Exon 1-targeting miRNA reduces the pathogenic exon 1 HTT protein in Huntington's disease models

Abstract

Huntington’s disease (HD) is a fatal neurodegenerative disease caused by a trinucleotide repeat expansion in exon 1 of the huntingtin gene (HTT) that results in toxic gain of function and cell death. Despite its monogenic cause, the pathogenesis of HD is highly complex, and increasing evidence indicates that, in addition to the full-length (FL) mutant HTT protein, the expanded exon 1 HTT (HTTexon1) protein that is translated from the HTT1a transcript generated by aberrant splicing is prone to aggregate and might contribute to HD pathology. This finding suggests that reducing the expression of HTT1a might achieve a greater therapeutic benefit than targeting only FL mutant HTT. Conversely, strategies that exclusively target FL HTT might not completely prevent the pathogenesis of HD.

We have developed an engineered microRNA targeting the HTT exon 1 sequence (miHTT), delivered via adeno-associated virus serotype 5 (AAV5). The target sequence of miHTT is present in both FL HTT and HTT1a transcripts. Preclinical studies with AAV5-miHTT have demonstrated efficacy in several rodent and large animal models by reducing FL HTT mRNA and protein and rescuing HD-like phenotypes and have been the rationale for phase I/II clinical studies now ongoing in the USA and Europe. In the present study, we evaluated the ability of AAV5-miHTT to reduce the levels of aberrantly spliced HTT1a mRNA and the HTTexon1 protein in the brain of two mouse models of HD (heterozygous zQ175 knock-in mice and humanized Hu128/21 mice).

Polyadenylated HTT1a mRNA and HTTexon1 protein were detected in the striatum and cortex of heterozygous zQ175 knock-in mice, but not in wild-type littermate control mice. Intrastriatal administration of AAV5-miHTT resulted in dose-dependent expression of mature miHTT microRNA in cortical brain regions, accompanied by significant lowering of both FL HTT and HTT1a mRNA expression at 2 months postinjection. Mutant HTT and HTTexon1 protein levels were also significantly reduced in the striatum and cortex of heterozygous zQ175 knock-in mice at 2 months after AAV5-miHTT treatment and in humanized Hu128/21 mice 7 months post-treatment. The effects were confirmed in primary Hu128/21 neuronal cultures.

These results demonstrate that AAV5-miHTT gene therapy is an effective approach to lower both FL HTT and the pathogenic HTTexon1 levels, which could potentially have an additive therapeutic benefit in comparison to other HTT-targeting modalities.

AAV5-miHTT is a gene therapy under development for Huntington's disease. Sogorb-Gonzalez et al. show in two mouse models that AAV5-miHTT lowers mRNA and protein levels of full-length mutant huntingtin as well as the toxic huntingtin exon 1, suggesting an additive therapeutic benefit compared to targeting full-length huntingtin alone.

Introduction

Huntington’s disease (HD) is a fatal genetic, neurodegenerative disease that manifests with progressive motor, cognitive and behavioural symptoms. HD is caused by an expansion of ≥40 CAG repeats in exon 1 of the huntingtin gene (HTT).^1,2 This CAG expansion is translated into an elongated polyglutamine (polyQ) tract within the N-terminal domain of the HTT protein. Similar to other related polyQ neurodegenerative disorders, the expanded polyQ tract confers a toxic gain of function, resulting in protein aggregation and neuronal death.^3,4 Large intracellular aggregates composed of mutant HTT, known as inclusion bodies, have been found throughout post-mortem brains and are considered a pathological hallmark of HD,^5-7 suggesting that prevention of mutant HTT aggregation could be a promising therapeutic strategy for the treatment of HD. During the last decade, this hypothesis has led to the development and clinical testing of several HTT-lowering therapies.^8-11 Antisense oligonucleotides, RNA interference compounds and zinc finger transcriptional repressors aim at reducing the levels of mutant HTT protein in the affected areas of the brain.^12-17

Increasing evidence indicates that, besides the mutant FL HTT protein, small amino-terminal fragments containing expanded polyQ tracts also form toxic aggregates and might contribute significantly to HD pathology.^18-22 Among those, the smallest mutant HTT fragment identified as an exon 1 HTT protein (HTTexon1) displayed the highest toxicity by a strong propensity to form nuclear aggregates and a reduction of motor function and lifespan in Drosophila.^23,24 In HD mouse models, HTTexon1 is the smallest identified HTT fragment.¹⁹ Transgenic mouse models expressing mutant HTTexon1 display the most severe phenotype,¹⁸ and conditional suppression of HTTexon1 expression reverses aggregate formation and motor decline,²⁵ suggesting that HTTexon1 reduction might be therapeutically efficacious for HD.

Until recently, the toxic HTTexon1 protein fragment was thought to be the result of a cascade of proteolytic cleavage events of the FL polyQ-containing HTT protein.²⁶ However, it has recently been demonstrated that in the context of HD, HTT pre-mRNA can undergo CAG-dependent alternative processing by which cryptic polyadenylation sites within intron 1 are activated, resulting in the production of the HTT1a transcript that encodes the pathogenic mutant HTTexon1 protein.^27-29 Analogous to the sequence in the correctly spliced FL HTT mRNA, the additional intron 1 sequence starts with a base pair that completes the codon for a proline residue, followed by a stop codon (TGA). Consequently, HTT1a is translated into the pathogenic HTTexon1 protein composed of the same amino acids as the N-terminus in the FL protein and ending in the LHRP amino acid sequence. Given that HTT1a production is CAG dependent, expansion of CAG repeats via somatic instability throughout the lifetime of a person with HD is expected to increase HTTexon1 production and, consequently, cause accelerated neurodegeneration.³⁰ This hypothesis is part of the two-step model of HD pathogenesis, in which somatic CAG repeat instability throughout the presymptomatic years of a mutation carrier triggers the increase in production of toxic drivers, such as HTTexon1, up to a cell type-specific lethal threshold that results in the neurodegeneration of vulnerable cell populations in the brain and disease manifestation.^28,31

