Significant advances have been made over the past two decades in our understanding of the basic biology of the nervous system and the molecular basis of neurologic disease. Identification of the complex regulatory mechanisms involved in the maintenance of normal cellular function and the roles assumed by different genes in the pathogenesis of specific diseases have paved the way for a new therapeutic approach based on the alteration of cellular phenotype by genotype manipulation. This new modality, designated gene therapy, has raised high expectations as a potential solution for a large spectrum of currently untreat- able conditions. Unfortunately, the rapid transfer from in vitro studies to clinical trials has so far yielded only anecdotal reports of success.
In this chapter, the current status of gene therapy as a regenerative tool for neurologic dis- ease will be reviewed. The principles that under- lie the methodology, the main gene delivery vehicle types (vectors), and the different thera- peutic approaches adopted for the nervous sys- tem will be outlined. This will be followed by a review of several examples of gene therapy use for neurologic diseases. We conclude this chap- ter by a detailed analysis of those obstacles, technical and conceptual, that still hinder the translation of in vitro efficacy to clinical cure.
Gene Therapy, a Novel Therapeutic Approach
Gene therapy is defined as the introduction of specific nucleic acid sequences into selected tar-
get cells for the treatment or prevention of dis- ease. The method was originally contemplated as a potential solution for the wide array of monogenic inherited disorders, such as cystic fibrosis and Duchenne muscular dystrophy, for which conventional pharmacotherapy is unable to provide any adequate response. Restoration of the function of a defective or missing gene product by the introduction of a correct copy of the gene seemed like a simple and feasible con- cept. It was soon to be appreciated that the approach should not be restricted to single-gene defect replacement therapy. Indeed, alteration of the cellular transcriptional status and in vivo production of a therapeutic protein may be implemented to achieve phenotypic changes in a wide spectrum of disorders. The list of potential applications thus expanded to encompass most neurologic disease states, acute and chronic, inherited and acquired, infectious and neoplas- tic (see Table 9.1).
Transition into clinical testing was then only a matter of time. Since the initiation of the first gene therapy clinical trial in 1989, nearly 1000 such trials have been approved worldwide.12 This time period has seen drastic changes in the types of diseases addressed. Most formidable is the slow drift from monogenic inherited disor- ders to neoplastic disease (in 2003, more than 60% of all gene therapy clinical trials were aimed at neoplastic disease). Two reasons underlie this change. First, the more immediately life-threat- ening nature of many of these conditions and the lack of efficient alternative therapies have facilitated their transition into clinical testing.
Second and perhaps more fundamental, is the current limited ability to achieve efficient gene
9
Gene Therapy to the Nervous System
Hillel Haim and Israel Steiner
133
expression in the desired cell population in vivo, to reverse disease phenotypes without causing unwanted side effects.
Thus, while experimentation in the field of gene replacement therapy goes on, the “fast lane” to the clinic has been taken up by cancer gene therapy trials. Accordingly, the number of tumor-destructive approaches increased, to include more cytotoxic genes and replication competent oncolytic viruses.13–15Although this class of diseases lies outside the scope of this review, the principles of vector formation and the limitations imposed on the success of in vivo gene transfer similarly apply.
Methods of Gene Delivery
Depending on the application, two modes of delivery may be used to transfer a chosen gene to the target tissue, the ex vivo and in vivo approaches. The ex vivo approach is based on the isolation and in vitro culture of selected cells, where they are manipulated and subsequently returned to the host. Although mainly applied for gene therapy of the immune system where isola- tion of a specific group of cells is possible,16the approach may also be used to generate “mini-fac- tories” that produce and locally secrete desired proteins. The genetically modified cells may thus
serve as delivery platforms of neurotrophic fac- tors or neurotransmitter-forming enzymes into the central nervous system (CNS).17,18
In contrast, the in vivo approach involves the direct administration of the gene-carrying vec- tor to the host. Delivery may be via simple local- ized application (such as stereotactic injection into the brain parenchyma), or by the vascular (systemic) route. The main drawback of this approach, and probably the most significant problem currently encountered in gene therapy to the nervous system, is the difficulty in target- ing the vector to the selected cell type. To date, this remains the most prominently unsur- mounted obstacle.
The Gene-Carrying Vectors
The efficient delivery of the gene to its nuclear target, where it may be expressed, serves as the most basic requirement in gene therapy. Two general types of vectors are used to package and deliver genes: Virus-based and synthetic gene delivery systems. In synthetic systems, the transgene forms part of a plasmid, which is replicated and subsequently purified from bac- teria. The only genetic material available for transcription in the target cells is therefore the transgene itself. Plasmid DNA may be delivered
Table 9.1. Gene therapy for neurodegenerative disorders: therapeutic approaches
Strategy Aim of therapy Targeted pathology Example disease Therapeutic gene References Introduction of Replacement of a Enzyme deficiency MPS VII β-Glucuronidase 1, 2 phenotypic changes missing gene Canavan’s disease Aspartoacylase 3 restricted to the function in
target cell monogenic disorders
Production of a Alteration of Enzyme deficiency Parkinson’s disease Tyrosine hydroxylase 4–6 secreted therapeutic cellular phenotype
protein to achieve a in non-monogenic localized or systemic disorders effect
Decreased neuronal Parkinson’s, Alzheimer’s, GDNF, NGF, CNTF viability Huntington, ALS,
ischemia, traumatic
injury 7–9
Neuronal
hyperexcitability Seizures GAD 10
Potassium channels 11 ALS, amyotrophic lateral sclerosis; CNTF, ciliary neurotrophic factor; GAD, glutamic acid decarboxylase; GDNF, glial cell line-derived neurotrophic factor; MPS, mucopolysaccharidosis; NGF, nerve growth factor.
