Recent Gene Therapy Advancements for Neurological Diseases

Abstract

The past few years have seen rapid advancements in vector-mediated gene transfer to the nervous system and modest successes in human gene therapy trials. The purpose of this review is to describe commonly-used viral gene transfer vectors and recent advancements towards producing meaningful gene-based treatments for central nervous system (CNS) disorders. Gene therapy trials for Canavan disease, Batten disease, adrenoleukodystrophy, and Parkinson’s disease are discussed to illustrate the current state of clinical gene transfer to the CNS. Preclinical studies are under way for a number of diseases, primarily lysosomal storage disorders, using a newer generation of vectors and delivery strategies. Relevant studies in animal models are highlighted for Mucopolysaccharidosis IIB and Krabbe disease to provide a prelude for what can be expected in the coming years for human gene transfer trials, using recent advancements in gene transfer technology. In conclusion, recent improvements in CNS gene transfer technology are expected to significantly increase the degree of disease rescue in future CNS-directed clinical trials, exceeding the modest clinical successes that have been observed so far.

Introduction

Neurological disorders are among the most difficult to treat with traditional pharmacological approaches due to the complexity of the nervous system and the physical barriers (such as the blood-brain barrier, BBB) that limit the distribution of many compounds into the central nervous system (CNS) after peripheral administration. Vector-mediated gene transfer is an alternative strategy to treat CNS disorders that holds much promise, especially for monogenic diseases.

Although significant advancements have been made using oligonucleotides to treat a variety of diseases and the use of either naked DNA or nanoparticle delivery of DNA, this review will focus on the development of viral vectors and their use in treating neurological disorders. The intent of this review is to provide a basic background on the most commonly used viral vectors and their uses in preclinical and clinical research, so that a reader can gain a concise view of where the field stands and seek additional detailed information in the cited references.

Gene Delivery Vectors

Viruses have the natural ability to deliver genetic material to cells, which makes them an excellent vector for gene delivery. Lentivirus, Herpes Simplex Virus, Adenovirus and Adeno-Associated virus (AAV) are some of the commonly used vectors for gene delivery to the CNS (Table 1).

Table 1.

Vectors Used in Gene Therapy.

Viral Vector	Retroviruses, HIV	Herpesvirus HSV-1	Adenovirus	Adeno-Associated Virus (AAV)
Family	Retroviridae	Herpesviridae	Adenoviridae	Parvoviridae
Disease associated with parent virus	Yes	Yes	Yes	No
Transgene capacity	8 kb	150 kb	36 kb	4.7 kb ssAAV 2.2 kb scAAV
Genome	ssRNA	dsDNA	dsDNA	ssDNA
Inserts into DNA	Yes	No (circular episome)	No	No (circular episome)
Transduction of non-dividing cells	Yes	No	Yes	Yes
Innate immunity	Yes	Yes	Yes	Limited

Retroviruses are a family of single-stranded (ss) RNA viruses that use reverse transcription to produce DNA which is randomly inserted into the host genome as a part of their life cycle. These viruses code for the gag, pol, and env genes which are contained in a capsid surrounded by a phospholipid bilayer (Leis et al., 1988). Lentivirus is a complex retrovirus that can transduce both non-dividing and proliferating cells (Lewis and Emerman, 1994). Retroviruses can integrate their genetic material into the host genome of quiescent cells of the CNS and confer long-term gene expression (Escors and Breckpot, 2010; Jakobsson and Lundberg, 2006; Wong et al., 2006). Pseudotyping, which involves modifying the envelope proteins on the capsid surface, can change the tropism of these viruses and also improve transduction efficiency (Colin et al., 2009; Desmaris et al., 2001; Greenberg et al., 2007; Jakobsson et al., 2006; Rahim et al., 2009). Lentiviruses have a packaging capacity of 8 kb of foreign DNA once the gag and pol genes are removed. The biggest advantage of retroviral vectors is their ability to incorporate the target gene of interest into the host chromosome of dividing cells, which ensures faithful transmission of the transgene to progeny cells. This advantage is critical when targeting a dividing cell population, such as hematopoietic cells.

