Review Article | Published:

Gene therapy for neurological disorders: progress and prospects

Nature Reviews Drug Discovery volume 17, pages 641659 (2018) | Download Citation

  • An Erratum to this article was published on 12 September 2018

This article has been updated


Adeno-associated viral (AAV) vectors are a rapidly emerging gene therapy platform for the treatment of neurological diseases. In preclinical studies, transgenes encoding therapeutic proteins, microRNAs, antibodies or gene-editing machinery have been successfully delivered to the central nervous system with natural or engineered viral capsids via various routes of administration. Importantly, initial clinical studies have demonstrated encouraging safety and efficacy in diseases such as Parkinson disease and spinal muscular atrophy, as well as durability of transgene expression. Here, we discuss key considerations and challenges in the future design and development of therapeutic AAV vectors, highlighting the most promising targets and recent clinical advances.


In the past several years, there has been much progress towards realizing the potential of AAV vectors for therapeutic gene delivery for neurological disorders. Several key advances have converged. Mutations for many, if not most, monogenic neurological disorders have now been identified and, together with an ever-increasing understanding of their mechanisms of disease pathogenesis, have revealed novel therapeutic targets. Although many of these genetic targets are not readily 'druggable' with conventional drug modalities, they are potentially druggable with adeno-associated viral (AAV) gene therapy, which enables both gene replacement to address loss-of-function mutations and gene silencing to address gain-of-function mutations (for example, replacing survival motor neuron protein (SMN) for spinal muscular atrophy (SMA)1,2,3 or silencing superoxide dismutase 1 (SOD1) for amyotrophic lateral sclerosis (ALS)4,5,6 or huntingtin (HTT) for Huntington disease7,8,9) (Table 1).

Table 1: Examples of promising molecular targets for AAV gene therapy

Transgenes that encode therapeutic proteins2,3,10,11,12,13,14,15,16,17,18,19,20,21, microRNAs (miRNAs)4,5,6,7,8,9, antibodies22,23 and Cas9 – guide RNA24,25 for gene editing have been successfully delivered to the central nervous system (CNS) with AAVs in mice and other species. In addition, multiple capsids have been used across species to successfully target a variety of tissues and cell types within the CNS, including neurons, astrocytes and oligodendrocytes26,27,28,29,30,31,32,33,34,35,36,37,38.

Important advances in optimizing delivery of AAVs into the CNS have been made. In preclinical studies, various routes of AAV delivery, including intraparenchymal (IPa)39, intrathecal (IT)5,15,40, intracerebroventricular (ICV)16, subpial41 and intravenous (IV) administration3,20,42,43, have been shown to achieve therapeutic levels of gene expression in the CNS. Intramuscular (IM) administration has been demonstrated to provide a depot for monoclonal antibody production and delivery44,45, with the theoretical potential to provide a source of antibodies that can cross the blood–brain barrier (BBB) in sufficient quantities, albeit to a limited extent (see below), for clinical benefit. Notably, in 2016, novel, engineered AAV capsids were reported to have unprecedented ability to transfer genes to the CNS in the adult mouse after systemic administration46, with a >40-fold enhancement over the previous standard AAV9, potentially transforming our ability to treat neurodegenerative diseases with AAV gene therapy.

A 'one-and-done' therapeutic approach that provides durable pharmacology and efficacy is attractive, particularly with neurological disorders that require more complex CNS dosing strategies, and is a key feature of AAV gene therapy. Importantly, early gene therapy clinical trials with IPa CNS administration may not have provided sufficient biodistribution to the intended brain targets, but improvements in AAV delivery have enabled more widespread therapeutic transgene expression. Recent clinical trials have demonstrated the relative safety of AAV gene therapy in >200 patients with various CNS diseases, including IPa, IT and IV administration (Table 2). Importantly, evidence of clinical efficacy in several CNS disorders, including Parkinson disease (PD)47, aromatic-L-amino-acid decarboxylase (AADC) deficiency48 and type I SMA, has recently been reported. Durable transgene expression in neurons of the putamen has also been demonstrated, with >4 years of sustained expression of AADC in patients with PD49 and >15 years of sustained expression in nonhuman primates (NHPs)50, supporting the one-and-done treatment concept. Finally, recent progress in chemistry, manufacturing and controls (CMC) using both mammalian (HEK293) and insect (Sf9) cells has enabled the production of AAV gene therapies at commercial quality and scale51,52. Together, these advances have resulted in a resurgence of interest in gene therapy for treating neurological disorders affecting the CNS.

Table 2: Ongoing AAV gene therapy clinical trials for neurological disorders

In this Review, we focus on AAV gene therapy for CNS disorders, which to date represents the vast majority of CNS gene therapy programmes, although other viral vector platforms have been tested clinically (for example, lentivirus)53,54. We discuss recent progress in gene therapy for treating neurological disorders, focusing on the key features of AAV gene therapy vectors that need to be optimized for a successful therapy, including capsid selection, vector genome design and selection, and the delivery and dosing paradigm, using examples from preclinical and clinical studies. Manufacturing challenges and advances are discussed, particularly in the context of the different routes of administration and associated drug requirements. Finally, the most promising CNS drug targets for AAV gene therapy are described, along with recent clinical experience with several of these targets and diseases.

AAV capsid design and selection

The AAV capsid is the determinant of several key features of an effective or successful AAV gene therapy, including tissue tropism and distribution as well as susceptibility to interference by circulating neutralizing antibodies. Both natural and engineered capsids have been explored in efforts to identify novel AAV vectors with properties better suited for the treatment of a given neurological disorder. The natural diversity of AAVs isolated from a broad range of species has provided researchers with a large collection of AAV capsids with unique, but largely overlapping, characteristics. As discussed below, beyond the repertoire provided by nature, engineered capsids are expanding the arsenal of vectors for gene delivery to the CNS. Although no laboratory-modified AAV vectors have yet entered clinical trials targeting the brain or spinal cord, one modified AAV2 variant55 is currently undergoing trials for several retinal diseases (NCT02935517, NCT03316560, NCT02599922 and NCT02416622), and a second laboratory-derived variant56 is being used in a haemophilia trial (NCT03307980). Undoubtedly, more will follow as an increasing number of AAV capsids are customized for improved efficacy and safety in preclinical models and humans.

When delivered to the CNS by IPa injection, AAV vectors provide durable transduction of neurons, leading to their use in several clinical trials and hundreds of basic and applied neuroscience studies. However, different serotypes vary in terms of their efficiency of transduction of specific neuron populations, the extent of glial transduction, the degree of spread from the injection site and their ability to transduce various cell types through alternative cerebrospinal fluid (CSF) or intravascular routes of administration. Below, we provide an overview of the CNS distribution and tropism characteristics that exist within the naturally occurring capsids and highlight several recent efforts in which capsid engineering has been used to overcome barriers that stand in the way of applying AAV vectors to a wider range of CNS diseases, including those that would benefit from broader gene delivery throughout the CNS.

Naturally occurring capsids

The first AAV serotypes (AAV1–AAV4) were identified in the 1960s from adenovirus stocks57,58,59. Following the cloning of the AAV2 genome into a bacterial plasmid, the AAV2 serotype became the initial focus of AAV vector biology research. However, each of these serotypes was found to have unique transduction profiles and receptor binding properties60,61, spurring the search for novel AAV serotypes with different biological properties. More than 100 natural AAV variants comprising at least 8 serotypes have been identified from humans, NHPs and other vertebrates62,63,64,65. For neurological applications, AAV2, which provides long-term expression in CNS neurons although with relatively restricted biodistribution, has been the most well-studied serotype29. Other serotypes (such as AAV1, AAV5, AAV9, AAVrh.10 and AAV-DJ8) that spread more broadly in the CNS and transduce cells with higher efficiency than AAV2 have been evaluated following IPa26,27,30,32,34,36,66, IT33,67,68, intracisternal68,69,70,71, ICV11,12,16,70 and intravascular dosing13,33,67,72,73,74,75,76.

The extent of biodistribution after direct CNS administration is influenced by multiple factors that are governed by the capsid itself, including receptor interactions77 and anterograde and retrograde axonal transport78. In general, the major challenge with translation of AAV delivery for application to CNS disorders is biodistribution and homogeneity of cellular transduction, particularly in adults. To address this delivery challenge and enable application to neurological disorders widely affecting the brain, capsid engineering will be critical.

Engineered capsids for enhanced CNS transduction

Adeno-associated viruses are dependoviruses that co-evolved with viruses such as adenoviruses or herpes simplex viruses (HSVs), which do not typically infect the CNS. Therefore, although adeno-associated viruses have a natural propensity for latency in the absence of a helper virus co-infection (a useful property for durable gene expression in the CNS), it is unlikely that adeno-associated viruses were subjected to selection pressures that would lead to their optimization as delivery vehicles for CNS gene transfer. Fortunately, laboratory-driven rational design, high-throughput screening and directed evolution strategies can be used to customize the properties of the AAV capsid for any of a number of routes of administration, which should open new options for treating neurological diseases.

Rational design and mutational analysis. Understanding structure–function relationships (for example, identification of key residues that affect capsid stability, receptor binding or trafficking) is critical for making targeted improvements to capsid function. Several AAV–receptor interactions have been mapped onto the AAV surface79,80,81 (see Box 1), and mutations that disrupt these interactions typically reduce transduction efficiency82,83,84. Interestingly, not all such mutations are deleterious, as a low-affinity O-linked α2,3-sialic-acid-binding mutant of AAV4 (AAV4.18) displays a new and unique tropism for subventricular zone neural stem cells when administered by ICV injection into neonatal mice85. In a tour de force alanine-scanning mutational analysis, Adachi et al.86 used barcoding-based biodistribution assays to study the contribution of specific AAV9 amino acids to numerous capsid characteristics. Leveraging these data and prior studies, the authors mapped the galactose-binding site of AAV9 and grafted ten key amino acids into a heparin-binding-deficient AAV2 mutant, resulting in an AAV2-based capsid that shares several tropism characteristics with AAV9, including enhanced liver, heart and Lec2 cell transduction86. In a recent intriguing finding that may help shed light on the mechanism of BBB crossing used by the rh.10 AAV serotype, Albright et al.87 identified a short linear region of the rh.10 capsid that is sufficient, when grafted into the AAV1 capsid, to convert AAV1 into a more BBB-penetrant variant. Whether the homologous region of AAV9 also shares this activity and whether the same region of rh.10 can enhance the BBB crossing efficiency of other AAV serotypes remains unreported. With the advancing knowledge of capsid structure–function relationships, rational design holds substantial promise for achieving improved biological profiles for therapeutic application. Nevertheless, designing capsids that more efficiently target specific cell types has proved challenging; for this purpose, directed evolution holds considerable promise.

Box 1: The AAV capsid and its interactions

The adeno-associated viral (AAV) capsid is a 25 nm icosahedral protein shell assembled from three proteins, VP1, VP2 and VP3, in an approximate 1:1:10 ratio. The VP proteins are encoded by a single open reading frame within the AAV cap gene through alternative splicing and codon usage. The AAV capsid protects the AAV genome, and its amino acid sequence and resulting structural features determine its tropism, spread and antigenicity through interactions with cellular receptors, glycans and blood proteins, including antibodies.

The core of the AAV capsid forms a β-barrel structure that is highly conserved across AAVs, but the surface of the capsid harbours more diversity. As primary receptors, AAV capsids utilize a range of proteoglycans, including heparin sulfate (AAV2, AAV3, AAV3b and AAV6), α2,3 and α2,6 N-linked sialic acid (SIA) (AAV1), α2,3 O-linked SIA (AAV4), α2,3 N-linked SIA (AAV5) and heparin sulfate and α2,3 and α2,6 SIA (AAV6) or terminal N-linked galactose (AAV9)77. Secondary receptors have also been identified for several AAVs (for example, fibroblast growth factor receptor (FGFR) and possibly αvβ5 integrin for AAV2 (Ref. 210); hepatocyte growth factor receptor for AAV3 (Ref. 211); platelet-derived growth factor receptor (PDGFR) for AAV5 (Ref. 212); and laminin receptor for AAV2, AAV3, AAV8, and AAV9 (Ref. 213)). These receptor interactions influence the biodistribution and tropism of the AAV capsid. In addition, a possible AAV receptor (AAVR) has been identified recently as a necessary cofactor for transduction by most, but not all, AAVs vectors214. Interestingly, AAV2 and AAV5 interact with distinct AAVR ectodomains whereas AAV1 and AAV8 interact with multiple domains215. Whether the AAV–AAVR interactions are important for binding or post-attachment trafficking is still an area of active investigation214.

Safely transporting the recombinant genome through the extracellular milieu and making contact with cell surface receptors is not the capsid's only task; it must also traffic intracellularly through endosomes and deliver the packaged genome to the host cell nucleus, feats it accomplishes through an acidic pH-dependent exposure of a buried amino-terminal phospholipase A2 domain and multiple nuclear trafficking signals216,217,218. Surprisingly, the AAV capsid may also have post-genome release effects on transcription. Several transduction-defective AAV capsid mutations have been identified that block mRNA expression from the recombinant genome without altering the efficiency of DNA uncoating in the nucleus219,220. The continued study of AAV vector biology is critical for informing AAV capsid engineering efforts aimed at optimizing the AAV capsid for specific gene therapy applications.