(transfected) either alone (naked) or complexed with agents that enhance cellular uptake (such as cationic polymers and liposomes). Although this vector type does not express any toxic or immunogenic foreign protein products, gene transfer is generally inefficient and transient.
In comparison, virus-based systems rely on the natural ability of these infectious agents to package a gene, deliver it to the host cell, and mediate its expression. The gene of interest is incorporated into the modified viral genome, harnessing viral structural and enzymatic com- ponents to facilitate delivery and expression.
Because the efficiency of gene transfer by viruses exceeds that of the nonviral approach, virus- based vectors currently serve as the most com- mon type of vehicle used in gene therapy at large and for gene transfer to the nervous system in particular.
Viral Vectors
Viral delivery systems are based on modifying the genome of replicative viruses to obtain repli- cation-defective vectors. Such alterations gener- ally include the deletion of coding sequences essential for virus replication and their replace- ment with the gene of interest. Packaging of the modified genome is performed in specialized cells, which provide in trans those deleted struc- tural and enzymatic components necessary for particle assembly and expression of the carried genome in the target cell. Viral vector infection (designated transduction) results in an abortive cycle, because of the lack of genes necessary for virus replication.
The list of viruses engineered into gene-carry- ing vectors is continuously growing. From the current arsenal of viral vectors, each retaining several of the properties inherent to the virus from which it was constructed, the vector most appropriate for the specific gene therapy appli- cation may be tailored (Table 9.2). Monogenic inherited disorders, such as certain errors of metabolism, require the long-term expression of the transgene, as provided by vectors that achieve stable integration into the host cell chro- mosomes. In contrast, transient disorders such as acute infections, tissue ischemia, or neo- plasms, require shorter time frames of therapeu- tic gene expression. In the following section, a brief outline of the currently most popular virus-based vectors will be presented, with refer-
ence to specific properties, advantages, and dis- advantages of each.
Retroviruses
Retroviruses are a large family of enveloped RNA viruses, so named because the genome is reverse transcribed in the infected cell to double-stranded DNA (dsDNA). They are char- acterized by the establishment of a chronic infection state, generally well tolerated by the host, but slowly progressing diseases such as malignancy and immunodeficiency may ensue.
Several members of the retrovirus family have been exploited to serve as vectors: i) C-type retroviruses (also referred to as oncoretro- viruses), including the murine leukemia virus (MuLV) group and the avian leukosis virus (ALV), ii) lentiviruses, including the simian, feline, and human immunodeficiency viruses (SIV, FIV, and HIV, respectively), and iii) spumaviruses such as the human foamy virus.
After binding of the viral envelope glycopro- tein to the cell-surface receptor, the viral genome enters the host cell and is reverse tran- scribed in the cytoplasm. Subsequently, the proviral DNA enters the nucleus and integrates into the host cell chromosomes. Viral particles contain two copies of the single-stranded RNA (ssRNA) genome (approximately 9 kb in size), which are incorporated into the assembling virions by the cis-acting packaging sequence (y).
Retrovirus genomes are flanked by long termi- nal repeat (LTR) sequences, involved in cis in reverse transcription, transcriptional control, and chromosomal integration. The LTRs frame the three genes gag, pol, and env, encoding structural proteins, enzymatic components, and the surface glycoprotein, respectively. This structural framework greatly facilitates the con- struction of retrovirus-based vectors: The gag, pol, and env genes are deleted and replaced by the exogenous transgene of choice. For vector assembly, these missing gene functions are pro- vided in trans in specialized cells that express these proteins transiently (packaging cells) or stably (producer cells).
To minimize the chances that replication- competent retroviruses (RCRs) will be gener- ated by homologous recombination, these trans-expressed genes are placed on separate constructs, carrying little sequence homol- ogy.19,20 In addition, abolishment of the viral
LTR promoter prevents the generation of full- length proviral RNA, further increasing vector safety.21
Because the retroviral vector genome is chro- mosomally integrated, the transgene is theoreti- cally expressed for the lifetime of the cell and is passed to daughter cells during mitosis.
A potential disadvantage of this retroviral fea- ture is a functional disturbance of the chromo- somal material into which it is inserted. This may result in the loss of cell cycle control and malignant transformation (see below).
Lentiviruses. A significant disadvantage of the oncoretrovirus-based vectors is their inabil- ity to transduce nondividing cells.22In vivo gene delivery to postmitotic cells, most notably in the nervous system, is thus rendered irrelevant.
Lentiviruses, however, have evolved the ability to harness the cellular nuclear import machin- ery to transport the viral genome into the nucle- oplasm23 and are thus able to efficiently transduce postmitotic cells. Although several members of the lentivirus family were devel- oped as gene delivery vehicles, most efforts have focused on the best-characterized lentivirus, the human immunodeficiency virus type 1 (HIV-1).
