Research ArticleHuntington’s Disease

Huntingtin suppression restores cognitive function in a mouse model of Huntington’s disease

See allHide authors and affiliations

Science Translational Medicine  03 Oct 2018:
Vol. 10, Issue 461, eaar3959
DOI: 10.1126/scitranslmed.aar3959

Rescuing cognition in Huntington’s disease

Huntington’s disease (HD) is a neurodegenerative disorder caused by mutation in the HTT gene. The encoded mutated protein, called huntingtin, acquires a toxic function causing motor, cognitive, and psychiatric impairments. Southwell and colleagues show that intracerebral injection of antisense oligonucleotides (ASOs) specifically inhibiting the expression of mutant Htt improved cognition and reduced anxiety and depressive behaviors in symptomatic HD mice. Moreover, HTT-targeting ASOs reduced huntingtin expression in nonhuman primates. The results suggest that ASO-based therapies might be effective for treating the cognitive impairments associated with HD.

Abstract

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disorder caused by a mutation in the huntingtin (HTT) protein, resulting in acquisition of toxic functions. Previous studies have shown that lowering mutant HTT has the potential to be broadly beneficial. We previously identified HTT single-nucleotide polymorphisms (SNPs) tightly linked to the HD mutation and developed antisense oligonucleotides (ASOs) targeting HD-SNPs that selectively suppress mutant HTT. We tested allele-specific ASOs in a mouse model of HD. Both early and late treatment reduced cognitive and behavioral impairments in mice. To determine the translational potential of the treatment, we examined the effect of ASO administration on HTT brain expression in nonhuman primates. The treatment induced robust HTT suppression throughout the cortex and limbic system, areas implicated in cognition and psychiatric function. The results suggest that ASOs specifically targeting mutated HTT might have therapeutic effects on HD-mediated cognitive impairments.

INTRODUCTION

Huntington’s disease (HD) is an autosomal dominant neurodegenerative disease caused by expansion of a polyglutamine-encoding cytosine adenine guanine (CAG) tract in exon 1 of the HTT gene to greater than 35 repeats (1). HD is characterized by progressive loss of voluntary motor control, psychiatric disturbance, and cognitive decline (2). Although moderately effective symptomatic treatments are commonly used, HD is incurable with no disease-modifying treatments available (3). HD has a complex pathogenesis that is difficult to target therapeutically (4). The mutant huntingtin (muHTT) resulting from the aberrant HTT gene is the sole cause of this complex pathogenesis and is targetable with gene suppression technologies (57), including antisense oligonucleotides (ASOs).

ASOs are, in general, chemically modified oligonucleotides that are designed to bind to RNA by Watson-Crick base pairing and catalyze a variety of downstream events (8). The traditional way ASOs are used for gene suppression is induction of ribonuclease H (RNase H)–mediated transcript degradation (8). ASOs delivered into the cerebrospinal fluid (CSF) distribute broadly throughout the mammalian central nervous system (CNS) (5, 9, 10) and are freely taken up by neurons, glia, and ependymal cells throughout the brain and spinal cord (1114). ASOs also have a long half-life in the brain (15), making them well suited to chronic CNS disorders. Intrathecal (IT) delivery of ASOs has been found to be well tolerated in clinical safety trials for multiple neurological disorders (1618), including a Food and Drug Administration–approved ASO therapy for spinal muscular atrophy. ASOs targeting HTT have demonstrated therapeutic benefit far outlasting the period of HTT lowering in mouse models of HD (5, 19), and one HTT ASO has just successfully completed a first in man phase 1/2a clinical safety trial (NCT02519036).

In addition to total HTT lowering approaches, ASOs can be designed to selectively suppress muHTT. The vast majority of HD mutation carriers are heterozygous, with one normal and one muHTT gene (1). Population genetics studies have identified single-nucleotide polymorphisms (SNPs) tightly linked to the CAG expansion and predicted that targeting as few as three to five of these “HD-SNPs” would provide an allele-specific muHTT gene silencing therapy for 80 to 90% of HD patients worldwide (2025). We have previously developed ASOs targeting SNPs common to A and B haplogroup HTT, collectively, most commonly found on CAG-expanded chromosomes in Caucasian/European (20, 22), Latin American (24), and Black South African (25) populations, and absent on C haplogroup HTT, most commonly found on control chromosomes in these populations (20, 23). Detailed screening led to the identification of allele-specific ASOs targeting two different HD-SNPs. These ASOs have high potency and selectivity against muHTT, and each could provide an allele-specific muHTT suppression therapy for about 50% of the HD population (26). What is not known is the potential for these allele-specific ASOs to alter a disease.

Although HD is often considered a motor disease, and patients with HD exhibit overt motor dysfunction, the symptoms reported to be of greatest concern are often cognitive and psychiatric disturbances (27). For any HD therapy to truly alter the quality of life for patients, altering the course of cognitive and psychiatric decline would be of enormous value. However, most preclinical work assessing potential therapeutics have focused primarily on motor end points. There is a need to understand the ability of a therapeutic to affect all aspects of human disease, including mood and cognition. This is particularly relevant for potentially disease-modifying therapeutics.

Here, we address key questions regarding the potential of ASOs to alter psychiatric and cognitive decline in HD using both a total HTT-suppressing ASO and allele-specific muHTT ASOs. First, we performed a series of experiments in a humanized rodent model of HD to determine whether ASO-mediated HTT suppression could alter cognitive and psychiatric end points. In these studies, we examine the degree of benefit and the relevance of time of intervention of total HTT- and muHTT-targeting ASOs on cognitive and psychiatric end points. Second, we determine in a larger primate brain whether the current ASO therapies and delivery techniques have a reasonable chance at suppressing HTT RNA in the regions implicated in cognitive and psychiatric disease. The work presented here supports the use of ASOs as a therapeutic for HD and provides a rationale for including cognitive and psychiatric end points in future human trials.

