Using non-coding small RNAs to develop therapies for Huntington's disease

Article metrics

Abstract

Huntington's disease (HD) is caused by an expansion of CAG triplets at the 5′ end of the HD gene, which encodes a pathologically elongated polyglutamine stretch near the N-terminus of huntingtin. HD is an incurable autosomal-dominant neurodegenerative disease characterized by movement disorder, as well as emotional distress and dementia. The newly discovered roles of the non-coding small RNAs in specific degradation or translational suppression of the targeted mRNAs suggest a potential therapeutic approach of post-transcriptional gene silencing that targets the underlying disease etiology rather than the downstream pathological consequences. From pre-clinical trials in different HD animal models to cells from HD patients, small RNA interference has been applied to ‘allele-non-specifically or allele-specifically’ silence the mutant HD transgene or endogenous mutant HD allele. Silencing the mutant HD transgene significantly inhibits neurodegeneration, improves motor control, and extends survival of HD mice. With future improvement of mutant allele selectivity (preserving the expression of the neuroprotective wild-type allele), target specificity, efficacy and safety, as well as optimization of delivery methods, small non-coding RNA-based therapeutic applications will be a promising approach to treat HD.

Introduction

Huntington's disease (HD) is an autosomal dominant, chronic neurodegenerative disease characterized by movement disorders (chorea), emotional distress and dementia. Medium spiny neurons in the striatum are the most vulnerable cells affected in the disease. Patients with HD typically show symptoms beginning at 30–50 years of age. The disease progresses without remission for approximately 10–25 years and invariably results in death. About 30 000 Americans suffer from the disease, and approximately 200 000 others are at risk of inheriting it from an affected parent. In 2008, tetrabenazine (marketed as Xenazine) became the first drug approved by the US Food and Drug Administration to treat HD-associated chorea. To date, no known drug or therapy provides more than symptomatic amelioration.

Symptoms of HD were noted as far back as the sixteenth century. The term ‘chorea’, Latin (and ultimately Greek) for ‘dance’, was used to describe the uncoordinated movement characteristic of the syndrome. Because of the unknown nature of this disease, HD victims were sometimes accused of being possessed by devils, and some were even persecuted as witches during the seventeenth century. Only in 1872 was a detailed clinical description, including the hereditary nature, of HD published by Dr George Huntington in The Medical and Surgical Reporter. More than a century later, a major advance in our understanding of the syndrome came from a genetic study of HD patients in two Venezuelan villages around Lake Maracaibo. These isolated populations were chosen because of an exceptionally high prevalence of the HD syndrome. In 1983, the US–Venezuela Huntington's Disease Collaborative Research Project demonstrated that there is a single causal gene present on chromosome 4. In 1993, this ‘HD gene’ was mapped to position 4p16.3, making it the first autosomal disease locus to be identified by genetic linkage analysis. Mutant and wild-type alleles were cloned and sequenced, revealing the molecular culprit: an expanded run of tandem CAG triplets at the 5′-end of the gene.1 Consequently, the mutant ‘huntingtin’ protein has over 36 tandem glutamine residues near its N-terminus, whereas the wild-type protein has 15 to 35. HD was recognized as one of over 20 genetic diseases caused by an expanded triple-nucleotide repeat. Besides HD, fragile X syndrome, myotonic dystrophy type 1 (DM1), and several types of spinocerebellar ataxias have this etiology.

As it is caused by a single, highly penetrant mutation at a defined locus, HD is potentially amenable to gene therapy using non-coding RNA. Though there are several types of non-coding RNA—for example, transfer RNA, ribosomal RNA, small nucleolar RNAs, microRNAs (miRNAs), short small interference RNAs (siRNAs), and piwi-interacting RNA—what they all have in common is that they do not encode proteins. In the past, some of these molecules were regarded as non-functional RNAs. Only in the past decade have scientists recognized the roles of non-coding small RNAs in the modulation of translational efficiency. This process consists of efficient and specific cleavage, degradation, or translational suppression of targeted mRNAs. Consequently, these non-coding RNAs reduce or even prevent the translation of target RNAs into proteins. Remarkably, RNAi is found in organisms from nematodes to humans, and its molecular mechanism is conserved through evolution. In all these organisms, a double-stranded RNA of 19–23 ribonucleotides directs gene silencing.

A fundamental problem in gene therapy for HD can be addressed using RNAi. It is known that most HD patients are heterozygous at the HD locus, carrying one mutant and one wild-type allele. Although mutant huntingtin protein is toxic to neurons, its wild-type counterpart is protective against apoptotic neuronal death. An effective gene therapy must silence the deleterious mutant allele without eliminating expression of the beneficial wild-type one. Because of the efficiency and specificity with which RNAi silences target genes, it has the potential to be such a ‘silver bullet’. Our hope is that RNAi can silence the disease-causing gene, thereby preventing pathophysiological changes at their source, whereas preserving the expression of the wild-type allele. This strategy is exemplary of a paradigm in Western medicine; it is better to prevent the initiation of a disease-causing pathway than to interfere with downstream processes or simply to remediate symptoms.

RNAi-based technology has been used to silence the mutant HD transgene in animal models for HD. In early experiments, the mutant transgene was silenced without considering endogenous allelic variation in HD patients. In a series of RNAi applications in different HD transgenic rodent models, it was proven that silencing the mutant HD transgene can significantly inhibit neurodegeneration, improving motor control, and in some cases, it extended the lifespan of HD mice (Figure 1).2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 More recently, strategies have been devised to specifically silence the mutant allele, whereas preserving expression of its wild-type counterpart. Because of the lack of a transgenic animal model expressing both alleles of human wild-type and mutant HD gene, allele-specific silencing is conducted in cells from HD patients currently. Allele-specific silencing of the mutant species has been achieved in these cells with both endogenous wild-type and mutant HD (Figure 1).15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26

Figure 1
figure1

Schematic illustration of therapeutic approaches using small non-coding RNA-mediated interference in HD transgenic animals and fibroblast cells from HD patients.

Application of RNAi in HD transgenic animal models

To date, all published studies on non-coding RNAs therapy in HD transgenic mice have used animals expressing short fragments of human huntingtin protein (instead of the full-length huntingtin) with an expanded CAG repeat. These mice exhibit a rapidly progressing syndrome reminiscent of HD in humans. All animals that model HD exhibit neuronal loss, formation of nuclear inclusion bodies in neuron, and, at the behavioral level, declining motor control. The pathology of certain transgenic mice includes selective loss of middle spiny neurons from the striatum, a physiological process characteristic of the syndrome in humans. As neurodegeneration proceeds, particularly quickly in transgenic mice expressing the 5′-fragment of human HD protein (e.g., HD-N171-82Q, CAG140, R6/2, R6/1 and HD190qG), these animals were the first to be used for the proof-of-principle experiments with RNAi-based therapies (Figure 1).2, 3, 4 Most of these studies used RNAi targeted at a human HD transgene. If these RNAi eventually go to clinical trials, this interfering small RNA molecule will silence both the mutant and the wild-type huntingtin alleles. A potential drawback with such ‘non-allele-specific silencing’ is the loss of the essential wild-type protein. In some studies, RNAi was designed to target both the human transgene and the endogenous mouse HD gene to evaluate whether the body can tolerate losing a certain percentage of endogenous wild-type huntingtin at the same time mutant HD is silenced.2, 3, 4

To determine the potential clinical efficacy of RNAi at silencing mutant HD and reducing its toxicity, mice were made to take up short hairpin RNA (shRNA) or synthetic siRNA. There are two strategies for introducing RNAi in vivo (Figure 2). One is to use a viral vector-mediated shRNA and miRNA (or shRNA imbedded in an miRNA scaffold) constructed from lentivirus, adenovirus, or adeno-associated virus adeno-associated virus (AAV)1, AAV2/1, or rAAV5 (Figure 2).2, 3, 4, 6, 7, 8, 10, 11, 12, 13, 14 The second is to use synthetic siRNA introduced with Lipofectamine 2000 (Invitrogen, Carlsad, CA, USA) or cholesterol (Figure 2).5, 9 Synthetic double-stranded RNA and virus-borne shRNA were designed to silence the mutant human HD allele present in transgenic mice. Intrastriatal or intraventricular injections are two methods used to deliver these siRNA/shRNAs into the brain of the mouse. RNAi against the 5′-portion of the transgene (mutant human HD transcript) successfully inhibited expression of the mutant HD gene and slowed neurodegeneration in these animals (Figure 1).

Figure 2
figure2

RNAi biogenesis and therapeutic interventions to silence mutant HD. In mammalian cells, primary miRNA/shRNA (pri-miRNA/pri-shRNA) with large single-stranded flanking sequences is transcribed by RNA polymerase II as hairpins structure in the nucleus. These hairpins are processed by the enzyme drosha into precursor miRNA (pre-miRNA). They are exported to the cytoplasm by exportin-5 and cleaved by dicer. After being separated into two strands, the guide strand of the miRNA or siRNA loaded into the RISC complex and resulted in translation repression or degradation of the targeted mRNA.86 To silence mutant HD, viral vector-mediated miRNAs, or shRNAs or shRNAs embedded in a miRNA scaffold, siRNAs and chemically modified ASOs were introduced into cells. Designed to target at either expanded CAG, SNP sites in HD cells or mutant transgene in HD transgenic mice, these small non-coding RNAs achieved silencing of mutant HD.

