Gene suppression approaches to neurodegeneration

Gene suppression approaches refer to targeted molecular genetic therapies that serve to lower the expression of specific genes. The term “gene silencing” has also been used to describe these methods; however, this term should be considered a misnomer as complete gene inactivation does not occur, and might not be desirable. Such techniques have made enormous progress over the last 20 years, and show great promise for the treatment of inherited neurodegenerative diseases arising from a known genetic mutation.

Neurodegenerative diseases that result from the toxic gain-of-function of a mutant protein or non-coding RNA that are without a significant loss-of-function are ideal candidates for gene suppression approaches. There are many strategies to lower the amount of toxic disease proteins that result from single gene mutations, including both post-transcriptional inhibition such as RNA interference (RNAi), anti-sense oligonucleotides (ASOs) and catalytic nucleic acids, and genome editing techniques such as zinc finger proteins (ZFPs) and CRISPR/Cas9. The great benefit of interventions at the gene suppression level, as opposed to interventions aimed at the toxic protein itself, is that the plethora of potentially negative downstream cellular pathogenic effects that may arise from the abnormal functioning of a single protein are all reduced as a consequence of treatment.

In this review we will use examples from Huntington’s disease (HD), in respect of which a worldwide collaborative research effort has led to the swift progression of gene suppression approaches, from in-vitro and in-vivo development through to a clinical trial of a Huntingtin-lowering therapy that is currently in progress [1]. Huntington’s disease is an autosomal dominant, neurodegenerative condition caused by a CAG expansion in exon 1 of the gene encoding the Huntingtin (HTT) protein. The presence of the mutant gene leads to the adult onset of chorea, psychiatric symptoms and cognitive decline, and is fatal after 15–20 years [2]. In addition, examples of gene suppression approaches in amyotrophic lateral sclerosis (ALS) [3] will be discussed. However, lessons learnt from development of gene suppression technologies in these conditions could equally be applied to any dementia or neurodegenerative condition in which the responsible disease protein is known. Indeed the potential of lowering tau protein for the treatment of Alzheimer’s disease (AD) and other tauopathies has been demonstrated recently [4], as well as targeting alpha-synuclein (SNCA) [5] and leucine-rich-repeat kinase 2 (LRRK2) [6] for Parkinson’s disease (PD), and ataxin-2 for the treatment of spinocerebellar ataxia type 2 (SCA2) [7], sporadic ALS and frontotemporal dementia (FTD) [8].

Gene suppression can also be achieved at the transcriptional level, using effectors that bind to specific DNA sequences. One such approach uses ZFPs which contain a zinc finger domain that can be manipulated synthetically to bind a DNA sequence of interest, fused to a functional protein domain. Zinc finger nucleases (ZFNs) cleave DNA at a particular target site, and zinc finger transcriptional repressors (ZFTRs) are transcription factor DNA-recognition motifs fused to a transcriptional repressor domain. Delivery of such agents would be through the use of viral vectors.

Potential advantages of this approach are the avoidance of on-target and off-target toxicities of RNA therapeutics and the elimination of all disease-causing mRNA splice variants, not just those that possess the target mRNA sequence [42]. The prospect of germ-line treatment, which would benefit future generations, is also possible. In the case of HD, ZFPs delivered intra-parenchymally using AAV vectors in a transgenic mouse model of HD (R6/2) reduced mutant HTT expression (without affecting wild-type HTT expression) and led to improvements in neuropathology and motor deficits [43]. The pharmaceutical company Sangamo, in conjunction with Shire Pharmaceuticals, are planning to take this approach forwards into clinical trials.

Other engineered nucleases include mega-nucleases, transcription activator-like effector nucleases (TALENs) and clustered regulatory interspaced short palindromic repeat (CRISPR)/Cas9 systems [44]. CRISPR/Cas9 is an endogenous mechanism within prokaryotic cells that recognises and destroys foreign DNA. This system can be manipulated to incorporate a synthetic guide RNA (gRNA) strand that targets a particular DNA location for cutting, followed by the insertion of DNA (e.g. stop codons) that essentially inactivates the mutant allele [45]. The technique has been demonstrated successfully in HD patient-derived fibroblasts, leading to the dramatic reduction of mutant HTT RNA and mHTT protein [46].

In addition to repressing gene transcription, the ability of these methods to target nuclease-induced scission and repair offers the exciting possibility of actually correcting genetic mutations (e.g. through the excision of the expanded CAG repeat in HD)—leading to true genome editing. There is thus the exciting potential to treat not just HD but any genetic cause of neurodegenerative dementia using these techniques.

ASOs can be developed to target mRNA from both the mutant allele and the wild-type allele equally or to favour the mutant allele over the wild-type allele. The former, non-allele-specific, approach has the advantages of addressing the entire HD population and maintaining flexibility for ASO design across the entire gene, increasing the likelihood of identifying a specific, potent and well-tolerated ASO. The potential disadvantage for non-allele-specific ASOs lies in reduction of the wild-type allele. An allele-specific approach can be achieved by targeting either the CAG expansion or a sequence containing a specific single nucleotide polymorphism (SNP) found only on the mutant allele. Allele-specific approaches have the advantage of preferential reduction of the mutant protein. Potential disadvantages include blunted potency and tolerability due to severe limitations on sequence space, undesirable reduction of other proteins containing CAG repeats not linked to disease (with CAG-targeting ASOs) and addressing only a subset of the disease population (with SNP-targeting ASOs).