RNA-targeting approaches designed exclusively to lower the FL (mutant) HTT protein might not necessarily reduce intracellular levels of the highly toxic HTTexon1 protein. In fact, most therapeutics in clinical development for HD target sequences downstream of exon 1 and are therefore unlikely to reduce the translation of aberrantly spliced HTT1a. Currently, there are two ongoing clinical trials that use an engineered miRNA specifically to target a sequence in HTT exon 1 (miHTT), delivered via adeno-associated serotype 5 virus (AAV5) (ClinicalTrials.gov identifiers NCT04120493 and NCT05243017). Briefly, this approach is based on an AAV5-delivered DNA construct encoding for engineered pri-miRNA hairpins, which are processed into mature miHTT molecules and loaded into the RNA interference silencing complex (RISC) for recognition and binding to the target sequence, resulting in translational repression or enzymatic cleavage of target mRNA.¹⁴ Proof-of-concept studies have demonstrated that intrastriatal administration of AAV5-miHTT is safe and results in adequate brain biodistribution and long-term lowering of FL mutant HTT protein in HD rodents^14,32-35 and in large animal models,^36-38 reducing HD-like phenotypes. In the present study, we evaluated the ability of AAV5-miHTT to reduce the levels of HTT1a transcripts in the brain of two HD mouse models, transgenic heterozygous zQ175 knock-in (KI) mice and humanized Hu128/21 mice.^39,40 In general, intrastriatal injection of AAV5-miHTT caused a dose-dependent lowering of both FL HTT and HTT1a mRNA and protein levels in the striatum and cortex of these HD mouse models. These data demonstrate that, in addition to FL HTT, AAV5-miHTT also effectively reduces the highly pathogenic HTTexon1 protein in HD.

Materials and methods

AAV5-miHTT gene therapy

The AAV5 vector carrying the miHTT cassette was produced using a baculovirus-based AAV production system (uniQure) as described previously.⁴¹ The miRNA expression cassette, comprising the miRNA sequence targeting the HTT exon 1 sequence driven by the chimeric chicken-beta actin promoter,¹⁴ was inserted into an AAV vector genome backbone flanked by two intact non-coding AAV2 inverted terminal repeats.

Q175 knock-in mouse injection and tissue collection

The zQ175 KI HD mouse model was generated by spontaneous expansion of the human–mouse chimeric Q140 line.^40,42 Wild-type (WT) mice and heterozygous zQ175 KI were treated at 5 months of age by stereotaxic bilateral intrastriatal injection, followed by an 8-week observation period before sacrifice. A total of four groups were included in the study (n = 8–10 animals per group): WT control (injected with formulation buffer, n = 10) and heterozygous zQ175 KI mice injected with formulation buffer (n = 10) or with a low dose (5.2 × 10⁹ genome copies per mouse, n = 8) or a high dose (1.3 × 10¹¹ genomes copies per mouse, n = 10) of AAV5-miHTT. Mice were sacrificed after 2 months of treatment, and striatal regions, both cortical hemispheres and cerebellar regions were collected and stored at −80°C until analysis. For additional information, see Supplementary material.

Hu128/21 mouse injection and tissue collection

The efficacy of AAV5-miHTT treatment was also investigated in the Hu128/21 mouse model of HD³⁹ and in their Hu21/21 counterparts. Hu128/21 mice were generated by intercrossing YAC128 and BAC21 models of HD with the Htt^−/− background.^43,44 There were five experimental groups per genotype (Hu21/21 or Hu128/21): vehicle-treated, AAV5-GFP (green fluorescent protein) treated (single dose, 1.3 × 10¹¹ genome copies per mouse), and AAV5-miHTT treated at three doses: low (5.2 × 10⁹ genome copies per mouse), mid (2.6 × 10¹⁰ genome copies per mouse) or high (1.3 × 10¹¹ genome copies per mouse) (n = 6–8 animals per group). Surgery was performed as previously reported.³³ At 4 months after injection, mice were sacrificed and the brains removed, placed on ice for 1 min to increase tissue rigidity, and subsequently microdissected in different brain regions, which were preserved in RNAlater (Ambion) overnight at 4°C and stored at −80°C until use.³³ In the present study, hippocampal tissues were used for analysis. For additional information, see Supplementary material.

Primary Hu128/21 cortical neuron culture

Primary cortical neurons from Hu128/21 mice were used to test the ability of AAV5-miHTT to reduce HTTexon1 in vitro. At day 5 of maturation, AAV5-miHTT was added to the media at three doses, multiplicity of infection 5 × 10⁴, 5 × 10⁵ or 5 × 10⁶ (genome copies per cell). Vehicle-treated cells and cells incubated with AAV5-GFP at a multiplicity of infection of 5 × 10⁶ were used as controls. Each treatment condition was performed in triplicate, and each experiment was replicated independently. Following a treatment duration of 10 days, the morphology and viability of cells was assessed qualitatively prior to harvest and collection of cells. Cell pellets were snap frozen and kept at −80°C until use. For additional information, see Supplementary material.

Analytical methods

For description of quantitative RT-PCR, 3′RACE-PCR and homogeneous time-resolved fluorescence (HTRF) methods, see Supplementary material.

Statistical analysis

Statistical analysis was performed with GraphPad Prism v.9 software. Sample outliers were identified using Grub’s test and removed from further analysis using an unpaired t-test, Pearson’s correlation and one-way ANOVA, with Tukey’s or Sidak’s multiple comparisons post hoc tests as indicated. Graphs were arranged using GraphPad Prism v.9. P-values of <0.05 were considered statistically significant (ns, non-significant; *P < 0.05, **P < 0.005, ***P < 0.0005 and ****P < 0.0001).

Results

Htt1a transcripts are expressed in the cortex of heterozygous zQ175 KI mice

To investigate the target engagement of aberrantly spliced Htt1a transcript by an engineered miRNA targeting HTT exon 1, a suitable animal model, closely mimicking the genetic context and HTT aberrant splicing in HD patients, is required. We selected the zQ175 KI mouse model of HD, a well-validated model for evaluation of therapeutic interventions^34,45,46 and for HTTexon1 protein analysis.⁴⁶ In zQ175 KI mice, exon 1 and part of intron 1 sequences of murine huntingtin (Htt) have been replaced with the human HTT exon 1 sequence carrying expanded (>180) CAG repeats⁴⁰ (Fig. 1A). Heterozygous zQ175 KI mice contain a chimeric human–mouse expanded Htt allele, which is subject to alternative Htt mRNA processing,^27,45 and a WT murine allele.

Figure 1.