AAV1 constructs expressing a U6 promoter-transcribed shRNA were one of the most commonly used delivery method for introducing RNAi to silence mutant transgenes (Figure 2). ShHD2.1 (AAV.sh2.1) targeting nucleotides 416–436 of human HD mRNA (exon 2) was injected bilaterally into the striatum of 4-week-old HD-N171-82Q mice.27 This mouse model contains exon 1 to exon 3 of human HD mRNA and has a lifespan of approximately 5.5 months.28 It displays motor deficits including shortened stride length and impaired performance on an accelerating rotarod, as well as E48-positive neuronal inclusions. Intrastriatal delivery of AAV.sh2.1 reduced intracellular levels of mutant huntingtin protein in mice 1 week after injection, and inhibited mutant HD mRNA levels by 51–55% in 4 months after injection. Thus, it seems that the silencing effect of AAV–shRNA is long lasting. Histology revealed a decreased number of inclusion bodies after treatment. Most important, host mice demonstrated improved motor endurance on the rotarod both 6 weeks and 14 weeks after the injection, and had longer stride length than HD-N171-82Q animals injected with control AAV.shLacZ.27 Although there was no extension of lifespan and no weight normalization in AAV.sh2.1-treated mice, this is the one of the first proof-of-principle animal trials demonstrating that shRNA in vivo application can improve the behavioral performance of HD transgenic mice. The R6/1 was generated by introducing exon 1 of the human HD gene with an expanded 115–150 CAG repeats, causing rapid neurological decline with death at the age of 4–5 months.29 AAV-hiHUNT-1 (an shRNA whose target was within the 5′-untranslated region of human mutant huntingtin) was injected into the striatum of R6/1 mice 6–8 weeks old. Mutant HD mRNA and protein were reduced in the transduced area compared with contralateral injection of rAAV–shRNA control vector.10 Decreased size and number of neuronal inclusion bodies and normalization of preproenkephalin and DARRP-32 mRNA level were found in R6/1 brain unilaterally treated with rAAV5–shRNA. The onset of rear paw clasping was delayed for 2 weeks after bilateral injection of rAAV5–shRNA.10 The HD190QG transgenic mouse expresses N-terminal huntingtin with 190 CAG repeats fused with enhanced green fluorescent protein (EGFP).30 This animal also showed progressive motor abnormality and neuropathology in the striatum. rAAV–shRNA targeting the EGFP (rAAV5–shEGFP) instead of HD was injected into the striatum of this transgenic mouse at 12 weeks of age.12 rAAV5–shEGFP reduced aggregate formation, suppressed insoluble protein accumulation and restored DARPP-32 expression in the striatum at 24 weeks.12

In the above three HD transgenic mice, the loss of neurons is not restricted to the striatum; it occurs throughout the brain. Intrastriatal injection of virus-borne mutant HD caused closely confined neuronal death in the striatum with a distribution similar to human HD.31, 32 The resulting HD models had rapid onset with progressive robust neuropathology and motor impairment, which facilitate experimental testing.6, 9, 13, 14 More important, these methods could be developed into pre-clinical studies in non-human primates.20 Injection of AAV1/2-HD-70 into the striatum resulted in substantial loss of projection neurons, cholinergic interneurons and NAPDH-d interneurons.6 In that model, the level of mRNA for mutant huntingtin increased 150-fold within 2 weeks, but dropped to near endogenous levels by 5 weeks. AAV-shHD2 targeting human huntingtin mRNA was injected 2 weeks after the first injection of AAV1/2-HD-70. ShHD2 demonstrated substantial neuroprotection with increased numbers of NeuN and calbindin-positive cells, and decreased numbers of Fluoro jade B-positive cells. In addition, AAV1/2-shHD2 injection reduced impairment of spontaneous exploratory forepaw in this HD model.6 Another rat model of HD was created by giving an intrastriatal injection of a lentivirus vector expressing HD-171aa-82Q.6, 13 Many of the resulting molecular changes were reversed by inducing the expression of lentivirus shRNA targeting the mutant transgene, including preservation of DARPP-32 and NeuN-positive cells, decreasing EM48-/ubiquitin-positive inclusions, and recovering succinate dehydrogenase function and the rate of cerebral glucose metabolism. In addition, high-capacity adenoviral vector (HC-AdHB07) was injected into the striatum to generate an acute HD model expressing HD-171aa-128Q.14 Huntingtin aggregates in neuronal cells were noted in the vicinity of injection sites. HC-AdHB04 (expressing shRNA target mutant HD) or HC-AdHB05 (expressing anti-EGFP shRNA) was co-injected with HC-AdHB07. Mutant huntingtin aggregates were efficiently reduced by co-injection with HC-AdHB04, but not with HC-AdHB05. A similar inhibitory effect on abundance of huntingtin immunoreactive aggregates was found in brain slices from R6/2 animals treated with HC-AdHB04.14

Some studies have addressed whether post-symptomatic injection of RNAi was also beneficial to HD. Doxycycline-regulated shRNA in a lentiviral vector (lentivirus-sihtt1.1) were delivered into the striatum of an HD acute rat model generated by lentivirus-carried HD-N171-82Q.13 Two months after lentivirus-HD-N171-82Q delivery, when pathology began to appear, doxycycline was administered, turning on sihtt1.1 expression. Inducing shRNA synthesis rescued these rats from depletion of DARPP-32-positive cells and partially cleared EM48 or ubiquitin-positive inclusions.13 In the AAV1/2-HD70 model, the substantial loss of neural immunoreactivity was evident 2 weeks after intrastriatal injection of AAV1/2-HD70. However, AAV-shHD2 injected at that time prevented HD-70-induced neurodegeneration and behavioral impairment.6 It was suggested that shRNA-based therapy could reverse HD pathology, even when disease symptoms become obvious.6, 13 Introducing the transgene by viral injection resulted in dramatic pathological changes in the striatum, but these effects were not confined to the medium spiny neurons. In some cases, it affected neuron types that are largely uninvolved in HD. Injection of AAV1/2-HD70 into rat striatum resulted in quite a rapid expression of HD70 in large interneurons and weaker immunoreactivity in medium-sized neurons.6 AAV-HD70 caused significant striatal interneuron death, including neuropeptide-Y, parvalbumin and choline acetyltransferase immunoreactive interneurons.6 Whether this lack of neuron type specificity will compromise the potential of this technique for testing an RNAi-based therapy must be further investigated in animal models, especially non-human primate models.

Although virus-delivered shRNA demonstrated beneficial effects of inhibiting neurodegeneration and improving motor function, cytotoxicity was also associated with AAV–shRNA injection. AAV.sh2.1 injection impaired rotarod performance of wild-type mice at 10 weeks, although this effect resolved by 18 weeks.27 Later studies also found that AAV–shRNA causes sequence-specific neurotoxicity to the mouse striatum and results in short-term deficits in motor behavior.7, 8 AAV–miRNA instead of AAV–shRNA was a successful alternative strategy to infect HD transgenic mice with virus carrying an shRNA vector.7, 8 AAV-mi2.4 is an artificial miRNA-based cassette that can generate shRNA, targeting the same sequence within exon 2 of both introduced and endogenous HD transcripts.7, 8 Due to its reduced production of antisense RNA from AAV-mir2.4, it generated less toxicity, but achieved effects similar to those of AAV-sh2.4, to silence mutant HD and extend the lifespan of HD-171-82Q mice.7, 8 As in HD-171-82Q mice, AAV-mir2.4 injected into CAG140 heterozygous knock-in mice produced a similar increase in DARPP-32 immunoreactive cells, indicating lower cytotoxicity in medium spiny neurons from AAV-mir2.4 treatment. Although there is controversy regarding the immunoreaction as judged by real-time analysis of CD11b mRNA in these two transgenic mice, AAV-mi2.4 did not elicit the same extent of microglial activation as AAV-sh2.4, according to immunohistochemical staining of Iba1-positive cells.7, 8 This artificial AAV–miRNA-based RNAi delivery system might provide a new tool to efficiently silence the mutant HD gene and improve the safety of AAV–shRNA.7, 8