The genomic structure of lentiviruses closely resembles that of the oncoretroviruses. In addi- tion, they possess two unique regulatory pro- teins, tat and rev, and several accessory proteins
that facilitate viral replication and the persist- ence of infection.24–26Lentiviral vectors deleted of all accessory genes have been constructed, without affecting vector production effi- ciency.27,28
As would be expected of a vector derived from this deadly pathogen, biosafety concerns top the list of aims during vector construction. Splitting the genome into separate constructs, which are cotransfected into the packaging cell line signif- icantly reduced chances of RCR generation.29,30 As with oncoretrovirus-based vectors, biosafety is further increased by inactivating the LTR pro- moter, thus abolishing transcription of full- length proviral RNA.31
A unique property of retroviruses is the ability to incorporate the envelope glycoprotein of a different virus into their lipid envelope, acquir- ing its host range for infection in a process designated pseudotyping (see later under “cell- specificity of gene transfer”). HIV-1-based vec- tors pseudotyped with the vesicular stomatitis virus envelope glycoprotein (VSV-G) have been extensively studied for in vivo transduction of the CNS. Such vectors mediate high-efficiency transduction of both neurons and glial cells of rodent28,32 and nonhuman primate brains.33 Transgene expression persists over extended time periods with no associated brain pathology detected.29Furthermore, the therapeutic efficacy
Table 9.2. Vector properties
Vector
Adenovirus Retroviruses Lentivirus AAV HSV Pasmid DNA
Particle size (nm) 70 100 100 20 200 Variable
Genome type and dsDNA (36) RNA (8.8) RNA (9.6) ssDNA (4.7) dsDNA(152) — size (kb)
Packaging capacity (kb) 6–8 (up to 7 7 5 30 (up to
30 kb in gutless 150 kb in Infinite
vectors) a mplicons)
Gene transfer to Yes No Yes Yes Yes Minimal
nondividing cells
Chromosomal No Yes Yes Yes (random No No
integration and 19q)
Preexisting immunity Yes No No Yes Yes No
Immune response +++ +/− +/− +/− + None
Typical TU 1013 109 109 1010 1010 1013
concentration (MI)
Major safety concerns Inflammatory Insertional Insertional Insertional Inflammatory None response, mutagenesis, mutagenesis, mutagenesis response,
cytotoxicity generation generation cytotoxicity of RCRs of RCRs
dsDNA, double-stranded DNA; ssDNA, single-stranded DNA; TU, transducing units; RCR, replication-competent retroviruses.
of this vector to treat neurodegenerative disease in several animal models was demonstrated, including Parkinson’s disease,34 metachromatic leukodystrophy 35and type VII mucopolysaccha- ridosis.2
Several other lentiviruses have been devel- oped to serve as gene transfer vehicles and recently an FIV-based vector was successfully used to correct cognitive impairments in a mouse model of mucopolysaccharidosis type VII (MPS VII).36
Adenoviruses
Adenoviruses are double-stranded nonen- veloped DNA viruses, which cause upper respi- ratory tract infections in humans. They efficiently infect a wide variety of cell types, both dividing and nondividing. More than 50 differ- ent human adenoviral serotypes are currently known, of which serotypes 2 and 5 have been most extensively manipulated for gene transfer purposes. Because these two serotypes are also the most common to which humans are exposed, alternative serotypes and nonhuman adenoviruses are currently developed as vectors to surmount potential problems associated with preexisting immunity.37
Adenoviral infection is initiated by attach- ment to the cell surface coxsackie and aden- ovirus receptor, followed by endocytosis, transport to the nucleus, and translocation into the nucleoplasm where the viral genome remains in the nonintegrated (episomal) state.
The replication cycle is then initiated by the sequential transcription of the early (E) genes, DNA replication, and finally the transcription of the late (L) family of genes, which is followed by encapsidation of the newly synthesized genome and viral release by cell lysis.
The first generation of replication-incompe- tent adenovirus-based vectors was constructed by deleting the E1 gene, the main activator of viral transcription.38,39Exogenous sequences of up to 8 kb could thus be introduced in its place.
Nevertheless, despite E1 removal, low-level expression of additional genes from these vec- tors still occurred, eliciting strong cellular and humoral immune responses and resulting in the clearance of transduced cells by cytotoxic T lym- phocytes.40,41This heralded construction of sec- ond- and third-generation vectors containing additional deletions, characterized by decreased
expression of viral proteins and improved toxic- ity profiles.42The most extensively deleted ade- noviral vector, designated “gutless,” is stripped of nearly all viral genes, retaining only the inverted terminal repeats that flank the genome, the packaging signal, and a transgene cassette.43 Up to 30 kb of exogenous sequences may be accommodated into the vector. Trans-comple- mentation of all viral genes is thus necessary for vector assembly and is achieved by coinfection of the packaging cells with a helper virus.
“Gutless” vector preparations are minimally contaminated by replication-competent recom- binants. They are associated with a significantly reduced inflammatory response,44,45 produce therapeutic levels of secreted transgene prod- ucts in experimental animal models,46 and are currently the vector of choice for adenovirus- mediated gene delivery. Nevertheless, “gutless”
vector purification procedures still require improvement in the ability to scale up to con- centrations necessary for in vivo delivery.
The episomal status of the adenoviral genome in the transduced cell leads to gradual extinction of transgene expression because of intranuclear degradation and dilution in a dividing cell pop- ulation. Attempts to overcome the transient expression of the transgene have been made by constructing hybrid adeno/adeno-associated viruses (AAV) and adeno/retrovirus vectors, allowing chromosomal integration of the genome.47,48 Transduction efficiency by these hybrid vectors is, however, low.