Detection of the Htt1a transcript in the cortex of zQ175 knock-in mice. (A) Schematic representation of wild-type (WT) huntingtin (Htt) allele and chimeric mutant Htt allele in heterozygous zQ175 knock-in (KI) mice. (B) 3′RACE RT-PCR together with intron 1 Htt primers showed the presence of polyadenylated Htt1a mRNA in zQ175 KI brain cortex, but not in WT mice. (C) Detection of polyadenylated HTT1a mRNA in the cortex of zQ175 KI mice by oligo-dT RT-PCR assay. Spliced exon 1–exon 2 transcripts (top) were detected in frontal cortex of both WT and zQ175 KI mice, whereas exon 1–intron 1 product (bottom) was detected only in zQ175 KI mice, but not in WT. (D) Relative expression of FL-Htt in the frontal cortex of zQ175 KI mice and WT detected by TaqMan qPCR with primer-probe sets exon1–2 (unpaired t-test, ****P < 0.001) and exon64–65 (unpaired t-test, ***P = 0.003). (E) Relative expression of Htt1a mRNA in cortical tissue of zQ175 KI and WT mice detected by TaqMan qPCR with primer–probe sets intron 1 (unpaired t-test, ****P < 0.0001) and human exon 1–intron 1 (unpaired t-test, ****P < 0.0001). In D and E, bars show the mean ± standard error of the mean (SEM).

Initially, to evaluate the levels of polyadenylated Htt1a transcripts in heterozygous zQ175 KI mice, molecular assays were performed with RNA isolated from the cortex.²⁷ 3′RACE-PCR (rapid amplification of complementary DNA ends) together with intron 1 Htt primers showed the exclusive presence of polyadenylated short Htt1a mRNA in the cortex of heterozygous zQ175 KI, but not WT littermate control mice (Fig. 1B). The sequence of the HTTintron1-containing amplified product was verified by Sanger sequencing (data not shown). End point PCR with primers for the exon 1–intron 1 boundary confirmed the presence of Htt1a in the cortex of heterozygous zQ175 KI and its absence in WT littermates (Fig. 1C). Full-length (FL-Htt) transcripts, detected with primers that spanned exon 1–exon 2 junction, were identified in both heterozygous zQ175 KI and WT mice (Fig. 1C).

To quantify the expression of the Htt1a transcript, a quantitative RT-PCR (qPCR) was performed with four different sets of primers. Levels of FL-Htt were determined with two different assays (primers spanning the exon 1–2 or exon 64–65 junctions, respectively), and Htt1a was quantified selectively by amplifying sequences within intron 1 (upstream of the first cryptic polyA site) and human exon 1–intron 1 (Fig. 1D and E). Significantly lower levels of FL-Htt mRNA were detected in heterozygous zQ175 KI mice in comparison to WT mice when normalized to three housekeeping genes [Fig. 1D; exon 1–2 unpaired t-test, two-tailed, t(16) = 7.944, ****P < 0.001; exon 64–65 unpaired t-test, two-tailed, t(16) = 4.662, ***P = 0.003]. This has been observed previously with a different assay and is likely to be attributable to the alternative processing of mutant Htt mRNA.⁴⁵ Consistent with end point PCR, we quantified high levels of the Htt1a transcript in the cortex of heterozygous zQ175 KI mice, but not in WT mice [Fig. 1E; intron 1, unpaired t-test, two-tailed, t(16) = 12.97, ****P < 0.0001; human exon 1–intron 1, unpaired t-test, two-tailed, t(16) = 20.79, ****P < 0.0001]. All together, these results obtained by different qualitative and quantitative assays confirmed the presence of the polyadenylated Htt1a transcript in the context of HD in the brain of zQ175 KI mice.

HTTexon1 protein is present in the striatum and cortex of heterozygous zQ175 KI mice

Polyadenylated Htt1a mRNA is translated to generate the HTTexon1 protein, which is prone to aggregate and is considered the most toxic species of mutant HTT.^23,24 Quantification of HTTexon1 was performed by HTRF with the MW8 antibody. The MW8 antibody was raised against the HTTexon1 carboxy-terminal domain and has previously been shown to be specific for HTTexon1 in HTT bioassays.^19,46,47 Specific combinations of different antibodies, as previously described,⁴⁷ were used to preferentially measure soluble mutant HTT (FL-HTT and HTTexon1; 2B7–4C9 antibody pair), soluble HTTexon1 (2B7–MW8 antibody pair) and total FL-HTT (mutant and WT; MAB5490–MAB2166 antibody pair), in addition to aggregated HTTexon1 (4C9–MW8 antibody pair) in both heterozygous zQ175 KI and WT mice at 7 months of age. Significantly higher levels of soluble mutant HTT and soluble HTTexon1 were detected in the striatum and cortex of heterozygous zQ175 KI mice, when compared with background levels in WT mice (Fig. 2A and B; n = 10; mutant HTT protein: one-way ANOVA, F = 117.5, P < 0.0001; HTTexon1 protein: one-way ANOVA, F = 76.68, P < 0.0001). However, and in contrast to the lower FL Htt mRNA levels quantified in the cortex of heterozygous zQ175 KI mice, there was no significant difference between the levels of FL-HTT protein in cortical samples [Fig. 2C; n = 10; unpaired t-test: two-tailed, t(17) = 0.3087, P = 0.7613]. This indicates that that comparable FL-HTT protein expression is found in zQ175 KI and WT mice at 7 months of age (Fig. 2C), in accordance with a recent report.⁴⁶ FL-HTT protein in the striatum was not determined owing to the limited amount of tissue sample for analysis in this brain region. Aggregated HTTexon1 was detected in both striatum and cortex in heterozygous zQ175 KI mice, but not in WT mice (Fig. 2D; n = 10; one-way ANOVA, F = 2607, P < 0.0001). Overall, levels of all mutant HTT protein species were significantly higher in the striatum compared with the cortex. These results confirm the presence of HTTexon1 in the brain in zQ175 KI mice at 7 months of age.

Figure 2.