RESULTS

ASOs induced sustained suppression of total and mutant HTT in HD mice

Nonsymptomatic HD mice were treated at 6 weeks of age with one of three allele-specific muHTT targeting ASOs previously identified to be active and well tolerated in acute studies (14). The treatments include allele-specific ASOs muHTT1, muHTT2, and muHTT3; a total human HTT-targeting ASO (ASO hHTT; Table 1); or phosphate-buffered saline (PBS) vehicle (figs. S1 and S2). muHTT2 was not tolerated in humanized HD Hu97/18 mice in this long-term study, inducing progressive hindlimb ataxia and decreased survival (fig. S3). This effect appears to be unique to this oligonucleotide and not related to HTT lowering and thus was not pursued further.

Table 1 Summary of ASOs.

Black, PS; yellow, MOE; blue, cEt. Target SNP is underlined for allele-selective ASOs. NHP, nonhuman primate.

View this table:

To determine the duration of action of the ASOs, we quantified HTT protein [wild type (wt) and mutant (mu)] in the anterior forebrain of treated mice (ASO or vehicle) with tissues collected at various points up to 1 year of age (Fig. 1, A and B). ASO muHTT3 was the least active, inducing a maximum of 71% muHTT suppression with a duration of action of 7 to 11 weeks. ASO hHTT had intermediate activity, with a maximum 78% muHTT and 82% wtHTT suppression and a duration of action of 28 to 37 weeks. ASO muHTT1 was the most active with a maximum 89% muHTT suppression and a duration of action of >46 weeks. As expected, no significant suppression of wtHTT was observed after treatment with ASOs muHTT1 or muHTT3 (Fig. 1, A and B). Similar reductions were observed in the late-intervention study with mice treated at 6 months of age (fig. S4). At this age, Hu97/18 mice display robust motor deficits compared to WT mice. However, because of the weight gain observed in Hu18/18 mice (fig. S5), consistent genotypic differences are not observed and it is not possible to evaluate treatment effects on motor performance (figs. S6 and S7).

Fig. 1 muHTT and hHTT ASOs reduce HTT protein concentration in the brain.

Quantification of (A) muHTT and (B) wtHTT in the anterior forebrain of both hemispheres from Hu97/18 mice (n = 2, four measurements) treated with ASO or PBS at 6 weeks of age at the indicated collection ages using allelic separation immunoblotting. HTT protein densities (wt or mu) were normalized to the mean value in PBS-treated animals of the same age to determine the percentage of HTT protein remaining after ASO treatment. *Different from PBS treatment for same allele using least mean squares analysis; ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001.

muHTT suppression rescues cognitive deficits

To assess cognitive performance, we performed novel object location (NOL; Fig. 2A) and novel object preference (NOP; Fig. 2B) learning assays at 9 months of age, when PBS-treated mice show clinical signs of the disease. It has been previously shown that Hu97/18 mice displayed normal learning in both assays at 3 months of age, spatial learning deficits as assessed by NOL testing from 6 months of age, and object recognition deficits as assessed by NOP testing from 9 months of age (28). Thus, early intervention was initiated before the onset of either reported learning deficit (6 weeks of age), whereas the late intervention (6 months of age) was performed after the onset of some, but not all, learning deficits.

Fig. 2 muHTT suppression rescues cognitive deficits.

Schematic representation of (A) the NOL learning assay and (B) the NOP learning assay performed at 9 months of age in mice. (C) Bar graph showing the percentage of time spent investigating the target (novel position) object in Hu18/18 and Hu97/18 mice after early treatment with PBS or the different ASOs indicated in the figure. (D) Bar graph showing the percentage of time spent investigating the target (novel object) object in Hu18/18 and Hu97/18 mice after early treatment with PBS or the different ASOs indicated in the figure. (E) Bar graph showing the percentage of time spent investigating the target (novel position) object in Hu18/18 and Hu97/18 mice after late treatment with PBS or the different ASOs indicated in the figure. (F) Bar graph showing the percentage of time spent investigating the target (novel object) object in Hu18/18 and Hu97/18 mice after late treatment with PBS or the different ASOs indicated in the figure. Asterisk (*) indicates difference from PBS Hu97/18 for trial difference versus treatment solution for fixed-effects analysis at P < 0.05.

As expected, PBS-treated Hu97/18 mice exhibited cognitive deficits, failing to show a preference for the target object in either assay, where the respective control animals (Hu18/18 mice) performed normally (Fig. 2, C to F). These differences are unlikely to be related to differences in activity because all mice showed similar duration of total object investigation during trial 1 (fig. S8). Early treatment with all three ASOs prevented development of spatial learning deficits (solution for fixed effect: hHTT, P = 0.0169; muHTT1, P = 0.0419; muHTT3, P = 0.0243; Fig. 2C). Similarly, late treatment with both ASOs tested restored normal spatial learning in Hu97/18 mice (solution for fixed effect: hHTT; P = 0.0492; muHTT1, P = 0.0352; Fig. 2E). Thus, spatial learning deficits were prevented (presymptomatic) by both allele-specific (muHTT1 or muHTT3) and the total (hHTT) HTT ASO. Both the allele-selective (muHTT1) and nonselective (hHTT) ASOs reversed deficits when treatment was initiated late (6 months of age). Although the phenotype is less robust than in the NOL test, both total and allele-selective HTT suppression, regardless of timing of intervention, improved NOP performance with treated Hu97/18 animals exhibiting similar preferences as Hu18/18 controls (Fig. 2, D to F).

muHTT and hHTT suppression rescues anxiety-like behavior

Anxiety-like behavior was assessed using time spent in the center of an open field at 3 and 9 months of age and time spent in open arms of an elevated plus maze at 6 months of age. No effect of genotype or treatment was observed in total distance traveled or mean velocity, indicating that activity was similar in Hu18/18 and Hu97/18 mice and unaffected by ASO treatment (figs. S9, A to D, and S10, A to D). However, assessment of exploratory behavior demonstrated a genotype effect. PBS-treated Hu97/18 mice less frequently entered and spent substantially less time in the center of the open field compared to the Hu18/18 mice at all time points evaluated (Fig. 3, A to D, and fig. S9, E to H). As expected, PBS-treated mice, independent of genotype, displayed reduced activity and increased anxiety-like behavior with age (3 months old versus 9 months old; Fig 3, A to D). This is consistent with a previous phenotyping study in this mouse strain (28). Thus, Hu97/18 mice exhibit a clear anxiety-like phenotype compared to Hu18/18 mice, independent of age and motor performance.