Adeno-Associated Viruses
AAVs are members of the parvovirus family and are not associated with any known disease in humans. A productive replication cycle in the host cell only occurs upon coinfection with a helper virus, such as adenovirus or herpes sim- plex virus. Six human AAV serotypes are cur- rently known, of which AAV-2 is the most intensively studied for gene therapy purposes.
The tiny AAV virions (20 nm in diameter) carry a ssDNA molecule, of either plus or minus orientation. Inverted terminal repeats flank the genome containing two open reading frames, rep and cap, each encoding for a number of pro- teins. Rep produces the regulatory proteins required for genome replication and cap encodes for the structural proteins of the cap- sid. After cell entry via the endocytic pathway,
the genome is transported to the nucleus and forms dsDNA by either annealing with a com- plementary strand from a second virus or by de novo formation by the host cell machinery.49In the absence of rep, the vector will randomly integrate into the host cell chromosomes (as a single provirus or as head-to-tail concatamers by nonhomologous recombination), whereas in its presence, the genome is directed to a specific site in chromosome 19. It should be noted, however, that only a fraction of all intranuclear vector genome copies undergo integration, the rest remain episomal. Following the same prin- ciples of vector construction as detailed above, the rep and cap genes are replaced by a trans- gene cassette and the missing functions required for vector replication and encapsida- tion are trans-complemented in the packaging cells.50–52
AAV-based vectors are extensively used in gene transfer studies to the brain. High trans- duction efficiency of quiescent cell, minimal toxicity (because of the absence of virus-derived genes), and stable chromosomal integration of the genome render the vector appealing for many applications. In addition, different AAV serotypes selectively transduce different cells in the CNS, enabling cell-type-specific targeting of gene delivery.53
A significant disadvantage of AAV-based vec- tors is the limited size of the transgene sequence that may be inserted (up to 5 kb). This capacity, however, may be doubled by the use of the trans-splicing or homologous recombination approaches, taking advantage of the con- catamerization of vector genomes that occurs before integration.54
Herpes Simplex Virus Type 1
Herpes simplex virus type 1 (HSV-1) is a dsDNA human neurotropic virus. The interest in HSV- 1-based vectors is therefore largely concentrated on gene transfer applications to the nervous sys- tem. Viral particles (approximately 200 nm in diameter) are composed of a lipid envelope con- taining 12 different glycoproteins that mediate entry into the host cell, which surrounds the viral capsid that engulfs the 152-kb genome.
Between the capsid and envelope, the tegument layer contains several proteins that trans-acti- vate viral gene expression and regulate the host cell translation machinery.55
The HSV-1 life cycle depends on the type of host cell infected. In epithelial cells, the virus goes through a lytic replication cycle, whereas in neurons it may enter a latent state, persisting episomally in the nucleus. In lytic infection, viral protein expression is conducted in a highly ordered manner, involving the sequential tran- scription of the immediate-early (IE), early (E), and the late (L) classes of genes. After nuclear entry of the genome, the Vmw65 tegument pro- tein trans-activates transcription of the IE genes, leading to E gene expression, whose products enable DNA replication and transcription of the L genes that encode for viral structural compo- nents. The encapsidated viral genome then buds through the nucleus, acquires the tegument pro- teins and envelope within the cytoplasm, and is secreted from the cells in vesicles, culminating in the death of the infected cell.
In neurons, HSV-1 infection may assume a latent form, far less characterized than the lytic cycle.56After uptake of the virus by axon termi- nals, it is retrogradely transported to the cell nucleus where it remains as an episome in a latent state. The events that lead to the establish- ment of latency and the presence or absence of de novo viral protein synthesis are yet unre- solved issues. Latency is, however, known to be associated with the expression of the viral latency associated transcripts (LATs), which are assumed to inhibit the expression of lytic repli- cation cycle genes.57 From this dormant state, after stimuli such as UV irradiation or stress, the virus may reactivate and go through the lytic infection cycle.
Two major types of gene delivery vectors based on HSV-1 have been constructed: Recombinant and amplicon-based. Recombinant vectors are rendered replication-defective by the deletion of one or more of the five IE genes and the transgene is inserted into the deleted genome by homolo- gous recombination.58Such vectors are transcrip- tionally silent other than expression of the transgene, mimicking the latent infectious state.
Propagation of the vector is performed in cell lines that trans-complement the missing IE genes. Because nearly half of the HSV-1 genes are not essential for virus replication in cell culture, the recombinant vector genome may accommo- date up to 30 kb of foreign DNA.59,60
The amplicon vector system is based on bac- terially produced plasmids, which contain no viral genes but only the Escherichia coli origin of replication, the HSV-1 origin of replication and
packaging signal, and the inserted transgene.
Packaging of the amplicon in viral particles is enabled by trans-complementation using a packaging signal-deleted bacterial artificial chromosome (bac) that expresses the entire HSV-1 genome61–63or by multiple cosmids that span the genome.64 Amplicon-based packaging systems, although producing vector stocks with a decreased content of replication-competent virus, suffer from low titers.