Herpes Simplex Virus I (HSV-I) is a member of the Herpesviridae family which carries a large, linear, double-stranded (ds) DNA genome encoding about 80 viral genes (Roizman, 1996). The glycoproteins on the capsid facilitate entry of the virion into host cells (Shieh et al., 1992), and once inside the cell, the virus uncoats and circularizes its linear genome into an episome in the nucleus without integration (Garber et al., 1993; Strang and Stow, 2005). Properties of HSV-1 that make it a suitable vector in treating CNS disorders include high transduction efficiency, large transgene capacity (152 kb), its ability to enter a state of latency in neurons, and its high neural tropism via retrograde axonal transport to dorsal root ganglia and trigeminal ganglions (Bearer et al., 2000; Berges et al., 2007). Replication-conditional or attenuated HSV vectors are capable of transgene expression in dividing cells but not in non-dividing cells (Wu et al., 1996). The application of this vector is limited due to its high inflammatory profile but it has been used in gene delivery to cancers such as glioblastoma multiforme (Andreansky et al., 1997; Lou, 2003; Markert et al., 2009; Mineta et al., 1995).

Another vector used in clinical applications is Adenovirus. Although it shows a high neural tropism (Jager and Ehrhardt, 2009), the use of this vector in the CNS is limited due to its high toxicity in the CNS (Lozier et al., 2002; Thomas et al., 2001). This vector is more commonly seen in cancer treatments and is capable of packaging 25 kb of foreign DNA (Kochanek et al., 1996).

Arguably the most prominent vector for gene delivery to the brain is AAV. It is an ssDNA virus whose genome consists of 3 capsid genes, 4 replication genes, and a gene called AAP contained within 3 open reading frames, flanked by inverted terminal repeats (ITRs) that serve in genome replication and packaging. Recombinant AAV (rAAV) is generated by flanking up to ~4.7 kb of foreign DNA with the ITRs and providing the viral genes in trans, to package the foreign DNA (but not viral genes) within the AAV capsid (Dong et al., 1996; Lai et al., 2010; Wu et al., 2010). The risks associated with AAV are highly reduced due to its non-pathogenic nature (Atchison et al., 1965) and its dependence on co-infection with a helper virus such as adenovirus in order to replicate and execute a lytic infection (Goncalves, 2005; McCarty et al., 2004). The viral capsid proteins facilitate binding of the virion to cell surface receptors, which can vary considerably between the >100 AAV variants that have been identified (Gao et al., 2005). The availability of many different capsids with differing tropisms provides a level of cell/organ specificity that can be modulated by the choice of capsid. High neural tropism of some serotypes adds to other advantages of AAV which include its ability to confer long-term transgene expression in non-dividing cells and its ability to deliver genes as an extra-chromosomal episome which highly reduces the possibility of insertional mutagenesis and oncogenesis (Choi et al., 2006; Duan et al., 1998).

For a detailed review of viral vectors used in gene therapy, see Gray et al., 2010b and Lentz et al., 2011.

Current State of Clinical Research

Two primary strategies have emerged as a means to advance gene therapy into the clinical arena. In vivo gene transfer focuses on using viral or chemical reagents to deliver genes directly to patients through a single injection. Ex vivo gene transfer relies on externally transducing cultured cells with a therapeutic vector and placing them back into patients (Suhr and Gage, 1993). This section will focus on current examples of CNS gene delivery trials made possible through rAAV and lentiviral vectors (Table 2). Additionally, shortcomings associated with each approach will be discussed to emphasize areas where vector improvements may lead to a more meaningful clinical outcome.

Table 2.

CNS Gene Delivery Trials Via rAAV and Lentiviral Vectors.