Directed evolution. Directed evolution has been the most successful approach to date for developing AAV capsids with improved biological characteristics. The power of this approach stems from it being high throughput, as millions of sequences can be tested against each other with the most potent sequences being enriched over iterative selections, and the lack of a need for prior knowledge of an enhancement mechanism. Applications include enhancing transduction efficiency, increasing resistance to pre-existing neutralizing antibodies, reducing immunogenicity, achieving specificity for particular tissues or cell types and improving axonal transport or spread.

The first step in applying directed evolution is to generate highly diverse capsid libraries (Fig. 1). This generation has been accomplished through homology-based recombination and shuffling56,88,89,90,91,92, peptide insertion46,93,94,95,96 or error-prone PCR97,98. Key to the success of this approach is that each capsid carries its own specific instruction manual (that is, the cap gene DNA sequence). Selective pressure is then applied in vivo or in vitro, and iterative rounds of selection enrich for the most potent variants. Sequences that pass the selection process have been recovered via one of several approaches, including PCR46,96, adenovirus-based AAV genome replication56,87 or fluorescence-activated cell sorting followed by PCR95 (Fig. 1).

Figure 1: AAV engineering through high-throughput selection.
Figure 1

The schematic shows the major steps used to evolve adeno-associated viral (AAV) capsids through high-throughput screening and selection. A library of capsids is generated by introducing diversity within the AAV genome (part a). The library of capsid-containing genomes is then used to generate a virus library (part b), which is subsequently subjected to selection pressure (part c) — for example, for those that escape antibody neutralization, transduce cells in vitro or distribute to the brain or other organs. The capsid sequences are recovered by PCR, cell-type-specific Cre-based amplification or replication with adenovirus (part d). Multiple rounds of selection can be used to enrich for the most effective capsids. Following selection, enriched capsids are screened with barcoding-based methods and/or tested individually in animal models (part e). CREATE, a novel Cre-based cell-type-specific selection method; FACS, fluorescence-activated cell sorting; SCHEMA, a structure-guided approach to generating recombination libraries.

For CNS applications, directed evolution or high-throughput selection has resulted in the identification of AAV capsids with enhanced transduction of the CNS following IPa administration (AAV-DJ)88, increased transduction of neural stem and progenitor cells (SCH9 and AAV4.18)85,99, enhanced retrograde transduction (rAAV2-retro)94, selective transduction of brain endothelia (AAV‐BR1)96 and greatly increased transduction of the adult CNS after IV administration (AAV-PHP.B and AAV-PHP.eB)46,100. In particular, in 2016, Deverman et al.46 reported several AAV capsids that efficiently transduced the adult mammalian CNS via the vasculature. The most notable variant, AAV-PHP.B, delivered >40-fold more AAV genomes to the CNS than AAV9 and transduced >50% of all neuron and astrocyte populations sampled after IV dosing of adult C57BL/6 mice. The CNS tropism of AAV-PHP.B has been replicated in other mouse studies43,101,102 and extended to the rat103, but not the marmoset104. A recent study reported the mouse strain-dependence of AAV-PHP.B as well as the toxicity of AAV-PHP.B.GFP in a rhesus monkey105. However, the contribution of the expression of a non-self-protein such as green fluorescent protein (GFP) to the observed toxicity remains to be determined, especially because GFP expression is well known to cause severe toxicity in NHPs106. Notably, although the toxicity was observed within a short time (by day 5) following AAV-PHP.B.GFP administration, GFP expression in the liver was clearly evident at that time. In a recently reported study in cynomolgus monkeys, AAV-PHP.B and a related variant expressing a self-protein were well tolerated and, relative to AAV9, showed enhanced transduction of several brain and spinal cord regions107,108. More recently, the same group that discovered AAV-PHP.B has reported an enhanced AAV-PHP.B variant (AAV-PHP.eB) that shows further improvements in neuronal transduction100. These AAV capsids were identified using a novel Cre-based, cell-type-specific selection method (CREATE). CREATE enables the selective recovery of sequences that transduce a specific cell population of interest within the CNS, thereby ensuring that the capsids recovered crossed the adult mouse BBB and, within the cell type of interest, mediate conversion of packaged genomes into double-stranded DNA, a late rate-limiting step required for AAV transduction109.

To date, only one laboratory using a mouse model of mucopolysaccharidosis type VII (MPS VII) has reported disease-specific changes to the brain that blocked gene delivery by AAV9 but not an AAV engineered for CNS transduction in this model110. Nonetheless, disease-specific tropism changes certainly warrant close attention to ensure successful translation to the clinic.

Tissue and cell-type selectivity

The AAV capsid and the promoter within the vector genome (below) play major roles in determining cell tropism and expression. AAV1, AAV2, AAV4, AAV5, AAV7, AAV8, AAV9 and rh.10 capsids have been studied by multiple laboratories for transduction and tropism in the CNS in several preclinical species. Similar to AAV2, capsids from most of these serotypes impart a neuronal tropism when injected directly into brain parenchyma, although species75, region30,36 and cell-specific differences in transgene expression have been noted28,111. AAV1, AAV5, AAV8, AAV9 and the less well-studied variants hu.11, hu.32, hu.48R3 and rh.8 transduce astrocytes and oligodendrocytes to a considerable extent28,34,111. Notably, owing to strong neuronal tropism, expression in glia may be underappreciated when ubiquitous promoters are used and only fully revealed with promoters that restrict expression to specific glial populations111,112,113. In contrast to the transduction of neural cells, microglia, pericytes and endothelial cells have proved relatively refractory to transduction following direct brain injection, although several engineered AAV vectors transduce brain endothelia when delivered into the vasculature46,96. Interestingly, AAV4 appears to have a propensity for transducing the ependymal cells that line the walls of the lateral ventricles after injection into or near the ventricle26,35. It should be noted that the virus purification method can also influence AAV tropism when administered to the brain27,114, highlighting the importance of attention to AAV production protocols when comparing in vivo results. Nonetheless, it is clear that the AAV capsid affects cell tropism and can more or less direct the therapeutic payload to the cell type of interest for a particular CNS disease application.

Vector genome design

The interior of the AAV capsid holds the recombinant vector genome (Box 2), which comprises the AAV inverted terminal repeat, transgene, promoter and post-transcriptional regulatory elements (Fig. 2). An AAV genome for gene replacement, for example, in the case of a monogenic disorder such as Friedreich ataxia, would contain a transgene that is a native or optimized codon sequence for the therapeutic protein115. For gene suppression, such as in the case of Huntington disease, the transgene may comprise a primary microRNA (pri-miRNA) or short hairpin RNA (shRNA) to inhibit the target of interest via RNA interference (RNAi)116 or alternatively may comprise a sequence for gene editing117 or repression. Antibodies or antibody fragments may also be expressed22,23. In all cases, selection of gene regulatory elements that determine where (which cells) the therapeutic transgene is expressed and to what degree (strength of expression) is critical for providing an effective and safe therapy.

Box 2: The recombinant AAV genome

The wild-type adeno-associated viral (AAV) genome is a single-stranded DNA (ssDNA) genome flanked by inverted terminal repeats (ITRs) with a total length of approximately 4,700 bp. The ITRs are the only AAV genome sequences required for the generation of recombinant virus. ITRs function during viral genome replication as packaging signals and during transduction by priming second-strand synthesis and episome formation in host cells109. Recombinant AAV vector genomes are typically constructed with AAV2 ITRs, which can be used to cross-package the genome into other AAV capsid serotypes26,60,221, making it a fairly simple 'plug-and-play' task to generate any AAV capsid serotype or variant by co-transfection with an AAV2-based genome.

Including ITRs, the maximum recombinant genome size has a strict size limit of just under 5 kb (Refs 222, 223). Within this space constraint, a cDNA, gene or RNA interference cargo can be inserted along with regulatory elements to control expression. Following uncoating in the nucleus, conversion of the ssDNA genome to a double-stranded genome is a key rate-limiting step for transduction109. Second-strand synthesis requires either host cell DNA synthesis or co-transduction with complementary positive-stranded and negative-stranded AAV genomes224. AAV genomes less than half the wild-type genome length package as dimers that can refold as self-complementary genomes225, and McCarty et al. applied this knowledge to design single ITR terminal resolution site mutants that force self-complementary DNA formation226. These self-complementary AAV genomes provide more rapid and more efficient transduction and have been used in type I spinal muscular atrophy and haemophilia trials1,227.

Figure 2: Recombinant AAV genome design.
Figure 2

The schematics show the basic design parameters for a therapeutic recombinant adeno-associated viral (AAV) genome for transgene expression (parts a,b) and gene knockdown (part c). Modular elements including enhancer promoters (green), introns, upstream open reading frames (uORFs; navy), transgenes (blue) and 3′ untranslated regions (UTR; purple) and polyadenylation signals (poly(A); grey) can be chosen to optimize where, when and how much transgene is expressed. a | AAV vectors can package single-stranded AAV (ssAAV) genomes that are up to nearly 5 kb in length or self-complementary AAV (scAAV) genomes up to 2.3 kb in length, which fold back on themselves to form a double-stranded genome through the mutation of one of the terminal resolution sites (mTRS). b | ssAAV genomes can carry full-length antibodies by expressing the heavy and light chains from a single multicistronic transcript through the use of 2A ribosomal skipping sequences and furin protease cleavage sites. c | To treat diseases with toxic gain-of-function mutations, artificial primary microRNAs (pri-miRNAs) and short hairpin RNAs (shRNAs) can be used for gene knockdown. Both pol III and ubiquitous, cell-type-specific or synthetic pol II promoters can be used to control the expression of the pri-miRNA or shRNA. miRNA, microRNA; TS, terminal sequence.

The transgene

The most common therapeutic application of AAV vectors has been for gene replacement by delivery of a transgene. Gene replacement as a therapeutic strategy in the human CNS was first evaluated for Canavan disease by delivering human aspartoacylase (ASPA) cDNA to multiple brain sites with AAV2 (Refs 118, 119). The transgene was well tolerated and showed evidence of efficacy, with reduced brain atrophy and clinical stabilization over a 5-year follow-up study119. Transgene expression for monogenetic gene deficiencies is being evaluated for several lysosomal storage disorders, giant axonal neuropathy (GAN; NCT02362438) and type I SMA. However, transgene delivery is not limited to the replacement of gene function for monogenetic disorders. Indeed, despite its more complex aetiology, PD has been the target of numerous AAV-based clinical trials with several gene therapy strategies being tested for safety and efficacy, including delivery of AADC directly to the putamen to enhance the conversion of 3,4-dihydroxyphenylalanine (l-DOPA) to dopamine49,120, glutamic acid decarboxylase delivery to the subthalamic nucleus to increase local GABA production121 and neurturin122 delivery to the striatum to provide neurotrophic support for the remaining dopaminergic neurons. Numerous additional promising AAV-based transgene delivery approaches are progressing through preclinical studies10.

To date, CNS-targeted gene therapy trials have exclusively used human cDNAs and thus have not introduced new non-self antigens. In one recent exception, a null patient was included in the IT AAV9 GAN trial (NCT02362438). Promising preclinical results with non-self proteins include full-length humanized antibodies44,45,123, and gene editing124,125,126 and gene regulation machinery127 may also comprise new therapeutic strategies in humans. If proved safe, and coupled to more effective capsids and routes of delivery, these non-self transgenes would greatly expand the options available for treating CNS disorders. However, caution is warranted owing to the potential severe consequences of T cell-mediated immune reactions in the CNS106,128.

Antibodies. Antibody therapies being evaluated for neurodegenerative diseases in the clinic currently include anti-amyloid129 and anti-tau22,23 monoclonal antibodies for Alzheimer disease. A major challenge for systemic administration of such antibody therapies with passive immunization is the BBB, which greatly restricts entry of large proteins such as antibodies from the peripheral circulation. Only 0.1% of administered monoclonal antibody reaches the brain130. Direct CNS administration of an AAV vector encoding an antibody would provide considerably greater drug levels in the CNS with potentially greater therapeutic benefit, although safety would of course need to be evaluated and demonstrated. Administration of AAV rh.10 vectorized phosphorylated-tau monoclonal antibody in a mutant tau transgenic mouse has recently been reported to result in a substantial reduction of tau pathology, supporting the potential of this therapeutic approach22. However, further work will be needed to optimize transgene configurations for antibody expression in the brain. Nonetheless, AAV-mediated antibody expression in the CNS is a promising alternative to traditional passive immunization.

Gene editing. Gene editing and engineered transcriptional regulators targeted to single genes are also being explored as potential therapeutics. Zinc-finger protein (ZFP) transcription factors and nucleases and a rapidly growing CRISPR–Cas toolkit are expanding potential options for regulating endogenous transcription of individual genes and editing genes131,132,133,134,135,136,137, RNA138,139 and single bases140,141,142. Clinically, ZFPs are attractive owing to their fairly short gene length and similarity to endogenous human proteins. A ZFP-based gene-editing platform designed to insert the IDS gene into the active albumin locus for MPS II has entered a clinical trial and, as of the time of writing, has not been associated with any reported adverse effects143. This study will be an important proof of concept for in vivo gene editing. However, success in this trial will not necessarily provide a straightforward path to gene editing in the CNS. CNS gene editing will require further safety data demonstrating the tolerance of irreplaceable CNS cells to expression of a ZFP. Off-target editing remains a concern, as does the expression of a non-self-protein, which may elicit cell-mediated immune responses. Notably, pre-existing acquired immune responses against Cas9 may be common in the human population128. Transient editing machinery expression may mitigate these challenges. In addition, the success of gene-editing approaches that require co-transduction by multiple vectors will likely require more efficient gene delivery vectors.