Although application of synthetic siRNA causes concerns regarding its quick degradation in vivo, one study injected siRNA into the lateral ventricle of R6/2 mice brain and found unexpectedly long-lived effects.5 The R6/2 transgenic mouse is one of the most thoroughly studied animal models for HD. It was generated by introducing exon 1 of the human HD gene with 150 expanded CAG repeats into mice, and causes rapid neurological decline with death at age of 3 months.29 siRNA-HDexon1 was designed to target human HD exon 1 and was injected intraventricularly with lipofectamine into the brains of mice 2 days after birth. Inhibition of mutant HD mRNA was found 4 to 7 days later. Reduction of mutant huntingtin aggregation in striatum and increased motor control were found in mice 8 weeks after treatment with siRNA-HDexon1, compared with mock-treated or untreated R6/2 mice brains of the same age.5 siRNA-HDexon1 injection also delayed the onset of the feet-clasping response, and the body weight of the animals remained relatively stable up to the age of 8 or 9 weeks when compared with untreated R6/2 mice. At 12 weeks, an age when symptoms were already advanced, siRNA-HDexon1-treated animals exhibited more activity in the open-field test than untreated controls. In addition, synthetic siRNA-HDexon1 extended the transgenic lifespan of the animals by 15 days.5 The long-lasting effect of a single injection of siRNA might indicate that the molecule's ability to cause physiological changes goes beyond simple silencing of gene expression. It seems that the early loss of mutant HD expression has long-lasting effects that persist after protein synthesis resumes. Another type of siRNA application was conducted in an acute HD model generated by injection of AAV vector expressing 365 N-terminal amino acids from the human HD with 100 CAG repeats (AAV-htt100Q).9 These animals suffered a debilitating rapid-onset syndrome in which striatal neurodegeneration and behavioral impairment were evident within 2 weeks. Cholesterol-conjugated synthetic siRNA against this HD transgene prolonged survival, ameliorated deficits in neuron function and reduced aggregates and inclusion bodies in mice injected with AAV-htt100Q.9

Researchers have also tested whether silencing wild-type huntingtin has a detrimental effect on disease progression when mutant HD is silenced at the same time. As described above, intrastriatal injection of lentivirus carrying the harmful htt171-82Q caused test rats to lose DARPP-32 and NeuN-positive cells, and to accumulate EM-48 and ubiquitin-positive inclusions.13 Lentivirus-mediated shRNAs (sihtt 3 and 6), which target mRNA of both mutant human HD and endogenous mouse HD, reduced the number of EM48/ubiquitin-positive inclusion bodies and preserved more DARPP-32 and NeuN-positive cells.13 Lentivirus-mediated sihtt 13, which targets only the mRNA of endogenous mouse HD, has no detrimental effects on GABAergic neuron survival or huntingtin inclusions for up to 2 months.13 Moreover, lentivirus-mediated sihtt 6 (silencing both endogenous mouse HD and mutant human HD) has long-term effects similar to those of sihtt1 (silencing only mutant human HD) in reducing HD pathology. sihtt 6 did not cause striatal toxicity or striatal vulnerability up to 9 months after injection. DARPP-32-positive cells are well preserved, and ubiquitin-positive inclusion is absent at 1, 3 and 9 months in the striatum of sihtt6-treated htt171-82Q rats. Similar non-allele-specific silencing effects were obtained in experiments on HD transgenic mice expressing human HD-N171-82Q.8 In these investigations, test mice were injected with an artificial miRNA that targets a sequence common to both the foreign and endogenous HD gene.7, 8 At 7 weeks of age, animals treated with AAV-mi2.4 performed better on the rotarod and survived longer than control-treated HD-N171-82Q mice. The endogenous mouse HD gene and the mutant human HD gene responded to the AAV-mir.2.4 silencing effort similarly. After bilateral intrastriatal injection with AAV-mi2.4, transcripts of both HD genes were 60% (4 weeks) and 75% (13 weeks) lower in injected mice than in control animals.8 When assessed at 10, 14 and 18 weeks of age, the rotarod performance of the test animals had not declined as much as that of control-treated HD-N171-82Q mice at 14 and 18 weeks.8 In addition, AAV-mi2.4-treated mice displayed a trend of improved survival. At 20 weeks, although 45% of control-treated HD mice survived, over 75% of AAV-mi2.4-treated mice were still alive. The results of both these experiments demonstrate that partial inactivation of wild-type huntingtin might be tolerated, whereas concomitant silencing of its mutant counterpart is beneficial.8, 13 As these studies are based on silencing the endogenous wild-type HD gene and mutant transgene, the silencing effect might be different when both HD genes are endogenous. In the previous case,13 the introduced transgene is expressed at a level 25-fold higher than in the endogenous species.13 In human HD, both mutant and wild-type HD alleles are equally expressed. Moreover, major depletion of wild-type huntingtin could still interfere with neuronal function, especially under stress conditions or simply over the long term. We can expect to see future experiments designed to address how effectively and how long non-allele-specific targeting sufficiently silences mutant HD gene effects without causing detrimental toxicity from partially losing wild-type HD, and to what extent this loss can be tolerated.

The potential off-target effects of AAV-mediated HD silencing are illustrated by the different responses of R6/1 mice to unilateral intrastriatal injection with rAAV-siHUNT-1 and rAAV-siHUNT-2. rAAV-siHUNT-1 did not strongly silence transcripts besides those of HD. By contrast, rAA-siHUNT-2 caused depletion of the mRNAs encoding the striatum-specific proteins preproenkephalin and DARPP-32.10, 11 This negative impact on transcripts in R6/1 mice was also reported with rAAV-HD6 and HD7 (two catalytically active hammerhead ribozymes) against the same region of human HD mRNA, also targeted by siHUNT-2.11 The mechanism by which same off-target effects from these different modalities are currently unclear.11 In contrast, AAV-shHD2, which targets a region of the human HD mRNA, similar to that targeted by siHUNT-2, was not found to cause off-target reduction of proenkephalin in the rat HD model by AAV-HD70 injection.6 The reason might come from the responses of different strains of mice to the same interfering RNA. Off-target effects and RNAi-induced toxicity that are problematic in some animals may be unimportant in others.5, 6, 7, 8, 10, 11

All these findings should remind us that in the design of future clinical trials, there might be differences in off-target effects among individual patients. In anticipation of this complexity, preclinical studies on RNAi therapy should be conducted in HD mice with various genetic backgrounds. In addition, HD is a chronic neurodegenerative disease; the transgenic animal model with the short HD fragment, with its quick disease onset and progression, resembles some features of this disease, but not the slow disease progression of HD. A mouse model that carries a full-length HD gene (e.g., YAC128 or BACHD) should also be tested.33, 34, 35, 36 These mice exhibit more gradual disease progression and loss of specific neuron type that is reminiscent of the chronic symptoms and pathology we see in human HD patients. However, these full-length HD mice also have the drawbacks of not losing weight, and because of their blindness in the late stages of disease, it is difficult to evaluate their behavioral performance. Consequently, performing studies on both types of HD transgenic mice is a reasonable approach to evaluate the potential of this RNAi-based therapy. It will determine whether virus-borne shRNA or synthetic siRNA with single or multiple injections can elicit effects in the transgenic mouse model expressing full-length HD similar to those we have seen in rapid disease transgenic models using HD fragments.

Allele-specific silencing of mutant huntingtin

Because of the heterozygotic nature of the HD gene in the majority of patients, the wild-type huntingtin protein is essential to cell survival and has a protective role against cell death induced by mutant HD or/and stress; the mutant HD is harmful, with greatest toxicity to striatal neurons; the ideal strategy for gene therapy would silence the mutant allele and preserve wild-type huntingtin. The technology of RNA interference might be a promising tool, as it offers the requisite efficiency and specificity to satisfy both imperatives. siRNAs can be designed to discriminate between two alleles that differ at a single nucleotide, that is, a single-nucleotide polymorphism (SNP).37 There are two sites in the mutant allele to which RNAi can target: mutant HD allele genetically linked SNP sites or expanded CAG repeats.

Researches of the human genome database and large number of patient samples have revealed many polymorphic sites within the HD gene. Some of these SNPs occur at a high frequency among HD groups of substantial size. There are totally 190 SNPs known in the introns and exons at the HD locus at chromosome 4p16.3. Analysis of HD alleles from 65 patients of European origin revealed that disease-associated SNPs form a cluster of similar haplotypes (haplogroup A) found on 95% of CAG-expanded chromosomes. Two variants of haplogroup A (A1 and A2) are dramatically and specifically enriched on HD chromosomes, and are therefore markers for persons at risk of the disease.38 In a study of 327 European Caucasian HD patients, 86% were heterozygous at one or more of the 26 SNPs analyzed.21 In these cases, allele-specific RNAi is potentially applicable should the siRNA target polymorphic sites specifically linked to CAG expansion. However, some important questions remain about targeting mutant HD linked SNP to approach allele-specific silencing: (i) is it possible to find one or a few SNPs that are common and have a sufficiently strong linkage to the CAG expansion to be generally useful for a relatively larger HD population and can be tested in large-scale clinical trials to obtain permits from drug regulatory agencies? Alternatively, is allele-specific RNAi therapy only for personal medication? (ii) On an individual basis, can one rapidly identify SNPs with sufficiently tight linkage to CAG expansion? (iii) Can one design an siRNA that silences the mutant allele enough to render it innocuous, but preserves sufficient expression of the normal species to execute essential functions? (iv) What does the future hold for generation of new transgenic HD animal models for pre-clinical trials of allele-specific silencing? Although there is a long road ahead, researchers have already made a significant start. The most promising results come from several studies on fibroblasts from HD patients. In cultures of these cells, RNAis have successfully silenced the mutant HD allele, whereas preserving expression of its wild-type counterpart (Figure 1).