Disease	Phase	Vector/Gene	Route	Reference
Batten Disease	I	AAVrh10/CLN2	Intracranial	Worgall et al., 2008
Canavan’s	II	AAV2/ASPA	Intracranial	Leone et al., 2012; McPhee et al., 2006
Parkinson’s	I	AAV2/CERE-120	Intracranial	Gasmi et al., 2007; Kordower et al., 2006
	I	AAV2/ GAD65/67	Intracranial	Kaplitt et al., 2007
	II	AAV2/AADC	Intracranial	Christine et al., 2009; Eberling et al., 2008; Muramatsu et al., 2010
Adrenoleukodystrophy	I	HIV-1/ABCD1	Ex vivo	Cartier and Aubourg, 2008; 2010

The first CNS gene therapy trial utilizing AAV began in 2001 for Canavan’s disease -- an inherited leukodystrophy that leads to toxic accumulation of N-acetyl-aspartate (NAA) in the CNS, interfering with the formation of myelin sheaths (Janson et al., 2002; Kumar et al., 2006; Leone et al., 1999). Thirteen of the 28 patients received intraparenchymal injections of an AAV2 vector carrying the enzyme ASPA into six brain regions. Five and ten years post-procedure, no long-term adverse effects were detected in treated patients, and host immune responses to the vector were minimal. Decreased levels of NAA accumulation in the brain suggested a functional copy of ASPA could be delivered to the CNS (Leone et al., 2012; McPhee et al., 2006). Though modest phenotypic improvements could be observed in treated patients such as hand-grasping and increased cognitive recognition, all patients still relied on assistance to walk, converse, and even acknowledge people around them (Leone et al., 2012). A similar result was observed in the 2008 trial to treat late infantile neuronal ceroid lipofuscinosis (LINCL), or Batten disease. Ten patients were injected with twelve stereotaxic injections of CLN2/AAV2 vector into six brain regions (Worgall et al., 2008). While four subjects exhibited mild humoral responses to the vector, the remaining subjects exhibited significantly decreased rates of decline compared to untreated control subjects (Worgall et al., 2008). However, like the Canavan trial, only modest phenotypic improvements were observed in these patients without complete reversal of disease pathology. In conclusion, though these studies were some of the first to demonstrate the safety of rAAV-mediated gene delivery for CNS disorders, the breadth of gene transfer achieved with direct brain injections appeared insufficient to achieve a substantial rescue.

Parkinson’s disease (PD) is a neurological disease triggered by the loss of dopaminergic neurons in the substantia nigra and the striatum. In the context of disease pathology, PD is a good candidate for gene therapy. However the involvement of several genes and neurotrophic factors has made PD anything but a simple therapeutic goal (Gray et al., 2010). Despite these setbacks, PD gene therapy has pushed its way into seven different phase I and phase II clinical trials, demonstrating the high enthusiasm for gene therapy (Xiao et al., 2012). Six out of the seven trials utilized AAV2 as a vehicle for delivering neurotrophic factors such as GDNF or glutamic acid decarboxylase (GAD) while the seventh utilizes a lentiviral vector (Feng and Maguire-Zeiss, 2010). Treated patients have reported improvements in motor function, but the treatment efficacy has been difficult to evaluate in the absence of an extended phase III, placebo-controlled study.

Lastly, for disorders where loss of a secreted protein results in CNS defects, ex vivo gene therapy offers a way to reprogram isolated patient cells for functional protein synthesis. An example is a 2009 trial to treat adrenoleukodystrophy (ALD) -- a demyelinating disorder caused by the loss of an enzyme critical for metabolizing fatty acid chains in the CNS. Autologous CD34+ cells were isolated from two ALD boys, genetically reprogrammed using a lentiviral vector carrying the missing ABCD1 gene, and then reinjected into the patients. Follow-up studies showed that demyelination was halted 14–16 months following treatment, and 24–30 months later ALD protein expression was still retained in several cell types (Cartier et al., 2012; 2010).

While several clinical trials have occurred, the degree of patient improvement has been limited. A major limitation in all of these trials is the current vector technology and the ongoing challenge of how to widely, efficiently, and safely transduce cells within the CNS. Vector technology has made considerable advancements over what is currently used in the clinic, and these improvements are poised to greatly improve treatment efficacy. Some of these advancements are outlined below.