Detection of HTTexon1 protein and other huntingtin (HTT) protein species in zQ175 knock-in mice. (A) Levels of soluble mutant HTT protein (FL-HTT and HTTexon1) in striatum and frontal cortex of wild-type (WT) and zQ175 knock-in (KI) mice measured by homogeneous time-resolved fluorescence (HTRF) with 2B7 and 4C9 antibodies and represented as the mean ± SEM (one-way ANOVA, Tukey’s multiple comparisons test, ***P = 0.009, ****P < 0.001). (B) Levels of soluble HTTexon1 protein in striatum and frontal cortex of WT and zQ175 KI mice measured by HTRF with 2B7 and MW8 antibodies and represented as the mean ± SEM (one-way ANOVA, Tukey’s multiple comparisons test, ****P < 0.001). (C) Levels of soluble full-length HTT protein (mutant and WT) in frontal cortex of WT and zQ175 KI mice measured by HTRF with MAB5490 and MAB2166 antibodies and represented as the mean ± SEM (unpaired t-test, ns, P = 0.7613). (D) Levels of aggregated HTTexon1 protein in striatum and frontal cortex of WT and zQ175 KI mice measured by HTRF with 4C9 and MW8 antibodies and represented as the mean ± SEM (one-way ANOVA, Tukey’s multiple comparisons test, ****P < 0.001).

AAV5-miHTT administration results in dose-dependent expression of HTT exon 1-targeting miHTT in heterozygous zQ175 KI mice

The discovery that HTTexon1 protein is mostly the result of aberrant splicing suggests that, depending on the HTT target sequence, RNA-targeted therapeutic approaches designed to lower the levels of FL-HTT might not necessarily reduce intracellular levels of the highly toxic HTTexon1. However, approaches targeting HTT within the exon 1 sequence, such as AAV5-miHTT¹⁴ (Fig. 3A), could result in the lowering of HTTexon1 and therefore might have a therapeutic advantage in comparison to strategies that lower FL-HTT exclusively. To investigate the efficacy of AAV5-miHTT to lower the HTTexon1 protein in vivo, heterozygous zQ175 KI mice received bilateral intrastriatal injections with AAV5-miHTT at a low (1×) or a high (25×) dose, or with formulation buffer (vehicle, Veh) as the control (n = 8–10 per group) at 5 months of age. The timing of administration was chosen to investigate the effects of the therapeutic intervention at an age when this mouse model displays significant HTT aggregation.⁴⁷ Two months after AAV5-miHTT treatment, mice were sacrificed, and brain areas were separately dissected for analysis of miHTT microRNA expression and huntingtin mRNA (RNA analysis) and mHTT protein analysis (Fig. 3B). Owing to the small volume, striata were used for protein analysis exclusively, whereas other areas from the left and right hemisphere were used for RNA and protein analysis, respectively. Biodistribution and transduction efficiency of AAV5-miHTT treatment was determined by measuring the expression of the mature miHTT microRNA by RT-qPCR in the frontal cortex, caudal cortex and hippocampus. A significant dose-dependent increase in expression of miHTT microRNA was measured in the frontal cortex, caudal cortex and hippocampus of zQ175 KI mice at 2 months postinjection (Fig. 3C–E; n = 10, one-way ANOVA, Tukey’s multiple comparisons test: frontal cortex: F = 16.84, P < 0.0001, Veh versus Low P = 0.9483, Veh versus High P < 0.0001, Low versus High P = 0.0003; caudal cortex: F = 14.48, P < 0.0001, Veh versus Low P = 0.9325, Veh versus High P = 0.0001, Low versus High P = 0.0007; hippocampus: F = 23.02, P < 0.0001, Veh versus Low P = 0.8088, Veh versus High P > 0.0001, Low versus High P < 0.0001). The low dose led to an average of 5 × 10⁵ molecules/µg RNA, whereas high-dose animals had levels of an average of 7 × 10⁶ molecules/µg RNA in frontal cortex (Fig. 3C).

Figure 3.

Intrastriatal administration of AAV5-miHTT results in dose-dependent expression of HTT exon 1-targeting miHTT. (A) Schematic representation of AAV5-delivered expression cassette including Pol II promoter, exon 1-targeting miHTT transgene and polyA signal. The transgene is processed into pre-miRNA hairpin with the same structure as has-miR-451 precursor, then processed into an miHTT guide strand that is complementary to HTT exon 1 sequence upstream CAG repeat expansion. (B) Representation of bilateral intrastriatal injection and collection of colour-coded brain tissues for RNA and protein analysis. (C–E) miHTT transgene expression in frontal cortex (C), caudal cortex (D) and hippocampus (E) from the left hemisphere determined by custom TaqMan RT-qPCR and represented as miHTT molecules per microgram of RNA (mean ± SEM) per dose group (n = 8–10, one-way ANOVA, Tukey’s multiple comparisons test, ***P < 0.005, ****P < 0.001).

AAV5-miHTT treatment results in efficient lowering of mutant HTT protein species in zQ175 KI mice

To evaluate the efficacy of AAV5-miHTT to reduce the pathogenic HTTexon1 fragment, we assessed both Htt1a RNA and HTTexon1 protein levels in heterozygous zQ175 KI mice after treatment (Fig. 3B). RNA extracted from frontal cortex was used with specific TaqMan assays with primers spanning the exon 1–2 and exon 64–65 junctions to determine levels of FL-Htt mRNA, whereas primers within intron 1 and the human exon 1–intron 1 junction were used for quantification of Htt1a mRNA (Fig. 4A and B). Two months after intrastriatal administration of AAV5-miHTT, we observed a significant 20% reduction of FL-Htt mRNA in the frontal cortex of the high-dose-treated mice compared with the vehicle group (Fig. 4A; n = 10, one-way ANOVA, Tukey’s multiple comparisons test: exon 1–2: F = 3.651, P = 0.0406, Veh versus Low P = 0.4101, Veh versus High P = 0.0316, Low versus High P = 0.4355; exon 64–65: F = 7.303, P = 0.0032, Veh versus Low P = 0.5030, Veh versus High P = 0.0329, Low versus High P = 0.0032); given that miHTT is not engaging with mouse Htt, exon-spanning primers underestimate the degree of FL-Htt mRNA lowering. Moreover, AAV5-miHTT treatment resulted in significant dose-dependent lowering of the Htt1a transcript in the frontal cortex, with 15% and 35% reduction observed with a low and high dose of AAV5-miHTT, respectively (Fig. 4B; n = 10, one-way ANOVA, Tukey’s multiple comparisons test: intron 1: F = 3.931, P = 0.00328, Veh versus Low P = 0.9270, Veh versus High P = 0.0374, Low versus High P = 0.1092. human exon 1–intron 1: F = 20.95, P < 0.0001, Veh versus Low P = 0.0272, Veh versus High P < 0.0001, Low versus High P = 0.0074).

Figure 4.