Fig. 3 muHTT suppression normalizes anxiety behavior.

Anxiety-like behavior was assessed during (A to D) a 10-min exploration of a brightly lit open field at the indicated ages or (E and F) a 5-min exploration of an elevated plus maze at 6 months of age in the animal groups indicated in figure. (A) Entries into and (B) time spent in the center of the open field in Hu18/18 and Hu97/18 mice treated early with PBS or with the ASOs indicated in the figure. (C) Entries into and (D) time spent in the center of the open field in Hu18/18 and Hu97/18 mice treated late with PBS or with the ASOs indicated in the figure. (E) Entries into and (F) total time spent in the open arms of the elevated plus maze in Hu18/18 and Hu97/18 mice treated early with PBS or with the ASOs indicated in the figure. Asterisk (*) indicates difference between PBS Hu97/18 and PBS Hu18/18, and “#” indicates difference from PBS Hu97/18 by differences of least mean squares. *P < 0.05, **P < 0.01, ***P < 0.001.

Early intervention with all ASOs resulted in prevention of anxiety-like behavior at 3 months of age compared to PBS-treated Hu97/18 mice (Fig. 3A) [difference of least mean squares square root transformed (center entries: hHTT, P < 0.0001; muHTT1, P = 0.0251; muHTT3, P = 0.0010; and center time: hHTT, P = 0.0008; muHTT1, P = 0.0063; muHTT3, P < 0.0001)]. By 9 months of age, Hu97/18 mice treated early with ASO hHTT or muHTT3 showed decreased number of center entries, suggesting attenuation of benefit when HTT returned to baseline. Consistent with this, Hu97/18 mice treated with ASO muHTT1, which exhibits continued reduction of muHTT protein at the time of assessment, continued to display rescue of anxiety-like behavior compared to PBS-treated Hu97/18 mice [difference of least mean squares square root transformed (center entries: hHTT, P = 0.0637; muHTT1, P = 0.0016; muHTT3, P = 0.0580; and center time: hHTT, P = 0.1735; muHTT1, P = 0.0072; muHTT3, P = 0.2732)] (Fig. 3, A and B). When intervention is initiated in mice already displaying anxiety-like behavior (late intervention), both hHTT and muHTT1 restored normal exploratory behavior, inducing a significant increase in center time compared to PBS-treated Hu97/18 mice. However, center entries at 3 and 9 months of age were similar in these groups [difference of least mean squares (center entries: hHTT, P = 0.0598; muHTT1, P = 0.0610; and center time: hHTT, P = 0.0499; muHTT1, P = 0.0054)] (Fig. 3, C and D). However, both ASOs reverted the decline in anxiety behavior, and muHTT1 also reverted the decline in exploratory behavior observed in PBS-treated Hu97/18 mice between 3 and 9 months of age [change in differences of least mean squares square root transformed (center entries: hHTT, P = 0.0435; muHTT1, P = 0.0169; and center time: hHTT, P = 0.0735; muHTT1, P = 0.0082)].

PBS-treated Hu97/18 mice also displayed anxiety-like behavior during elevated plus maze exploration, less frequently entering and spending less total time in the open arms (Fig. 3, E and F). Early treatment with muHTT1, but not with hHTT or muHTT3, rescued this phenotype, inducing a substantial increase in open arm entries and time compared to PBS-treated Hu97/18 mice [difference of least mean squares square root transformed (open arm entries: hHTT, P = 0.1614; muHTT1, P = 0.0268; muHTT3, P = 0.9907; and open arm time: hHTT, P = 0.7416; muHTT1, P = 0.0459; muHTT3, P = 0.6247)] (Fig. 3, E and F). As expected, no differences were observed in the late-intervention group because assessment occurred before treatment initiation (fig. S10, E and F). Together, these data indicate that HTT suppression can prevent onset of anxiety and reduce the decline in normal exploratory behavior if administered to mice already exhibiting an anxiety-like phenotype. Because benefit at later time points after early intervention was limited to the more potent, longer-lasting allele-specific ASO (muHTT1), this suggests that either robust HTT suppression and/or continued HTT suppression is required for benefit to these phenotypes.

muHTT suppression may attenuate depressive-like behavior in mice

Hu97/18 mice exhibit nonprogressive depressive-like behavior during forced swimming detected as early as 3 months of age (28, 29). Because this phenotype is nonprogressive and the stress of forced swimming could alter subsequent behavioral performance, ASO-treated mice were evaluated for immobility during forced swimming at the end of the study at 12 months of age as a measure of depressive-like behavior. Consistent with previous studies, increased time spent immobile was observed in PBS-treated Hu97/18 mice compared to Hu18/18 mice (Fig. 4, A and B). Early treatment with ASO muHTT1 resulted in performance not only similar to Hu18/18 mice when tested >10 months after a single treatment but also not different from PBS-treated Hu97/18 mice. Treatment with ASO hHTT or muHTT3 did not reduce immobility time (differences of least mean squares square root transformed compared to Hu97/18: hHTT, P = 0.1301; muHTT1, P = 0.0945; muHTT3, P = 0.9838) (Fig. 4A).

Fig. 4 muHTT suppression may ameliorate depressive behavior.

Depressive-like behavior was assessed at 12 months of age after early (A) and late (B) intervention in the animal groups indicated in figure by measuring the time spent immobile during the final 5 min of a 6-min forced swim. Asterisk (*) indicates difference between PBS Hu97/18 by differences of least mean squares. *P < 0.05, **P < 0.01.