Gene silencing. Another therapeutic application of AAV vectors has been gene silencing. RNAi is a natural biological process by which small interfering RNAs (siRNAs), typically 19–21 nucleotides in length, can direct sequence-specific degradation of target mRNA, leading to reduced expression of the corresponding protein. Several clinical trials in >800 patients have demonstrated that RNAi can be harnessed in humans by synthetic siRNAs for the suppression of targeted genes and is generally well tolerated (for example, NCT01437059 and NCT01559077). Although RNAi is durable for weeks, the need for continuous exposure to the miRNA for target suppression necessitates repeated dosing of synthetic oligonucleotides, which is particularly challenging for CNS diseases. By contrast, artificial miRNAs or shRNAs expressed from a single administration of AAV gene therapy can provide durable gene silencing and represent an extension of the clinical experience with synthetic siRNAs. The endogenous miRNA biogenesis pathway can be harnessed at multiple points (for example, at the level of the pri-miRNA or shRNA) to inhibit gene expression via the RNA-induced silencing complex (RISC)144. Multiple head-to-head comparisons demonstrate that artificial miRNA cassettes that enter the pathway at the pri-miRNA level result in a safer and/or more efficient production of the inhibitory siRNA of interest than shRNAs that enter the pathway at the shRNA level7,9. shRNAs have been shown to disrupt miRNA biogenesis, whereas artificial pri-miRNAs do not9. Notably, in vivo, shRNAs caused striatal and Purkinje cell toxicity in mice after intrastriatal and intracerebellar administration, respectively, whereas artificial pri-miRNA expression did not7,9. Artificial pri-miRNA cassettes have provided safe and efficacious CNS expression of mature miRNAs targeting HTT7, ataxin 1 (Ref. 145) and ataxin 3 (Ref. 146) in animal models of Huntington disease, spinocerebellar ataxia type 1 and spinocerebellar ataxia type 3, respectively, demonstrating proof of concept for this approach to harness RNAi for therapy. Thus, expressed RNAi using artificial pri-miRNAs is the preferred approach for an AAV gene therapy targeting inhibition by RNAi.

Regulating expression

Transgene and RNAi expression from AAV genomes is highly customizable through the use of an expanding set of transcriptional and post-translational regulatory elements. However, the regulatory elements discussed below are all constitutive and therefore provide no post-administration control over transgene 'dosing' in the event of transgene-related adverse effects. The holy grail of regulated gene expression would be to achieve drug-regulatable control over gene expression without a protein coding regulator. Efforts to date have not yet yielded a translatable approach, but such a regulatable system would be a monumental advance for the field of gene therapy and should be investigated further.

Controlling transcription with gene regulatory elements. Promoters and associated enhancers are the principal drivers of transgene transcription and thereby control how much, in what cell types and under what conditions the transgene is expressed. Owing to their ability to drive high levels of transcription across most cell types, strong ubiquitous promoters such as the cytomegalovirus (CMV) immediate-early enhancer and promoter, artificial CMV-chicken β-actin promoter with β-globin splice acceptor (CAG) and chicken β-actin (CBA) promoter have been chosen for gene transfer applications10. However, strong promoters may provide artificially high levels of expression, especially in cells transduced by multiple copies of the AAV genome resulting in potentially toxic levels of even a self-protein19. For many applications (for example, FXN delivery for Friedreich ataxia or SMN1 for SMA), transgene expression that more closely matches the endogenous protein levels should be sufficient and would likely be better tolerated. Several trials have used the comparatively weaker phosphoglycerate kinase (PGK) promoter (MPS IIIA and B)147. Beyond controlling expression levels, gene regulatory elements can be used to restrict expression to target cell types for optimum therapeutic benefit and reduced off-target expression. Enhancer-promoter elements that restrict expression to several neuronal and glial cell types in the CNS are available148,149,150. For expression of longer transgenes, several gene regulatory fragments and synthetic promoters have been developed, including a shorter CBA promoter (CBh)151, an MECP2 fragment151 and an artificial JeT promoter152, which is being used in the case of gigaxonin expression in the ongoing clinical trial. In addition, cytokine-responsive and hypoxia-responsive enhancer-promoter elements have been used to modulate expression in response to disease states153,154,155,156,157. The development of transcriptional regulatory elements remains an active area of research that will continue to expand the options available for optimizing transgene expression.

Post-transcriptional regulation. Transgene expression control can be further refined using elements that affect RNA (for example, by modulating splicing, nuclear export and stability), translation or protein stability. The transgene is often provided as a cDNA owing to the limited AAV packaging capacity, but the inclusion of an intron, when possible, can improve mRNA processing158,159. The strength of the polyadenylation signal can be used to affect expression160.

Transgene expression in an off-target cell population can be dramatically reduced by including sequences complementary to endogenous miRNAs that are selectively expressed in off-target cell populations. For example, incorporating three miRNA target sequences that are complementary to miR-122, which is expressed in the liver, and miR-1, which is expressed in skeletal muscle, reduced expression by >30-fold in these 'off-target' organs161 while having only negligible effects on expression in the CNS and pancreas, where miR-122 and miR-1 are not highly expressed. Incorporating miR-142-3p target sequences, which selectively degrade the transgene mRNA in antigen-presenting cells, reduced transgene immunogenicity following IM delivery162. Because these sequences are short, they can be multiplexed for greater refinement of expression161.

Translational control. An additional level of control that may be precisely fine-tuned is the efficiency of translation of the open reading frame (ORF). Translational start sites that conform to consensus sequences first described by Kozak provide maximum likelihood for translation initiation in the absence of competing ORFs. Intriguingly, Ferreira et al.163 have described a panel of two amino acids upstream of the ORF and transgene ORF pairs that can be used to fine-tune expression of the transgene over three orders of magnitude. This technique has a fairly unexplored potential for improving the safety of transgenes that may function at low expression levels. Transgene expression can also be optimized through codon modification2,164,165.

Dosing optimization

Along with the AAV capsid and genome elements, the route of administration and dosing paradigm dictate levels and homogeneity of expression across different tissues and regions of interest. For the CNS, multiple routes of delivery have been evaluated in preclinical studies, including IPa, IT, ICV, subpial, IV and IM, and several of these are being used in ongoing clinical trials. Each route of administration has pros and cons that are context-specific, depending on the particular CNS disease application, tissue or cell type of interest for transgene expression and the level and homogeneity of transgene expression required.

Direct CNS delivery

IPa, IT, ICV and subpial routes of administration have been evaluated preclinically for biodistribution of AAV gene therapy to the CNS with direct delivery. Each route of administration, together with the choice of capsid, may be best suited for certain disease applications, depending on the targeted CNS region of interest and the specific cell types within this region.

Intraparenchymal delivery. Initial CNS gene therapy trials in humans have leveraged IPa administration to bypass the BBB and deliver genes directly into the brain region of interest (Fig. 3). The one-and-done feature of AAV gene therapy is essential given the required surgical procedure necessary for delivery. Most importantly, direct IPa administration of AAV vector to the CNS has been well tolerated49,120,121,166. Additional advantages of this route of administration include minimal biodistribution to peripheral organs, which can reduce immunogenicity, and substantially lower vector doses than required for other routes of administration.

Figure 3: Delivery of AAV gene therapy with intraparenchymal administration.
Figure 3

Before (part a) and after (part b) aromatic-L-amino-acid decarboxylase (AADC) gene transfer images using [18F]-6-fluoro-L-meta-tyrosine positron emission tomography (PET) as a measure of AADC enzyme activity from an early clinical trial using standard stereotactic delivery. Focal increases in enzyme activity (green arrow) are visible. Parts a and b are reproduced from Valles, F. et al. Qualitative imaging of adeno-associated virus serotype 2–human aromatic L-amino acid decarboxylase gene therapy in a phase I study for the treatment of Parkinson disease. Neurosurgery 2010, 67(5): 1377–1385, by permission of Oxford University Press (REF.228).. Subsequent developments in magnetic resonance guided imaging allow for greater physical coverage of the putamen and the potential for more widespread transgene expression. Coronal images showing progression of thalamic infusion (part c) and axial images showing progression of a putamen infusion, using a trajectory through the occipital lobes (part d), demonstrate diffuse coverage of the target structures evident in the magnetic resonance images acquired during infusion of a solution containing dilute magnetic resonance contrast in nonhuman primate putamen. Parts c and d adapted with permission from Journal of Controlled Release, 240, Bankiewicz, et al. AAV viral vector delivery to the brain by shape-conforming MR-guided infusions, 434–442, Copyright (2016), with permission from Elsevier (REF.229). AAV, adeno-associated viral.

IPa administration distributes drug directly to the brain region and neurons of interest. Substantially higher local levels of AAV vector and vector genomes can be achieved compared with IV67 or IT69 dosing in NHPs, both of which show relatively reduced levels of transgene delivery to CNS neurons owing to low penetration into the adult brain parenchyma. IPa delivery of AAV vector in multiple preclinical studies has demonstrated robust delivery of transgene to striatal medium spiny neurons and cortical pyramidal neurons following infusion into the putamen167 and thalamus168. Direct administration not only exposes putaminal and thalamic neurons and glia to AAV vectors, but also other regions of the brain through axonal transport via extensive topographic connections to the cortex and more distant sites, thereby achieving more widespread distribution of AAV vectors169,170,171. In NHPs, IPa delivery of AAVrh.10 encoding human arylsulfatase A has similarly demonstrated transgene expression and safety at therapeutic doses172. Importantly, IPa delivery has been used safely in clinical trials in patients with MPS IIIB with 16 intracerebral injections per subject of AAV5 encoding the transgene for human α-N-acetylglucosaminidase173.

Convection enhanced delivery (CED), which involves directly infusing molecules into the parenchyma under positive pressure, can be used to increase the distribution volume throughout the desired region of the brain174. A number of drugs, including AAV2, have not only been distributed to a substantially greater extent by CED IPa dosing than passive diffusion, but are also well tolerated when administered directly into the putamen of patients with advanced PD (NCT01973543).

The success of the IPa route of administration with CED is volume-dependent. A sufficient volume of infusate that contains the AAV vector needs to be administered to cover the anatomical region of interest. This volume dependency has been an important issue in AAV-based clinical trials in PD in which suboptimal volumes were delivered into the putamen, which is >4,000 mm3 in size in patients with PD. CED is based on pressure-induced temporal expansion of the extracellular space of the brain parenchyma that allows fluid containing AAV vector to fill a tissue volume on average two to three times the volume of the infusate. Thus, to cover 80% of the human putamen, at least 1,500 μl of fluid must be administered. The volume of infusion used in previous gene therapy trials in PD ranged from only 40 μl to 150 μl and therefore had no chance of providing gene expression beyond 10% of the putamen, most likely resulting in substantial underdosing. With the development of magnetic-resonance-guided AAV delivery methods and near-real-time monitoring of vector infusion in the putamen using a magnetic resonance imaging (MRI) tracer, much larger volumes are currently being investigated, ranging from 450 μl to 1,500 μl, which have resulted in substantially greater coverage of the putamen47.

The choice of capsid can also be used to tailor how focused or broadly the AAV vectors are distributed within the brain following IPa delivery. Several AAV vectors, including AAV2, AAV6 and AAV-DJ88, bind heparin sulfate proteoglycans (HSPGs) and, as a consequence, diffuse over smaller areas of the brain than AAV vectors lacking HSPG binding (for example, AAV1, AAV5, AAV8, AAV9 and rh.10), which results in expression over larger areas of the mouse brain169. For certain disease applications such as AAV2–AADC for PD, the restricted spread of AAV2 is highly desirable to limit the expression of AADC to the putamen. In an early clinical trial for Canavan disease, AAV2 carrying ASPA was injected into six white matter sites to increase the extent of transduction119. Although some evidence of efficacy was noted in this trial, the coverage provided by this multiple-injection strategy was probably inadequate for maximal therapeutic benefit. Recently, an AAV2 variant lacking the ability to bind HSPG has been described that results in greater IPa spread than AAV2 in rats175.

Intrathecal, intracerebroventricular and subpial delivery. Other routes of direct CNS administration include injections into the CSF compartment. IT33,67,68,70,71, IC68,69,70 and ICV11,12,16,70 dosing have been tested in multiple preclinical studies and have begun to be evaluated in the clinic. Subpial injections have also been studied, but only preclinically to date41; therefore, the translation of this dosing paradigm for AAV vectors remains to be seen. IT administration of AAV to the CNS has been well tolerated in multiple preclinical studies33,67,68,70,176 and is particularly useful for delivering AAV vectors to motor neurons and sensory neurons of the dorsal root ganglia. Previous studies have demonstrated that AAV vectors, including AAV9 and AAVrh.10, distribute primarily to motor neurons in the spinal cord after IT administration in NHPs. IT dosing regimens can be varied by adjusting the volume delivered and the rate of infusion as well as the site of infusion and concentration of vector. Distribution within the brain after CSF dosing of AAV vector in NHPs has a notable perivascular distribution and distribution to certain cell types such as Purkinje neurons of the cerebellum69. Compared with IPa dosing, IT or ICV dosing results in lower tissue levels of the vector genome and transgene and a more heterogeneous distribution. The limited clinical experience to date (Table 2) comprises a Batten disease trial with IT AAV9 to deliver a transgene encoding ceroid-lipofuscinosis neuronal protein 6 (CLN6; NCT02725580), a giant axonal neuropathy trial with IT AAV9 to deliver a transgene encoding gigaxonin (NCT02362438) and an expanded access study with IT AAVrh.10 to deliver an miRNA targeting SOD1 (R. Brown, personal communication). Total doses are generally higher than those used with IPa administration but considerably lower than those used with IV administration.