In the case of HD, embryonic knockout of Htt is lethal [79] and its complete inactivation in the rodent brain causes a progressive neurological phenotype [80]. However, as mentioned previously, partial HTT lowering has been shown to be well tolerated in NHPs [30, 35] and complete inactivation of HTT in adulthood has been shown to be well tolerated in mice carrying the human HTT transgene [81]. Intuitively, strategies to silence the mutant allele whilst leaving the normal allele unaffected may be desirable to prevent any potential longer-term detrimental effects from loss of wild-type protein function. It is noteworthy, however, that ASOs do not ever achieve complete gene suppression. Even at high doses of ASO, residual HTT will be produced, mitigating risk of non-allele-specific approaches.

ASOs complementary to the expanded CAG have been produced which have up to 6-fold selectivity for mHTT over HTT [82–84]; selectivity is thought to arise as longer CAG transcripts provide binding sites for multiple ASOs. One potential drawback of this strategy is that other polyQ-containing proteins may be affected, and the potency of CAG targeting therapies is not as great as for other approaches. BioMarin Pharmaceuticals have, however, very recently carried out a study showing that an ASO with sequence (CUG)7 successfully targeted the expanded CAG repeat in mutant Htt mRNA in two different mouse models of HD and that this led to phenotypic improvement [85].

SNPs have been identified that reside only on the mutant HTT allele [86, 87], with siRNAs designed to target these specific SNPs. Three such SNPs have been identified that could be used to treat a majority of patients in Europe and the USA [88]. A recent study has shown the feasibility of allele selective suppression in HD patient primary ex-vivo myeloid cells when using siRNA to target one of these potential SNPs, although the other SNPs tested did not achieve any allele selectivity [89]. This was thought to be due to the nucleotide sequence and tertiary structure of the transcript surrounding the base mismatch, and therefore a lower proportion of HD patients have targetable SNPs using this siRNA approach than originally thought.

As mentioned previously, ASOs could also be used in a SNP-targeting approach. An advantage of ASOs over siRNAs is that they can target SNPs present in intronic sequences, due to their capacity to bind pre-mRNA in the nucleus rather than only mature mRNA transcripts in the cytoplasm. ASOs targeting mutant-allele linked SNPs have been designed that lead to 50-fold selectivity of the mHTT allele over the normal allele [90–92]. Three to five ASOs targeting different SNP sites would be needed to treat 80% of HD patients [93]. Wave Life Sciences are initiating clinical trials of two ASOs targeting SNP-containing regions of HTT mRNA; this involves screening potential trial subjects for the presence of these SNPs before allocation to one or other of the compounds [94].

The first human clinical trial of intrathecal delivery of ASOs for the treatment of a neurodegenerative condition was carried out in patients with amyotrophic lateral sclerosis (ALS) [3]. Inherited mutations in SOD1 account for 13% of familial ALS and 2% of all ALS, and cause disease through a toxic gain-of-function of SOD1 protein. An ASO targeting SOD1 mRNA was shown to reduce both wild-type and mutant SOD1 (i.e. non-allele selective) in transgenic mice and in human cell models of ALS [95, 96]. A phase 1 trial carried out by Ionis Pharmaceuticals has since demonstrated the safety and tolerability of intrathecal infusion of this drug in ALS patients and has proven that the drug is distributed to spinal cord tissues following injection and is detectable in the CNS 3 months following injection [3]. Having demonstrated the feasibility of this approach, Ionis Pharmaceuticals, in conjunction with Biogen, is currently undertaking a phase 1/2a clinical study of a more potent ASO (ISIS-SOD1_Rx) in ALS associated with a SOD1 mutation [97].

As mentioned previously, another phase 1/2a clinical trial is also currently being undertaken by Ionis to test the safety and tolerability of a MOE-modified ASO targeting human HTT delivered by lumbar intrathecal bolus administration to early-stage HD patients. The trial includes neuroimaging assessments and exploratory endpoints to assess the effect on cognitive, motor and neuropsychiatric symptoms [1]. A clinical trial of AAV2-delivered anti-HTT miRNA, which has shown promise in NHPs [29], is also being planned to initiate in the next few years [98].