Treatment initiated later, at 6 months of age, with both the total HTT or allele-selective ASO resulted in time immobile in Hu97/18 mice similar to Hu18/18 control mice (Fig. 4B), but again, not different from PBS-treated Hu97/18 mice (differences of least mean squares square root transformed: hHTT, P = 0.5844; muHTT1, P = 0.1656). Together, these data suggest that the nonprogressive depressive-like behavior in Hu97/18 mice might involve multiple factors not dependent solely on the presence of muHTT at the time of testing.

muHTT suppression ameliorates forebrain atrophy and prevents loss of striatal DARPP-32

To assess HD-like neuropathological changes, we assessed forebrain weight, stereological striatal and cortical volume, and striatal dopamine- and cyclic AMP-regulated phosphoprotein (DARPP-32) immunoreactivity at 12 months of age, either 10.5 (in the early-treatment group) or 6 (in the late-treatment group) months after treatment. Consistent with previous reports (28), 12-month-old PBS-treated Hu97/18 mice had smaller forebrains than PBS-treated Hu18/18 mice (Fig. 5, A and B). Mean forebrain loss after early intervention was not different in ASO-treated compared to PBS-treated Hu97/18 mice (differences of least mean squares square root transformed: hHTT, P = 0.7077; muHTT1, P = 0.1899; muHTT3, P = 0.6479) (Fig. 5A). In the late-intervention group, muHTT1 induced a significant preservation in forebrain weight, whereas hHTT had no effect (differences of least mean squares square root transformed: hHTT, P = 0.2669; muHTT1, P = 0.0023) (Fig. 5B), suggesting that ongoing muHTT suppression during the period of forebrain atrophy susceptibility may be required for preservation.

Fig. 5 muHTT suppression ameliorates aspects of forebrain atrophy and increases striatal DARPP-32.

Neuropathology and brain histology were performed at 12 months of age after ASO treatment at 6 weeks or 6 months of age. Bar graphs showing the effect of early (A, C, E, and G) and late (B, D, F, and H) PBS or ASO treatment on weight of postperfusion forebrains (A and B), stereological cortical volume (C and D), stereological striatal volume (E and F), and striatal DARPP-32 immunoreactivity (G and H) in Hu18/18 and Hu97/18 mice treated with PBS or with the different ASOs indicated in the figure. Asterisk (*) indicates difference between PBS Hu97/18 and PBS Hu18/18, and “#” indicates difference from PBS Hu97/18 by differences of least mean squares analysis. *P < 0.05, **P < 0.01. IOD, integrated optical density.

Compared to PBS-treated Hu18/18 mice, PBS-treated Hu97/18 mice also have cortical and striatal atrophy but not cerebellar atrophy (Fig. 5, C to F, and fig. S11). Early or late intervention with muHTT1 prevented cortical atrophy [early treatment: 12.5% loss PBS, 15.2% loss hHTT, 3.2% loss muHTT1, 15.3% loss muHTT3 (differences of least mean squares square root transformed: hHTT, P = 0.8788; muHTT1, P = 0.0276; muHTT3, P = 0.4467); and late treatment: 13.2% loss PBS, 8.0% loss hHTT, 4.6% loss muHTT1 (differences of least mean squares square root transformed: hHTT, P = 0.1205; muHTT1, P = 0.0012)] (Fig. 5, C and D). Conversely, only late intervention with muHTT1 significantly increased Hu97/18 striatal volume [early treatment: 8.8% loss PBS, 13.4% loss hHTT, 4.5% loss muHTT1, 11.8% loss muHTT3 (differences of least mean squares square root transformed: hHTT, P = 0.2227; muHTT1, P = 0.1228; muHTT3, P = 0.3132); and late treatment: 11.8% loss PBS, 9.6% loss hHTT, 4.7% loss muHTT1 (differences of least mean squares square root transformed: hHTT, P = 0.4152; muHTT1, P = 0.0032)] (Fig. 5, E and F).

Early or late intervention with allele-specific ASOs muHTT1 or muHTT3 induced an increase in Hu97/18 striatal DARPP-32, whereas total HTT ASO hHTT did not [(differences of least mean squares square root transformed) early intervention: hHTT, P = 0.4597; muHTT1, P = 0.0306; muHTT3, P = 0.0062; and late intervention: hHTT, P = 0.0801; muHTT1, P = 0.0341)] (Fig. 5, G and H). For this measure, benefit from allele-specific muHTT suppression was observed long after HTT had returned to basal amounts (~8 months). This result suggests that protection from striatal DARPP-32 loss is enhanced by allele-specific muHTT suppression.

HTT ASOs reduced HTT in cortical and limbic systems in the NHP brain

In the clinic, HTT ASOs are currently being delivered to HD patients via lumbar puncture (LP). To determine whether ASOs delivered by LP can target the regions implicated in cognitive and psychiatric control in a larger brain, we delivered ASOs targeting NHP HTT (ASO NHTT1 and NHTT2) by LP to cynomolgus monkeys. Human ASOs hHTT, muHTT1, and muHTT3 do not target NHP HTT, and thus, ASOs homologous to NHP HTT were used. To determine whether the effects are similar across ASOs, two independent sequences were included. We have previously delivered NHP ASOs via constant IT infusion (ASO NHTT2) and demonstrated targetability of cord and cortex with more limited exposure to striatum (5). Here, animals received weekly bolus injections of HTT ASO (each animal received ascending doses of 4, 8, 12, and 16 mg, for a total of 30 mg) or vehicle for 4 weeks, with tissue collected 4 weeks after the last dose (Fig. 6A). Distribution of two HTT ASOs, as determined by measurement of ASO concentrations in different brain regions, was similar between the two ASOs, despite differing oligonucleotide sequences and length (Fig. 6B and Table 1). Two animals had extremely low drug concentration (<1 μg/g in spinal cord), likely due to technical failure to deliver ASOs to the IT space. Consistent with the distribution of drug, HTT ASOs reduced HTT mRNA in all cortical regions assessed (Fig. 6C). HTT was also reduced in limbic regions, including hippocampus, amygdala, and cingulate and entorhinal cortices (Fig. 6C). Together, these data suggest that, in a larger brain, ASOs delivered by LP can effectively suppress HTT in cortical and limbic systems.

Fig. 6 HTT suppression in cortical and limbic structures in the NHP brain.