Many animal models of different neurologic diseases have been tested for therapeutic effi- ciency of gene transfer using HSV-1 vectors, including neuroprotection,65,66enzyme replace- ment therapy for monogenic disorders,67 and immune modulation for demyelinating dis- ease.68Long-term expression from the episomal vector is difficult to achieve. However, in one report, transgene expression in dorsal root gan- glia was detected for up to 18 months, from vec- tors in which the transgene, controlled by the MuLV LTR promoter, was inserted directly upstream of the LAT promoter region.69 Similarly, using the LAT promoter, continuous systemic release of nerve growth factor (NGF) from an intraarticularly administered vector was observed for at least 6 months in rhesus macaques.70
Nonviral Vectors
Whereas the use of viruses as a gene delivery vehicle carries the advantage of mediating high- efficiency gene transfer and the potential to tar- get specific cell types, several significant disadvantages are noted:
a. An immune response is always mounted by the introduction of a foreign protein.
b. Safety risks result from the use of integrating viral vectors.
c. Potential reversion of vectors to replication- competent viruses may ensue.
d. There are limitations on the large-scale pro- duction of preparations containing high con- centrations of transducing units.
e. The capacity of most viral vector genomes to accommodate large exogenous sequences is limited.
These shortcomings have prompted efforts to produce alternative nonviral systems. The safety and versatility of these latter are, however,
traded off by the lack of the functions that viruses have evolved to package, protect, and deliver their genome through the target cell membrane and into the nucleus. A more thor- ough dissection of the virus infection cycle was thus warranted, to identify and adopt those
“ready-made” functions that viruses possess, which render them such efficient gene delivery vehicles. In this respect, the most important functions required are stabilization and pro- tection of the gene in the extra- and intracellu- lar milieu, efficient and selective uptake into the cytoplasm, delivery of the genome into the nucleoplasm, and release from the packaging formulation.
The efficiency of plasmid DNA transfer across the cellular membrane is generally extremely low. To enhance transmembrane transport, both mechanical and chemical approaches are used.
Mechanical means, based on increasing mem- brane permeability to the DNA include: i) Cell bombardment with high-velocity gold or tung- sten microparticles complexed to plasmid DNA.71,72Routinely used for vaccination proto- cols that require limited amounts of intra- muscular gene expression, application of the methodology for more diffuse and high-level transgene delivery seems unlikely. ii) Increased extracellular pressure was shown to augment uptake of DNA by cells of the cardiovascular sys- tem.73 This method was more recently success- fully applied to render vein grafts resistant to hypercholesterolemia-induced atherosclerosis.74 iii) Transient permeabilization of the plasma membrane by electrical pulses (electroporation).
The use of low-voltage, high-frequency pulses significantly reduces the considerable levels of cell mortality previously associated with the treatment and enabled efficient gene expression in mouse skeletal muscle.75iv) Ultrasound-medi- ated gene transfer, based on the same principle as electroporation, has also been shown to increase skeletal muscle transfection efficiency in vivo.76
Despite the wide array of membrane perme- ability-enhancing strategies, mechanical means are generally expected to induce some degree of tissue damage and are therefore currently not considered appropriate for gene transfer to the nervous system. In contrast, chemical agents are constantly gaining momentum as the strategy of choice to increase transfection efficiency.
Electrostatically based interactions between the negatively charged DNA and positively charged
formulation are used to form complexes that increase cellular deposition, attachment, and uptake of the gene. Among the earliest formula- tions, most notable are DEAE-dextran and cal- cium phosphate. Extensively used for in vitro gene transfer, these agents, however, cause sig- nificant toxicity and gene expression levels among cells is extremely variable.
A major improvement in nonviral gene trans- fer technology was introduced by the lipid-based delivery systems. Liposomes are submicron- sized aggregates, composed of amphiphilic lipids that self-assemble to an organized structure sur- rounding a watery interior. Depending on the composition of the polar head group, liposomes are classified as anionic, cationic, zwitterionic, or nonionic. Cationic liposomes, which associate in aqueous medium with the negatively charged plasmid DNA to form overall positively charged
“lipoplexes,” are the most widely used.
Incorporation of “fusogenic” lipidic elements increases mixing of the lipid phases of the lipo- some and endosomal membrane, thus enhanc- ing DNA escape from the endosome before fusion with the nuclease-containing lysosomes occurs.77–79Although liposomes of different lipid composition are continuously developed to enhance transmembrane delivery, attempts are also made to increase the low efficiency step of plasmid translocation from the cytoplasm into the nucleus.80 Utilization of virus-derived nuclear localization polypeptide signals, which are conjugated to the vector formulation, has been shown to significantly augment transfec- tion efficiency.81,82Moreover, the transfection of nondividing cells, which are more refractory to polynucleotide uptake relative to dividing cells, was significantly enhanced, opening new possi- bilities for gene transfer to the quiescent cells of the nervous system.
The ability of liposomes to associate with a variety of molecular structures has been further exploited to provide lipoplexes with additional functions. For example, conjugation to polyeth- ylene glycol (PEG) has been shown to protect plasmid DNA from intracellular degradation83 and increase stability and circulation time upon systemic administration.84,85Targeting moieties have also been incorporated into “PEGylated”
liposomes, enabling brain-directed delivery of the systemically injected vector,84,86as detailed later.
Although liposomal formulations are rou- tinely used in the clinic for the delivery of anti-
fungal and chemotherapeutic drugs, it should be noted that their use for gene transfer applica- tions has been associated with some degree of cytotoxicity.87 As an alternative, protein-based agents are gaining interest. Cationically charged polypeptides, such as poly-L-lysine, condense DNA and increase transfection efficiency.88This formulation allows the chemical conjugation of targeting moieties for specific cell-type target- ing.88,89 Attempts at assembling more complex polypeptide structures, composed of different functional building blocks, have been initiated.