Patients with SMA type 1 (infantile onset) have been treated with lumbar intrathecal bolus injections of an MOE-modified ASO (Nusinersen), and have shown dramatic improvement [34]. This is an important result for the field, but it is worth noting that Nusinersen differs mechanistically from other ASOs because it does not trigger RNAse H1-induced cleavage of the target mRNA, and therefore does not lead to “gene suppression”. Infants with SMA type 1 have homozygous gene mutations or deletion of SMN1, leading to a lack of survival motor neuron (SMN) protein and the death of motor neurons, which are particularly reliant on it. Nusinersen in fact alters the splicing of the mRNA of another gene, SMN2, that usually encodes a non-functional SMN protein, and in doing so increases the amount of functional SMN available in the cell. The phase 3 ENDEAR trial of this drug showed such promising efficacy in interim analysis that urgent work has already begun on an extended access programme to widen the availability of the drug to those not in a clinical trial [99]. The drug (marketed as Spinraza™) received expedited approval by the FDA [100] for all forms of SMA in paediatric and adult patients, and is administered by lumbar intrathecal bolus injection at 4-month intervals. The success of Nusinersen is encouraging; however, infantile SMA is an aggressive and rapidly progressive disease which manifests in the first few weeks or months of life, and is generally fatal over 1–2 years. Therefore positive treatment effects could be easily and rapidly discerned in the clinical trial setting. The outcome of gene suppression therapies for most other neurodegenerative diseases will most probably take longer to elucidate, as positive effects may not be so striking.

Recently DeVos et al. [4] have shown that treatment with tau-reducing ASOs prevents neuronal loss in a mouse model of tauopathy. Tau is a microtubule-associated protein in neurons, which becomes hyper-phosphorylated, misfolds and forms insoluble neurofibrillary tangles (NFTs) in AD. The close temporal and spatial relationship between tau pathology and neurodegeneration is thought to underlie the widespread neuronal loss and progressive dementia seen in this condition. Indeed mutations in the MAPT gene, which encodes tau protein, are causative for another neurodegenerative disease—frontotemporal lobar degeneration (FTLD). This study demonstrated improved survival and histopathology in a mouse model of tauopathy, and also showed tau protein reduction in CNS tissues and CSF in non-human primates following intrathecal bolus administration of ASOs targeting MAPT mRNA. This opens up the exciting potential for gene suppression treatments of tau as a therapeutic avenue for AD and other tauopathies such as progressive supranuclear palsy (PSP), and clinical trials of tau-lowering therapies are imminent.

The possibility of treatment with ASOs that reduce pathogenic splice variants of C9orf72 is also under investigation for the treatment of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). In the future, an ASO that suppresses alpha-synuclein gene (SNCA) expression to reduce alpha-synuclein protein might have utility in limiting the progression of PD [5]. In addition, an ASO targeting leucine-rich-repeat kinase 2 (LRRK2) (mutations of which are the major cause of familial late-onset PD) has been shown to reduce the spread of alpha-synuclein pathology in the mouse brain [6]. There may be great benefit in preventing the cortical dissemination of alpha-synuclein pathology, to target the cognitive decline and dementia that will eventually afflict more than 80% of PD patients and is one of the most feared consequences for those developing PD.

Spinocerebellar ataxia type 2 (SCA2) is an autosomal-dominant, adult-onset neurodegenerative disease caused by a polyglutamine expansion in the ataxin-2 gene (ATXN2). Scoles et al. [7] recently carried out a screen of multiple ASOs directed against ATXN2 mRNA (non-allele specific) and tested their most promising candidate in two mouse models. Following intracerebroventricular injection, this ASO localised to Purkinje cells and reduced cerebellar ATXN2 expression and levels of cerebellar ATXN2. This was also associated with significantly improved motor function [7].

An ASO lowering ATXN2 has also been shown to increase the lifespan and improve motor symptoms in a mouse model of TDP-43 proteinopathy [8]. Nearly all patients with ALS have aggregates of TDP-43 (an RNA-binding protein) in their brain and spinal cord, and TDP-43 pathology is also found in 50% of cases of FTD. Genetic lowering of ataxin-2 (using ataxin-2 knockout mice crossed with TDP-43 transgenic mice) leads to a reduction in TDP-43 aggregation. However, therapies that suppress TDP-43 are not viable due to the protein’s vital cellular functions. The finding that ATXN2 lowering markedly extends survival in these mice lends promise to this approach for the treatment of sporadic ALS and FTD, and is the first example of an ASO therapy targeting a modifier gene that is not directly causative of the disease [8].

Challenges with drug delivery of gene suppression agents have been researched extensively and largely overcome. In the case of ASOs, lumbar intrathecal bolus delivery is proving to be a viable option, and in the case of RNAi, modification of viral vectors has allowed for tolerability and efficacy. Administration of viral vectors either through directed brain injection/infusion to target specific sites, or through peripheral vascular injection leading to global CNS effects, has been tolerated in animal models.

The next few years will see further expansion in the clinical evaluation of gene suppression therapies for the treatment of dementia, both in terms of the number of neurodegenerative diseases targeted and the number of patients treated. Thus there is real hope that, in the face of aging populations and the ever-increasing incidence of dementia, we will in the near future be able to offer effective treatments to sufferers. Moreover, with advances in biomarker technology that can detect the increasingly understood evolution of pathology in asymptomatic individuals, the prospect of early disease interception to prevent the symptomatic expression of these neurodegenerative disorders is a tantalising possibility.