Advancements in Vector Technology for In Vivo Gene Transfer

An ideal vector would provide widespread distribution of the transgene, large packaging capacity, stable gene expression, high transduction efficiency, target cell specificity, and low immunological responses in order to overcome current limitations. Except for the packaging capacity, recent advancements have been made for AAV in each of these regards. Improvements in other vectors have also been made in most of these areas, but will not be discussed in detail. For more information on Adenovirus and HSV vectors, see Lentz et al., 2011.

Identification of the optimal capsid is critical to effectively target the desired cells and tissues. Over 100 AAV variants have been identified with varying tropisms providing a broad toolkit for optimized delivery to target tissue/organs (Gao et al., 2005; Wu et al., 2006). Several variants of AAV have been used in CNS applications including AAV1, AAV2, AAV4, AAV5, AAV6, AAV8, and AAV9 (Gray, 2012). Although AAV2 is the most studied serotype in clinical applications, it performs relatively poorly compared to more recently characterized serotypes. Direct injection in the brain parenchyma with AAV1 and AAV9 provides an almost exclusive neuronal tropism, a mix of neurons and glia with AAV5, and mostly astrocytes with AAV4 (Burger et al., 2004; Cearley and Wolfe, 2006; Davidson et al., 2000). More recently, AAV9 has been shown to cross the BBB and transduce neurons and glia in the brain and spinal cord following intravenous injection (Duque et al., 2009; Foust et al., 2009; Gray et al., 2011b). Current challenges to intravenous delivery of AAV9 include high peripheral tropism, high doses required, and limited neuronal transduction in non-human primate cells due to the existence of pre-existing neutralizing antibodies (NAbs) (Gray et al., 2011b). To circumvent these problems AAV9 can be injected into the cerebrospinal fluid (CSF) via the lumbar cistern or cisterna magna. Using this route of administration still provides widespread CNS gene transfer in non-human primates, but avoids anti-AAV NAbs, greatly reduces the peripheral biodistribution, and uses a considerably lower dose (Samaranch et al., 2012; Gray et al., in press) compared to intravenous injection.

Tissue/cell specificity can be further manipulated by engineering the viral capsid beyond what is found in nature. Strategies to modify the capsid include rational mutagenesis, peptide insertion, and directed evolution/DNA shuffling. By introducing rational mutations in the capsid, novel variants of AAV can be produced which are not only optimized for transduction but also increase target specificity, prevent the production of circulating antibodies to the vector, and/or improve intracellular trafficking. As an example, AAV9.47 is a variant of AAV9 that is liver-detargeted but still retains high CNS tropism (Pulicherla et al., 2011).

To increase the specificity of the viral vector, known ligands for target cells can be incorporated on the capsid. This approach has been used to modify AAV2’s ability to deliver genes to the CNS. Unmodified AAV2 normally has limited capability for axonal transport, but when peptides derived from an NMDA receptor agonist and a dynein binding motif were incorporated on the capsid, a 10–100 fold increase of retrograde transport delivery to the CNS was observed (Xu et al., 2005). Similarly, novel CNS-targeted AAV vectors have been created by incorporating peptides obtained through phage-display biopanning (Chen et al., 2009). When these peptides were incorporated into an AAV2 capsid, peptides generated in disease models specifically targeted the CNS vasculature in those models but not in wild-type mice, and vice versa. This demonstrates the high specificity that these approaches can confer upon the vectors.

As opposed to rational mutagenesis, directed evolution is a process of selecting novel capsid variants without a priori knowledge of physical determinants. This process was developed by Schaffer and Maheshri (Maheshri et al., 2006; Schaffer and Maheshri, 2004) and allows for the recovery of clones that are highly selective for a given characteristic, above that of the parent serotypes. Error-prone polymerase chain reaction of a single serotype, DNA shuffling by randomly mixing multiple serotypes, and a combination of these two methods are used to generate library clones with various mutations (Stemmer, 1994a; 1994b). These library clones are pooled together and subjected to multiple rounds of selective pressure. This technique was successful in the isolation of two clones for delivering therapeutic genes to areas of the brain that suffered damage from kainic acid seizures after intravenous injection (Gray et al., 2010a). These clones had the benefit of near-complete loss of tropism for liver, heart, and muscle conferring a more favorable biodistribution profile, leading to a higher degree of safety.