(A) Schematic of study design. Cynomolgus monkeys received weekly LP delivery of NHP HTT ASOs (NHTT1 or NHHT2) or vehicle (aCSF). Tissue is collected 4 weeks after the final dose. (B) Drug concentration in spinal cord and cortex. (C) Monkey HTT mRNA in spinal cord and cortical and limbic regions quantified by quantitative polymerase chain reaction and expressed as means ± SEM percentage (%) of HTT relative to vehicle-treated controls. Asterisk (*P < 0.05, **P < 0.01, ***P < 0.001) denotes statistically significant differences, with changes relative to same region aCSF [one-way analysis of variance (ANOVA) and Dunnett’s multiple comparison tests].

DISCUSSION

HD is a disease of the whole brain, with patients suffering motor, cognitive, and psychiatric disturbances. Here, we show that ASO therapy ameliorated psychiatric and cognitive impairments in a mouse model of HD. In addition, cognitive and psychiatric dysfunctions were not only prevented but also, in some paradigms, reversed. Moreover, with a clinically relevant route of delivery, HTT ASOs distribute to regions involved in psychiatric and cognitive function in an NHP brain, suggesting that the same approach might be beneficial in the clinical setting.

Our results build upon previous data demonstrating that lowering HTT ameliorated motor phenotypes in HD mice. Genetic inactivation of the muHTT transgene in a symptomatic conditional HD model was shown to reverse HD neuropathology and recover motor performance (30), indicating that reducing muHTT expression could be both a preventative and a restorative therapy. Suppression of HTT by RNA interference (RNAi) has shown benefit to a variety of HD-like signs in several transgenic HD mouse models reviewed here (31). Although these studies support the idea that reducing HTT expression could be an effective strategy for treating HD, there are limitations to the translatability to effective therapeutics. In our previous study showing that HTT-targeting ASOs ameliorated motor and anxiety phenotypes, the drug was delivered via a 2- to 4-week intracerebroventricular (ICV) infusion, and allele selectivity was achieved with compounds that would not be allele-selective in human patients. Here, we increased the translatability of the treatment by delivering ASOs via a more therapeutically feasible bolus injection and by using ASOs that selectively suppress either mutant or total HTT. Furthermore, we demonstrated a further benefit of HTT suppression on cognition, anxiety, and depressive-like behaviors. Because the cognitive and psychiatric symptoms in HD patients are often reported as being the most detrimental, a benefit in these areas would have high therapeutic impact.

Given the compelling data in HD rodent models, described here and previously (5, 19), there is a reasonable expectation that, with a disease-modifying therapy such as gene suppression approach targeting muHTT, most aspects of disease may be altered if the regions implicated are targeted. Although the striatum is the site of the earliest and most marked pathology (32, 33), HD is a disease of the whole brain with clear changes broadly through much of the CNS. Layer-specific cortical neurodegeneration is a key feature of the HD brain (34). Genetic inactivation of the muHTT transgene in the cortex of BACHD HD model mice was shown to be more protective than inactivation in the striatum, though combined inactivation of striatal and cortical muHTT was found to be most beneficial (35). In addition, hypothalamic neurodegeneration and atrophy are observed from early stages in HD (36), and selective removal of muHTT from hypothalamic neurons in BACHD mice ameliorated the anxiety phenotype (37). Finally, adult hippocampal neurogenesis is impaired in the YAC128 mouse model of HD (38). These findings suggest that, to be maximally efficacious, an effective HD gene therapy should achieve broad CNS distribution. Here, mimicking the human treatment paradigm by delivering ASOs via LP, we were able to suppress HTT throughout the cortex and limbic structures of NHPs, suggesting that similar treatment could be effective in reducing HTT in regions involved in cognition, anxiety, and depression in human patients. Minimal reduction was observed in the striatum, suggesting that more potent ASOs or higher doses would be necessary for more effective targeting of this region.

Here, we suppressed muHTT with an allele-selective approach that could be used clinically. The most obvious approach to allele-specific gene silencing is to target the causative mutation. However, targeting the HD mutation itself, which is an expansion of a normally occurring CAG repeat tract, could result in off-target knockdown of other CAG tract–containing genes (39). Fortuitously, the HD mutation is rarely the only difference between the wt and muHTT alleles in individuals. Population genetics studies have identified SNPs tightly linked to the CAG tract expansion (2025). Only ~3% of SNPs in the HTT gene are present in mature mRNA, with the remaining 97% being intronic (10). Thus, only a very small proportion of HTT SNPs are available for binding with RNAi reagents that use the cytosolic Dicer/RISC machinery (40). Conversely, ASOs that work through RNase H to degrade the transcript are active both in the nucleus and the cytosol and can therefore target 100% of HTT SNPs, including those in introns.

We have previously developed ASOs that target HD-SNPs (14, 26, 41, 42). Our development pipeline has included identification of therapeutically relevant SNPs and validation of their susceptibility to ASO-mediated muHTT suppression (26), generation of a humanized mouse model of HD, Hu97/18, recapitulating HD-SNP heterozygosities (28), optimizing ASO design for enhanced selectivity (41), in vitro screening for potency, selectivity, and tolerability of ASOs in primary Hu97/18 neurons (42), and in vivo lead selection and evaluation of drug-like properties in Hu97/18 brain (14). Cumulatively, these studies led to identification of the ASOs tested here. They have shown high potency against muHTT and excellent acute CNS tolerability and could each provide an allele-specific muHTT silencing therapy for about 50% of the HD population. Here, we have shown, in a humanized HD mouse model, that these compounds can alter many aspects of disease. Thus, SNP-targeting ASOs have the potential to provide benefit to human HD patients.

Our results show that temporary suppression of HTT expression exerts a sustained benefit in the cognitive NOL and NOP behavioral tests. With regard to cognitive benefits, muHTT knockdown with muHTT3 lasted only 7 to 11 weeks after dosing, with muHTT protein returning to basal amounts about 5 months before cognitive testing, demonstrating a lasting benefit to cognitive performance by transient early muHTT suppression. This is consistent with a previous report of persistent benefit to motor and anxiety-like phenotypes after transient HTT suppression with a total HTT ASO (5). In contrast, this effect was not observed in the anxiety tests, indicating that there are aspects of disease where sustained benefits are reachable with transient HTT suppression and others where suppression would optimally be continual. One example highlighting the importance of time of intervention is the effect of HTT suppression on neuropathology. muHTT1 prevented all aspects of pathology with late treatment and was less effective with a single early treatment. This is consistent with a previous report of lack of preservation of brain mass in BACHD mice after treatment initiation early in disease at 2 months of age (5). Understanding how the time of intervention affects different aspects of the disease is critical to understand to inform clinical trial design.

