the allele-selective potential of the miSNP67T construct, we included in this study a second LV vector named LV-mtHTT-67C. The AAV5-miSNP67T-155 expression product perfectly matches LV-mtHTT-67T transcripts and has one nucleotide mismatch with LV-mtHTT-67C transcripts at the single nucleotide polymorphism (SNP) rs362307. Therefore, this system enabled us to test the allele selectivity based on only a single SNP. The mutant HTT protein that is expressed from both LV vectors is known to cause HD-like neuronal dysfunction. 22 Two months post-injection, rats were euthanized and the brain tissues were analyzed for AAV5 vector distribution, mature miHTT expression, on-target silencing efficiency measured by mutant HTT aggregate formation and DARPP-32-associated neuronal dysfunction. To evaluate AAV5 vector distribution in the HD rat brain, we performed immunohistochemistry (IHC) against GFP on the fixed striato-cortical sections (Figure 2a). For all AAV5 constructs, we observed broad striatal with partial cortical GFP distribution. Consistent with the latter, we observed~1800 times more miHTT molecules in the AAV5-miHTT-155-injected rats compared with the saline control ( Figure 2b).
Optimized miHTT-451 and miSNP50T-451 expressed from AAV5 vectors induced the mutant HTT mRNA knock-down in HD rats Although the engineered mmu-miHTT-155 constructs generates the favorable miHTT effector sequence and is suitable for pilot experiments addressing efficacy and distribution, it is not optimal for clinical studies in humans due to a high risk of immune response to GFP and possibility of off-target silencing caused by the passenger strand. 18 Previously, we showed that the silencing efficacy and processing of the miHTT construct are influenced by the pre-miHTT precursor. 18 We identified the miHTT-451 precursor that showed no passenger strand in vivo opposite to the miHTT-101 and miHTT-135 precursors ( Figure 3a). Moreover, the miHTT-451 construct showed efficacious HTT knock-down and no signs of toxicity. Therefore, in this study we included the miHTT-451 precursor for the total HTT knock-down. For the allele-selective approach, we optimized the miSNP50T construct designed to preferentially bind to HTT messenger RNA (mRNA) carrying the U isoform of SNP rs362331, which is associated at high frequencies with HD. 25 This miSNP50T construct, which exhibits stronger efficacy than miSNP67T in vitro, was not available when the pilot study was initiated and therefore, it was introduced at this stage. Moreover, for both constructs the cytomegalovirus promoter together with GFP was replaced by the strong cytomegalovirus immediate-early enhancer fused to chicken βactin (CAG) promoter, which has been shown to be effective in the brain. 26 The resultant constructs are named miHTT-451 and miSNP50T-451 ( Figure 3b).
To further establish mechanistic efficiency, safety, as well as allele selectivity in vivo, we generated AAV5-miHTT-451 and AAV5-miSNP50T-451 vectors and co-injected them bilaterally in the striatum of rats with a LV vector encoding a chimeric mutant HTT sequence with an 82Q chain and carrying target regions for the miHTT and miSNP50T constructs, named LV-mtHTT-50T (Figures 3b and c). In order to address the allele-selective potential of the miSNP50T construct, we included a second LV vector, named LV-mtHTT-50C. The AAV5-miSNP50T-451 expression product was designed to have one nucleotide mismatch with the LV-mtHTT-50T transcripts and two nucleotide mismatches with the LV-mtHTT-50C transcripts. Therefore, this system enabled us to test the allele selectivity based on only a single SNP. As negative controls, we included a GFP expressed from the CAG promoter and saline control. Two months post-injection, rats were euthanized and the brains were processed to assess on-target efficacy, HD-like neuronal dysfunction, and immune reaction.