A major rate-limiting step in gene expression, and conversion of the genome into a stable episome in ssAAV, is the synthesis of the second strand for transcription following the uncoating of AAV in the cell (McCarty et al., 2003). Transduction efficiency has been increased over 100-fold with the use of self-complementary (sc) AAV (McCarty et al., 2001). scAAV vector genomes are composed of complementary copies of the DNA insert linked in cis through a mutated AAV inverted terminal repeat (ITR). The main disadvantage of using scAAV is the reduced packaging capacity of 2.3 kb. This limitation has led to the development of new compact promoters and polyadenylation elements (Gray et al., 2011a).

Advancements in Ex Vivo Gene Transfer: Lentiviral Correction of Bone Marrow Stem Cells

Successful treatment options for lysosomal storage diseases include enzyme replacement therapy (ERT) (Desnick and Schuchman, 2002), hematopoietic cell transplantation (HCT) (Peters et al., 2003; Souillet et al., 2003), or a combination of the two as seen in Hurler’s syndrome (Cox-Brinkman et al., 2006). The main problem with ERT is the efficacy of the enzyme to cross the BBB which limits the access of systemically injected enzyme. Allogeneic HCT can be used to overcome this. Hematopoietic cells, such as activated lymphocytes, monocytes, and microglial precursors, are able to cross the BBB and can be used to deliver enzymes to the CNS (Asheuer et al., 2004; Priller et al., 2001). Following hematopoietic stem cell (HSC) transplantation, donor derived cells migrate to the CNS and differentiate to form perivascular and parenchymal microglia. These cells could mediate the cross-correction of the enzyme in neighboring neurons and glial cells. However, the benefit is limited to a small subset of lysosomal storage diseases (LSDs) and is not effective in patients with overt neurological or aggressive infantile forms (Rovelli, 2008). Another limitation of allogeneic HCT is the development of Graft-vs-Host Disease (GVHD) or Host-vs-Graft Disease (HVGD), where the immune cells are activated and cause wide-spread inflammation (Hwang et al., 2007).

Transplantation of autologous HSCs modified with lentiviral vectors to express the missing enzyme or protein may circumvent the problems associated with allogeneic HSCT (Biffi et al., 2011; Cartier and Aubourg, 2008; Cartier and Aubourg, 2010; Cartier et al., 1996).

Random insertion of the transgene into the host genome, initially seen as advantage of lentiviral vectors due to the propagation of the transgene among dividing cells, has the risk of insertional mutagenesis as seen in the patients treated for X-linked SCID who developed leukemia as a side effect of treatment (Bokhoven et al., 2009; Hacein-Bey-Abina et al., 2003; Pike-Overzet et al., 2007). Lentiviral vectors have been made significantly safer by reducing or eliminating its integration into the host genome. Directing integration to heterochromatin regions minimizes the risk of gene activation (Gijsbers et al., 2010) and increases the safety of lentiviral vectors. Other advancements in lentiviral vectors are self-inactivating (SIN) mutations, which knock out the promoter activity of the LTR (Miyoshi et al., 1998; Zufferey et al., 1998), as well as non-integrating lentiviral (NIL) vectors, which carry mutant integrase or mutations in their LTRs that inhibit integrase binding (Apolonia et al., 2007; Philippe et al., 2006; Sarkis et al., 2008). These viral genomes exist in linear and circular forms episomally (Cara and Reitz, 1997; Philpott and Thrasher, 2007). These new technologies aim to improve safety of the vectors and overcome risks of cancer. For a detailed review on gene transfer in HSC, see Cartier and Aubourg, 2010.

Gene Therapy for Lysosomal Storage Disorders

Lysosomal storage diseases (LSDs) are a set of inherited, metabolic disorders characterized by reduced or absent levels of soluble lysosomal enzyme activity. Over time, loss of enzyme activity results in accumulation of toxic metabolic substrates, which in 70% of LSD patients results in debilitating neurological symptoms (Gritti, 2011; Neufeld, 1991; Sands and Davidson, 2006). Genzyme Corporation currently offers recombinant enzyme replacement therapy (ERT) for four different LSDs -- Gaucher disease, Fabry disease, Mucopolysaccharidosis I, and Pompe disease. ERT requires regular infusions of enzyme, whereas gene therapy offers a permanent source of the missing enzyme within the body after a single administration.