Intravenous delivery

In contrast to IPa, IT or ICV administration, IV administration of AAV vectors has the potential to transfer genes to the entire CNS non-invasively and with a more uniform distribution. The reports by Foust et al.72 and soon thereafter by Duque et al.73 that AAV9 can cross the BBB and transduce mouse neurons and glia in neonatal mice initiated a wave of interest in IV delivery of AAV gene therapy for CNS transduction. The finding was soon replicated and extended to other species67,74,75,76, and several other naturally occurring AAV vectors, including rh.10 and rh.8, were also found to transduce the neonatal177 as well as adult CNS178 after IV administration, although to a much more limited extent in adult than in neonatal animals.

Despite the promise of non-invasive gene delivery throughout the CNS via engineered AAV vectors, several notable aspects must be kept in mind. First, tropism differences across species may pose particular challenges for capsid engineering and selection for human therapeutic use. IV administration of self-complementary AAV9 led to transduction of mouse and NHP CNS, although the latter was less efficient and largely directed towards astrocytes75. Second, IV injection exposes the virus to potential antibody neutralization in subjects who have been pre-exposed to natural AAV infections. An estimated 90% of adult humans have been exposed to AAVs, and a smaller but substantial fraction harbour neutralizing antibodies against AAV capsids, which can have a profound negative impact on AAV vector transduction. Modifying capsid surface epitopes through capsid engineering may enable the identification of capsids that better evade neutralizing antibodies87,92,179. Third, IV administration typically requires higher total doses to achieve efficient transduction than IPa, IT or ICV administration; therefore, the scale of manufacturing capacity must be sufficient to support these greater material requirements. Fourth, IV administration results in broad and substantial vector distribution to multiple peripheral tissues and organs, which comes with the associated biological impact of the potential for immune responses against the capsid, nascent unpackaged genome and transgene that needs to be considered from a safety standpoint. In this regard, Wilson and colleagues have recently reported that high-dose IV AAV administration in NHPs with two capsids similar to AAV9 — PHP.B and AAVhu68 — caused life-threatening hepatotoxicity and internal haemorrhage105,180, although we have not observed such severe toxicity with similar capsids at comparable doses in adult NHPs (D.W.Y.S., unpublished data)107,108. The exact reason for these discrepant safety findings in NHPs will require more investigation. In addition, in juvenile NHPs and piglets, Wilson and colleagues reported that high-dose IV AAV administration with the AAVhu68 capsid and a transgene encoding human SMN1 caused axonopathy, dorsal root ganglion toxicity and ataxia180. More importantly, however, clinical trials of IV AAV9 encoding SMN have demonstrated safety as well as remarkable efficacy1. In our view, the recent progress in capsid engineering and manufacturing technologies (described below) has the potential to enable widespread gene delivery to the adult CNS via IV delivery and provide safe and effective AAV therapies for neurodegenerative diseases that affect multiple regions of the brain, such as Alzheimer disease or frontotemporal dementia.

Intramuscular delivery

IM delivery of AAV vectors may provide an alternative approach for delivering antibody therapeutics targeting the CNS. Local AAV transduction of muscle could provide a depot for transgene expression (that is, for sustained release of the antibody of interest), thus reducing the need for monthly IV administration as is the case with current passive immunization protocols44,45. Although the BBB still represents a considerable barrier, the major advantage of such an approach would be the durability of antibody expression following a single injection of AAV vector.


Successful AAV gene therapy for the treatment of neurological disorders requires a production platform that reproducibly provides vector that is of sufficient and comparable quality across multiple batches and is scalable for late-stage clinical development as well as commercialization. A major challenge for AAV gene therapies has been the scalability of any manufacturing process to meet these material demands. Traditional production systems in mammalian cells such as HEK293 cells result in high-quality vector but rather limited yields. By contrast, production of AAV vectors using baculovirus–Sf9 insect cells has been shown to increase vector yields by approximately 10-fold to 100-fold52. However, with some capsids, potency of baculovirus–Sf9-produced vectors may be lower than vectors produced in mammalian cells, but this can be successfully overcome by molecular engineering181. For CNS applications that require limited quantities of vector such as an intraparenchymally administered AAV vector, mammalian systems may suffice. However, for IV and possibly IT administered AAV vectors, it has been a major challenge for mammalian systems to provide the quantities needed for clinical development, let alone scale up for commercialization. Production systems that can be scaled to address these needs include the baculovirus–Sf9 system52, the recombinant HSV-based system182,183,184 and the adenovirus–HeLa cell system185,186,187.

Disease and target selection

There is a growing list of neurological disorders and targets for which AAV gene therapy offers potential therapeutic application. Table 1 shows a list of disorders and associated targets compiled on the basis of criteria that support successful translational work and drug development and that have a favourable benefit-to-risk profile. These criteria are similar to what would be required for any neurological drug development programme, but can be operationally defined for AAV gene therapy: strong target validation; the ability to safely achieve therapeutic tissue levels of vector at the target CNS location as well as transduction of the relevant cell type or types; an adequate therapeutic index for pharmacological effects and safety; high unmet medical need; biomarkers and clinical assessments that will allow for measurement of pharmacological activity and clinical outcomes; and the ability to identify patients and importantly intervene at an appropriate time in the course of disease progression.

Rare, monogenic disorders still dominate this development landscape. The most clinically advanced and successful programmes in neurological, ocular and systemic disorders are for monogenic disorders188. The success and safety profile of AAV gene therapy in these disorders has broadened interest in the range of potential targets. These include genetic risk factors such as the APOE allele for Alzheimer disease189 and pharmacological targets such as AADC for PD47,49,120. Targets without strong genetic, pharmacological or pathophysiological validation, such as trophic factors in PD, have shown less promise in clinical trials, although delivery challenges and the timing of intervention may have limited their potential efficacy190.


The potential to successfully deliver an AAV vector to the CNS is one of the most important questions in choosing a disorder and target to pursue with gene therapy. Direct delivery to discrete brain regions has markedly improved with the advent of MRI-guided imaging, offering the ability to tailor and optimize dosing in near real time for each subject. Cell therapy clinical trials also suggest the capability of safe delivery directly to the spinal cord191. Although direct delivery has upfront risk and costs, it circumvents the issues of pre-existing immunity and potential adverse effects related to systemic distribution and expression in non-target tissues. Broader CNS delivery may require CSF-based dosing using IT or ICV routes. IV dosing has been successful to date in infants with type I SMA in whom very modest levels of transgene expression are likely sufficient for therapeutic benefit1. Systemic dosing for CNS diseases in older children and adults needs to take into account the high doses required for gene transfer to the adult CNS and pre-existing immunity to AAV capsids192,193,194.


After the acute dosing period, the concept of therapeutic index in AAV gene therapy is a function of both the transgene and the mechanism of action of the vector. Most clinically advanced AAV gene therapy programmes involve gene replacement, and only modest increases of expression and functional protein may be adequate for clinical benefit1. Potential toxic effects from gene replacement may occur owing to expression in non-target tissues or overexpression of the therapeutic transgene within the targeted cells. For example, mutations in MECP2 (encoding methyl-CpG-binding protein 2) cause Rett syndrome in females, but duplications also cause a developmental disorder with intellectual impairment in boys195,196,197,198,199,200. MeCP2 overexpression in mouse models causes behavioural abnormalities, learning impairments and seizures201. A successful therapy for Rett syndrome would therefore require that transgene expression stays within a very narrow therapeutic window. Potential toxic effects related to gene silencing will also depend on the function of the endogenous protein, timing of the intervention and the level of expression of the miRNA and resultant silencing. HTT is involved in neurogenesis, and disruption of HTT results in embryonic lethality in mice202,203. However, silencing of both wild-type and mutant HTT by approximately 45% in the striatum of adult rhesus macaques appears safe and is not associated with neuronal pathology or motor deficits204.

Although AAV gene therapy has been well tolerated to date in a range of diseases, the inability to regulate expression remains an important safety consideration for target and disease selection. Additionally, immune responses and durable expression may preclude subjects from participating in other clinical trials after receiving gene transfer. Disease states with high unmet clinical need and a favourable benefit-to-risk profile therefore remain most appropriate for AAV gene therapy.

Clinical development challenges

The clinical development challenges for neurological disorders are not specific to AAV gene therapy. A relative lack of biomarkers or the ability to biopsy affected CNS tissue has traditionally presented a challenge in assessing target engagement and pharmacodynamic effects. Advances in CSF biomarkers and brain imaging such as positron emission tomography (PET) tracers for amyloid-β and tau have enabled earlier detection of disease and measurement of pharmacodynamic effects that may facilitate clinical trials. However, even relatively rapidly progressive neurological disorders of interest for gene therapy, such as monogenic forms of ALS, must still be evaluated clinically or with neurophysiological measures and may require ≥12 months to assess treatment effects205. The window to intervene is also uncertain in many neurological disorders. Adult-onset neurodegenerative disorders, in particular, may progress for decades preclinically (that is, without symptoms or clinical signs manifesting)206,207. Therefore, potential biological targets must be carefully vetted for the feasibility of clinical development, including the ability to identify subjects who can be treated early enough in the course of their disease for potential benefit from AAV therapy, and the availability of assessments for pharmacological and clinical effects.

Clinical trials

Trial design

Clinical trials with AAV gene therapy address most of the same drug development questions as trials with other treatment modalities: appropriate patient population, safety, dose selection, choice of end points and efficacy and ultimately overall benefit-to-risk ratio. AAV gene transfer clinical trials do not have a particular paradigm but have some common features shaped by the generally rare patient populations of interest, routes of delivery and expectations about the durability of expression and magnitude of the treatment effect. Key features include a dose range that is limited to doses that are expected to be pharmacologically active, because re-dosing may not be possible, and the ability to assess long-term safety and efficacy from the first subjects treated due to long-term expression of the therapeutic transgene.

Most completed AAV gene therapy trials in neurological disorders have had small sample sizes of <30 participants, with the exception of PD10. The small sample sizes are due to small incident and prevalent populations for many of the diseases of interest to date and the early stage of development of the interventions. Small sample sizes are also supported by expected large treatment effects with monogenic disorders as has been observed with type I SMA (Ref. 1). Many of the completed and ongoing trials are open-label, uncontrolled trials, which are appropriate for early trials in which safety and delivery methodology may be the key questions or for diseases with a uniform natural history or progression. These factors are not unique to gene therapy interventions. As the field progresses into larger populations with various functional, cognitive and behavioural outcomes, there may be a need for more rigorous, placebo-controlled trials. Careful consideration of the type of control group necessary for each individual clinical trial will be required.

The expected durability of gene therapy has also blurred the lines of the traditional phase I, II and III clinical trials. Because neurons in the adult brain are terminally differentiated, cell turnover, except through disease progression, should be minimal. In theory, subjects are thus continually 'treated' once dosed and, in practice, emerging clinical data support long-term, durable expression of transgenes. There is therefore little justification for a traditional single-ascending-dose or multiple-ascending-dose trial, and subjects treated in the first-in-human study will necessarily provide valuable long-term data on safety, pharmacology and efficacy. The evolving paradigm has an initial phase I study that assesses safety and dose selection over a limited range of vector doses followed by a trial or trials to address efficacy. The current inability to re-dose or clinically regulate expression has limited the initial dose range to doses that are expected to be pharmacologically active but well tolerated. A first-in-human trial therefore typically involves only two to three dose levels separated by half-log increments, and thus dose range must be defined by preclinical pharmacology, efficacy and toxicology studies.

Ongoing trials

Several AAV gene therapy clinical trials for neurological disorders are ongoing (Table 2; see website, accessed 5 June 2018), including trials in SMA, PD, MPS and Batten disease. Clinical trials in SMA and PD illustrate the range of patients, doses, routes of administration, end points and trial designs used, and some examples and experiences will therefore be discussed in detail below.

Type I SMA. Type I SMA is a progressive, fatal, monogenic motor neuron disease with an onset in infancy208. It manifests in failure to achieve motor milestones and results in mechanical ventilation and death by 2 years of age. In one recently completed trial (NCT02122952), 15 infants received high or low doses of AAV9 carrying self-complementary DNA encoding the SMN protein by IV administration1. The mean age was 6.3 months in the low-dose group and 3.4 months in the high-dose group. Transient increases in hepatic transaminases occurred in four subjects post-dose, which were attributed to immune responses to the AAV9 capsid. Prophylactic prednisolone treatment was used in subjects 2–15 after aminotransferase elevations were seen in the first subject. As of August 2017, all 15 patients in this study were alive and event-free at 20 months of age, exceeding the expected survival from a natural history study. Some dose-dependent improvements in motor function and motor milestones were observed, although the higher-dose group was also treated at a younger age of ≤6 months.

This trial illustrates the potential safety concerns associated with the high doses of vector that are needed with systemic delivery of AAV, approximately 1015 vg in infants. Increases in hepatic transaminases, also seen in haemophilia trials, were readily managed with prednisolone. The safety at higher doses required to treat adults or older children is unclear, and the therapeutic index for acute adverse effects may be capsid-serotype-specific. The effects of a higher dose and lower age in the high-dose cohort cannot be separated, but the data suggest not surprisingly that earlier treatment of SMA is important. The objective outcomes of mortality or need for ventilation, uniform natural history and apparent large treatment effect support the open-label design. However, a trial with an antisense oligonucleotide in the same population used a randomized controlled trial design209.