To measure AAV5 vector DNA in striata, we performed real-time quantitative PCR (qPCR) with primers directed towards the CAG promoter ( Figure 3d). We observed high and comparable vector DNA in the striatal homogenates from AAV5-GFP-, AAV5-miHTT--451-and AAV5-miSNP50T-451-injected rats. Similar to TaqMan assays (data not shown), we detected a background level of 3000 genome copies in the saline control, which was considered for calculations of final vector DNA. Furthermore, the latter inversely correlated with the human-specific HTT mRNA knockdown from the same tissue homogenate demonstrated by TaqMan RT-qPCR (Figure 3e). We detected 62.3% ± 21.6 HTT mRNA knock-down in AAV5-miHTT-451-and 81% ± 4.3 HTT mRNA knock-down in AAV5-miSNP50T-451-injected rats as compared with the saline control. We observed similar HTT mRNA knock-down by AAV5-miSNP50T-451 in LV-mtHTT-50T-injected rats compared with LV-mtHTT-50C-injected rats where 65.5% ± 9.2 HTT mRNA reduction was detected.
AAV5 viruses delivering the miHTT-451 or miSNP50T-451 precursors did not induce an overt immune response via GFAP and Iba1 activation Although AAV-based gene therapy is an attractive approach for the delivery of gene products, such as therapeutic miRNAs, and overall has demonstrated low toxicity to date, activation of the immune response should be addressed for each therapeutic candidate in a given target tissue. To evaluate the immune response to our AAV5 vectors expressing therapeutic miRNA precursors in the brain, we analyzed the activation of microglia and infiltration of astrocytes in the injected areas by IHC using an anti-ionized calcium-binding adapter molecule 1 (Iba1) and antiglial fibrillary acidic protein (GFAP) antibody, respectively (Figures 5a and b). We found similar levels of immunoreactivity in AAV5-miHTT-451-and AAV5-miSNP50T-451-treated rats compared to the saline control, indicating no apparent activation of immune reaction via microglia and astrocytes.
AAV5-miHTT-451 and miSNP50T-451 constructs do not generate a passenger strand in vivo Previously, we showed that the miHTT-451 construct does not generate a passenger strand and is not toxic probably due to higher specificity in vivo (Figure 6a). 18 The absence of the passenger strand from the miR-451 scaffold provides a better safety profile for therapeutic shRNAs or miRNAs since it reduces the chance for off-target suppression of other genes by the passenger strand. 18,27,28 To establish the absence of a passenger strand for the miSNP50T-451 construct in vivo, we analyzed the processing patterns in the striatum of AAV5-miSNP50T-451injected rats by next-generation sequencing. We observed On-target knock-down efficacy of AAV5-miHTT gene therapy for HD J Miniarikova et al consistent processing patterns between the analyzed samples ( Figure 6b). The most abundant reads were 23 nucleotides long and belonged to the miSNP50T guide strand (Figure 6c). We did not observe any reads belonging to the passenger strand, confirming the absence of a passenger strand associated with the miR-451 precursor.
Based on this study, we identify the AAV5-miHTT-451 construct as the most efficient candidate for the gene therapy approach targeting HTT by showing a robust suppression of crucial HD pathological features in the rat model of HD.