A noteworthy polymorphism that is linked to CAG expansion is the Δ2642 triplet deletion in exon 58 of the HD gene.39, 40, 41, 42, 43 The typical sequence beginning at this position is 5′-GAG.GAG.GAG.GAG-3′; that for Δ2642 is 5′-GAG.GAG.GAG-3′. The deletion of a codon causes the loss of one of the four tandem glutamate residues in the huntingtin protein. The three-glutamate species occurs substantially more frequently among HD alleles (38%) than in those without CAG expansion (7%).39 One skin fibroblast from an HD patient was identified that carried both an HD allele marked by the Δ2642 deletion and its wild-type counterpart. These cells were transfected with each of four siRNAs designed to target the polymorphic site. One of these molecules efficiently and specifically silenced the Δ2642-marked, CAG-expanded allele in this HD fibroblast,17 and similar silencing results were obtained from a later study of this HD fibroblast.18

Only a minority of mutant HD alleles is marked by the Δ2642 deletion mutation. Ongoing research aims to identify additional SNP sites in the HD gene that could be used for siRNA silencing directed specifically to alleles with the CAG expansion.16, 19 As the HD locus is large and both isoforms are over 10 kb, SNPs in that gene are usually distant from the CAG repeat at the 5′-terminal. For this reason, it is difficult to identify SNPs that are linked to the harmful triplet expansion in mutated HD alleles. Two methods have been used to identify allele-specific SNPs. One is to use SNP-specific reverse transcription primers to selectively generate cDNA from a single allele of HD. A second round of PCR is conducted on the resulting cDNA, using primers spanning the CAG repeat sequence.16 Using this method, 11 known SNP heterozygote sites in fibroblast cells from 21 different HD patients were allele-specifically determined.16 For example, fibroblasts from one patient had the rs363125 SNP with cytosine at that position in the mutant HD allele, but adenine in its wild-type counterpart (Figure 1). On the basis of this information, siRNA guide strand targeting at mutant allele with rs363125 SNP can silence the mutant HD while preserving expression of the wild-type one.16 Another method used the circularization method to link the SNP with CAG repeats.19 A primer flanking the SNP with a KasI site at each end was used to amplify DNA ranging from SNP to CAG expansion. The key was to circularize this PCR product by intramolecular KasI ligation. A PCR spanning SNP and CAG repeat was then conducted on two wild-type and mutant individual ligation products. The length of CAG repeats was evaluated on the PCR products by sequencing. As the allele-specific SNP site and CAG repeats are adjacent, it was easy for direct sequencing to identify the SNP-(CAG)n linkage.19 This method was used in later studies with large sample sizes.20 Evaluation of 225 human HD and control samples from American and European carriers revealed that over 48% of patients were heterozygous at 1 of 24 identified SNP sites. Incidence of the U isoform of rs362307 SNP at exon 67 on HD alleles greatly exceeded that on control samples.20 Seven out of sixteen HD blood samples evaluated have an expanded CAG-linked heterozygous U isoform at the rs362307 site, an approximate 50% linkage. Moreover, disease-associated SNPs at sites rs363125, rs362273 and rs362307 are so frequent within this sample that five siRNAs that specifically silence the harmful allele have the potential to treat three-quarters of these HD patients.20 In addition, a computational and experimental analysis was conducted in 327 unrelated European Caucasian HD patients at 26 SNP sites in the HD gene. It predicted that over 85% of HD patients could benefit from shRNA-based therapies that target six to seven SNP sites (Figure 1).21

Researchers are also trying to adapt RNAi technology to target the expanded CAG repeat that makes the mutant HD allele harmful. As discussed earlier, standard techniques using shRNA and siRNA are not applicable to a sequence as long and regular as (CAG)30. The number of tandem CAG triplets at the 5′-end of the HD, the critical difference between wild-type and mutant HD genes, is over 36 in deleterious alleles, but generally fewer than 30 in healthy ones. As RNAi molecules with only 19–23 bases act efficiently, but do not elicit a strong immune response, it seems too short to span the CAG repeat and distinguish between mutant and wild-type alleles. However, longer CAG trinucleotide repeat sequences have been predicted to form a secondary hairpin structure (Figure 3).44, 45, 46, 47, 48, 49, 50 The stability of this hairpin conformation might differ between the wild-type and mutant alleles. The short stem structure formed by the interaction between short CAG repeats and CCG repeats in wild-type allele is relatively more stable and resistance to RNA-induced silencing complex (RISC)-medicated cleavage. The long-hairpin stem structure composed of pure expanded CAG repeats was not stable enough to form a barrier for RISC activity. Therefore, siRNA targeting at CAG can selectively silence mutant HD allele, based on the decreased stability and increased susceptibility of CAG repeat hairpins to interact with RISC.22 Pursuing this reasoning, single-strand antisense oligoribonucleotides (ASOs), or two strands of siRNA (with or without unmatched sites), were designed to silence HD alleles with unusually long CAG repeats.18, 22, 23, 24, 25, 26 ASOs contain a single nucleic acid strand that were heavily modified to enhance stability, target binding and biodistribution.18, 23, 24 Locked nucleic acids-modified ASOs with chemical conjugation with peptide (peptide nucleic acid) were initially reported to have up to six-fold allele selectivity to potently silence the mutant, but not the wild-type HD allele in fibroblasts from HD patients.18 However, after comparing with 10 chemical modification methods, even the most efficient chemically modified ASO had limited potency and were unable to reach greater than eight-fold allele selectivity.24 Recently, it was found that using mismatch-containing duplex siRNAs, robust selectivity (>10- to 20-fold) and potencies for inhibiting mutant huntingtin expression (<10 nM) in two HD skin fibroblast cell lines were achieved.25, 26 Adapting endogenous miRNA-like targeting with mismatches design in siRNA guide strand produced a better selectivity. Inhibition of mutant HD allele by siRNAs with position 9 mutation in guide strand can reach high potency and 30-fold greater selectivity.25 In another study, siRNA guide strand with position 13 and 16 mismatch increased silencing discrimination between the mutant and wild-type HD alleles.26 MiRNA-like duplexes designed to target the CAG repeats might be an alternative candidate for achieving allele selectivity (Figures 1, 2 and 3).25, 26

Figure 3
figure3

Schematic image of inhibition of mutant huntingtin expression by targeting at expanded CAG repeats. Among the three sections of HD mRNA secondary structure (base, central and terminal sections), the terminal section formed by the fold-back hairpin structure of pure CAG repeats is the most variable and flexible part.22 Compared with wild-type allele, mutant HD allele with long CAG repeats not only offers more complementary targeting sites, but also is more susceptible to ASOs or siRNA targeting. ASOs with chemical modifications or siRNA with central position mutation can selectively inhibit mutant HD translation.18, 23, 24, 25, 26

Original double-strand siRNA with sequence-exact complementary matching to a target is processed via the RISC and results in robust mRNA cleavage without selectivity. The mechanism of selective silencing of chemically modified ASOs and miRNA-like siRNA might be different from traditional siRNA processing. Both ASOs and siRNA with central-site mismatch inhibited HD protein without any significant effect on HD mRNA,24, 25, 26 which indicates translational repression rather than transcriptional degradation (Figure 3). For translational inhibition, it is best to have a target near the start codon of an mRNA.51, 52 As mutant CAG expansion is only 51 nucleotides away from the translational start site, mutant HD is a good candidate for translational repression by ASO or siRNA with central mismatch.24, 25, 26

Although from previous observations in non-allele-specific silencing in mice, it seems that partial knockdown of wild-type HD for a period could be tolerated in animals,13 wild-type huntingtin is essential for the survival and function of neurons. There is currently no transgenic mouse in which to test the feasibility of SNP-based allele-specific RNAi silencing as a therapy for HD; however, RNAi targeting at expanded CAG repeats will be practicable using current available transgenic HD models to obtain proof of principle with an in vivo system. Mouse endogenous huntingtin expression will be evaluated for its selectivity. For SNP-based RNAi therapy, it is difficult to construct the necessary transgenic animal to test the on-target and off-target effects of allele-specific silencing in. Transgenic mice can express both alleles linked with different CAG length. However, this human HD wild-type allele transgene will not function like endogenous wild-type mouse huntingtin. It will work only when the polymorphism site on the mutant allele targeted by the RNAi has the same isoform sequence in the human wild-type allele as it has in mice HD. In this case, if off-target effects are seen, not only will that be revealed by the wild-type allele construct, but there will also be a functional readout from mouse endogenous HD. Targeting mutant HD alleles linked to expanded CAG or SNPs are two promising strategies to selectively silence mutant, not the wild-type HD. However, there are more than 200 mRNA containing CAG repeats, and CAG expansion length or SNPs associated with HD alleles differ from patient to patient. Consequently, the efficiency of CAG-based allele-specific silencing may vary among different patients; SNP-based allele-specific RNAi therapies may be more useful in personalized medicine than as general remedies. Unlike the population limitation of SNP-based allele-specific silencing, CAG expansion-based allele-specific silencing might be developed into universal therapeutic approach that is applicable to HD and other polyQ diseases. However, experimental approaches exploring the allele-specific silencing efficiency and therapeutic potential of these methods need to be well designed and evaluated in vivo.