Although the studies outlined here were not explicitly designed to directly compare allele-selective HTT suppression versus nonallele-selective HTT suppression, we included two allele-selective ASOs and one previously characterized nonallele-selective ASO (5) in the pretreatment studies. One of the allele-selective compounds was more active and one less active than the nonallele-selective ASO. Therapeutic effects in several tests were observed with all three compounds, with a rank order of benefit from most potent to least potent regardless of allele selectivity, suggesting that the degree of muHTT suppression is the major driver for benefit. Preservation of striatal DARPP-32 was the only test in which we observed an improvement after treatment with both allele-selective ASOs, but not after treatment with the nonselective ASO, hinting at a moderate benefit of allele selectivity.

A limitation of this work is that, because of the inherent differences in the profiles of the compounds, we are unable to discern subtle benefits from allele selectivity and potential detriments from more robust total HTT suppression. One possible area for differences in allele-selective suppression versus nonallele-selective suppression that was not assessed here is motor deficits. Because of weight gain and the resulting decline in motor performance seen in both the control and HD genotypes, the Hu97/18 model is not well suited for assessment of muHTT-driven motor end points, and as expected, consistent genotype differences were not observed. Thus, although Hu97/18 mice are an ideal model for assessing cognitive and psychiatric dysfunction and neuropathology, they are not suited for assessing the effects of ASO-mediated HTT lowering on motor performance. Despite these limitations, together, these data support the use of ASOs for the treatment of HD, with the potential to improve cognitive and psychiatric disturbances in HD patients. With the most active compound, intervention before or after symptom onset proved beneficial, and the benefits were sustained up to 10.5 months after a single ASO injection. Our data highlight that, by directly targeting the underlying disease mechanism, we can achieve marked effects to the many aspects of the HD phenotype.

MATERIALS AND METHODS

Study design

The primary objective of this study was to determine the ability of HTT ASOs, particularly the allele-selective ASOs (14, 42), to alter an array of phenotypes in HD animals, with a focus on cognitive and psychiatric phenotypes. To determine whether HTT suppression could alter cognitive and psychiatric decline, a series of experiments were conducted in humanized HD mice. These mice exhibit robust and quantifiable psychiatric and cognitive dysfunction mimicking that was observed in human HD patients (28, 29). To determine the benefits of ASO treatment, mice were treated at 6 weeks of age with 300 μg of ASO delivered by ICV bolus injection with one of three allele-specific ASOs, two targeting rs7685686 [ASOs muHTT1 and muHTT2, previously called A16 and A21, respectively (14)] and one targeting rs6446723 [ASO muHTT3, previously called G1 (14)], an ASO hHTT [previously called HH1 (14)], or PBS vehicle.

Therapeutic end points were based on previous phenotyping studies in the same mouse strain (28, 29), as were power calculations for minimum number of animals for each end point. Cognitive function was evaluated by NOL and preference learning assays at 9 months of age. Anxiety-like behavior was evaluated during open-field exploration at 3 and 9 months of age and elevated plus maze exploration at 6 months of age. Depressive-like behavior was evaluated during forced swimming at 12 months of age. Body weight was assessed at 2-month intervals from 2 to 12 months of age. Motor behavior was evaluated by longitudinal accelerating rotarod and spontaneous climbing tasks at 2-month intervals from 2 to 12 months of age. Finally, neuropathology and brain histology were evaluated at 12 months of age (fig. S2).

To determine whether timing of intervention alters benefit, a second cohort of mice were treated with the more potent allele-selective ASO (muHTT1), hHTT, or vehicle (PBS) at 6 months of age and assessed under the same experimental paradigms as the early intervention (6 weeks of age). Other than timing of treatment, to control for potential confounds from handling and testing, the two cohorts (early and late intervention) were treated identically, thus allowing for direct comparison between them. The early time point was selected to be as early as possible, but still in a developed adult animal, before reported deficits to represent a premanifest population. Late intervention was selected to be when clear phenotype was present, but there was still active disease progression to represent an early/mid-symptomatic population. Thus, late intervention was started at 6 month of age when established anxiety and depressive phenotypes as well as some learning and memory deficits were present (28).

Frequency and timing of testing were carefully selected on the basis of both published (28, 29) and unpublished experience, to exhibit the most robust genotype effects in a given task, without influencing other tasks. Open-field exploration was performed twice because we have noted that repeat testing is acceptable with a substantial intertesting interval. Elevated plus maze was performed only once for each mouse, at 6 months of age, because we have previously shown Hu97/18 mice to demonstrate an anxiety-like phenotype in this test at this time point and we have also learned that any repetition of this test using the same mice alters performance. Similarly, cognitive testing was limited to one point because we have observed that mice once tested in the object learning paradigms perform substantially different than mice of the same age, genotype, and sex when encountering the paradigm for the first time. Hence, we allotted only the 9-month time point for this assessment. This is a time sufficiently past the loss of spatial learning at about 6 months and the earliest known time of loss of object recognition (28), so most probable to be able to demonstrate either an amelioration or an exacerbation of learning deficits. Depressive-like behavior was assessed at 12 months of age during forced swimming because this is a nonprogressive phenotype (29), for which the stress of the test precludes repeat testing or testing before other psychiatric or cognitive tests. Raw data are provided in table S6.

Mice and treatments

Male and female Hu97/18 and Hu18/18 mice (28), generated and bred within the transgenic animal facility of the University of British Columbia’s Centre for Molecular Medicine and Therapeutics, were maintained on a 12-hour light/12-hour dark cycle in a clean facility with free access to food and water. All experiments were performed with the approval of the Animal Care Committee of the University of British Columbia. ASOs were delivered by ICV bolus injection of 300 μg in a total volume of 10 μl of sterile PBS as previously described (14). Hu97/18 and Hu18/18 littermates were randomly assigned to treatment groups and divided into cohorts with similar genotype, sex, and treatment proportions. Final treatment groups were 24 to 36 animals per genotype for early intervention and 15 to 21 animals per genotype for late intervention. Two Hu97/18 mice per time point were sacrificed at 1.75, 2, 3, 4, 6, 8, 10, or 12 months of age for early intervention and at 10 months of age for late intervention for brain HTT quantitation. These time points were 1 and 2 weeks and 1.5, 2.5, 4.5, 6.5, 8.5, and 10.5 months after treatment for early intervention and 4 months after treatment for late intervention.