Normally when an enzyme or protein is first translated, it is trafficked to the endoplasmic reticulum and the Golgi apparatus to undergo post-translational modifications. A small percentage of modified proteins eventually become trafficked out of the lysosome and find their way to the cell surface to be secreted out of the cell. This secreted enzyme can be taken up by neighboring cells, providing cross-correction (Sands and Davidson, 2006). Gene therapy can take advantage of this system by transducing a small population of cells, correcting them to permanently produce the functional soluble enzyme and thereafter, by inherent cellular trafficking, pass on the therapeutic product to neighboring cells. Data from in vitro cross-correction and in vivo clinical studies have demonstrated that missing LSD enzyme expressed at levels 1–10% of normal levels can be therapeutic (Leimig et al., 2002; Sands and Davidson, 2006). Therefore, cross-correction allows even sub-optimal gene expression to make a strong impact on LSDs. However, there are a number of LSDs where the mutation is not in a soluble enzyme. Mutations in non-enzymatic proteins like GM2 activators or saposins can lead to an inability of cells to activate specific lysosomal enzymes (Sands and Davidson, 2006). Other LSD types are characterized by mutations in integral lysosomal membrane proteins or proteins that assist in the intracellular trafficking of lysosomal enzymes. LSDs caused by a missing soluble enzyme, that is capable of cross-correction, represent the easiest targets for gene therapy.

Several proof-of-principle studies have validated the therapeutic potential of cross-correction, as long as the transgene is widely distributed across the CNS. One example is Mucopolysaccharidosis (MPS) IIB disease -- an LSD caused by a deficiency in the alpha-N-acetyl-glucosaminidase (NAGLU) enzyme responsible for breaking down heparan sulfate in the lysosome. In 2011, Fu et al. (2011) demonstrated that a single intravenous injection of rAAV9-NAGLU significantly improved the behavioral phenotypes and increased survival of treated MPS IIIB disease mice. Histological staining of target CNS tissues showed diffuse global transduction and pathological correction, demonstrating that gene delivery to a small percentage of widely distributed cells could provide therapeutic enzyme activity levels across the CNS (Fu et al., 2011). In Krabbe disease, a neurodegenerative disorder characterized by a deficiency in galactocerebrosidase (GALC), Rafi et al. (2012) demonstrated the potential of utilizing multiple CNS injections to optimize delivery and overall cross-correction. By injecting neonate mice intracerebroventricularly, intracerebellarly, and intravenously with rAAVrh10-GALC, high enzyme activity was achieved in the brain and cerebellum, and moderate to high activity was detected in the spinal cord and the sciatic nerve. In addition, treated mice successfully lived up to 8 months and gained the ability to mate and care for new-borns, among other phenotypic corrections (Rafi et al., 2012).

The disease paradigm of LSDs lends itself well for gene therapy applications. Encouraging preclinical results from murine models of MPSIIB and Krabbe disease have demonstrated the capacity in which new global gene transfer approaches are proving beneficial for this family of diseases.

Summary

In an ideal scenario, a single administration of the vector (containing the therapeutic gene) would confer permanent correction of the target disease. Monogenic CNS disorders, especially LSDs that can benefit from cross-correction, are particularly amenable to this approach. Clinical trials spanning the prior decade have shown positive effects resulting from gene transfer, but the benefit to the patients has been relatively modest. The vector technology, both for in vivo and ex vivo gene transfer, has advanced rapidly in the last few years and new approaches have been developed to significantly improve the scope and efficiency of gene transfer. The effectiveness of these strategies for in vivo and ex vivo gene transfer is being realized in preclinical animal studies, and they are poised to rapidly transition into the clinical realm. Clinical successes, however modest, will continue to generate optimism and drive the field forward. If the preclinical successes now being realized translate to the clinical arena, the coming years should be quite exciting indeed.