Challenges and perspectives

Target and off-target effects

The greatest challenge to the advancement of an RNAi-based therapy is efficient silencing of the harmful HD allele with an interfering RNA that is not itself toxic to model animals. The foremost difficulty when designing a safe and effective RNAi-based therapy for HD arises from off-target effects of the RNAi. These constraints are particularly stringent for allele-specific silencing, where on- and off-target alleles differ by only one to three nucleotides in SNPs, or there is only the length difference of CAG repeats in two alleles.

As is to be expected, the mRNAs responsible for off-target effects have sequences resembling those of the target species. For the off-target effect, there can be partial gene silencing of mRNAs that match the introduced siRNA at as few as 11 nucleotides.53 Bioinformatic analysis reveals that off-target effects most often arise from binding of the siRNA with a seed region matching with the sequence in off-targeted mRNA's 3′-untranslated region.53, 54, 55, 56, 57 Within the molecule's total length of 19–23 nucleotide pairs, positions 2 through 8 at the 5′-end of the guide strands of RNAi are of exceptional importance to the recognition of and binding to target mRNAs.58, 59 However, the rules of mRNA on-target/off-target selection are too complex to be fully determined by the 16 348 possible combinations for nucleotides in positions 2–8.60, 61, 62 Optimized sequence design of guide strand or passenger strand will result in a stronger preference for the guide strand over the passenger strand in determining which mRNAs are silenced by the RISC machinery.60, 61, 62 To predict the potential off-target effects, a ‘BLAST search’ of the gene database enables the investigator to identify additional mRNAs with close sequence homology to the target mRNA. However, mRNAs that are perfectly complementary to the seed region of the siRNA are not always silenced by that interfering molecule.57 The current approach uses microarrays to determine the global pattern of mRNA expression, with and without changes affected by RNAi. Moreover, it is difficult to determine whether a gene is silenced by an siRNA–mRNA interaction or simply by non-specific changes due to systemic perturbations. In addition, some confounding phenomena arise when an siRNA mimics the function of a miRNA. MiRNAs are themselves endogenous regulatory elements that inhibit translation of target messages, but do not necessarily degrade them. In that situation, the cell's mRNA profile remains the same, whereas the spectrum of expressed proteins changes. Analysis by proteomics will reveal effects due to altered translation.

The off-target effects of RNAi were found in HD mice treated with one of shRNAs that targeted at mutant transgene. This shRNA injection caused not only the reduction of mutant HD mRNA, but also a depletion of the mRNAs encoding the striatum-specific proteins preproenkephalin and DARPP-32.10, 11 Similar off-targets effects on transcripts in HD mice were also reported with rAAV-HD6 and HD7 against the same region of human mutant HD mRNA.11 Moreover, the nature of off-target effects varies among both cell lines, rodent models and strains.6, 7, 8 ShRNA with this same target sequence did not cause off-target reduction of proenkephalin in the rat HD model by AAV-HD70 injection.6 Another example of this confounding variable is the different changes in CD11b regulation. Increasing of CD11 indicates the activation of microglial. When AAV1-mi2.4 was injected into B6C3F1/J and C57BL/6 mice, the B6C3F1/J mouse exhibited increased levels CD11b, but C57BL/6 animals were not affected by introducing that very RNAi species in a miRNA cassette.7, 8 This example underlines the complexity expected when developing RNAi therapies for HD. The clinical responses to RNAi-mediated gene silencing might vary among patients. The extent and nature of off-target effects will largely determine who tolerates the therapy and who does not. In anticipation of such variability, preclinical studies should be conducted using additional disease models.

To selectively silence mutant HD allele and not the wild type one, siRNA is designed to have mismatch at central sites of the guide strand. Unlike the mechanism that targeted mRNA that will be cleaved by argonaute 2 protein when duplex RNA is fully complementary to its mRNA target, the mismatches between the duplex RNA and mRNA target disrupt interactions at the catalytic center of argonaute 2 and make cleavage less efficient.25 The mechanism the mismatch siRNA will use is more like one involved in inhibition of ribosome-mediated translation by endogenous miRNAs (Figure 3). In HD mRNA secondary structure, it contains a base section composed of CAG interacting with CCG repeats, a central module formed by the interacting CAG repeats and the specific sequence, and a terminal section formed by the fold-back structure composed of pure CAG repeats.22 Because of the A:C mismatches instead of A:A mismatch, the stability of the base section of the hairpin structure is much higher than other hairpin parts. The most variable section of the HD mRNA hairpin is the terminal section of the hairpin structure containing pure CAG repeats (Figure 3). Compared with wild-type allele, mutant HD allele with long and flexible CAG repeats are more susceptible to siRNA targeting, especially the siRNA having a central mismatch at the guide strand. The expanded CAG repeat also offers multiple target sequences for fully complementary siRNAs or mismatch-containing siRNAs to target (Figure 3).25, 26 These multiple targeting can yield synergistic increases in potency for ASOs or siRNAs.25, 26, 63 The miRNA-like siRNA with mismatch design complementary to the CAG repeat seems a better candidate for achieving allele-specific selectivity in silencing mutant HD.25, 26

shRNA versus siRNA

Another major hurdle to developing an RNAi therapy is constructing a system that efficiently delivers these small RNAs. Currently, there are two major strategies with RNAi for HD: (i) viral vectors with DNA encoding an shRNA and (ii) synthetic siRNA or antisense RNA (Figure 2). Each approach has distinct advantages and disadvantages. The viral vector system has long-term effectiveness following a single injection of DNA. Because of its relatively low immunogenicity, AAV is the most widely used vector for making RNAi constructs for HD. One advantage is its ability to exist in a stable episomal form, resulting in lasting gene expression. After a single injection, shRNA expressed from an AAV vector was shown to cause long-term silencing of a mutant huntingtin transgene.8, 10, 12, 14, 27 The low immunogenicity of AAV as compared with lentivirus and its small size also make it an attractive vector for investigations involving RNAi. In addition, transduction selectivity in different neuron populations from different tropism of AAV serotypes will improve AAV transduction specificity.64, 65, 66, 67 The lentiviral vector proved to be a good choice for inducible RNAi expression.13 It produces long-lasting gene silencing, presumably a result of their effective in non-dividing neurons and integration into the host genome. This phenomenon is a mixed blessing, however. As integration of the foreign DNA into the host's chromosomes may cause harmful insertion mutations, it also poses a safety issue for future clinical trials. The major shortcoming of shRNA is its potential to interfere in the endogenous small RNA processing machinery. Before an shRNA can effect physiological changes, it must be processed by enzymes for endogenous miRNA biogenesis, for example, nuclear exportin-5 (Figure 2). As injected AAV–shRNA competes with and overloads the enzymes that process miRNA, it disrupts this essential molecular pathway. In one study on mice, injected shRNA depressed levels of mature liver-specific miRNAs so much as to kill the host animal.68 Significant buildup of unprocessed shRNA was detected, suggesting that the processing machinery was largely saturated.68 Reducing the dosage of this shRNA vector diminished toxicity to the host. In one study, HD mice were given intrastriatal injections with AAV–shRNA that target HD. There was significant neural cell death in the striatum of the animals both from an shRNA that targets the endogenous Hdh mRNA and from one targeting the mutant human HD transgene. Death of striatal neurons was also observed when the shRNA was designed to have a few mismatches relative to both alleles.7, 8 Moreover, this AAV–shRNA remained toxic to striatal cells, even as its dosage was reduced. Unlike the situation for liver tissue, unprocessed shRNA molecules did not accumulate to high levels in the striatum of these test animals. The character and vulnerability of brain cells in these studies was in contrasts with those qualities of liver cells in the other.7, 8, 68 A reasonable explanation is that the affected miRNA-processing machinery differs between liver cells and striatal cells.7, 8, 68 To reduce the toxicity of shRNA, the artificial miRNA driven by RNA polymerase II promoter was designed to silence the same HD target.7 Upon intrastriatal injection, this artificial miRNA effectively silenced target messages with fewer toxic effects than shRNA.7 Although the silencing effect was similar, artificial miRNA was relatively well tolerated by striatal cells, evidently because the harmful antisense species is less abundant.7

In summary, HD therapy must use AAV–shRNA or AAV–miRNA at a dosage that silences target genes, but does not overwhelm the ability of the cell to process endogenous small RNAs. Complicating this task is the difficulty of controlling expression levels of virus-borne shRNA in vivo. A compliant system must allow for fine adjustment of shRNA expression. The effectiveness of an inducible shRNA vector for HD has been demonstrated in vivo using lentivirus.13 Transcription from a doxycycline (tetracycline)-responsive promoter may lend itself to reversible repression, though such sensitivity and reversibility in processed RNAi have yet to undergo sophisticated evaluation in vivo.13