AAV5-miHTT shows dose-dependent lowering of full-length Htt and Htt1a mRNA and HTT and HTTexon1 protein in zQ175 knock-in mice at 2 months postinjection. (A) Expression level of FL-Htt mRNA in right frontal cortex of AAV5-miHTT-treated mice relative to vehicle group. (B) Expression level of Htt1a mRNA in right frontal cortex of AAV5-miHTT-treated mice relative to vehicle group. (C) Levels of soluble mutant HTT protein in striatum and cortex of AAV5-miHTT-treated mice relative to vehicle group and wild-type (WT) mice. (D) Levels of HTTexon1 HTT protein in striatum and cortex of AAV5-miHTT-treated mice relative to vehicle group and WT mice. (E) Levels of aggregated HTTexon1 HTT protein in striatum and cortex of AAV5-miHTT-treated mice relative to vehicle group and WT mice. (F) Levels of full-length HTT (mutant and WT) protein in frontal cortex of AAV5-miHTT-treated mice relative to vehicle group. In A–F, bars represent the mean ± SEM. Statistics: one-way ANOVA, Tukey’s multiple comparisons test (*P < 0.05, **P < 0.005, ***P < 0.0005 and ****P < 0.0001).

To quantify the HTTexon1 protein at 2 months post-treatment, striatal and frontal cortical tissues from zQ175 KI were lysed and analysed by HTRF, using specific antibody combinations to determine soluble mutant HTT (FL-HTT and HTTexon1; 2B7–4C9 antibody pair), soluble HTTexon1 (2B7–MW8 antibody pair), aggregated HTTexon1 (4C9–MW8 antibody pair) and total soluble FL-HTT (mutant and WT; MAB5490–MAB2166 antibody pair) levels (Fig. 4C–F). In comparison to vehicle-treated animals, high-dose AAV5-miHTT treatment resulted in a significant 45% and 28% reduction of soluble mutant HTT in the striatum and frontal cortex, respectively (Fig. 4C; n = 10, one-way ANOVA, Tukey’s multiple comparisons test: striatum: F = 53.53, P < 0.0001, Veh versus Low P = 0.5883, Veh versus High P < 0.0001, Low versus High P < 0.0001; cortex: F = 16.80, P < 0.0001, Veh versus Low P = 0.7325, Veh versus High P < 0.0001, Low versus High P = 0.0005). Likewise, we detected a significant lowering of soluble HTTexon1 in the striatum (36% knockdown) and cortex (24% knockdown) (Fig. 4D; n = 10, one-way ANOVA, Tukey’s multiple comparisons test: striatum: F = 15.31, P = 0.0002, Veh versus Low P = 0.7881, Veh versus High P = 0.0002, Low versus High P = 0.0014. cortex: F = 3.946, P = 0.0324, Veh versus Low P = 0.9538, Veh versus High P = 0.0394, Low versus High P = 0.0981). High-dose AAV5-miHTT treatment also resulted in ≤20% lowering of aggregated HTTexon1 protein in the striatum, but not in the cortex (Fig. 4E; n = 10, one-way ANOVA, Tukey’s multiple comparisons test: striatum: F = 31.73, P < 0.0001, Veh versus Low P = 0.6980, Veh versus High P < 0.0001, Low versus High P < 0.0001; cortex: F = 5.328, P = 0.0117, Veh versus Low P = 0.1196, Veh versus High P = 0.4710, Low versus High P = 0.0089). In general, and consistent with intrastriatal delivery of AAV5-miHTT, we observed a significantly higher magnitude of HTTexon1 lowering in the striatum compared with the cortex. Given that the miHTT was designed exclusively to target the human sequence upstream of the CAG repeat (within mutant allele in zQ175 KI mice), we did not expect, nor did we observe, a significant reduction of total FL HTT protein in the cortex of heterozygous zQ175 KI mice (Fig. 4F; n = 10, one-way ANOVA, Tukey’s multiple comparisons test, F = 2.995, P = 0.0698, Veh versus Low P = 0.6550, Veh versus High P = 0.2431, Low versus High P = 0.0652).

Potent suppression of HTTexon1 protein in humanized Hu128/21 mice upon AAV5-miHTT treatment

To validate the lowering of HTTexon1 by AAV5-miHTT treatment further, in a different disease model, we selected the humanized Hu128/21 mouse model of HD. Hu128/21 mice express two FL human HTT transgenes heterozygous for the HD mutation (128CAG and 21CAG repeats) on a Htt^−/− background.³⁹ In a previous study, we investigated the long-term tolerability and efficacy of AAV5-miHTT in Hu128/21 mice.³³ Following AAV5-miHTT striatal administration at 1 month of age, we reported a dose-dependent expression of miHTT and a potent dose-dependent lowering of both WT and mutant FL-HTT in the striatum (≤90%) and cortex (≤60%) at 4 and 7 months postinjection. AAV5-miHTT treatment ameliorated the loss of striatal volume and cognitive function in Hu128/21 mice.³³ In a subset of Hu128/21 animals of the same study, and in their Hu21/21 control counterparts, we now investigated HTT1a target engagement by AAV5-miHTT. We quantified HTTexon1 protein levels in the hippocampus of Hu128/21 and Hu21/21 at 7 months following intrastriatal administration of three ascending doses of AAV5-miHTT (Low, Mid and High). Based on the known biodistribution of AAV5-miHTT after intrastriatal administration, transduction levels in the hippocampus are expected to be comparable to those in frontal cortex (Fig. 3C–E). We observed a strong reduction of all the different HTT protein species measured, including HTTexon1, in a dose-dependent manner after AAV5-miHTT administration (Fig. 5A; soluble mutant HTT protein, 2B7–MW1: one-way ANOVA, Tukey’s multiple comparisons test, F = 16.00, P < 0.0001; GFP versus Low P = 0.9972, GFP versus Mid P = 0.0181, GFP versus High P < 0.0001; Low versus Mid P = 0.0199, Low versus High P < 0.0001, Mid versus High P = 0.0380; soluble HTTexon1 protein, 2B7–MW8: one-way ANOVA, Tukey’s multiple comparisons test, F = 16.86, P < 0.0001; GFP versus Low P = 0.9959, GFP versus Mid P = 0.3158, GFP versus High P < 0.0001; Low versus Mid P = 0.2733, Low versus High P < 0.0001, Mid versus High P = 0.0007; full-length HTT protein, MAB5490–MAB2166: one-way ANOVA, Tukey’s multiple comparisons test, F = 29.46, P < 0.0001; GFP versus Low P = 0.8375, GFP versus Mid P < 0.0001, GFP versus High P < 0.0001; Low versus Mid P = 0.0022, Low versus High P < 0.0001, Mid versus High P = 0.0137; aggregated HTTexon1 protein, 4C9–MW8: one-way ANOVA, Tukey’s multiple comparisons test, F = 53.26, P < 0.0001; GFP versus Low P < 0.0001, GFP versus Mid P < 0.0001, GFP versus High P < 0.0001; Low versus Mid P = 0.5912, Low versus High P = 0.0016, Mid versus High P = 0.0089). The effects were specific for the Hu128/21 animals, because no differences were observed in Hu21/21 controls (Fig. 5B–E). These data confirm the ability of AAV5-miHTT treatment to reduce the highly pathogenic HTTexon1 protein in HD mice carrying the human HTT gene.