Parkinson disease. Unlike SMA, clinical trials with PD use functional and patient-reported outcomes and have generally been placebo controlled. The placebo procedure must be sufficient to blind participants, family members and study staff who may influence outcomes and ratings, but it should minimize risk to placebo participants, who do not receive brain cannula placements or infusions.

AADC is the final enzymatic step in the production of dopamine. Trials of AAV2 carrying single-stranded DNA encoding human AADC (AAV2.hAADC)49,120 with intraputaminal infusion represent a pharmacological approach to treating advanced PD using relatively small amounts of vector with targeted administration. Gene transfer occurs primarily into medium spiny neurons of the putamen, which does not normally produce dopamine. Patients can then regulate the amount of levodopa they take to control the amount of dopamine that is produced by this new depot that bypasses degenerating dopaminergic neuronal projections from the substantia nigra. In early dose escalation trials, AAV2.hAADC was administered using stereotactic infusions into each putamen. Several of the early patients in this trial experienced intracranial haemorrhages, which were clinically minor or asymptomatic. The procedure was changed to move the entry point more lateral, away from the central veins, with no subsequent bleeds. Importantly, the AAV2.hAADC vector itself was well tolerated with no vector-related safety events observed. No systemic adverse effects have been observed with intracranial administration of AAV2, which is thought to have limited systemic biodistribution. Dose-related, durable increases in AADC enzyme activity were measured over 5 years using the tracer [18F]6-fluoro-L-meta-tyrosine and PET, demonstrating that enzyme activity can be restored and maintained using AAV2.hAADC gene transfer49.

A more recent clinical trial (NCT03065192) has further optimized vector delivery using real-time MRI to guide the infusion, thereby maximizing the safety and coverage of the putamen and minimizing distribution to non-target brain areas. Vector is mixed with gadoluminate, an MRI contrast agent, for visualization of infusate distribution (Fig. 3c,d). This method has enabled a more than tenfold increase in the volume and total dose of vector administered to translate into a mean coverage of >50% of the putamen47, a further increase in AADC enzyme activity as measured by 18F-DOPA PET scans (Fig. 3a,b) and improved clinical outcomes.

PD trials with AAV gene therapy to deliver AADC and other therapeutic transgenes49,120,121,122 have illustrated the safety of intracranial administration of AAV, the low doses required for pharmacology and efficacy and the limited systemic exposure with this approach. The increasing experience and progressive evolution of intracranial infusion technology will support AAV dosing not only for PD but also for other neurological disorders.

Summary and prospects

AAV vector-based therapies are a rapidly emerging approach for potentially treating or even preventing neurological diseases. AAV gene therapy is particularly well suited for highly validated CNS targets, especially genetic targets that are not readily druggable with conventional therapies. Multiple examples of successful proof of concept for AAV gene therapy exist in animal models of disease for both gene replacement to address loss-of-function mutations and gene silencing to address gain-of-function mutations. Clinical trials are underway to evaluate AAV vectors for delivery of therapeutic proteins and miRNAs to the CNS. To date, both safety and long-lasting pharmacology (over multiple years) have been demonstrated in clinical studies for neurological diseases such as PD, AADC deficiency and SMA. The durability of AAV gene transfer to non-dividing cells such as neurons after a single administration provides an important advantage for more complex dosing paradigms in which repeat dosing may be inconvenient or not feasible. In our view, the emerging database of clinical safety and efficacy of AAV gene therapy for CNS disorders illustrates the remarkable potential of AAV gene therapy to treat neurodegenerative diseases.

The greatest immediate consideration for realizing the full potential of AAV gene therapy for neurological diseases is delivery, governed primarily by the capsid, which dictates the tissues and cell types transduced. For the CNS, IPa, IT, ICV, subpial and IV administration have shown promise in preclinical studies for achieving the targeted distribution, and all of these routes of administration except subpial delivery have progressed to clinical trials. Each of these routes of administration, together with the capsid, has particular attributes and considerations, including total dose requirements and the impact of pre-existing immunity on vector transduction and capsid or transgene immunogenicity. It is likely that no one route of administration will be ideal for all CNS applications, but rather that the preferred route of administration for a particular target and disease will be driven by achieving the required target biodistribution profile and minimizing exposure to non-target tissues. Clearly, these different routes of administration will be associated with very different drug requirements and therefore require different production and manufacturing processes. Recent progress in AAV manufacturing using both mammalian (HEK293) and insect (Sf9) cells has enabled the production of AAV gene therapies at commercial quality and scale for a range of applications, including those in which larger drug quantities are needed.

In summary, the recent progress towards realizing the remarkable potential of AAV gene therapy for neurological disorders has resulted in a tremendous increase in interest and effort in this area. As advances in optimizing AAV capsids, vector delivery and transgene design continue, the breadth of CNS disease applications and targets will undoubtedly expand.

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Change history

  • 12 September 2018

    Details related to the copyright permissions for the images shown in Figure 3 have been added to the figure legend.


  1. 1.

    et al. Single-dose gene-replacement therapy for spinal muscular atrophy. N. Engl. J. Med. 377, 1713–1722 (2017). This seminal clinical study demonstrates the safety and early efficacy of the first intravenously administered AAV-based therapy designed to correct the SMN deficiency in the CNS.

  2. 2.

    et al. Intravenous scAAV9 delivery of a codon-optimized SMN1 sequence rescues SMA mice. Hum. Mol. Genet. 20, 681–693 (2011).

  3. 3.

    et al. Rescue of the spinal muscular atrophy phenotype in a mouse model by early postnatal delivery of SMN. Nat. Biotechnol. 28, 271–274 (2010).

  4. 4.

    et al. Delayed disease onset and extended survival in the SOD1G93A rat model of amyotrophic lateral sclerosis after suppression of mutant SOD1 in the motor cortex. J. Neurosci. 34, 15587–15600 (2014).

  5. 5.

    et al. Therapeutic rAAVrh10 mediated SOD1 silencing in adult SOD1(G93A) mice and nonhuman primates. Hum. Gene Ther. 27, 19–31 (2016).

  6. 6.

    et al. Therapeutic AAV9-mediated suppression of mutant SOD1 slows disease progression and extends survival in models of inherited ALS. Mol. Ther. 21, 2148–2159 (2013).

  7. 7.

    et al. Artificial miRNAs mitigate shRNA-mediated toxicity in the brain: implications for the therapeutic development of RNAi. Proc. Natl Acad. Sci. USA 105, 5868–5873 (2008).

  8. 8.

    et al. Silencing mutant huntingtin by adeno-associated virus-mediated RNA interference ameliorates disease manifestations in the YAC128 mouse model of Huntington's disease. Hum. Gene Ther. 25, 461–474 (2014).

  9. 9.

    et al. Nonallele-specific silencing of mutant and wild-type huntingtin demonstrates therapeutic efficacy in Huntington's disease mice. Mol. Ther. 17, 1053–1063 (2009).

  10. 10.

    , , , & Adeno-associated virus-based gene therapy for CNS diseases. Hum. Gene Ther. 27, 478–496 (2016).

  11. 11.

    et al. AAV gene transfer delays disease onset in a TPP1-deficient canine model of the late infantile form of Batten disease. Sci. Transl Med. 7, 313ra180 (2015).

  12. 12.

    et al. Comparative efficacy and safety of multiple routes of direct CNS administration of adeno-associated virus gene transfer vector serotype rh.10 expressing the human arylsulfatase A cDNA to nonhuman primates. Hum. Gene Ther. Clin. Dev. 25, 164–177 (2014).

  13. 13.

    , , , & Correction of neurological disease of mucopolysaccharidosis IIIB in adult mice by rAAV9 trans-blood-brain barrier gene delivery. Mol. Ther. 19, 1025–1033 (2011).

  14. 14.

    et al. Systemic AAV9 gene transfer in adult GM1 gangliosidosis mice reduces lysosomal storage in CNS and extends lifespan. Hum. Mol. Genet. 24, 4353–4364 (2015).

  15. 15.

    , , & Development of intrathecal AAV9 gene therapy for giant axonal neuropathy. Mol. Ther. Methods Clin. Dev. 9, 160–171 (2018).

  16. 16.

    et al. Whole body correction of mucopolysaccharidosis IIIA by intracerebrospinal fluid gene therapy. J. Clin. Invest. 123, 3254–3271 (2013).

  17. 17.

    et al. Systemic delivery of MeCP2 rescues behavioral and cellular deficits in female mouse models of Rett syndrome. J. Neurosci. 33, 13612–13620 (2013).

  18. 18.

    et al. Improved survival and reduced phenotypic severity following AAV9/MECP2 gene transfer to neonatal and juvenile male Mecp2 knockout mice. Mol. Ther. 21, 18–30 (2013).

  19. 19.

    et al. Development of a novel AAV gene therapy cassette with improved safety features and efficacy in a mouse model of Rett syndrome. Mol. Ther. Methods Clin. Dev. 5, 180–190 (2017).

  20. 20.

    et al. Systemic delivery of scAAV9 expressing SMN prolongs survival in a model of spinal muscular atrophy. Sci. Transl Med. 2, 35ra42 (2010).

  21. 21.

    et al. An AAV9 coding for frataxin clearly improved the symptoms and prolonged the life of Friedreich ataxia mouse models. Mol. Ther. Methods Clin. Dev. 1, 14044 (2014).

  22. 22.

    et al. Vectored intracerebral immunization with the anti-tau monoclonal antibody PHF1 markedly reduces tau pathology in mutant tau transgenic mice. J. Neurosci. 36, 12425–12435 (2016).

  23. 23.

    et al. AAV-mediated expression of anti-tau scFvs decreases tau accumulation in a mouse model of tauopathy. J. Exp. Med. 214, 1227–1238 (2017).

  24. 24.

    et al. In vivo interrogation of gene function in the mammalian brain using CRISPR-Cas9. Nat. Biotechnol. 33, 102–106 (2015).

  25. 25.

    et al. In vivo genome editing via CRISPR/Cas9 mediated homology-independent targeted integration. Nature 540, 144–149 (2016).

  26. 26.

    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 97, 3428–3432 (2000).

  27. 27.

    , , , & AAV8, 9, Rh10, Rh43 vector gene transfer in the rat brain: effects of serotype, promoter and purification method. Mol. Ther. 16, 89–96 (2008).

  28. 28.

    , , & Corticospinal tract transduction: a comparison of seven adeno-associated viral vector serotypes and a non-integrating lentiviral vector. Gene Ther. 19, 49–60 (2012).

  29. 29.

    , & Selective and rapid uptake of adeno-associated virus type 2 in brain. Hum. Gene Ther. 9, 1181–1186 (1998).

  30. 30.

    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. 10, 302–317 (2004).

  31. 31.

    Intracranial delivery of CLN2 reduces brain pathology in a mouse model of classical late infantile neuronal ceroid lipofuscinosis. J. Neurosci. 26, 1334–1342 (2006).

  32. 32.

    et al. Intraventricular brain injection of adeno-associated virus type 1 (AAV1) in neonatal mice results in complementary patterns of neuronal transduction to AAV2 and total long-term correction of storage lesions in the brains of β-glucuronidase-deficient mice. J. Virol. 77, 7034 (2003).

  33. 33.

    et al. Robust spinal motor neuron transduction following intrathecal delivery of AAV9 in pigs. Gene Ther. 19, 852–859 (2012).

  34. 34.

    et al. Expanded repertoire of AAV vector serotypes mediate unique patterns of transduction in mouse brain. Mol. Ther. 16, 1710–1718 (2008).

  35. 35.

    , , & Adeno-associated virus type 4 (AAV4) targets ependyma and astrocytes in the subventricular zone and RMS. Gene Ther. 12, 1503–1508 (2005).

  36. 36.

    et al. Comparative analysis of adeno-associated viral vector serotypes 1, 2, 5, 7, and 8 in mouse brain. Hum. Gene Ther. 18, 195–206 (2007).

  37. 37.

    et al. Strong cortical and spinal cord transduction after AAV7 and AAV9 delivery into the CSF of non-human primates. Hum. Gene Ther. 24, 526–532 (2013).

  38. 38.

    , , & Adeno-associated virus vector-mediated transduction in the cat brain. Gene Ther. 10, 1874–1881 (2003).

  39. 39.

    et al. T2 imaging in monitoring of intraparenchymal real-time convection-enhanced delivery. Neurosurgery 69, 154–163; discussion 163 (2011).

  40. 40.

    et al. Translational fidelity of intrathecal delivery of self-complementary AAV9-survival motor neuron 1 for spinal muscular atrophy. Hum. Gene Ther. 25, 619–630 (2014).

  41. 41.

    et al. Potent spinal parenchymal AAV9-mediated gene delivery by subpial injection in adult rats and pigs. Mol. Ther. Methods Clin. Dev. 3, 16046 (2016).

  42. 42.

    et al. Adeno-associated viral gene therapy using PHP.B:NPC1 ameliorates disease phenotype in mouse model of Niemann-Pick C1 disease. Mol. Ther. 25, 1–363 (2017).

  43. 43.

    et al. AAV-PHP.B-mediated global-scale expression in the mouse nervous system enables GBA1 gene therapy for wide protection from synucleinopathy. Mol. Ther. 25, 2727–2742 (2017).

  44. 44.

    et al. Antibody-based protection against HIV infection by vectored immunoprophylaxis. Nature 481, 81–84 (2011).

  45. 45.

    , , , & Broad protection against influenza infection by vectored immunoprophylaxis in mice. Nat. Biotechnol. 31, 647–652 (2013).