The other important strategy for introducing RNAi into the host cell or animal is to use synthetic siRNA (Figure 2). This technique has a major disadvantage when compared with virus-based methods; unlike virus constructs, synthetic siRNA is easily degraded in vivo. Nevertheless, siRNA has several properties that recommend it. Simple nucleic acids are less immunogenic than viral complexes. Although this property is not relevant to in vitro systems, it removes a major barrier to application in vivo (e.g., the ultimate goal—a clinical therapy). At the molecular level, siRNA interferes less with endogenous pathways of miRNA biogenesis (Figure 2). Chemical modification of siRNAs adds a new dimension to optimizing their utility. Their large size (molecular weight of 13 000 or more), strongly negative charge and hydrophilic nature make siRNAs different from other small-molecule pharmaceuticals. Various covalent modifications to siRNAs increase their stability, target specificity and the efficiency with which they silence the target mRNA. Moreover, these chemical derivatives of oligoribonucleotides produce less severe off-target effects and are less toxic to the host cells. As noted earlier, siRNA is easily degraded by nucleases in serum. Chemical changes to the sugars, backbone, or bases of nucleic acid make them more like conventional drugs with significantly increased stability, in-vivo potency and less immunostimulation.69, 70, 71, 72, 73, 74, 75 After comparison of 10 types of chemical modification—including peptide nucleic acid, locked nucleic acids, (S)-cET bridged-nucleic acid, carba-locked nucleic acid, ethylene nucleic acid, altritol nucleic acid, 2′-O-methoxyethyl, 2′-fluoro and 2′-fluoroarabino nucleic acid oligomers—for allele-selective inhibition, bridged-nucleic acids (includes locked nucleic acid) were highly successful and were considered a lead modification of oligomer for allele-specific silencing mutant HD.24 To apply siRNA or other synthetic nucleic acid to inhibit mutant HD, the most prudent and robust strategy is to synthesize and screen a substantial library of siRNA duplexes (or oligomers) for multiple sites in mutant HD mRNA to identify the most promising one. Further chemical modification of it will increase its potency and selectivity. Finally, the US Food and Drug Administration has no established code to regulate potential RNAi-based biological products. Technologies using synthetic siRNA are likely to be classified as ‘antisense therapies’. Virus-borne shRNA is apt to be classified as ‘gene therapy’. Because of several unfortunate events in past years, the latter branch of experimental medicine is strictly regulated. For this reason, siRNA-based therapies are more likely to obtain US Food and Drug Administration's approval.

Delivery

Besides off-target effects, another big challenge in the development of RNAi-based therapy for HD concerns delivery. In previous in vivo studies, viral and non-viral application and local intrastriatal or intraventricular administrations were successfully used to silence HD in vivo. Although virus-carried shRNA showed high efficiency in transducing cells in the striatum, coupling siRNA with lipids or cholesterol are other strategies for introducing HD RNAi into a living organism. In one study, siRNA was coupled to liposomes (Lipofectamine 2000) to form amorphous lipoplex particles. Intraventricular injection of Lipofectamine 2000–siRNA into postnatal day-2 mice countered both the generation of striatal nuclear inclusion and brain atrophy. The molecular and anatomical changes were correlated with a reduction in abnormal behavior and increased lifespan.5 As this cationic lipid-based transfection reagent is toxic to neurons, novel delivery strategies are developed. Another experimental system for siRNA-based therapy has been explored in mice transduced with the AAV–htt 100Q viral construct.9 Intrastriatal injection with a cholesterol-coupled HD siRNA silenced the mutant transgene for 3 days and relieved neuropathological symptoms for 1 week.9 In addition, lipophilic polypeptides such as stearylated octaarginine (steary1-R8), MPG-based particles and lipid-based artificial virus-like particles are all peptide-based particles with potential applications in siRNA therapy. They achieve transfection efficiency comparable to liposomal reagents, but are less detrimental to primary neurons and embryonic stem cells.76, 77, 78, 79 Moreover, siRNA coupled with cell-penetrating peptides, like aptamer and rabies virus glycoprotein provides a potential tool for cell type-specific targeting and in-vivo siRNA delivery.80, 81 SiRNA coupled with penetratin via a disulfide bond is far more effective than Lipofectamine 2000 in entering and silencing genes in primary neurons.82, 83, 84 Whether these methods could be successfully applied to deliver siRNA or ASOs in vivo with significant silencing effect and low toxicity to cells need to be further evaluated in developing siRNA-based therapy for HD.85

For local delivery, Alnylam Pharmaceuticals (Cambridge, MA, USA) and Medtronic (Minneapolis, MN, USA) are developing an implantable pump to infuse ALN-HTT (an RNAi-based drug against HD) into the brain. Using an experimental catheter and the commercially available Medtronic SynchroMed II pump, this technology permits the local delivery of siRNA to striatum and subsequent transport to distant regions of the brain. Early studies have shown that continuous infusion of an appropriate siRNA over 7 days caused an approximate 45% drop in the level of HD mRNA in the putamen of rhesus monkeys. One big advantage of implanted pumps is their ability to deliver drugs with much greater efficiency than other techniques. However, the long-term efficacy of gene silencing is not yet known. Another critical issue is the safety of the procedure. Although no clinical abnormalities were observed following 28 consecutive days of siRNA infusion into the monkeys’ brains, a comprehensive study must still be conducted. Ongoing research on this promising technology of implanted pumps is likely to reach the level of clinical trials in the not-so-distant future.

Conclusion

The rapidly developing technology of RNA interference has already proven useful in experimental therapies for HD. Findings from these investigations suggest that targeted gene silencing may be used in therapy for what is now an untreatable disease. As we know, it is difficult and time-consuming to develop conventional drugs. By comparison, siRNAs can be produced relatively easily and their sequences designed to efficiently and selectively silence target genes. Non-coding small RNA-based technology is a promising tool for the treatment of HD, and it certainly deserves future research and development. Many technical hurdles remain to the use of RNAi to develop a therapy for HD, improvement of target specificity and efficacy, optimization of local and systemic delivery, ensuring host safety, and so on. We should expect to see this rapidly developing technology become a major therapeutic modality for HD when these hurdles are successfully overcome. HD patients who suffer from devastating disease would someday benefit from therapies based on small interference RNA.

References

  1. 1

    The Huntington's Disease Collaborative Research Group. A novel gene containing a trinucleotide repeat that is expanded and unstable on Huntington's disease chromosomes. Cell 1993; 72: 971–983.

  2. 2

    Harper SQ . Progress and challenges in RNA interference therapy for Huntington disease. Arch Neurol 2009; 66: 933–938.

  3. 3

    Denovan-Wright EM, Davidson BL . RNAi: a potential therapy for the dominantly inherited nucleotide repeat diseases. Gene Therapy 2006; 13: 525–531.

  4. 4

    Boudreau RL, Davidson BL . RNAi therapy for neurodegenerative diseases. Curr Top Dev Biol 2006; 75: 73–92.

  5. 5

    Wang YL, Liu W, Wada E, Murata M, Wada K, Kanazawa I . Clinico-pathological rescue of a model mouse of Huntington's disease by siRNA. Neurosci Res 2005; 53: 241–249.

  6. 6

    Franich NR, Fitzsimons HL, Fong DM, Klugmann M, During MJ, Young D . AAV vector-mediated RNAi of mutant huntingtin expression is neuroprotective in a novel genetic rat model of Huntington's disease. Mol Ther 2008; 16: 947–956.

  7. 7

    McBride JL, Boudreau RL, Harper SQ, Staber PD, Monteys AM, Martins I et al. Artificial miRNAs mitigate shRNA-mediated toxicity in the brain: implications for the therapeutic development of RNAi. Proc Natl Acad Sci USA 2008; 105: 5868–5873.

  8. 8

    Boudreau RL, McBride JL, Martins I, Shen S, Xing Y, Carter BJ et al. Nonallele-specific silencing of mutant and wild-type huntingtin demonstrates therapeutic efficacy in Huntington's disease mice. Mol Ther 2009; 17: 1053–1063.

  9. 9

    DiFiglia M, Sena-Esteves M, Chase K, Sapp E, Pfister E, Sass M et al. Therapeutic silencing of mutant huntingtin with siRNA attenuates striatal and cortical neuropathology and behavioral deficits. Proc Natl Acad Sci USA 2007; 104: 17204–17209.

  10. 10

    Rodriguez-Lebron E, Denovan-Wright EM, Nash K, Lewin AS, Mandel RJ . Intrastriatal rAAV-mediated delivery of anti-huntingtin shRNAs induces partial reversal of disease progression in R6/1 Huntington's disease transgenic mice. Mol Ther 2005; 12: 618–633.

  11. 11

    Denovan-Wright EM, Rodriguez-Lebron E, Lewin AS, Mandel RJ . Unexpected off-targeting effects of anti-huntingtin ribozymes and siRNA in vivo. Neurobiol Dis 2008; 29: 446–455.

  12. 12

    Machida Y, Okada T, Kurosawa M, Oyama F, Ozawa K, Nukina N . rAAV-mediated shRNA ameliorated neuropathology in Huntington disease model mouse. Biochem Biophys Res Commun 2006; 343: 190–197.

  13. 13

    Drouet V, Perrin V, Hassig R, Dufour N, Auregan G, Alves S et al. Sustained effects of nonallele-specific Huntingtin silencing. Ann Neurol 2009; 65: 276–285.