DISCUSSION
In the current study, we have demonstrated the feasibility of total and allele-selective HTT silencing induced by therapeutic miRNAs delivered to the brain. Intracerebral administration of AAV5 vectors delivering the miHTT candidates for total HTT knockdown or the miSNP50T and miSNP67T candidates for the alleleselective approach resulted in suppression of mutant HTT aggregates and prevention of DARPP-32-associated neuronal dysfunction in rat striata. The AAV5-miHTT-451 construct recapitulated the in vitro observations by demonstrating the strongest efficacy in vivo.
The LV HD rat model very closely and rapidly reproduces the critical pathogenic features of HD: the mutant HTT aggregation and striatal neuronal dysfunction. Although this model does not permit study of improvement in HD-like behavioral symptoms, it provides a means to preselect therapeutic candidates to continue with high-cost preclinical studies in large animals that carry on for several months to years. Previous studies with shRNAs showed equivalent silencing efficacy of the chimeric HTT constructs in rats and HEK-293T cells compared with silencing of the endogenous full-length human HTT in HD-derived neuronal cultures, suggesting that accessibility of silencing agents to the chimeric HTT constructs resembles the natural situation. 23 Moreover, the rapid severity of DARPP-32-associated neuronal dysfunction challenges the read-out efficacy; but on the other hand, it strengthens confidence in case efficacy is demonstrated. (d) qPCR to determine AAV5 genome copies (gc) in the striatum of AAV5-miHTT-451 and AAV5-miSNP50T-451 injected rats (n = 3), two months post-injection. Primers directed to the CAG promoter were used and the gc values were calculated based on the standard curve and considering the background signal from the negative control. (e) TaqMan qPCR assay shows HTT mRNA knock-down in the striatum (n = 2-3) induced by the AAV5-miHTT-451 and AAV5-miSNP50T-451 expression products. Human HTT-specific exon-spanning primers were used and HTT values were subsequently normalized to GAPDH, an internal control set at 100%. All data were analyzed using one-way ANOVA. *Pp0.05; **Pp0.01; ***Pp0.001; ****P ⩽ 0.0001. The values were calculated as a mean ± s.d.
Most HD patients are heterozygous at the HTT locus and therefore many studies attempt to target only the mutant HTT. However, the nature of the HD mutation makes allele-selective HTT silencing challenging. To distinguish between the mutant and healthy HTT alleles, most of the therapeutic interventions so far have focused on targeting the heterozygous HD-associated SNPs or CAG expansions. 23,25,[29][30][31][32][33] However, not all HD patients have the identified common SNP isoforms linked to HD. Therefore, more than one product would be needed to fully address the HD patient population. For instance, one study showed that five therapeutic products targeting the isoforms of three different SNPs are required to treat three-quarters of the United States and European HD patient populations. 31 Moreover, as a personalized medicine, additional genotyping will be required to identify suitable HD candidates for a given clinical trial. Furthermore, the allele selectivity based on targeting the CAG expansions depends on a difference between the consecutive CAGs located on the mutant and healthy HTT alleles. Nevertheless, it has not shown to be more efficacious than the SNP approach, when simultaneously tested. 18,32 Ultimately, the presence of CAG repeats in more than 66 genes increases the chance for off-targeting, making it a less attractive target for clinical testing. 34 Consequently, the allele-selective approach still faces obstacles to advance into the clinical development.
In this study, we identified AAV5-miSNP50T-451 that showed allele-selective silencing of the HTT allele carrying the T isoform of SNP rs362331. The ongoing experiments with the Hu128/21 mice that express full-length human HTT with the U isoform of SNP rs362331 will provide more insights into efficacy and alleleselectivity of the AAV5-miSNP50T-451. The AAV5-miSNP67T-155 construct showed no allele-selectivity in vivo. mfold prediction analysis indicated that the secondary RNA structures of miHTT, miSNP67T and miSNP50T target sites do not form stable duplexes (data not shown), supporting their similar accessibility to the therapeutics. Previously, it has been reported that allele-selective potential of artificial miRNAs observed in vitro may not be retained in vivo. 35 In general, miRNA-mRNA interactions are determined by the sequence composition that influences the thermodynamics, such as the melting temperature, of the duplex RNA. 36 miSNP67T binds to 52%-rich GC region, which could enhance the miRNA: mRNA pairing stability and ultimately contribute to no alleleselectivity based on a single SNP observed in vivo.
On-target knock-down efficacy of AAV5-miHTT gene therapy for HD J Miniarikova et al employs this approach (https://clinicaltrials.gov/). Preclinical studies using HD animal models and non human primates indicate that there is a therapeutic window where total HTT knock-down is safe and yet beneficial for HD patients. [37][38][39] Importantly, this strategy would provide a treatment for all HD patients independent from their genotype. In this study, we showed that the miHTT construct processed from the miR-451 scaffold is able to strongly suppress mutant HTT aggregation and ultimately dramatically reduce the generation of striatal lesions at two months post-injection. Moreover, we did not observe immune reaction to the AAV5-miHTT-451 vector via astrocytes or microglia activation, suggesting a favorable safety profile.
Altogether, these data provide strong evidence for the therapeutic efficacy of the AAV5-miHTT-451 vector by inducing functional improvements downstream in the HD pathological process without raising safety concerns in a rodent model of HD. These promising results support our ongoing efforts for the development of an HD-modifying treatment by silencing the human HTT gene.