  46. 46.

    et al. Cre-dependent selection yields AAV variants for widespread gene transfer to the adult brain. Nat. Biotechnol. 34, 204–209 (2016). A novel, cell-type-specific, functional in vivo selection system enables the identification of several engineered AAV capsids that provide up to 40-fold greater CNS transduction than the previous standard, AAV9, after IV administration in adult mice.

  47. 47.

    et al. Intraputaminal AADC gene therapy for advanced Parkinson's disease: interim results of a phase 1b Trial [abstract]. Hum. Gene Ther. 28, OR12 (2017).

  48. 48.

    et al. Gene therapy for aromatic L-amino acid decarboxylase deficiency. Sci. Transl Med. 4, 134ra61 (2012).

  49. 49.

    et al. Long-term evaluation of a phase 1 study of AADC gene therapy for Parkinson's disease. Hum. Gene Ther. 23, 377–381 (2012).

  50. 50.

    et al. Persistent expression of dopamine-synthesizing enzymes 15 years after gene transfer in a primate model of Parkinson's disease. Hum. Gene Ther. Clin. Dev. 28, 74–79 (2017).

  51. 51.

    , & Production of recombinant adeno-associated virus vectors using suspension HEK293 cells and continuous harvest of vector from the culture media for GMP FIX and FLT1 clinical vector. Mol. Ther. 24, 287–297 (2016).

  52. 52.

    & Manufacturing clinical grade recombinant adeno-associated virus using invertebrate cell lines. Hum. Gene Ther. 28, 350–360 (2017).

  53. 53.

    et al. Hematopoietic stem cell gene therapy with a lentiviral vector in X-linked adrenoleukodystrophy. Science 326, 818–823 (2009).

  54. 54.

    et al. Lentiviral hematopoietic cell gene therapy for X-linked adrenoleukodystrophy. Methods Enzymol. 507, 187–198 (2012).

  55. 55.

    et al. High-efficiency transduction of the mouse retina by tyrosine-mutant AAV serotype vectors. Mol. Ther. 17, 463–471 (2009).

  56. 56.

    et al. Selection and evaluation of clinically relevant AAV variants in a xenograft liver model. Nature 506, 382–386 (2014).

  57. 57.

    , & Adenovirus-associated defective virus particles. Science 149, 754–756 (1965). The first identification of small viruses now known as AAVs as contaminants of adenovirus preparations.

  58. 58.

    , & Studies of small DNA viruses found in various adenovirus preparations: physical, biological, and immunological characteristics. Proc. Natl Acad. Sci. USA 55, 1467–1474 (1966).

  59. 59.

    , , & Physicochemical characterization of adeno-associated satellite virus type 4 and its nucleic acid. J. Virol. 1, 980–987 (1967).

  60. 60.

    , , , & Cloning of adeno-associated virus type 4 (AAV4) and generation of recombinant AAV4 particles. J. Virol. 71, 6823–6833 (1997).

  61. 61.

    , , , & Adeno-associated virus (AAV)-3-based vectors transduce haematopoietic cells not susceptible to transduction with AAV-2-based vectors. J. Gen. Virol. 81, 2077–2084 (2000).

  62. 62.

    et al. Clades of adeno-associated viruses are widely disseminated in human tissues. J. Virol. 78, 6381–6388 (2004).

  63. 63.

    et al. Adeno-associated viruses undergo substantial evolution in primates during natural infections. Proc. Natl Acad. Sci. USA 100, 6081–6086 (2003).

  64. 64.

    et al. Identification of mouse AAV capsid-specific CD8+ T cell epitopes. Mol. Ther. 12, 1023–1033 (2005).

  65. 65.

    et al. Novel caprine adeno-associated virus (AAV) capsid (AAV-Go.1) is closely related to the primate AAV-5 and has unique tropism and neutralization properties. J. Virol. 79, 15238–15245 (2005).

  66. 66.

    et al. Comparative transduction efficiency of AAV vector serotypes 1–6 in the substantia nigra and striatum of the primate brain. Mol. Ther. 18, 588–593 (2010).

  67. 67.

    et al. Systemic gene delivery in large species for targeting spinal cord, brain, and peripheral tissues for pediatric disorders. Mol. Ther. 19, 1971–1980 (2011).

  68. 68.

    , , & Global CNS gene delivery and evasion of anti-AAV-neutralizing antibodies by intrathecal AAV administration in non-human primates. Gene Ther. 20, 450–459 (2013).

  69. 69.

    et al. Adeno-associated virus serotype 9 transduction in the central nervous system of nonhuman primates. Hum. Gene Ther. 23, 382–389 (2012).

  70. 70.

    et al. Evaluation of intrathecal routes of administration for adeno-associated viral vectors in large animals. Hum. Gene Ther. 29, 15–24 (2018).

  71. 71.

    et al. Widespread gene transfer in the central nervous system of cynomolgus macaques following delivery of AAV9 into the cisterna magna. Mol. Ther. Methods Clin. Dev. 1, 14051 (2014).

  72. 72.

    et al. Intravascular AAV9 preferentially targets neonatal neurons and adult astrocytes. Nat. Biotechnol. 27, 59–65 (2009). This report is one of two studies to first demonstrate that AAV vectors can cross the BBB and provide widespread gene expression throughout the neonatal and adult CNS; this study focuses on the mouse.

  73. 73.

    et al. Intravenous administration of self-complementary AAV9 enables transgene delivery to adult motor neurons. Mol. Ther. 17, 1187–1196 (2009). This report is one of two studies to first demonstrate that AAV vectors can cross the BBB and provide widespread gene expression throughout the neonatal and adult CNS; this study includes the cat.

  74. 74.

    et al. Expansive gene transfer in the rat CNS rapidly produces amyotrophic lateral sclerosis relevant sequelae when TDP-43 is overexpressed. Mol. Ther. 18, 2064–2074 (2010).

  75. 75.

    et al. Preclinical differences of intravascular AAV9 delivery to neurons and glia: a comparative study of adult mice and nonhuman primates. Mol. Ther. 19, 1058–1069 (2011). This study highlights the species-specific tropism differences between mouse and NHPs that may complicate translational efforts of engineered AAV capsids.

  76. 76.

    et al. AAV9 transduction in the central nervous system of non-human primates. Hum. Gene Ther. 23, 382–389 (2012).

  77. 77.

    , & Parvovirus glycan interactions. Curr. Opin. Virol. 7, 108–118 (2014).

  78. 78.

    et al. Local and retrograde gene transfer into primate neuronal pathways via adeno-associated virus serotype 8 and 9. Neuroscience 193, 249–258 (2011).

  79. 79.

    et al. Characterization of the adeno-associated virus 1 and 6 sialic acid binding site. J. Virol. 90, 5219–5230 (2016).

  80. 80.

    et al. Identification of a heparin-binding motif on adeno-associated virus type 2 capsids. J. Virol. 77, 11072–11081 (2003).

  81. 81.

    et al. Identification of amino acid residues in the capsid proteins of adeno-associated virus type 2 that contribute to heparan sulfate proteoglycan binding. J. Virol. 77, 6995–7006 (2003).

  82. 82.

    et al. Single amino acid changes can influence titer, heparin binding, and tissue tropism in different adeno-associated virus serotypes. J. Virol. 80, 11393–11397 (2006).

  83. 83.

    , , , & Terminal N-linked galactose is the primary receptor for adeno-associated virus 9. J. Biol. Chem. 286, 13532–13540 (2011).

  84. 84.

    , , , & Identification of the galactose binding domain of the AAV9 capsid. J. Virol. 86, 7326–7333 (2012).

  85. 85.

    , , & Unique glycan signatures regulate adeno-associated virus tropism in the developing brain. J. Virol. 89, 3976–3987 (2015).

  86. 86.

    , , , & Drawing a high-resolution functional map of adeno-associated virus capsid by massively parallel sequencing. Nat. Commun. 5, 3075 (2014). In this tour de force study, the authors use alanine-scanning mutagenesis and AAV genome barcoding to map onto the linear AAV capsid sequence key functional features, including glycan binding, circulating half-life and liver and other organ transduction.

  87. 87.

    et al. Mapping the structural determinants required for AAVrh.10 transport across the blood-brain barrier. Mol. Ther. 26, 510–523 (2018).

  88. 88.

    et al. In vitro and in vivo gene therapy vector evolution via multispecies interbreeding and retargeting of adeno-associated viruses. J. Virol. 82, 5887–5911 (2008).

  89. 89.

    et al. Engineering and selection of shuffled AAV genomes: a new strategy for producing targeted biological nanoparticles. Mol. Ther. 16, 1252–1260 (2008).

  90. 90.

    et al. Combinatorial engineering of a gene therapy vector: directed evolution of adeno-associated virus. J. Gene Med. 8, 155–162 (2006).

  91. 91.

    et al. Directed evolution of adeno-associated virus to an infectious respiratory virus. Proc. Natl Acad. Sci. USA 106, 3865–3870 (2009).

  92. 92.

    , , & Directed evolution of adeno-associated virus yields enhanced gene delivery vectors. Nat. Biotechnol. 24, 198–204 (2006).

  93. 93.

    et al. Random peptide libraries displayed on adeno-associated virus to select for targeted gene therapy vectors. Nat. Biotechnol. 21, 1040–1046 (2003).

  94. 94.

    et al. A designer AAV variant permits efficient retrograde access to projection neurons. Neuron 92, 372–382 (2016).

  95. 95.

    et al. In vivo-directed evolution of a new adeno-associated virus for therapeutic outer retinal gene delivery from the vitreous. Sci. Transl Med. 5, 189ra76 (2013).

  96. 96.

    et al. A brain microvasculature endothelial cell-specific viral vector with the potential to treat neurovascular and neurological diseases. EMBO Mol. Med. 8, 609–625 (2016).

  97. 97.

    , , & Construction of diverse adeno-associated viral libraries for directed evolution of enhanced gene delivery vehicles. Nat. Protoc. 1, 701–706 (2006).

  98. 98.

    et al. Engineering liver-detargeted AAV9 vectors for cardiac and musculoskeletal gene transfer. Mol. Ther. 19, 1070–1078 (2011).

  99. 99.

    et al. In vivo selection of a computationally designed SCHEMA AAV library yields a novel variant for infection of adult neural stem cells in the SVZ. Mol. Ther. 26, 304–319 (2018).

  100. 100.

    et al. Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nat. Neurosci. 20, 1172–1179 (2017).

  101. 101.

    , , & Therapeutic delivery of butyrylcholinesterase by brain-wide viral gene transfer to mice. Molecules 22, E1145 (2017).

  102. 102.

    et al. Global representations of goal-directed behavior in distinct cell types of mouse neocortex. Neuron 94, 891–907.e6 (2017).

  103. 103.

    , , & Better targeting, better efficiency for wide-scale neuronal transduction with the synapsin promoter and AAV-PHP.B. Front. Mol. Neurosci. 9, 116 (2016).

  104. 104.

    et al. Intravenous administration of the adeno-associated virus-PHP.B capsid fails to upregulate transduction efficiency in the marmoset brain. Neurosci. Lett. 665, 182–188 (2018).

  105. 105.

    et al. The neurotropic properties of AAV-PHP.B are limited to C57BL/6J mice. Mol. Ther. 26, 664–668 (2018).

  106. 106.

    et al. AAV9-mediated expression of a non-self protein in nonhuman primate central nervous system triggers widespread neuroinflammation driven by antigen-presenting cell transduction. Mol. Ther. 22, 329–337 (2014). This study highlights the risk of expressing the GFP transgene in the CNS with inflammation and immune responses as well as ataxia after direct CNS delivery in NHPs.

  107. 107.

    et al. Translation of intravenous delivery of AAV gene therapy for the treatment of CNS diseases [abstract]. Hum. Gene Ther. 28, P209 (2017).

  108. 108.

    et al. Safety and increased transduction efficiency in the adult nonhuman primate central nervous system with intravenous delivery of two novel adeno-associated virus capsids [abstract 661]. Mol. Ther. (2018).

  109. 109.

    , , & Second-strand synthesis is a rate-limiting step for efficient transduction by recombinant adeno-associated virus vectors. J. Virol. 70, 3227–3234 (1996).

  110. 110.

    , , & Sialic acid deposition impairs the utility of AAV9, but not peptide-modified AAVs for brain gene therapy in a mouse model of lysosomal storage disease. Mol. Ther. 20, 1393–1399 (2012).

  111. 111.

    , , , & Efficient gene delivery and selective transduction of glial cells in the mammalian brain by AAV serotypes isolated from nonhuman primates. Mol. Ther. 17, 1692–1702 (2009).

  112. 112.

    et al. Glial promoter selectivity following AAV-delivery to the immature brain. PLoS ONE 8, e65646 (2013).

  113. 113.

    et al. Intracerebroventricular injection of adeno-associated virus 6 and 9 vectors for cell type-specific transgene expression in the spinal cord. Hum. Gene Ther. 25, 109–120 (2014).

  114. 114.

    et al. Recombinant adeno-associated virus purification using novel methods improves infectious titer and yield. Gene Ther. 6, 973–985 (1999). This paper describes a simple and universally applicable purification protocol for AAV comprising an iodixanol step gradient, which has become a commonly used alternative to caesium chloride gradients.

  115. 115.

    et al. Prevention and reversal of severe mitochondrial cardiomyopathy by gene therapy in a mouse model of Friedreich's ataxia. Nat. Med. 20, 542–547 (2014).