  14. 14

    Huang B, Schiefer J, Sass C, Landwehrmeyer GB, Kosinski CM, Kochanek S . High-capacity adenoviral vector-mediated reduction of huntingtin aggregate load in vitro and in vivo. Hum Gene Ther 2007; 18: 303–311.

  15. 15

    Zhang Y, Friedlander RM . Silencing Huntington's disease (HD) gene with RNAi. In: Erdmann VA and Barciszewski J (eds). RNA Technologies and Their Applications. Springer: Berlin, 2010, pp 131–160.

  16. 16

    van Bilsen PH, Jaspers L, Lombardi MS, Odekerken JC, Burright EN, Kaemmerer WF . Identification and allele-specific silencing of the mutant huntingtin allele in Huntington's disease patient-derived fibroblasts. Hum Gene Ther 2008; 19: 710–719.

  17. 17

    Zhang Y, Engelman J, Friedlander RM . Allele-specific silencing of mutant Huntington's disease gene. J Neurochem 2009; 108: 82–90.

  18. 18

    Hu J, Matsui M, Corey DR . Allele-selective inhibition of mutant huntingtin by peptide nucleic acid-peptide conjugates, locked nucleic acid, and small interfering RNA. Ann NY Acad Sci 2009; 1175: 24–31.

  19. 19

    Liu W, Kennington LA, Rosas HD, Hersch S, Cha JH, Zamore PD et al. Linking SNPs to CAG repeat length in Huntington's disease patients. Nat Methods 2008; 5: 951–953.

  20. 20

    Palfi S, Brouillet E, Jarraya B, Bloch J, Jan C, Shin M et al. Expression of mutated huntingtin fragment in the putamen is sufficient to produce abnormal movement in non-human primates. Mol Ther 2007; 15: 1444–1451.

  21. 21

    Lombardi MS, Jaspers L, Spronkmans C, Gellera C, Taroni F, Di Maria E et al. A majority of Huntington's disease patients may be treatable by individualized allele-specific RNA interference. Exp Neurol 2009; 217: 312–319.

  22. 22

    de Mezer M, Wojciechowska M, Napierala M, Sobczak K, Krzyzosiak WJ . Mutant CAG repeats of huntingtin transcript fold into hairpins, form nuclear foci and are targets for RNA interference. Nucleic Acids Res 2011; 39: 3852–3863.

  23. 23

    Hu J, Matsui M, Gagnon KT, Schwartz JC, Gabillet S, Arar K et al. Allele-specific silencing of mutant huntingtin and ataxin-3 genes by targeting expanded CAG repeats in mRNAs. Nat Biotechnol 2009; 27: 478–484.

  24. 24

    Gagnon KT, Pendergraff HM, Deleavey GF, Swayze EE, Potier P, Randolph J et al. Allele-selective inhibition of mutant huntingtin expression with antisense oligonucleotides targeting the expanded CAG repeat. Biochemistry 2010; 49: 10166–10178.

  25. 25

    Hu J, Liu J, Corey DR . Allele-selective inhibition of huntingtin expression by switching to an miRNA-like RNAi mechanism. Chem Biol 2010; 17: 1183–1188.

  26. 26

    Fiszer A, Mykowska A, Krzyzosiak WJ . Inhibition of mutant huntingtin expression by RNA duplex targeting expanded CAG repeats. Nucleic Acids Res 2011; 39: 5578–5585.

  27. 27

    Harper SQ, Staber PD, He X, Eliason SL, Martins IH, Mao Q et al. RNA interference improves motor and neuropathological abnormalities in a Huntington's disease mouse model. Proc Natl Acad Sci USA 2005; 102: 5820–5825.

  28. 28

    Schilling G, Becher MW, Sharp AH, Jinnah HA, Duan K, Kotzuk JA et al. Intranuclear inclusions and neuritic aggregates in transgenic mice expressing a mutant N-terminal fragment of huntingtin. Hum Mol Genet 1999; 8: 397–407.

  29. 29

    Mangiarini L, Sathasivam K, Seller M, Cozens B, Harper A, Hetherington C et al. Exon 1 of the HD gene with an expanded CAG repeat is sufficient to cause a progressive neurological phenotype in transgenic mice. Cell 1996; 87: 493–506.

  30. 30

    Kotliarova S, Jana NR, Sakamoto N, Kurosawa M, Miyazaki H, Nekooki M et al. Decreased expression of hypothalamic neuropeptides in Huntington disease transgenic mice with expanded polyglutamine-EGFP fluorescent aggregates. J Neurochem 2005; 93: 641–653.

  31. 31

    Senut MC, Suhr ST, Kaspar B, Gage FH . Intraneuronal aggregate formation and cell death after viral expression of expanded polyglutamine tracts in the adult rat brain. J Neurosci 2000; 20: 219–229.

  32. 32

    de Almeida LP, Ross CA, Zala D, Aebischer P, Deglon N . Lentiviral-mediated delivery of mutant huntingtin in the striatum of rats induces a selective neuropathology modulated by polyglutamine repeat size, huntingtin expression levels, and protein length. J Neurosci 2002; 22: 3473–3483.

  33. 33

    Slow EJ, van Raamsdonk J, Rogers D, Coleman SH, Graham RK, Deng Y et al. Selective striatal neuronal loss in a YAC128 mouse model of Huntington disease. Hum Mol Genet 2003; 12: 1555–1567.

  34. 34

    Gray M, Shirasaki DI, Cepeda C, Andre VM, Wilburn B, Lu XH et al. Full-length human mutant huntingtin with a stable polyglutamine repeat can elicit progressive and selective neuropathogenesis in BACHD mice. J Neurosci 2008; 28: 6182–6195.

  35. 35

    Spampanato J, Gu X, Yang XW, Mody I . Progressive synaptic pathology of motor cortical neurons in a BAC transgenic mouse model of Huntington's disease. Neuroscience 2008; 157: 606–620.

  36. 36

    Hodgson JG, Agopyan N, Gutekunst CA, Leavitt BR, LePiane F, Singaraja R et al. A YAC mouse model for Huntington's disease with full-length mutant huntingtin, cytoplasmic toxicity, and selective striatal neurodegeneration. Neuron 1999; 23: 181–192.

  37. 37

    Schwarz DS, Ding H, Kennington L, Moore JT, Schelter J, Burchard J et al. Designing siRNA that distinguish between genes that differ by a single nucleotide. PLoS Genet 2006; 2: e140.

  38. 38

    Warby SC, Montpetit A, Hayden AR, Carroll JB, Butland SL, Visscher H et al. CAG expansion in the Huntington disease gene is associated with a specific and targetable predisposing haplogroup. Am J Hum Genet 2009; 84: 351–366.

  39. 39

    Ambrose CM, Duyao MP, Barnes G, Bates GP, Lin CS, Srinidhi J et al. Structure and expression of the Huntington's disease gene: evidence against simple inactivation due to an expanded CAG repeat. Somat Cell Mol Genet 1994; 20: 27–38.

  40. 40

    Almqvist E, Spence N, Nichol K, Andrew SE, Vesa J, Peltonen L et al. Ancestral differences in the distribution of the delta 2642 glutamic acid polymorphism is associated with varying CAG repeat lengths on normal chromosomes: insights into the genetic evolution of Huntington disease. Hum Mol Genet 1995; 4: 207–214.

  41. 41

    Rubinsztein DC, Leggo J, Goodburn S, Barton DE, Ferguson-Smith MA . Haplotype analysis of the delta 2642 and (CAG)n polymorphisms in the Huntington's disease (HD) gene provides an explanation for an apparent ‘founder’ HD haplotype. Hum Mol Genet 1995; 4: 203–206.

  42. 42

    Novelletto A, Persichetti F, Sabbadini G, Mandich P, Bellone E, Ajmar F et al. Polymorphism analysis of the huntingtin gene in Italian families affected with Huntington disease. Hum Mol Genet 1994; 3: 1129–1132.

  43. 43

    Vuillaume I, Vermersch P, Destee A, Petit H, Sablonniere B . Genetic polymorphisms adjacent to the CAG repeat influence clinical features at onset in Huntington's disease. J Neurol Neurosurg Psychiatry 1998; 64: 758–762.

  44. 44

    Sobczak K, de Mezer M, Michlewski G, Krol J, Krzyzosiak WJ . RNA structure of trinucleotide repeats associated with human neurological diseases. Nucleic Acids Res 2003; 31: 5469–5482.

  45. 45

    Gacy AM, Goellner G, Juranic N, Macura S, McMurray CT . Trinucleotide repeats that expand in human disease form hairpin structures in vitro. Cell 1995; 81: 533–540.

  46. 46

    Broda M, Kierzek E, Gdaniec Z, Kulinski T, Kierzek R . Thermodynamic stability of RNA structures formed by CNG trinucleotide repeats. Implication for prediction of RNA structure. Biochemistry 2005; 44: 10873–10882.

  47. 47

    Sobczak K, Michlewski G, de Mezer M, Kierzek E, Krol J, Olejniczak M et al. Structural diversity of triplet repeat RNAs. J Biol Chem 2010; 285: 12755–12764.