HTT quantification

Mice were sacrificed with an overdose of tribromoethanol (Avertin), and brains were removed. A 1-mm section of anterior forebrain was collected from each hemisphere using a coronal rodent brain matrix (ASI Instruments). Tissue was lysed, and HTT protein was quantified in 40 μg of total protein using allelic separation immunoblotting as previously described (14, 26).

Behavior testing

All behavior tests were performed under white light during the animal’s dark phase by researchers blind to genotype and treatment. Methods for supplementary motor behavior assessments (rotarod, spontaneous climbing, and body weight) can be found in Supplementary Methods.

For elevated plus maze exploration, mice were videotaped during a 5-min exploration of an elevated plus maze with 30 cm × 10 cm arms, 20-cm walls on closed arms, and 50-cm elevation. Distance traveled, mean velocity, and entries into and time spent in the open arms were scored using EthoVision XT 7 (Noldus). Animals with <2 cm/s mean velocity were excluded from analysis for having failed to explore the maze. Trials in which EthoVision failed to track the animal more than 10% of the time were excluded from analysis.

Open-field exploration, NOL, and NOP: mice were videotaped during a 10-min exploration of a 50 cm × 50 cm brightly lit open field. Distance traveled, mean velocity, and entries into and time spent in the center of the field were scored using EthoVision XT 7. Animals with <2 cm/s mean velocity were excluded from analysis for having failed to explore the arena. Trials in which EthoVision failed to track the animal more than 10% of the time were excluded from analysis. After a 5-min intertrial interval (ITI), mice were returned to the field and recorded during a 5-min trial with two objects placed in the upper corners. After a 5-min ITI, mice were recorded during a 5-min trial with the object on the right moved to the lower corner (NOL). On the subsequent day, after a 5-min reacclimation to the field, mice were recorded during a 5-min trial with the two objects in their original positions. After a 5-min ITI, mice were recorded during a 5-min trial with the object on the right replaced with a novel object (NOP). Duration of investigation to each object was recorded with EthoVision XT 7, and the percentage of investigations to the target object on the right was scored for each trial. Animals that did not investigate both objects were excluded from analysis. Trials in which EthoVision failed to track the animal more than 10% of the time were excluded from analysis.

For the forced swim test, mice were recorded during a 6-min forced swim in a 25 cm × 19 cm cylinder filled with room temperature water. Time spent immobile (floating) during the final 5 min was scored.

Neuropathology and brain histology

Mice were anesthetized with 2.5% Avertin and perfused transcardially with PBS and 4% paraformaldehyde (PFA) in PBS. Brains were removed and postfixed overnight at 4°C in 4% PFA. Brains were then weighed, divided into forebrain and cerebellum, and weighed again. Forebrains were cryoprotected by equilibration in 30% sucrose, frozen and mounted in optimal cutting temperature (Sakura), and cut by cryostat into 25-μm free-floating coronal sections.

For stereological volumetric analysis, a series of sections spaced 200 μm apart and spanning the striatum were stained for neuronal nuclei using NeuN (1:1000; Millipore) primary and biotinylated goat anti-mouse secondary antibody (1:1000; Vector). Signal was amplified using the ABC Elite kit (Vectastain) and visualized with metal enhanced diaminobenzidine (DAB; Thermo Fisher Scientific). Forebrain structure volumes were determined by tracing the area using StereoInvestigator software (Microbrightfield) and applying the Cavalieri principle.

To test for DARPP-32 immunoreactivity, a series of four midstriatal sections spaced 200 μm apart were stained for DARPP-32 (1:1500; R&D Systems) using biotinylated goat anti-rat secondary antibody and DAB detection as above. The integrated optical density of striatal staining was quantified in each brain as previously described (43) and normalized to the mean value for PBS-treated Hu18/18 mice stained together.

NHP IT delivery

All procedures were accomplished using a protocol approved by the appropriate institutions Institutional Animal Care and Use Committee. NHP in life was performed at Northern Biomedical Research (Michigan) and subsequent tissue analysis at Ionis Pharmaceuticals. ASO dose administration was performed by LP to cynomolgus monkeys of both sexes. Each animal was provided dexmedetomidine hydrochloride intramuscularly (0.04 mg/kg) to induce sedation. About 15 min later, each animal received an intramuscular injection of ketamine hydrochloride (2.5 mg/kg) to maintain sedation. The LP was performed using a 25-gauge BD Whitacre spinal needle. Once confirmation of IT needle placement was made, a 1-ml dose of ASO or vehicle was administered over about 1 min using a syringe pump. Upon completion of dosing, the animals were provided atipamezole hydrochloride IM at a dose of 0.2 mg/kg to reverse anesthesia. Brain and spinal cord samples were harvested for the determination of ASO tissue concentrations and target mRNA expression and stored at −80°C until analysis. Monkey HTT mRNA quantification was performed as described previously (5, 9).

Statistical analysis

Mice that developed the natural seizure disorder that FVB mice are susceptible to, known as reactive mice, as determined by total brain weight over 500 mg and enlarged triangular forebrain, were excluded from analysis. The statistical analyses for Figs. 1 to 5 were conducted using linear mixed-effect models. In each data set, there were two hypotheses of primary interest: (i) Was there a significant treatment versus PBS (placebo) effect in the Hu97/18 group? (ii) Was there a significant performance difference between the PBS 97/18 group and the PBS 18/18 group? Litter membership was treated as a random effect in all analyses. Mouse ID was an additional random effect for all tests that involved repeated measurements of the same mice. In many cases (all noted in the results), a square root or logarithmic transform of the outcome measurement was used to improve the match between model residual data distributions and standard inference assumptions of normally distributed data with equal variance among treatment groups. In cases where about equal residual variance could not be attained with a transformation, unstructured residual covariance matrices were used to fit the model. Denominator degrees of freedom for all F tests were adjusted using the Kenward-Rogers correction (44), which often results in a noninteger degree-of-freedom value. All models were estimated using the MIXED procedure of SAS/STAT version 14.1

For repeated measurement models, we addressed the potential for bias due to mice sacrifices and missing data by including all available data from all mice that had at least a baseline measure, regardless of whether they had one or two measurements. In doing so, we preserve the validity of the mixed-effect model estimates under the assumption that any missing data are missing at random (ignorable) after conditioning on outcome and predictor data that are present and used in the model (45).