  116. 116.

    et al. Design, characterization, and lead selection of therapeutic miRNAs targeting huntingtin for development of gene therapy for Huntington's disease. Mol. Ther. Nucleic Acids 5, e297 (2016).

  117. 117.

    , , & CRISPR/Cas9 editing of the mutant Huntingtin allele in vitro and in vivo. Mol. Ther. 25, 12–23 (2017).

  118. 118.

    et al. Immune responses to AAV in a phase I study for Canavan disease. J. Gene Med. 8, 577–588 (2006).

  119. 119.

    et al. Long-term follow-up after gene therapy for canavan disease. Sci. Transl Med. 4, 165ra163 (2012).

  120. 120.

    et al. Results from a phase I safety trial of hAADC gene therapy for Parkinson disease. Neurology 70, 1980–1983 (2008). This article presents the initial results from one of the first CNS-targeted gene therapy trials and highlights the safety of IPa AAV delivery and AADC transgene expression in patients with PD.

  121. 121.

    et al. AAV2-GAD gene therapy for advanced Parkinson's disease: a double-blind, sham-surgery controlled, randomised trial. Lancet Neurol. 10, 309–319 (2011).

  122. 122.

    et al. Issues regarding gene therapy products for Parkinson's disease: the development of CERE-120 (AAV-NTN) as one reference point. Parkinsonism Relat. Disord. 13 (Suppl. 3), S469–S477 (2007).

  123. 123.

    et al. Intranasal antibody gene transfer in mice and ferrets elicits broad protection against pandemic influenza. Sci. Transl Med. 5, 187ra72 (2013).

  124. 124.

    et al. Genome editing abrogates angiogenesis in vivo. Nat. Commun. 8, 112 (2017).

  125. 125.

    et al. Nrl knockdown by AAV-delivered CRISPR/Cas9 prevents retinal degeneration in mice. Nat. Commun. 8, 14716 (2017).

  126. 126.

    et al. Dose-dependent prevention of metabolic and neurologic disease in murine MPS II by ZFN-mediated in vivo genome editing. Mol. Ther. 26, 1127–1136 (2018).

  127. 127.

    et al. Designed zinc finger protein transcription factors for single-gene regulation throughout the central nervous system [abstract 949]. Mol. Ther. (2018).

  128. 128.

    et al. Identification of pre-existing adaptive immunity to Cas9 proteins in humans. Preprint at bioRxiv (2018).

  129. 129.

    Anti-amyloid-beta monoclonal antibodies for Alzheimer's disease: pitfalls and promise. Biol. Psychiatry 83, 311–319 (2018).

  130. 130.

    Therapeutic antibodies for brain disorders. Sci. Transl Med. 3, 84ps20 (2011).

  131. 131.

    et al. Highly efficient Cas9-mediated transcriptional programming. Nat. Methods 12, 326–328 (2015).

  132. 132.

    et al. Epigenome editing by a CRISPR-Cas9-based acetyltransferase activates genes from promoters and enhancers. Nat. Biotechnol. 33, 510–517 (2015).

  133. 133.

    et al. Naturally occurring off-switches for CRISPR-Cas9. Cell 167, 1829–1838.e9 (2016).

  134. 134.

    et al. High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects. Nature 529, 490–495 (2016).

  135. 135.

    et al. Multiplex gene editing by CRISPR-Cpf1 using a single crRNA array. Nat. Biotechnol. 35, 31–34 (2017).

  136. 136.

    et al. Enhanced proofreading governs CRISPR-Cas9 targeting accuracy. Nature 550, 407–410 (2017).

  137. 137.

    et al. Engineered Cpf1 variants with altered PAM specificities. Nat. Biotechnol. 35, 789–792 (2017).

  138. 138.

    et al. RNA editing with CRISPR-Cas13. Science 358, 1019–1027 (2017).

  139. 139.

    et al. C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector. Science 353, aaf5573 (2016).

  140. 140.

    et al. Base editing with a Cpf1-cytidine deaminase fusion. Nat. Biotechnol. 36, 324–327 (2018).

  141. 141.

    , , , & Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533, 420–424 (2016).

  142. 142.

    et al. Programmable base editing of A*T to G*C in genomic DNA without DNA cleavage. Nature 551, 464–471 (2017).

  143. 143.

    et al. Update on phase 1/2 clinical trials for MPS I and MPS II using ZFN-mediated in vivo genome editing. Mol. Genet. Metab. 123, S59–S60 (2018).

  144. 144.

    RNAi the natural way. Nat. Genet. 37, 1163–1165 (2005).

  145. 145.

    , , , & RNAi or overexpression: alternative therapies for spinocerebellar ataxia type 1. Neurobiol. Dis. 56, 6–13 (2013).

  146. 146.

    et al. Silencing mutant ATXN3 expression resolves molecular phenotypes in SCA3 transgenic mice. Mol. Ther. 21, 1909–1918 (2013).

  147. 147.

    et al. Intracerebral administration of adeno-associated viral vector serotype rh.10 carrying human SGSH and SUMF1 cDNAs in children with mucopolysaccharidosis type IIIA disease: results of a phase I/II trial. Hum. Gene Ther. 25, 506–516 (2014).

  148. 148.

    , , & GFAP promoter elements required for region-specific and astrocyte-specific expression. Glia 56, 481–493 (2008).

  149. 149.

    et al. A viral strategy for targeting and manipulating interneurons across vertebrate species. Nat. Neurosci. 19, 1743–1749 (2016).

  150. 150.

    et al. rAAV-compatible MiniPromoters for restricted expression in the brain and eye. Mol. Brain 9, 52 (2016).

  151. 151.

    et al. Optimizing promoters for recombinant adeno-associated virus-mediated gene expression in the peripheral and central nervous system using self-complementary vectors. Hum. Gene Ther. 22, 1143–1153 (2011).

  152. 152.

    , , & Generation of a synthetic mammalian promoter library by modification of sequences spacing transcription factor binding sites. Gene 297, 21–32 (2002).

  153. 153.

    et al. Reduction of arthritis following intra-articular administration of an adeno-associated virus serotype 5 expressing a disease-inducible TNF-blocking agent. Ann. Rheum. Dis. 66, 1143–1150 (2007).

  154. 154.

    et al. Application of a disease-regulated promoter is a safer mode of local IL-4 gene therapy for arthritis. Gene Ther. 14, 1632–1638 (2007).

  155. 155.

    et al. An adeno-associated virus-based intracellular sensor of pathological nuclear factor-kappaB activation for disease-inducible gene transfer. PLoS ONE 8, e53156 (2013).

  156. 156.

    et al. Intracisternal delivery of NFkappaB-inducible scAAV2/9 reveals locoregional neuroinflammation induced by systemic kainic acid treatment. Front. Mol. Neurosci. 7, 92 (2014).

  157. 157.

    et al. Adeno-associated viral vector-mediated hypoxia-regulated VEGF gene transfer promotes angiogenesis following focal cerebral ischemia in mice. Gene Ther. 15, 30–39 (2008).

  158. 158.

    & Intervening sequences increase efficiency of RNA 3′ processing and accumulation of cytoplasmic RNA. Nucleic Acids Res. 18, 937–947 (1990).

  159. 159.

    et al. Interruption of coding sequences by heterologous introns can enhance the functional expression of recombinant genes. Gene Ther. 8, 649–653 (2001).

  160. 160.

    et al. Optimization of AAV expression cassettes to improve packaging capacity and transgene expression in neurons. Mol. Brain 7, 17 (2014).

  161. 161.

    et al. MicroRNA-regulated, systemically delivered rAAV9: a step closer to CNS-restricted transgene expression. Mol. Ther. 19, 526–535 (2010).

  162. 162.

    et al. Mir-142-3p target sequences reduce transgene-directed immunogenicity following intramuscular adeno-associated virus 1 vector-mediated gene delivery. J. Gene Med. 15, 219–232 (2013).

  163. 163.

    , & Tuning gene expression with synthetic upstream open reading frames. Proc. Natl Acad. Sci. USA 110, 11284–11289 (2013).

  164. 164.

    et al. Codon optimization of human factor VIII cDNAs leads to high-level expression. Blood 117, 798–807 (2011).

  165. 165.

    et al. Analysis of factor VIII mediated suppression of lentiviral vector titres. Gene Ther. 15, 289–297 (2008).

  166. 166.

    , & Adeno-associated virus (AAV) gene therapy for neurological disease. Neuropharmacology 69, 82–88 (2013).

  167. 167.

    et al. Widespread AAV1- and AAV2-mediated transgene expression in the nonhuman primate brain: implications for Huntington's disease. Mol. Ther. Methods Clin. Dev. 3, 16037 (2016).

  168. 168.

    et al. Efficient gene therapy-based method for the delivery of therapeutics to primate cortex. Proc. Natl Acad. Sci. USA 106, 2407–2411 (2009).

  169. 169.

    & A single injection of an adeno-associated virus vector into nuclei with divergent connections results in widespread vector distribution in the brain and global correction of a neurogenetic disease. J. Neurosci. 27, 9928–9940 (2007).

  170. 170.

    , , , & Adeno-associated virus serotypes 1, 8, and 9 share conserved mechanisms for anterograde and retrograde axonal transport. Hum. Gene Ther. 25, 705–720 (2014).

  171. 171.

    et al. Axonal transport of AAV9 in nonhuman primate brain. Gene Ther. 23, 520–526 (2016).

  172. 172.

    et al. Intracerebral gene therapy using AAVrh.10-hARSA recombinant vector to treat patients with early-onset forms of metachromatic leukodystrophy: preclinical feasibility and safety assessments in nonhuman primates. Hum. Gene Ther. Clin. Dev. 26, 113–124 (2015).

  173. 173.

    et al. Intracerebral gene therapy in children with mucopolysaccharidosis type IIIB syndrome: an uncontrolled phase 1/2 clinical trial. Lancet Neurol. 16, 712–720 (2017).

  174. 174.

    , , , & Convection-enhanced distribution of large molecules in gray matter during interstitial drug infusion. J. Neurosurg. 82, 1021–1029 (1995).

  175. 175.

    et al. Rationally engineered AAV capsids improve transduction and volumetric spread in the CNS. Mol. Ther. Nucleic Acids 8, 184–197 (2017).

  176. 176.

    et al. Intrathecal gene therapy corrects CNS pathology in a feline model of mucopolysaccharidosis I. Mol. Ther. 22, 2018–2027 (2014).

  177. 177.

    et al. Several rAAV vectors efficiently cross the blood–brain barrier and transduce neurons and astrocytes in the neonatal mouse central nervous system. Mol. Ther. 19, 1440–1448 (2011).

  178. 178.

    et al. Global CNS transduction of adult mice by intravenously delivered rAAVrh.8 and rAAVrh.10 and nonhuman primates by rAAVrh.10. Mol. Ther. 22, 1299–1309 (2014).

  179. 179.

    et al. Novel random peptide libraries displayed on AAV serotype 9 for selection of endothelial cell-directed gene transfer vectors. Gene Ther. 19, 800–809 (2012).

  180. 180.

    et al. Severe toxicity in nonhuman primates and piglets following high-dose intravenous administration of an adeno-associated virus vector expressing human SMN. Hum. Gene Ther. 29, 285–298 (2018).

  181. 181.

    et al. Direct head-to-head evaluation of recombinant adeno-associated viral vectors manufactured in human versus insect cells. Mol. Ther. 25, 2661–2675 (2017).

  182. 182.

    et al. A scalable method for the production of high-titer and high-quality adeno-associated type 9 vectors using the HSV platform. Mol. Ther. Methods Clin. Dev. 3, 16031 (2016).

  183. 183.

    et al. High-titer recombinant adeno-associated virus production utilizing a recombinant herpes simplex virus type I vector expressing AAV-2 Rep and Cap. Gene Ther. 6, 986–993 (1999).

  184. 184.

    , & Large-scale adeno-associated viral vector production using a herpesvirus-based system enables manufacturing for clinical studies. Hum. Gene Ther. 20, 796–806 (2009).

  185. 185.

    , , , & Selective Rep-Cap gene amplification as a mechanism for high-titer recombinant AAV production from stable cell lines. Mol. Ther. 2, 394–403 (2000).

  186. 186.

    , & A stable cell line carrying adenovirus-inducible rep and cap genes allows for infectivity titration of adeno-associated virus vectors. Gene Ther. 3, 1124–1132 (1996).

  187. 187.

    , , & Cell lines for the production of recombinant adeno-associated virus. Hum. Gene Ther. 6, 1329–1341 (1995).

  188. 188.

    et al. Gene therapy comes of age. Science 359, eaan4672 (2018).

  189. 189.

    et al. Intracerebral adeno-associated virus gene delivery of apolipoprotein E2 markedly reduces brain amyloid pathology in Alzheimer's disease mouse models. Neurobiol. Aging 44, 159–172 (2016).

  190. 190.

    & Clinical tests of neurotrophic factors for human neurodegenerative diseases, part 1: where have we been and what have we learned? Neurobiol. Dis. 97, 156–168 (2017).

  191. 191.

    et al. Transplantation of spinal cord-derived neural stem cells for ALS: analysis of phase 1 and 2 trials. Neurology 87, 392–400 (2016).

  192. 192.

    et al. Neutralizing antibodies against adeno-associated virus examined prospectively in pediatric patients with hemophilia. Gene Ther. 19, 288–294 (2012).

  193. 193.

    et al. Prevalence of neutralizing antibodies against adeno-associated virus (AAV) types 2, 5, and 6 in cystic fibrosis and normal populations: implications for gene therapy using AAV vectors. Hum. Gene Ther. 17, 440–447 (2006).