One such construct contains the E. coliβ-galac- tosidase protein (for DNA protection), an RGD motif (for targeting the cell-surface αVβ3 inte- grin receptor) and a poly-L-lysine tail (for DNA condensation).90 Intracortical injection of this chimeric protein-gene complex mediated effi- cient transgene delivery to both the excitotoxi- cally injured and noninjured brain of postnatal rats.91 Transgene expression was observed in neuronal and nonneuronal cells, with no evi- dence of inflammation or cytotoxicity 6 days postinjection. Modeling functional polypeptide units with tailor-made characteristics is certain to advance in coming years, replacing those functions provided by viruses with synthetic or human-derived substitutes.
Disease-Targeted Gene Therapy
General Points
There are many neurologic disorders that are amenable to gene therapy.92,93In fact, the spec- trum may include almost any neurologic condi- tion. Examples abound and include slow and acute viral infections,94immune-mediated CNS conditions and multiple sclerosis,95 traumatic injury to the brain, spinal cord and peripheral nervous system disorders,96,97 stroke,98 and epilepsy.10At present, however, three conditions receive most of the attention and are at the focus of intensive research: neoplasms, inherited metabolic diseases, and neurodegenerative dis- orders. Whereas in previous sections of this chapter we focused on the principles underlying vector construction and delivery, here we out- line the biological requirements for treatment of neurologic conditions.
The basic requirement to therapy is an under- standing of the molecular basis of normal
function of the diseased cell and the abnormal- ity that is responsible for the pathological phe- notype. Three neurodegenerative conditions will highlight this point. Although much is known about Alzheimer’s disease (AD), Parkinson’s disease (PD), and amyotrophic lat- eral sclerosis (ALS), the basic aspects of etiology and pathogenesis are still incompletely under- stood. The conditions are extremely prevalent and tend specifically to affect the older age groups. They are characterized by dysfunction and early death of selective neuronal subpopula- tions. In ALS, upper and lower motor neurons in the pyramidal nervous system are affected, lead- ing to paralysis and death. In PD, the pigmented, mainly nigrostriatal dopaminergic, cells are dis- eased causing severe motor incapacity and in AD the burden seems to rest on a wider range of neuronal subpopulations culminating in dementia. Whereas one approach considers these conditions as the result of an enhanced aging process of selective neuronal cell popula- tions, accumulating data underline either a bio- chemical basis of cellular dysfunction or the possible involvement of an environmental factor/s.
Gene-based therapy may be attempted in these and other neurodegenerative conditions using two different approaches: Nonspecific delay of cell death and specific disease-tailored strategies.
Delaying Neuronal Cell Death
This may be achieved by the CNS delivery of growth factors and neuroprotective genes, such as NGF,17 or the antiapoptotic HSV-1 gamma34.5,99respectively. The approach is non- specific and has two major advantages. First, there is no true requirement to fully understand the biochemical basis of the disorder. The thera- peutic gene has a general, nonspecific action, and is usually nonselective, namely, it has a broader range of cell populations that it can affect. Second, there is no need to specifically transfer the therapeutic gene to the affected cell type. In many inherited metabolic abnormali- ties, the intracellular defect calls for intracellular repair (either by providing a normal gene, or by silencing a toxic abnormal gene product), and thus gene therapy has to be targeted to trans- duce the defective cell from within. With agents such as cytokines or nerve trophic factors, the
gene products act from without and, therefore, theoretically, require only that the therapeutic gene product will reach the extracellular com- partment. For obvious reasons, this approach has occupied most of the current experimental thrust in ALS,100,101PD,7,102and AD,103using both the “effect from without” and “effect from within” strategies.
Specific Disease-Tailored Strategies Amyotrophic Lateral Sclerosis
Mutations in the gene coding for Cu/Zn super- oxide dismutase enzyme were identified in the familial form of ALS104and there is evidence that oxidative stress is involved in motor neuron cell damage in the much more prevalent, sporadic form of ALS.105Tissue culture and animal mod- els have been developed to mimic some of the biochemical changes and neuropathology of the disease and antioxidant enzymes (such as superoxide dismutase, catalase, and glutathione peroxidase) have demonstrated therapeutic effi- cacy in these models (reviewed in Pong106).
Parkinson’s Disease
Pharmacologic therapy of the disease is based, in part, on supplying the CNS with L-DOPA, the precursor of the neurotransmitter dopamine that is reduced in the disease. Providing the gene that encodes for tyrosine hydroxylase (TH), the enzyme responsible for the synthesis of dopamine, can alleviate extrapyramidal sympto- matology in experimental animal models.4,5
Alzheimer’s Disease
One of the systems severely affected in AD are the cholinergic neurons in the basal forebrain and loss of cholinergic function is closely corre- lated with the decline in cognitive performance observed in AD patients. Current pharma- cotherapy, only partially effective, uses anti- choline esterase compounds that reduce neurotransmitter clearance. NGF was previously believed to have a mere neurotrophic role in the survival of cholinergic neurons; however, it is slowly being discovered as a central regulator of
both phenotype and function of these cells. Not only does it enhance neurotransmitter synthesis and sprouting in the uninjured cholinergic neu- rons, but it also directly controls cholinergic neurotransmission by mechanisms independent of gene expression or cell signaling.107 Accordingly, NGF has been shown to improve function in animals bearing cholinergic lesions and has already entered the clinical trial phase for the treatment of AD.