In view of the multiple hypotheses tested on these mice, we calculated upper bounds of the false discovery rates (FDRs) using the method of Benjamini and Hochberg (46). To avoid a plausible lack of positive regression dependence among true null hypotheses, we used two separate P value pools to estimate the FDR for the hypotheses related to treatment effects and hypotheses related to genotype differences.

The statistical analyses for Fig. 6 were conducted using two-way ANOVA and Dunnett’s post hoc test for repeated measures or Bonferroni post hoc test for single time points, with GraphPad Prism 5 software (GraphPad). Error bars on graphs are SEM. Significance markers on figures are from post hoc analysis (ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001).

The statistical analyses for fig. S2 were conducted using a log-rank (Mantel-Cox) test for comparison of survival curves. The statistical analyses for figs. S3 to S9 were conducted using two-way ANOVA and Dunnett’s post hoc test for repeated measures or Bonferroni post hoc test for single time points, with GraphPad Prism 5 and 7 software (GraphPad).

Error bars on graphs are SEM. Significance markers on figures are from post hoc analysis (ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001). Results of statistical analyses are reported in tables S1 to S5.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/461/eaar3959/DC1

Methods

Fig. S1. Mechanism of ASO-mediated SNP-targeted allele-specific muHTT suppression.

Fig. S2. Experimental design.

Fig. S3. ASO tolerability.

Fig. S4. Comparison of HTT protein at 10 months of age in early- and late-intervention groups.

Fig. S5. Body weight.

Fig. S6. Accelerating rotarod.

Fig. S7. Spontaneous climbing.

Fig. S8. Total object investigation time during NOL trial 1.

Fig. S9. Supplementary open-field exploration.

Fig. S10. Supplementary elevated plus maze exploration.

Fig. S11. Cerebellum weight.

Table S1. Linear mixed-effect models analysis HTT protein levels.

Table S2. Linear mixed-effect models analysis: Early intervention.

Table S3. Linear mixed-effect models analysis: Late intervention.

Table S4. ANOVAs with multiple-comparison correction: Early intervention.

Table S5. ANOVAs with multiple-comparison correction: Late intervention.

Table S6. Raw data (Excel file).

REFERENCES AND NOTES

Acknowledgments: We thank W. Zhang, B. Felczak, S. Ko, J. Rupar, V. Kovalik, Q. Xia, and M. Wang for technical assistance, M. Amirabassi for administrative assistance, and S. Sanders, C. Kay, S. Warby, and J. Carroll for discussion and support. Funding: This work was supported by grants from Ionis Pharmaceuticals, The Canadian Institutes of Health Research (CIHR MOP-84438), and the Huntington Society of Canada (HSC). A.L.S. held postdoctoral fellowships from CIHR, HSC, the Michael Smith Foundation for Health Research, and the Huntington’s Disease Society of America. N.H.S. and N.S.C. held postdoctoral fellowships from CIHR. Author contributions: A.L.S. selected the ASOs and treatments, designed the therapeutic efficacy trial in Hu97/18 mice, managed cohort assembly, performed ICV ASO injections, developed the behavioral paradigms, supervised data collection, analyzed data, generated figures, drafted and revised the manuscript, and coordinated the collaboration between the University of British Columbia (UBC) and Ionis Pharmaceuticals. H.B.K. participated in the design, analyzed samples from the NHP study, and assisted with manuscript preparation, revision, statistical analysis, and figure assembly. D.L. performed statistical analyses and assisted with manuscript revision. N.H.S. assisted with ASO selection, therapeutic efficacy trial design, data interpretation, and presentation. M.P.P. assisted with therapeutic efficacy trial design, data interpretation, and manuscript revision. E.B.V. assisted with behavioral paradigm optimization, performed behavior testing, and assisted with data interpretation. N.S.C. assisted with HTT protein time course experimental design, data interpretation, and presentation. M.E.Ø. designed the ASOs used in this study. L.M.A. assisted with neuropathology and histology protocol design and performed neuropathology and histology experiments. Y.X. assisted with therapeutic trial design, performed ICV ASO injections, assisted with design of, and performed terminal tissue and sample collection. L.D.C. assisted with design of and performed terminal tissue and sample collection. H.F.-B. optimized protocols for and performed HTT quantification. C.N.D. assisted with HTT protein time course experimental design and performed HTT quantification. B.F. analyzed tissues from the NHP study and assembled figures. E.E.S. and P.P.S. designed ASOs, participated in NHP study design, contributed to data interpretations, and managed the UBC/Ionis Pharmaceuticals collaboration. L.A.R. assisted with therapeutic efficacy trial design, data interpretation, and manuscript revision. C.F.B. assisted with ASO selection and treatment paradigm, data interpretation, and manuscript revision. M.R.H. assisted with ASO selection and treatment paradigm, therapeutic efficacy trial design, data interpretation, and manuscript revision. Competing interests: H.B.K., M.E.Ø., B.F., E.E.S., P.P.S., and C.F.B. are employees of Ionis Pharmaceuticals. Ionis Pharmaceuticals synthesized the ASOs used in this study, assisted with experimental design, data interpretation, and manuscript revision, and funded work in the laboratory of M.R.H. M.R.H. is an employee of Teva Pharmaceuticals. Teva Pharmaceuticals played no role in this study. All other authors declare that they have no competing interests. Data and materials availability: All the data are included in the main text or in the Supplementary Materials.
View Abstract

Navigate This Article