  194. 194.

    et al. Adeno-associated virus antibody profiles in newborns, children, and adolescents. Clin. Vaccine Immunol. 18, 1586–1588 (2011).

  195. 195.

    et al. Chromosomal copy number changes in patients with non-syndromic X linked mental retardation detected by array CGH. J. Med. Genet. 43, 362–370 (2006).

  196. 196.

    et al. A partial MECP2 duplication in a mildly affected adult male: a putative role for the 3' untranslated region in the MECP2 duplication phenotype. BMC Med. Genet. 13, 71 (2012).

  197. 197.

    et al. Recurrent infections, hypotonia, and mental retardation caused by duplication of MECP2 and adjacent region in Xq28. Pediatrics 118, e1687-95 (2006).

  198. 198.

    et al. Duplication of the MECP2 region is a frequent cause of severe mental retardation and progressive neurological symptoms in males. Am. J. Hum. Genet. 77, 442–453 (2005).

  199. 199.

    et al. Submicroscopic duplication in Xq28 causes increased expression of the MECP2 gene in a boy with severe mental retardation and features of Rett syndrome. J. Med. Genet. 42, e12 (2005).

  200. 200.

    et al. Real-time quantitative PCR as a routine method for screening large rearrangements in Rett syndrome: report of one case of MECP2 deletion and one case of MECP2 duplication. Hum. Mutat. 24, 172–177 (2004).

  201. 201.

    et al. A mouse model for MeCP2 duplication syndrome: MeCP2 overexpression impairs learning and memory and synaptic transmission. J. Neurosci. 32, 3109–3117 (2012).

  202. 202.

    et al. Targeted disruption of the Huntington's disease gene results in embryonic lethality and behavioral and morphological changes in heterozygotes. Cell 81, 811–823 (1995).

  203. 203.

    et al. Huntingtin is required for neurogenesis and is not impaired by the Huntington's disease CAG expansion. Nat. Genet. 17, 404–410 (1997).

  204. 204.

    et al. Preclinical safety of RNAi-mediated HTT suppression in the rhesus macaque as a potential therapy for Huntington's disease. Mol. Ther. 19, 2152–2162 (2011).

  205. 205.

    et al. Randomized, double-blind, placebo-controlled trial of arimoclomol in rapidly progressive SOD1 ALS. Neurology 90, e565–e574 (2018).

  206. 206.

    et al. Huntington disease: natural history, biomarkers and prospects for therapeutics. Nat. Rev. Neurol. 10, 204–216 (2014).

  207. 207.

    et al. Clinical and biomarker changes in dominantly inherited Alzheimer's disease. N. Engl. J. Med. 367, 795–804 (2012).

  208. 208.

    et al. Natural history of infantile-onset spinal muscular atrophy. Ann. Neurol. 82, 883–891 (2017).

  209. 209.

    et al. Nusinersen versus sham control in infantile-onset spinal muscular atrophy. N. Engl. J. Med. 377, 1723–1732 (2017).

  210. 210.

    , & AlphaVbeta5 integrin: a co-receptor for adeno-associated virus type 2 infection. Nat. Med. 5, 78–82 (1999).

  211. 211.

    et al. Human hepatocyte growth factor receptor is a cellular coreceptor for adeno-associated virus serotype 3. Hum. Gene Ther. 21, 1741–1747 (2010).

  212. 212.

    et al. Identification of PDGFR as a receptor for AAV-5 transduction. Nat. Med. 9, 1306–1312 (2003).

  213. 213.

    et al. The 37/67-kilodalton laminin receptor is a receptor for adeno-associated virus serotypes 8, 2, 3, and 9. J. Virol. 80, 9831–9836 (2006).

  214. 214.

    et al. An alternate route for adeno-associated virus entry independent of AAVR. J. Virol. 92, e02213-17 (2018).

  215. 215.

    et al. AAV serotypes have distinctive interactions with domains of the cellular receptor AAVR. J. Virol. 91, e00391-17 (2017).

  216. 216.

    , & Mutational analysis of narrow pores at the fivefold symmetry axes of adeno-associated virus type 2 capsids reveals a dual role in genome packaging and activation of phospholipase A2 activity. J. Virol. 79, 2528–2540 (2005).

  217. 217.

    , , , & Surface-exposed adeno-associated virus Vp1-NLS capsid fusion protein rescues infectivity of noninfectious wild-type Vp2/Vp3 and Vp3-only capsids but not that of fivefold pore mutant virions. J. Virol. 81, 7833–7843 (2007).

  218. 218.

    , & Separate basic region motifs within the adeno-associated virus capsid proteins are essential for infectivity and assembly. J. Virol. 80, 5199–5210 (2006).

  219. 219.

    et al. Adeno-associated virus capsid proteins may play a role in transcription and second-strand synthesis of recombinant genomes. J. Virol. 88, 1071–1079 (2014).

  220. 220.

    et al. Mutants at the 2-fold interface of adeno-associated virus type 2 (AAV2) structural proteins suggest a role in viral transcription for AAV capsids. J. Virol. 90, 7196–7204 (2016).

  221. 221.

    et al. Cross-packaging of a single adeno-associated virus (AAV) type 2 vector genome into multiple AAV serotypes enables transduction with broad specificity. J. Virol. 76, 791–801 (2002).

  222. 222.

    , & Quantitative analysis of the packaging capacity of recombinant adeno-associated virus. Hum. Gene Ther. 7, 2101–2112 (1996).

  223. 223.

    , & Effect of genome size on AAV vector packaging. Mol. Ther. 18, 80–86 (2010).

  224. 224.

    et al. Nonrandom transduction of recombinant adeno-associated virus vectors in mouse hepatocytes in vivo: cell cycling does not influence hepatocyte transduction. J. Virol. 74, 3793–3803 (2000).

  225. 225.

    & Design and packaging of adeno-associated virus gene targeting vectors. J. Virol. 74, 4612–4620 (2000).

  226. 226.

    et al. Adeno-associated virus terminal repeat (TR) mutant generates self-complementary vectors to overcome the rate-limiting step to transduction in vivo. Gene Ther. 10, 2112–2118 (2003). This paper presents the first study to describe the now commonly used self-complementary AAV vectors that improve the rate and efficiency of transduction by overcoming the requirement for second-strand synthesis.

  227. 227.

    et al. Long-term safety and efficacy following systemic administration of a self-complementary AAV vector encoding human FIX pseudotyped with serotype 5 and 8 capsid proteins. Mol. Ther. 19, 876–885 (2011).

  228. 228.

    et al. Qualitative imaging of adeno-associated virus serotype 2-human aromatic L-amino acid decarboxylase gene therapy in a phase I study for the treatment of Parkinson disease. Neurosurgery 67, 1377–1385 (2010).

  229. 229.

    et al. AAV viral vector delivery to the brain by shape-conforming MR-guided infusions. J. Control. Release 240, 434–442 (2016).

  230. 230.

    et al. ApoE4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature 549, 523–527 (2017).

  231. 231.

    et al. Targeting of nonlipidated, aggregated apoE with antibodies inhibits amyloid accumulation. J. Clin. Invest. 128, 2144–2155 (2018).

  232. 232.

    et al. Gene delivery of human apolipoprotein E alters brain Abeta burden in a mouse model of Alzheimer's disease. Proc. Natl Acad. Sci. USA 102, 1211–1216 (2005).

  233. 233.

    et al. The antibody aducanumab reduces Abeta plaques in Alzheimer's disease. Nature 537, 50–56 (2016).

  234. 234.

    et al. Combination anti-Abeta treatment maximizes cognitive recovery and rebalances mTOR signaling in APP mice. J. Exp. Med. 215, 1349–1364 (2018).

  235. 235.

    , , & Progranulin gene therapy improves lysosomal dysfunction and microglial pathology associated with frontotemporal dementia and neuronal ceroid lipofuscinosis. J. Neurosci. 38, 2341–2358 (2018).

  236. 236.

    et al. Development of therapeutics for C9ORF72 ALS/FTD-related disorders. Mol. Neurobiol. 54, 4466–4476 (2017).

  237. 237.

    et al. Widespread central nervous system gene transfer and silencing after systemic delivery of novel AAV-AS vector. Mol. Ther. 24, 726–735 (2016).

  238. 238.

    et al. A phase I study of aromatic L-amino acid decarboxylase gene therapy for Parkinson's disease. Mol. Ther. 18, 1731–1735 (2010).

  239. 239.

    et al. Glucocerebrosidase gene therapy prevents alpha-synucleinopathy of midbrain dopamine neurons. Neurobiol. Dis. 82, 495–503 (2015).

  240. 240.

    et al. Antibody-aided clearance of extracellular alpha-synuclein prevents cell-to-cell aggregate transmission. J. Neurosci. 32, 13454–13469 (2012).

  241. 241.

    & Mechanism of anti-alpha-synuclein immunotherapy. J. Mov. Disord. 9, 14–19 (2016).

  242. 242.

    et al. An alpha-synuclein AAV gene silencing vector ameliorates a behavioral deficit in a rat model of Parkinson's disease, but displays toxicity in dopamine neurons. Brain Res. 1395, 94–107 (2011).

  243. 243.

    et al. shRNA targeting alpha-synuclein prevents neurodegeneration in a Parkinson's disease model. J. Clin. Invest. 125, 2721–2735 (2015).

  244. 244.

    , , , & Nonallele specific silencing of ataxin-7 improves disease phenotypes in a mouse model of SCA7. Mol. Ther. 22, 1635–1642 (2014).

  245. 245.

    et al. A new AAV10-U7-mediated gene therapy prolongs survival and restores function in an ALS mouse model. Mol. Ther. 25, 2038–2052 (2017).

  246. 246.

    , , & Neuroprotection by gene therapy targeting mutant SOD1 in individual pools of motor neurons does not translate into therapeutic benefit in fALS mice. Mol. Ther. 19, 274–283 (2011).

  247. 247.

    et al. Widespread spinal cord transduction by intrathecal injection of rAAV delivers efficacious RNAi therapy for amyotrophic lateral sclerosis. Hum. Mol. Genet. 23, 668–681 (2014).

  248. 248.

    , & TDP-43 proteinopathy and ALS: insights into disease mechanisms and therapeutic targets. Neurotherapeutics 12, 352–363 (2015).

  249. 249.

    et al. shRNA mediated knockdown of Nav1.7 in rat dorsal root ganglion attenuates pain following burn injury. BMC Anesthesiol. 16, 59 (2016).

Download references


The authors thank their colleagues at Voyager Therapeutics for numerous discussions on adeno-associated viral (AAV) gene therapy for central nervous system (CNS) disorders and the Parkinson disease team and P. Larson for their work on the delivery of AAV2–aromatic-L-amino-acid decarboxylase (AADC) gene therapy in patients with Parkinson disease with intraparenchymal brain administration. The authors also thank E. Smith, W. Yen and M. Lawrence for assistance with the figures, tables and text, respectively. B.E.D. was supported by the Beckman Institute for the CLARITY, Optogenetics and Vector Engineering Research Center (CLOVER) at the California Institute of Technology, the Friedreich's Ataxia Research Alliance (FARA) and FARA Australasia and the CHDI Foundation and is currently supported by the Stanley Center for Psychiatric Research at Broad Institute. K.S.B. was supported by the Michael J. Fox Foundation. B.M.R., S.M.P. and D.W.Y.S. are currently employees of Voyager Therapeutics, a CNS gene therapy company working on AAV vectors for the treatment of severe neurological diseases.

Author information

Author notes

    • Benjamin E. Deverman

    Present address: Stanley Center for Psychiatric Research at Broad Institute, Cambridge, MA, USA.


  1. Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA, USA.

    • Benjamin E. Deverman
  2. Voyager Therapeutics, Cambridge, MA, USA.

    • Bernard M. Ravina
    • , Steven M. Paul
    •  & Dinah W. Y. Sah
  3. Department of Neurological Surgery, University of California–San Francisco, San Francisco, CA, USA.

    • Krystof S. Bankiewicz


  1. Search for Benjamin E. Deverman in:

  2. Search for Bernard M. Ravina in:

  3. Search for Krystof S. Bankiewicz in:

  4. Search for Steven M. Paul in:

  5. Search for Dinah W. Y. Sah in:

Competing interests

The authors declare competing financial interests in the form of funding from Voyager Therapeutics, employment by Voyager Therapeutics and/or personal financial interests in Voyager Therapeutics.

Corresponding author

Correspondence to Dinah W. Y. Sah.



The protein shell of the virus that protects the adeno-associated viral genome and mediates entry into and trafficking within the host cell.


(IPa). Direct delivery of an agent into the tissue of interest.


(IT). A route of access into the spinal cord cerebrospinal fluid via the space under the arachnoid membrane.


(ICV). A route of access into the CSF via the cerebral ventricles (typically the lateral ventricle).


Specificity for a particular host tissue or cell type.


Capsid variants or groups of capsids that have distinct neutralization properties.


A route of access into the CSF via the cerebellomedullary cistern.


Genus of parvoviruses that are replication-incompetent in the absence of co-infection of the host cell with a second virus such as an adenovirus or HSV.

Self-complementary AAV

An AAV genome that has been modified by elimination of the 5′ terminal resolution site and can fold into double-stranded DNA without the requirement for DNA synthesis.

Convection enhanced delivery

(CED). Infusion of adeno-associated viral vectors or other molecules into the parenchyma under positive pressure to increase the distribution volume.

About this article

Publication history



Further reading