AD is associated with extracellular deposition of the alpha-amyloid protein in senile plaques and the brain microvasculature. It is the product of the alpha-amyloid precursor protein (APP) gene localized on chromosome 21, and is prima- rily, but not exclusively, expressed in neurons.
Interfering with APP expression may reduce the tissue accumulation of the neurotoxic alpha- amyloid protein. For this purpose, antisense oligonucleotides directed at the amyloid beta- peptide region of the APP gene have been used experimentally. Encouraging results from stud- ies with a mouse model of AD that overproduces amyloid beta protein indicate a reversal of cog- nitive defects after both intracerebroventricular and intravenous administration of such anti- sense oligonucleotides.108–110
Manipulating Postinjury CNS Regeneration
Injury to the adult CNS often results in perma- nent loss of cells and neuronal function. This is attributed, among others, to the failure of injured axons to regenerate (reviewed in Spencer and Filbin111 and Fournier and Strittmatter112). Extensive work in recent years has uncovered the nature of CNS inhibitory mechanisms that prevent brain tissue regener- ation and repair. These include the formation of the glial scar, the participation of several myelin-associated molecules that inhibit axonal regrowth, and the dividing status of neurons. To date, three major myelin-associ- ated inhibitors have been identified: Myelin- associated glycoprotein (MAG113), Nogo-A,114 and oligodendrocyte-myelin glycoprotein (OMgp115), all of which exert their effect via the Nogo receptor. Regenerative capacity in the adult CNS may therefore be potentially stimu- lated by the targeting of one or all of these fac- tors after injury. Such manipulation has so far been attempted only in vitro, and highlighted
the potential associated with an approach aimed at inhibiting the inhibition. Suppressing the activity of the Nogo receptor using an AAV vector expressing a dominant-negative form of the Nogo receptor has been shown to increase axon regeneration.116
Critical Issues
Targeting Gene Delivery
Whereas much effort is focused on the construc- tion of vectors with increased gene transfer effi- ciencies and higher safety profiles, the most significant challenge currently faced in gene therapy, prominently exemplified in the CNS, is the achievement of selective transgene expres- sion in a specific cell population within a chosen site. The risks associated with the promiscuous transfer of genetic material mandate the imple- mentation of such selectivity measures.
Conventional pharmacotherapy is based on the systemic administration of a therapeutic drug that is distributed throughout the body via the vascular route. With minimally restricted access to any site or compartment, the speci- ficity of effect is solely determined by the inter- action with the target molecule. Gene therapy thus differs from conventional pharmacother- apy in two major respects. First, the biodistrib- ution of a systemically administered vector is strongly biased by the structure and dynamics of the vascular system. Second, the vector itself does not mediate a physiologic effect, but rather serves as a precursor that is processed at the target site to produce the biologically active fac- tor. As a result, the specificity of effect is deter- mined at five different levels: i) Accessibility to the target cells, ii) entry into the target cell, iii) a permissive intracellular environment that enables processing of the vector to a functional gene, iv) expression of the gene by the host cell transcriptional and translational machin- ery, and v) effect of the transgene protein product.
These levels of specificity were translated into targeting strategies, based on: i) The administra- tion method, ii) selective cellular uptake of the gene-carrying vector, and iii) cell-type-specific transcription.
Modes of Vector Delivery to the CNS
The first consideration in deciding on a mode of vector delivery is the anatomical distribution of afflicted target cells. In PD, for example, the dis- eased cells are confined to a relatively restricted area, lying deep in the substantia nigra, whereas in AD the afflicted cells are more globally dis- tributed throughout the brain and in lysosomal storage diseases each and every cell in the tissue is involved.
A second consideration is the mode of action of the transgene product. Therapeutic agents that are secreted from the transgene-expressing cells (e.g., growth factors and neurotransmitter precursors) theoretically require gene transfer to a relatively small number of cells. In contrast, for those protein products that induce pheno- typic changes restricted to the cell in which they are produced (e.g., ion channels to provide pro- tection from excitotoxic damage), delivery should be conducted so as to achieve maximal gene transfer of the targeted population.
Direct injection. Despite the risks associated with invasive craniotomy-based procedures, as for today, the most common method of vector delivery to the brain used in clinical trials is direct intraparenchymal injection. This approach, how- ever, is associated with local tissue damage by the needle tract, causing secondary damage to the blood-brain barrier (BBB), local inflammation, and hemorrhage. Moreover, standard injection procedures enable vector diffusion over only a few millimeters away from the injection site.117 The resulting gene transfer is nonhomogenous with a short transduction gradient (i.e., cells prox- imal to the injection site are significantly more efficiently transduced than those further away).
To increase vector distribution to more dis- tant parts of the injected tissue, convection- enhanced delivery devices have been developed.