  48. 48

    Michlewski G, Krzyzosiak WJ . Molecular architecture of CAG repeats in human disease related transcripts. J Mol Biol 2004; 340: 665–679.

  49. 49

    Krol J, Fiszer A, Mykowska A, Sobczak K, de Mezer M, Krzyzosiak WJ . Ribonuclease dicer cleaves triplet repeat hairpins into shorter repeats that silence specific targets. Mol Cell 2007; 25: 575–586.

  50. 50

    Napierala M, Krzyzosiak WJ . CUG repeats present in myotonin kinase RNA form metastable "slippery" hairpins. J Biol Chem 1997; 272: 31079–31085.

  51. 51

    Braasch DA, Liu Y, Corey DR . Antisense inhibition of gene expression in cells by oligonucleotides incorporating locked nucleic acids: effect of mRNA target sequence and chimera design. Nucleic Acids Res 2002; 30: 5160–5167.

  52. 52

    Dias N, Dheur S, Nielsen PE, Gryaznov S, Van Aerschot A, Herdewijn P et al. Antisense PNA tridecamers targeted to the coding region of Ha-ras mRNA arrest polypeptide chain elongation. J Mol Biol 1999; 294: 403–416.

  53. 53

    Jackson AL, Bartz SR, Schelter J, Kobayashi SV, Burchard J, Mao M et al. Expression profiling reveals off-target gene regulation by RNAi. Nat Biotechnol 2003; 21: 635–637.

  54. 54

    Qiu S, Adema CM, Lane T . A computational study of off-target effects of RNA interference. Nucleic Acids Res 2005; 33: 1834–1847.

  55. 55

    Lin X, Ruan X, Anderson MG, McDowell JA, Kroeger PE, Fesik SW et al. siRNA-mediated off-target gene silencing triggered by a 7 nt complementation. Nucleic Acids Res 2005; 33: 4527–4535.

  56. 56

    Jackson AL, Burchard J, Schelter J, Chau BN, Cleary M, Lim L et al. Widespread siRNA “off-target” transcript silencing mediated by seed region sequence complementarity. RNA 2006; 12: 1179–1187.

  57. 57

    Birmingham A, Anderson EM, Reynolds A, Ilsley-Tyree D, Leake D, Fedorov Y et al. 3′ UTR seed matches, but not overall identity, are associated with RNAi off-targets. Nat Methods 2006; 3: 199–204.

  58. 58

    Doench JG, Sharp PA . Specificity of microRNA target selection in translational repression. Genes Dev 2004; 18: 504–511.

  59. 59

    Haley B, Zamore PD . Kinetic analysis of the RNAi enzyme complex. Nat Struct Mol Biol 2004; 11: 599–606.

  60. 60

    Schwarz DS, Hutvagner G, Du T, Xu Z, Aronin N, Zamore PD . Asymmetry in the assembly of the RNAi enzyme complex. Cell 2003; 115: 199–208.

  61. 61

    Khvorova A, Reynolds A, Jayasena SD . Functional siRNAs and miRNAs exhibit strand bias. Cell 2003; 115: 209–216.

  62. 62

    Reynolds A, Leake D, Boese Q, Scaringe S, Marshall WS, Khvorova A . Rational siRNA design for RNA interference. Nat Biotechnol 2004; 22: 326–330.

  63. 63

    Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel DP . MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell 2007; 27: 91–105.

  64. 64

    Burger C, Gorbatyuk OS, Velardo MJ, Peden CS, Williams P, Zolotukhin S et al. Recombinant AAV viral vectors pseudotyped with viral capsids from serotypes 1, 2, and 5 display differential efficiency and cell tropism after delivery to different regions of the central nervous system. Mol Ther 2004; 10: 302–317.

  65. 65

    Paterna JC, Feldon J, Bueler H . Transduction profiles of recombinant adeno-associated virus vectors derived from serotypes 2 and 5 in the nigrostriatal system of rats. J Virol 2004; 78: 6808–6817.

  66. 66

    Taymans JM, Vandenberghe LH, Haute CV, Thiry I, Deroose CM, Mortelmans L et al. Comparative analysis of adeno-associated viral vector serotypes 1, 2, 5, 7, and 8 in mouse brain. Hum Gene Ther 2007; 18: 195–206.

  67. 67

    Davidson BL, Stein CS, Heth JA, Martins I, Kotin RM, Derksen TA et al. Recombinant adeno-associated virus type 2, 4, and 5 vectors: transduction of variant cell types and regions in the mammalian central nervous system. Proc Natl Acad Sci USA 2000; 97: 3428–3432.

  68. 68

    Grimm D, Streetz KL, Jopling CL, Storm TA, Pandey K, Davis CR et al. Fatality in mice due to oversaturation of cellular microRNA/short hairpin RNA pathways. Nature 2006; 441: 537–541.

  69. 69

    Bumcrot D, Manoharan M, Koteliansky V, Sah DW . RNAi therapeutics: a potential new class of pharmaceutical drugs. Nat Chem Biol 2006; 2: 711–719.

  70. 70

    Allerson CR, Sioufi N, Jarres R, Prakash TP, Naik N, Berdeja A et al. Fully 2′-modified oligonucleotide duplexes with improved in vitro potency and stability compared to unmodified small interfering RNA. J Med Chem 2005; 48: 901–904.

  71. 71

    Hoshika S, Minakawa N, Matsuda A . Synthesis and physical and physiological properties of 4′-thioRNA: application to post-modification of RNA aptamer toward NF-kappaB. Nucleic Acids Res 2004; 32: 3815–3825.

  72. 72

    Dande P, Prakash TP, Sioufi N, Gaus H, Jarres R, Berdeja A et al. Improving RNA interference in mammalian cells by 4′-thio-modified small interfering RNA (siRNA): effect on siRNA activity and nuclease stability when used in combination with 2′-O-alkyl modifications. J Med Chem 2006; 49: 1624–1634.

  73. 73

    Czauderna F, Fechtner M, Dames S, Aygun H, Klippel A, Pronk GJ et al. Structural variations and stabilising modifications of synthetic siRNAs in mammalian cells. Nucleic Acids Res 2003; 31: 2705–2716.

  74. 74

    Jackson AL, Burchard J, Leake D, Reynolds A, Schelter J, Guo J et al. Position-specific chemical modification of siRNAs reduces “off-target” transcript silencing. RNA 2006; 12: 1197–1205.

  75. 75

    Amarzguioui M, Holen T, Babaie E, Prydz H . Tolerance for mutations and chemical modifications in a siRNA. Nucleic Acids Res 2003; 31: 589–595.

  76. 76

    Tonges L, Lingor P, Egle R, Dietz GP, Fahr A, Bahr M . Stearylated octaarginine and artificial virus-like particles for transfection of siRNA into primary rat neurons. RNA 2006; 12: 1431–1438.

  77. 77

    Futaki S, Ohashi W, Suzuki T, Niwa M, Tanaka S, Ueda K et al. Stearylated arginine-rich peptides: a new class of transfection systems. Bioconjug Chem 2001; 12: 1005–1011.

  78. 78

    Crombez L, Charnet A, Morris MC, Aldrian-Herrada G, Heitz F, Divita G . A non-covalent peptide-based strategy for siRNA delivery. Biochem Soc Trans 2007; 35: 44–46.

  79. 79

    Simeoni F, Morris MC, Heitz F, Divita G . Peptide-based strategy for siRNA delivery into mammalian cells. Methods Mol Biol 2005; 309: 251–260.

  80. 80

    Chu TC, Twu KY, Ellington AD, Levy M . Aptamer mediated siRNA delivery. Nucleic Acids Res 2006; 34: e73.

  81. 81

    Kumar P, Wu H, McBride JL, Jung KE, Kim MH, Davidson BL et al. Transvascular delivery of small interfering RNA to the central nervous system. Nature 2007; 448: 39–43.

  82. 82

    Jankowski MP, McIlwrath SL, Jing X, Cornuet PK, Salerno KM, Koerber HR et al. Sox11 transcription factor modulates peripheral nerve regeneration in adult mice. Brain Res 2009; 1256: 43–54.

  83. 83

    Davidson TJ, Harel S, Arboleda VA, Prunell GF, Shelanski ML, Greene LA et al. Highly efficient small interfering RNA delivery to primary mammalian neurons induces MicroRNA-like effects before mRNA degradation. J Neurosci 2004; 24: 10040–10046.

  84. 84

    Muratovska A, Eccles MR . Conjugate for efficient delivery of short interfering RNA (siRNA) into mammalian cells. FEBS Lett 2004; 558: 63–68.

  85. 85

    Schwartz EI . Potential application of RNAi for understanding and therapy of neurodegenerative diseases. Front Biosci 2009; 14: 297–320.

  86. 86

    Seyhan AA . RNAi: a potential new class of therapeutic for human genetic disease. Hum Genet 2011; 130: 583–605.

Download references

Author information

Correspondence to R M Friedlander.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • non-coding small RNAs
  • therapy
  • Huntington's disease
  • mutant HD
  • wild-type HD
  • allele specific

Further reading