Figure 5.

Striatal AAV5-miHTT treatment lowers levels of different huntingtin (HTT) protein species, including HTTexon1, in the hippocampus of Hu128/21 mice. (A) Four different HTT homogeneous time-resolved fluorescence (HTRF) assays were used, measuring soluble mutant HTT protein (2B7–MW1), soluble HTTexon1 protein (2B7–MW8), full-length HTT protein (MAB5490–MAB2166) and aggregated HTTexon1 (4C9–MW8), all of them demonstrating dose-dependent lowering after AAV5-miHTT treatment in Hu128/21 hippocampi (data are represented as a percentage of Veh-treated groups). (B–E) Data from the four different HTT HTRF assays in Hu21/21 and Hu128/21, expressed in arbitrary units (AU), with dose-dependent lowering in Hu128/21 but not in Hu21/21 mice, in which the analytes were detected only at background levels. In A–E, bars represent the mean ± SEM. Statistics: one-way ANOVA, Tukey’s multiple comparisons test (*P < 0.05, **P < 0.005, ***P < 0.0005 and ****P < 0.0001).

Potent suppression of soluble HTT protein species in humanized Hu128/21 cortical neurons upon AAV5-miHTT

To assess the effects of AAV5-miHTT treatment further on different soluble HTT species, including HTTexon1, in the Hu128/21 model system, we used primary cortical neurons derived from Hu128/21 mice.³⁹ Primary cortical Hu128/21 neurons were transduced with AAV5-miHTT at three different multiplicities of infection (5 × 10⁴, 5 × 10⁵ and 5 × 10⁶); AAV5-GFP (multiplicity of infection 5 × 10⁶) and vehicle-treated Hu128/21 neurons served as controls. Cells were harvested 10 days after transduction, and soluble HTT levels were determined by HTRF using specific antibody pairs.⁴⁷ In both independent experiments performed, we observed a dose-dependent reduction of total soluble mutant HTT (mutant FL HTT and HTTexon1) (MW1–4C9; average lowering ≤77%; experiment 1: one-way ANOVA, Sidak’s multiple comparisons test, F = 13.89, P = 0.0015; GFP versus Low P = 0.9275, GFP versus Mid P = 0.0397, GFP versus High P = 0.0024; Low versus Mid P = 0.1650, Low versus High P = 0.0077, Mid versus High P = 0.3054; experiment 2: one-way ANOVA, Sidak’s multiple comparisons test, F = 23.96, P = 0.0002; GFP versus Low P = 0.9987, GFP versus Mid P = 0.0068, GFP versus High P = 0.0005; Low versus Mid P = 0.0083, Low versus High P = 0.0006, Mid versus High P = 0.1738; Fig. 6A), soluble HTTexon1 (2B7–MW8 antibody pair; average lowering ≤71%; experiment 1: one-way ANOVA, Sidak’s multiple comparisons test, F = 39.14, P < 0.0001; GFP versus Low P = 0.0148, GFP versus Mid P = 0.0011, GFP versus High P < 0.0001; Low versus Mid P = 0.3122, Low versus High P = 0.0014, Mid versus High P = 0.0207; experiment 2: one-way ANOVA, Sidak’s multiple comparisons test, F = 24.07, P = 0.0002; GFP versus Low P = 0.7912, GFP versus Mid P = 0.0195, GFP versus High P = 0.0008; Low versus Mid P = 0.0058, Low versus High P = 0.0003, Mid versus High P = 0.1094; Fig. 6B), and total soluble FL HTT (mutant and WT) (MAB5490–MAB2166 or D7F7–MAB5490 antibody pair; average lowering ≤87%; MAB5490–MAB2166 experiment 1: one-way ANOVA, Sidak’s multiple comparisons test, F = 12.41, P = 0.0022; GFP versus Low P > 0.9999, GFP versus Mid P = 0.0083, GFP versus High P = 0.0016; Low versus Mid P = 0.0108, Low versus High P = 0.0020, Mid versus High P = 0.7262; MAB5490–MAB2166 experiment 2: one-way ANOVA, Sidak’s multiple comparisons test, F = 20.03, P = 0.0004; GFP versus Low P > 0.9999, GFP versus Mid P = 0.0083, GFP versus High P = 0.0016; Low versus Mid P = 0.0108, Low versus High P = 0.0020, Mid versus High P = 0.7262; D7F7–MAB5490 experiment 1: one-way ANOVA, Sidak’s multiple comparisons test, F = 27.02, P = 0.0002; GFP versus Low P = 0.9621, GFP versus Mid P = 0.0279, GFP versus High P = 0.0002; Low versus Mid P = 0.0947, Low versus High P = 0.0005, Mid versus High P = 0.0166; D7F7–MAB5490 experiment 2: one-way ANOVA, Sidak’s multiple comparisons test, F = 46.29, P < 0.0001; GFP versus Low P = 0.3048, GFP versus Mid P = 0.0027, GFP versus High P = 0.0002; Low versus Mid P = 0.0003, Low versus High P < 0.0001, Mid versus High P = 0.2268; Fig. 6C and D). These data confirm that AAV5-miHTT leads to a dose-dependent reduction of soluble HTT species, including HTTexon1.

Figure 6.