The pressure gradient-dependent flow current of the vector allows the injection of larger prepa- ration volumes with more homogeneously dis- tributed transduction and minimal tissue damage.118–120A setback of this approach is the minimal control over vector flow, which may find its way to unwanted brain regions. Indeed, several studies have shown that diffusion of the injected vector particles throughout the brain is minimally affected by physical tissue barriers, but rather by the interaction between the virus
and tissue components (both cellular and extra- cellular matrix).118,120,121
A fundamental shortcoming of the use of reporter genes for in vivo biodistribution studies should be noted. Marker proteins, such as the E.
coliβ-galactosidase, fluorescent proteins, and the firefly luciferase are most often used to determine vector distribution throughout the tissue. Results of such assays reflect the final product of a long sequence of events, starting with arrival at the site and the cell-entry process, through a complex series of intracellular events and culminating in translation and accumulation of the reporter pro- tein. Therefore, differential processing of the gene product may theoretically bias apparent vector distribution. Indeed, intracerebrally injected ade- noviral or AAV vectors expressing either the β- galactosidase or the HSV-1 thymidine kinase (TK) gene under identical promoter sequences produced significantly varying results.120,122,123
HSV-1-TK immunoreactivity was detected over larger areas of the brain and persisted for several months longer. Transgene-specific properties (and assay sensitivity) therefore bias any study of vector distribution, underscoring reservations that should be applied to reporter gene-based studies. A more prudent approach is thus recom- mended, including both analyses of physical par- ticle distribution (e.g., by fluorescently labeled virions or in situ hybridization for virus-specific sequences) and the use of assays to detect the protein product specific for the application.
The intravascular route. The ability to sys- temically administer brain-targeted vectors is an appealing alternative to the invasive intra- parenchymal procedures. This approach is facil- itated by the tremendously high vascularity of the brain, in which every neuron is directly adja- cent to a capillary. However, access of vectors from the intravascular compartment and into the brain parenchyma is hindered by the cere- bral microvascular endothelium forming the BBB. Transfer across the BBB may be facilitated by temporary disruption of this barrier using an osmotic shock124 or by drug-induced open- ing,125,126but such methods are associated with chronic neuropathologic changes because of increased access of circulating neurotoxins and potentially pathogens to the brain.127,128
An alternative approach, exploiting receptor- mediated transport systems which actively transport large molecules such as insulin and
transferrin across the BBB, was recently intro- duced.129By conjugating a monoclonal antibody to the transferrin receptor to a PEGylated liposome, the complexes undergo receptor- mediated transcytosis across the microvascular endothelial cells, followed by receptor-mediated endocytosis into neurons beyond the BBB.86 Efficient brain delivery of a plasmid expressing the TH gene was shown to result in the normal- ization of striatal TH levels and reverse motor impairments in a neurotoxin-induced rat model of PD. Similarly, targeting the insulin receptor in primates yielded global brain expression of reporter genes restricted to neurons that express the insulin receptor.6
Despite these advances, the most significant disadvantage of systemic vector administration is the biased bodily distribution of these nano- sized particles. Sequestration of the large major- ity of vectors by organs that have a sinusoidal capillary system (i.e., liver, spleen, bone mar- row) requires both higher vector doses and the prevention of transgene expression in these organs.46 In addition, exposure to circulating components of the immune system is associated with vector inactivation and priming of an immune response (see later).
Intraventricular and intrathecal administra- tion. Intraventricular administration of gene- carrying vectors is generally regarded as a noneffective means to achieve neuronal gene transfer in the brain of adult animals. Trans- duction is limited to the ependymal cells lining the ventricles, to the choroid plexus, and only a minimal distance into the brain parenchyma itself.130,131 Nevertheless, this approach can be used to produce transgene products that are secreted into the CSF and may penetrate into the parenchyma.1,130,132 Unfortunately, not all gene products are able to cross tissue barriers in such a manner.133
Transneuronal transfer. The transneuronal portal of entry is used by various neurotropic viruses such as HSV-1, rabies virus, and certain enteroviruses to invade the CNS. Indeed, viruses have been exploited as transneuronal tracers in order to study brain anatomy.134 Although the efferent transsynaptic transfer of the replication- defective vectors used in gene therapy is not pos- sible, this portal may be used to pass secreted transgene products to higher-order neurons.
Thus, vectors administered in the periphery (intramuscularly or subcutaneously) may express the encoded transgenes in the dorsal root ganglia
or anterior horn cell bodies and the secreted pro- tein be accessible to second-order neurons in the spinal column.135Similarly, import of an adenovi- ral vector through the olfactory tracts was shown to result in transgene expression in directly related structures in the brain.136
Interestingly, brain access of the vector itself is not an absolute necessity for entry of the product. In a study by Hennig et al.,137although the intraocularly administered AAV genome remained in the injected eye, efferent transfer of the encoded lysosomal β-glucuronidase enzyme into the CNS occurred, resulting in a reduction of lysosomal storage in brain areas unrelated to the ocular tracts.
Cell Specificity of Gene Transfer
After administration, a fundamental require- ment is the ability to direct and restrict gene expression selectively to the cell type of interest.
This may be achieved by either selective uptake of the vector by specific cells (targeted entry), or by selective transgene expression through the utilization of cell-specific promoters (transcrip- tional targeting). These approaches are comple- mentary; whereas targeted entry better addresses the safety issues concerning vector- induced cellular effects, transcriptional target- ing provides a secondary line of defense to limit expression to the desired cell type.
Targeted entry. Restriction of vector uptake to selected cell types is based on exploiting sur- face components unique to the target cell, which serve as receptors to the gene-carrying vector. Because the basic principles of cell entry are derived from the study of viruses, attempts at retargeting viral vectors preceded that of their nonviral counterparts. Cellular tropism of viruses is determined by an interaction between viral glycoproteins and cell-surface expressed receptors. Manipulation of this interaction therefore stands at the basis for all retargeting attempts, which may be categorized into:
1. Selective use of viral glycoproteins that medi- ate infection of the desired cell type.
2. Genetic engineering of viral glycoproteins to express retargeting moieties.
3. Use of bispecific bridging molecules with specificity to both viral surface and target cell membrane components.