HTT target sequences
Homo sapiens HTT mRNA, complete CDS (gb|L12392.1|HUMHDA) obtained from http://blast.ncbi.nlm.nih.gov/Blast.cgi, was used to identify target sequences for the artificial miHTT targeting exon 1, the miSNP50T construct binding to the U isoform of SNP rs362331, and the miSNP67T construct targeting the U isoform of SNP rs362307.

DNA constructs
To generate the miHTT-155 and miSNP67T-155 constructs, their guide sequences targeting HTT transcripts were embedded into the engineered murine pre-miR-155 backbone of pcDNA6.2-GW/EmGFP-miR (Invitrogen, Carlsbad, CA, USA) by annealing complementary oligonucleotides (Qiagen, Valencia, CA, USA) followed by ligation into the linearized pcDNA6.2 plasmid. 18 To generate miHTT-451 and miSNP50T-451 constructs, the guide and passenger strand sequences were incorporated in the hsa-pri-miR-451a scaffold. 200 nucleotides long 5′ and 3′ encompassing flanking regions were included with EcoRV and BamHI restriction sites and the complete sequences were ordered from GeneArt gene synthesis (Invitrogen). In Figure 5. No apparent activation of microglia and astrocytes following AAV5-miHTT-451 and AAV5-miSNP50T-451 injections in HD rats. (a) IHC against Iba1 to show microglial activity. A representative picture of the right hemisphere is shown. (b) IHC against GFAP to show the astrocyte activity. A representative picture of the right hemisphere is shown. Figure 6. Next-generation sequencing analysis of the pre-miHTT-451 and pre-miSNP50T-451 processing patterns shows no passenger strand in vivo. (a) Sequence distribution (%) of reads mapping to pre-miHTT-451 in vivo. 18 The predicted guide strand is indicated in red and a mismatch with the reference sequence in light blue. (b) Sequence distribution (%) of reads mapping to pre-miSNP50T in vivo. The predicted guide strand is indicated in red and a mismatch with the reference sequence in light blue. (c) The length distribution of reads mapping to the pre-miSNP50T precursor (n = 2). For the read alignments, up to 3 mismatches with the reference sequence were allowed. Reads represented with less than 3% were excluded from the figure. these constructs, the cytomegalovirus promoter was replaced by the CAG promoter (Inovio, Plymouth Meeting, PA). 18 AAV5 vector production AAV5 vectors expressing miHTT, miSNP50T, miSNP67T and GFP were produced by baculovirus-based AAV production system (uniQure, Amsterdam, The Netherlands). The artificial miRNA cassettes were cloned in a transfer plasmid in order to generate an entry plasmid. The expression cassettes were inserted in a recombinant baculovirus vector by homologous recombination and clones were selected by plague purification. The recombinant baculovirus containing the cassettes were further amplified till the passage 6 and clones screened for the best production and stability by PCR and qPCR. To generate AAV5, infections with different recombinant baculoviruses containing the vector genome, the replicon enzyme and the capsid protein were performed. The cells were lysed 72 h after infection and the crude lysate was treated with Benzonase (50 U ml − 1 ; Merck, Darmstadt, Germany) for 1 h at 37°C. AAV5 was purified on AVB Sepharose column (GE Healthcare, Little Chalfont, UK) using AKTA Explorer purification system (GE Healthcare) and the genome copy titer was determined by qPCR.

Lentiviral vector production
LVs were produced in HEK-293T cells with the four-plasmid system, as previously described. 40 Human immunodeficiency virus type-1 vectors were pseudotyped with the vesicular stomatitis virus glycoprotein envelope, concentrated by ultracentrifugation and resuspended in phosphate-buffered saline (PBS, Gibco, Life Technologies, Zug, Switzerland) supplemented with 1% bovine serum albumin (BSA, Sigma-Aldrcich, Buchs, Switzerland). The viral particle content of each batch was determined by p24 antigen enzyme-linked immunosorbent assay (p24 ELISA, RETROtek, Kampenhout, Belgium). Viral stocks were stored at − 80°C. Lentiviral vectors encoding the mutant HTT fragments were used at 300 ng of p24. Based on transcriptional start site from the PGK promoter and the presence of polyA+ signal in the 3′LTR of the lentiviral vector, the mRNAs expressed from the LV-mtHTT-50T and LV-mtHTT-67 constructs are expected to be~1993 bp and 2059 bp, respectively. 41

Animals
Adult male Sprague Dawley rats (200 g; OFA, Charles River, Oncins, France) were housed in a temperature-controlled room and maintained on a 12 h day/night cycle. Food and water were available ad libitum. All experiments were carried out in accordance with the European Community directive (86/609/EEC) for the care and use of laboratory animals as well as the Swiss animal welfare laws under the authorization no. VD 2487 and 2889 from the Service de la consommation et des affaires vétérinaires du Canton de Vaud, Switzerland.

Stereotaxic injections
The animals were anesthetized using a solution containing 75 mg kg − 1 ketamine (Ketasol, Graeub, Bern, Switzerland) and 10 mg kg − 1 xylazine (Rompun, Bayer Health Care, Uznach, Switzerland) administered i.p. The rat was placed in the stereotaxic frame (model 963 Ultra Precise Small Animal Stereotaxic Instrument, David Kopf Instruments, Tujunga, CA, USA). Viral vectors were stereotaxically injected into the striatum of animals through a 34-gauge blunt-tip needle (Phymep, Paris, France) linked to a Hamilton syringe (Hamilton Medical AG, Bonaduz, Switzerland) by a polyethylene catheter. The stereotaxic coordinates for an intrastriatal injection were: 0.5 mm rostral to bregma, 3 mm lateral to midline and 5 mm from the skull surface. For the infections, 300 ng p24 antigen of lentiviral vectors were mixed with 6.5 × 10E 10 gc of AAV5 in 4 μl and injected at 0.2 μl min − 1 . The injections were not performed in a blind manner. The rats received 2 mg ml − 1 acetaminophen (Dafalgan, UPSA, Agen, France) in water for the 3 days following the intervention.