Soluble huntingtin (HTT) protein expression in primary Hu128/21 neurons treated with AAV5-GFP or AAV5-miHTT. Cells were transduced with AAV5-miHTT at a multiplicity of infection of 10⁵, 10⁶ or 10⁷. Control groups included vehicle-treated and AAV5-GFP multiplicity of infection 10⁷-treated cells. Five days post-transduction, cell pellets were collected to determine soluble HTT protein levels, analysed by homogeneous time-resolved fluorescence (HTRF) using different antibody pairs. Two independent experiments were performed, with three biological replicates per condition. HTT expression levels (biological replicates and mean ± SEM) are represented as a percentage of vehicle-treated control groups: (A) total soluble mutant HTT (FL-HTT and HTTexon1; MW1–4C9 antibody pair); (B) soluble HTTexon1 (2B7–MW8 antibody pair); and total soluble full-length HTT (mutant and WT) with (C) MAB5490–MAB2166 antibody pair or (D) D7F7–MAB5490 antibody pair. Each experiment was evaluated independently by one-way ANOVA followed by multiple comparison tests, corrected using Sidak’s test (*P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 versus GFP controls).

Discussion

The work presented here demonstrates that targeting the exon 1 sequence of HTT with an AAV5-delivered engineered miRNA is an effective approach to suppress both FL-HTT protein (mutant and WT) and the pathogenic HTTexon1 fragment in the brains of HD mice. These results are an important extension of our previous studies, which showed that intrastriatal delivery of AAV5-miHTT resulted in widespread miHTT expression throughout the brain and a significant dose-dependent reduction of mutant and normal HTT protein in HD iPSC-derived neurons, HD mice and (mutant HTT) in HD transgenic minipigs.^32-34,36,37 This, together with the long-term miHTT expression, the favourable safety profile and the beneficial prevention of neurological phenotypes, supported the initiation of phase I/II clinical trials of AAV5-miHTT therapy in HD patients (ClinicalTrials.gov identifiers NCT04120493 and NCT05243017).

We have previously reported that AAV5-miHTT treatment led to motor phenotypic rescue in R6/2 mice,³⁴ a transgenic mouse with an expanded CAG repeat in a human HTT exon 1 transgene,¹⁸ and improved median survival of R6/2 mice up to 4 weeks.³⁴ However, the finding of HTTexon1 generation by aberrant splicing described by Sathasivam et al.²⁷ raised the question of whether the therapeutic efficacy of AAV5-miHTT would be the same in an HD-associated splicing context.²⁷ To our knowledge, this is the first study to demonstrate the efficacy of an HTT-lowering agent successfully to reduce the pathogenic HTTexon1 protein generated by aberrant splicing in vivo, adding therapeutic value to the AAV5-miHTT gene therapy for HD. In contrast to most nucleic acid-based HTT-lowering approaches that have been designed so far, AAV5-miHTT targets HTT exon 1, thereby lowering both HTTexon1 and FL-HTT.

In this study, we used two mouse models of HD, heterozygous zQ175 KI mice and humanized Hu128/21 mice. In both mouse models, we observed a significant reduction of the toxic HTTexon1 protein with comparable doses of AAV5-miHTT, with the effect being slightly larger in Hu128/21 mice, especially in the primary cultures. These results could be attributable to differences in technical aspects, such as in vitro versus in vivo assessments, or in biological aspects, such as splicing ratio and/or rate of somatic instability leading to different levels of HTTexon1 accumulation and/or in the proportion of soluble versus aggregated HTTexon1 forms between the mouse models.

In the present study, aberrant splicing ratio and levels of FL-Htt mRNA, HTT1a mRNA and protein in 7-month-old heterozygous zQ175 mice were in agreement with previous studies.^27,28,45-47FL-Htt transcript levels of ∼70%–80% of WT levels have been found in the striatum of 2-month-old heterozygous zQ175 mice⁴⁵; the reduced FL-Htt mRNA levels with respect to WT mice were attributed to a consequence of incomplete splicing between exon 1 and exon 2 in the mutant allele, and not to changes in the WT murine allele, and were accompanied by increased Htt1a mRNA levels, in line with our observations. Here, the observed reduction in FL-Htt mRNA levels in zQ175 versus WT were not reflected at the protein level, which remained comparable. This could be attributable to endogenous compensatory mechanisms and would require further investigation. The more potent reduction of Htt1a/HTTexon1 mRNA/protein than of FL-Htt/FL-HTT mRNA/protein by AAV5-miHTT was expected, because heterozygous zQ175 mice were used, and AAV5-miHTT does not target the WT mouse allele in this model. To our knowledge, the proportion of FL-HTT and HTT1a transcripts in Hu128/21 have not been characterized previously. In the present study, we did not measure the levels of HTT1a mRNA in Hu128/21 brain tissue owing to the lack of sample material; therefore, the splicing ratio could not be assessed. Previous studies have reported the presence of HTT1a transcripts and protein in other human HTT mouse models, YAC128 and BACHD,^27,48 which were used to generate the Hu128/21 mouse model on an Htt^−/− background.³⁹ We confirmed the strong effects of AAV5-miHTT in Hu128/21 cortical neurons, where dose-dependent reduction in soluble HTTexon1, total soluble mutant HTT and FL-HTT was observed, in line with the in vivo results in this model.

Age-dependent effects (at the time of delivery of the therapeutic agent or at the time of testing) on the different degree of therapeutic efficacy between zQ175 and Hu128/21 mice can also not be excluded. The time-dependent aspect of HTT protein expression in heterozygous zQ175 mice has been well studied.^46,47 HTT aggregation in different brain regions, including striatum and cortex, is detected as early as 1 month of age, increasing up to 6–12 months in heterozygous zQ175 mice. In contrast, soluble HTTexon1 decreases with age, suggesting HTTexon1 recruitment into aggregates, whereas FL-HTT protein remains relatively constant until 6 months of age and is reduced by 12 months. Despite the very high levels of HTT protein aggregation at the time of intervention (5 months of age), AAV5-miHTT significantly reduced HTTexon1 and mutant HTT aggregates 2 months later. Previous studies with a zinc-finger protein targeting mutant HTT indicated that the effect on HTT aggregation was significantly higher with early (2 months) as opposed to late (6 months) intervention in heterozygous zQ175 mice, suggesting that early interventions might be more efficacious.¹⁵ In Hu128/21 mice, the time dependence of HTTexon1 expression and HTT aggregation has been less characterized. Nonetheless, in two independent HD mouse models and using two different analytical methods, AAV5-miHTT treatment caused a significant lowering of both FL-HTT and HTTexon1 in the striatum, cortex and hippocampus. This was confirmed in Hu128/21 primary neuronal cultures, where the observed effect was stronger than in the in vivo studies.

Striatal AAV5-miHTT administration led to target engagement in the striatum, cortex and hippocampus, albeit to a larger extent in the target region (striatum). Given that the cortex and hippocampus were not the direct targets of AAV5-miHTT administration, transgene levels in cortical and hippocampal areas indicate that AAV5 vector was actively distributed by axonal transport via corticostriatal pathways⁴⁹ and largely via polysynaptic projections between striatum and hippocampus.⁵⁰ A comparable spread of the vector from striatum (injection area) to cortical and hippocampal areas has been shown previously in mice and large animals.^{33,34,36-38,49} Based on our previous studies using comparable doses of AAV-miHTT,^33,34 miHTT levels in the striatum are expected to be approximately eight times higher than in the frontal cortex and the hippocampus. HTTexon1 levels were also significantly higher in the striatum than in frontal cortex, the former being the main affected brain area in HD patients.^47,51 As we have reported previously,³⁶ higher striatal miHTT expression is generally accompanied by more pronounced HTT lowering in striatum when compared with cortical or other brain areas.

The generation of HTT1a transcripts by aberrant splicing has been shown to occur in a CAG repeat length-dependent manner.^27-29 Hence, in contrast to FL-HTT protein, levels of the pathogenic HTTexon1 might increase and accumulate over time if CAG repeat length expands via somatic instability, as is known to occur in HD patients.³⁰ We did not measure somatic CAG repeat instability in the brain of zQ175 KI and Hu128/21 mice, but uninterrupted repeat regions of CAG repeats, such as in zQ175 KI mice, with a pure CAG expansion of ∼190 repeats,^40,42 are more prone to expand via somatic instability, as in HD patients with the HTT loss-of-interruption variant.⁵² In contrast, Hu128/21 mice, generated by intercrossing of BAC21 mice with YAC128 mice on the Htt^−/− background,³⁹ do not display any obvious somatic instability. The CAG tract length, potential differences in somatic instability, degree of aberrant splicing and HTTexon1 production between these two mouse models might explain the relatively stronger effects of AAV5-miHTT on Hu128/21 than on zQ175 KI mice. Further studies are needed to investigate whether the efficacy of AAV5-miHTT to reduce HTTexon1 is robust upon somatic CAG expansion and how this is influenced by the time of intervention. To answer those questions, treatments would need to be performed at different time points relative to the development of HD pathology in the brain.

Most current disease-modifying therapies for HD are based on HTT-lowering technologies.^10,53 However, lack of efficacy with intrathecal infusion of an antisense oligonucleotide has raised concern about the potential of HTT-lowering therapies as a treatment for HD in general.⁵⁴ One important hurdle that all HTT-lowering interventions need to overcome is the effective delivery of therapeutics into the deep brain structures, such as the striatum, characterized by high levels of HTTexon1 aggregation and neurodegeneration. Importantly, the data reported in the present study indicate that the location of the target sequence within the HTT gene can also contribute to discrepancies in therapeutic efficacy between drug candidates. Lowering of FL-HTT by targeting sequences distant from the CAG-containing exon 1 might not prevent the generation of highly toxic, aberrantly spliced HTT species. In this study, we did not perform a direct comparison with therapeutic approaches that do not target HTT exon 1; however, previous in vitro work indicates that a, HTT antisense oligonucleotide targeting exon 42 of the FL-HTT transcript leads to dose-dependent reduction of FL-HTT, but not of HTT1a transcripts.⁴⁸ We also have preliminary data to indicate that HTT-lowering strategies downstream of HTT exon 1 have no effect on HTTexon1 protein (Bates group, manuscript in preparation). Hence, when designing therapeutic HTT-lowering strategies for HD, it might be of therapeutic advantage to choose a target sequence to reduce not only FL protein (both WT and mutant), but also the more pathogenic HTTexon1 protein.

Engineered miRNAs targeting sequences close to the repeat expansion have also been effective in other CAG-repeat expansion disorders, such as spinocerebellar ataxia 3.⁵⁵ Another aspect of importance for disease modification is the question of the relative contribution of HTTexon1 aberrant splicing versus HTT proteolytic cleavage in the generation of small toxic N-terminal fragments leading to toxicity and aggregation.²⁶ The expectation is that both mechanisms contribute to the generation of toxic HTT fragments, which would be supportive of therapeutic strategies such as AAV5-miHTT, targeting both the HTT1a mRNA and the FL-HTT mRNA.

Previous studies have shown a differential effect of small interfering RNAs and antisense oligonucleotides on RNA foci or insoluble HTT mRNA clusters present in the nucleus.^56,57 Although we did not evaluate this in the present work, other studies indicate that endogenous microRNAs are active both in the cytoplasm and in the nucleus.^58,59 Therefore, miHTT would be expected to be active in the nucleus, although its effect on RNA foci or HTT clusters would need to be established.

In conclusion, our data demonstrate that exon 1-targeting miRNA delivered by AAV5 is an effective approach to reduce the levels of the pathogenic HTTexon1 fragment in an HD-associated splicing context. These results, together with all previous studies demonstrating the efficacy and safety of AAV5-miHTT in preclinical studies, support continued clinical development of AAV5-miHTT gene therapy in HD patients. Therapeutics that target the exon 1 HTT sequence address several primary causes of HD pathology and progression (the mutant FL-HTT, the pathogenic HTTexon1 fragment and the correction of somatic CAG repeat instability), suggesting that they could achieve a greater efficacy for the treatment of HD.

Data availability

The authors confirm that all the data supporting the findings of this study are included within the article. Raw data will be shared by the corresponding author on request.

Funding

This study was funded by uniQure biopharma BV, Amsterdam, The Netherlands.

Competing interests

A.S., A.V. and M.E. are employees of uniQure and may own stock and/or stock options. Filed patent applications pertaining to the results presented in this paper include the following: RNA interference induced HTT gene suppression (WO2016/102664, resulting in at least US 10,174,321, US 10,767,180 and EP 3237618B1), A companion diagnostic to monitor the effects of gene therapy (PCT/EP2019/081759), Method and means to deliver miRNA to target cells (PCT/EP2019/081822) and Targeting misspliced transcripts in genetic disorders (PCT/EP2020/075871); the last three have not yet been published.

Supplementary material

Supplementary material is available at Brain online.