Histology
The rats were euthanized by an overdose of sodium pentobarbital 150 mg kg − 1 (B-Braun Mediacal SA, Sempach, Switzerland) and transcardially perfused at a rate of 20 ml min − 1 with 100 ml of 1x PBS and then with 300 ml of 4% PFA. Brains were removed and post-fixed by incubation in 4% PFA for 24 h at 4°C and cryoprotected by incubation in 20% sucrose (Sigma-Aldrich, Buchs, Switzerland) in 1x PBS for 24 h and in 30% sucrose in 1x PBS for 24 h. Brains were frozen on dry ice and then stored at − 80°C. For RNA analysis, the animals were euthanized and the brains were snapfrozen in liquid nitrogen.

Image acquisition and quantification
Images were obtained with a digital camera (3CCD Hitachi HV-F202SCL) on a slide scanner microscope (×20 objective, Zeiss axioscan Z1). Measurements were made with Image J software (http://rsb.info.nih.gov/ij/, NIH, Bethesda, USA) for aggregates and Zeiss Zen Lite 2 (blue Edition; Carl Zeiss AG, Germany) for DARPP32 lesion. An Image J macro was developed to count the number of aggregates. The aggregates were identified on the basis of their intensity relative to the background, their size and their circularity. The same intensity, size and circularity parameters were used for all samples.

Quantification of mtHTT aggregates and DARPP-32 lesions
The number of mutant HTT aggregates in the striatum was automatically counted in 1 in 8 sections with analyze particle plugin on ImageJ (Supplementary Material and Method 1). Cut-off with minimum size and maximum pixel area size was applied to exclude non-aggregates in the ROI. The total number of aggregates was estimated by multiplying these densities by 8. The loss of DARPP-32 expression was analyzed by collecting digitized images sections (1 in 8 sections separated by 200μm). Lesion areas in each section were determined as regions poor in DARPP-32 staining relative to the surrounding tissue. The lesion volume for each animal was quantified with the following formulae: lesion in mm 3 = (mean size in μm 2 × number of counted section × inter-section distance)/1 × 10E9. Means and s.d.'s were calculated for each group. The quantifications were performed in a blind manner. isolation, the striatal powder was homogenized in Trizol using gentleMACS Dissociator (Miltenyi Biotec) and total RNA was isolated from Trizol according to the manufacturer's protocol (Invitrogen). To remove genomic DNA, RNA was treated with dsDNase (Thermo Scientific, Waltham, MA, USA). Quantitative real-time PCR reaction was performed to detect HTT mRNA knock-down using TaqMan gene expression assays (Applied Biosystems, Foster City, CA, USA), human HTT (Hs00918134_m1), rat GAPDH (Rn01462662_g1); and miHTT expression levels using custom TaqMan microRNA expression assay or U6 snRNA (001973; Applied Biosystems). The expression level of each gene was normalized to either endogenous GAPDH or U6 snRNA levels. Fold change/ percentages in HTT mRNA knock-down or miHTT expression were calculated based on 2^DDCT method (Supplementary Material and Method 2).
For the genomic DNA isolation, the striatal powder was processed using DNeasy96 Blood and Tissue kit (Qiagen). qPCR reaction was performed to detect AAV5 vector titers using a standard line and SYBR Green protocol (Applied Biosystems). Forward primer sequence: 5′-GAGCCGCAGCCA TTGC-3′ and reverse primer sequence: 5′-CACAGATTTGGGACAAAGGAAGT -3′. Based on the standard line, genome copies per microgram DNA were calculated.

Next-generation sequencing data analysis
The raw sequencing data were produced as previously described. 18 Small RNA raw data sets were analyzed using the CLC Genomics Workbench 6 (Qiagen). All reads containing ambiguity N symbols, reads shorter than 10 nucleotides, longer than 45 nucleotides and reads represented less than 10 times were discarded. The obtained unique small RNA reads were aligned to the reference sequences of the pre-miHTT and pre-miSNP50T constructs with a max. of 3 nucleotide mismatches allowed. The percentages of reads based on the total number of reads matching the reference sequences were calculated.