| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
REVIEW ARTICLE |
1 Department of Neurology
2 Department of Molecular Genetics and Biochemistry, University of Pittsburgh, Pennsylvania.
3 Geriatric Research, Education and Clinical Center, Pittsburgh VA Healthcare System, Pennsylvania.
 |
Abstract |
|---|
Gene therapy was first formally proposed more than 30 years ago (1). The original notion was that genetically determined disorders might be completely "cured," as opposed to simply "treated" if the defective gene were identified and replaced with a corrected copy of the normal gene. Despite its heuristic appeal, correction of even monogenic recessive disease by gene therapy has remained an elusive goal. Only one condition, X-linked severe combined immunodeficiency (SCID) in children, has been successfully treated by gene therapy (2), and even that success has been marred by the development of a lymphoproliferative disorder in 1 of every 10 successfully treated children. Of the common neurodegenerative conditions such as AD, PD, and MND, only a small percentage of cases develop as a result of gene mutations. While these genetically determined forms of neurodegenerative disease have proven extremely useful in unraveling the molecular etiology of the conditions, they do not provide direct targets for therapeutic intervention in treating the far-more-common acquired forms of the disease. However, as the field of gene therapy has developed over the last 30 years, it has become apparent that genes can be used as therapeutic agents to treat conditions other than genetic disease.
The treatment of type I diabetes can be used as an analogy. Even though details of the pathogenetic sequence were not known, the final common pathway involves lack of insulin; for several decades insulin derived from animal sources was used to effectively treat the disease. In the last two decades isolation of the animal peptide was replaced by human insulin produced in vitro in bacteria engineered to contain the insulin gene. Gene transfer takes this process one step further, using the therapeutic gene directly to affect the local or focal expression of a polypeptide to achieve a desired therapeutic effect in polygenic diseases or diseases with no significant genetic component. There are several reasons why this approach is particularly applicable to disease of the brain and the nervous system. The blood–brain barrier limits the penetration of systemically administered macromolecules into brain, and even macromolecules injected directly into the ventricles penetrate only a short distance into brain parenchyma. In many cases, the regional specialization of brain function dictates that a therapeutic intervention may be best achieved by the local expression of a transgene product such as a neurotrophic or antiapoptotic factor. In addition, the widespread and redundant use of a limited repertoire of neurotransmitters and receptors in diverse pathways in the nervous system means that the local production of neurotransmitters achieved by therapeutic gene transfer may be used to achieve desired outcomes while avoiding unwanted adverse side effects that would result from activation of the same receptors in other pathways by a systemically administered drug.
In this review, we will briefly describe the principal techniques that have been employed to accomplish gene transfer, consider the progress that has been made in animal models of the principal categories of neurologic disease, and review the status of the few cases that are moving towards clinical trial.
|
Techniques of Gene Transfer |
|---|
Gene delivery in either case is accomplished through the use of "vectors" that may be derived from viruses (viral vectors) or constructed using nonviral elements (nonviral vectors). The simplest nonviral vectors are plasmids (sometimes referred to as "naked plasmids"), and consist simply of the gene of interest, a promoter element required to drive gene expression, and sequences that allow the propagation of the plasmid in bacteria. Plasmid DNA can be complexed with any one of a number of lipid formulations to create a particle, "liposome", that may enter cells more efficiently than naked plasmids. An advantage of plasmids or liposomes as gene transfer vectors is that manufacture is straightforward, and can be easily standardized to produce a pharmaceutical agent for human use. Unfortunately, naked plasmids have proven to be effective only in transfer of genes into muscle following direct inoculation, and liposome-mediated gene transfer has not proven particularly useful for neuronal cells in vitro or in vivo.
Viral vectors exploit the natural biology of viruses to deliver DNA to the nucleus of the target cell. Genetic modification is used to reduce or remove viral pathogenicity while leaving the targeting characteristics of the viral vector intact. Five different viruses have been modified to create the vectors that are used in most of the gene transfer studies reported to date (Table 1). Retroviral vectors, based principally on the Moloney murine leukemia virus, were the first viral vectors developed. Retroviral vectors, like the parental virus, must transduce actively dividing cells in order for the cDNA transcribed from the viral RNA to be incorporated into the host cell genome. The major nervous system application of ex vivo gene therapy involves retroviral transduction of host fibroblasts to produce nerve growth factor to treat AD, and will be considered in some detail below. Most of the gene transfer applications directed at brain, peripheral nerve, or muscle have employed vectors that do not require host cell division. The vectors employed include those that do not integrate into the host cell genome, such as adenovirus (Ad) and herpes simplex virus (HSV)-based vectors, vectors that do integrate into the genome of nondividing cells such as lentiviral vectors (LV), and the adeno-associated virus (AAV)-based vectors, which clearly integrate in a site-specific manner in vitro, but whose behavior in this regard in vivo has not been fully established. The issue of integration into the host cell genome has important practical implications. In the human trial using retroviral transduction of bone marrow-derived stem cells with a retroviral vector to treat X-linked SCID, insertion of the vector sequences into or near an oncogene in those cells resulted in the development of leukemia in 2 patients.
|
LV vectors were originally designed as replication-defective hybrid viral particles, containing the core proteins and enzymes of HIV-1 and the envelope of a different virus (6). Vector particles are assembled by viral proteins expressed in trans from viral constructs devoid of most of the viral cis-acting sequences (packaging constructs). The viral cis-acting sequences are linked to an expression cassette for the transgene. Both types of constructs are introduced into a cell to produce vector particles (7). The efficiency and safety of this process depends on the segregation of cis and trans acting functions of the viral genome (8). Unlike other retroviruses, LVs contain an integrase function that allows the integration of the viral genome into the chromosomes of nondividing cells (9). LV vectors have proven highly efficient in transducing cells in the nervous system. They appear to induce very little immune response, and with the incorporation of appropriate regulatory elements, provide long-term gene expression at very high levels (10).
HSV, a human pathogen that causes the common cold sore and infections of the conjunctiva, is a double-stranded DNA virus with a capsid and a surrounding tegument and envelope. HSV has a 152 kb genome, and HSV vectors can potentially accommodate up to approximately 44 kb of DNA insert. HSV has undergone numerous manipulations to generate HSV replication-defective vectors designed for gene delivery. One strategy has been to delete essential immediate early genes, a process that renders the vector replication incompetent. The first replication-incompetent HSV vector was generated by removing the infected cell polypeptide-4 gene (11). Analogous to Ad vectors, growth of replication-incompetent HSV vectors in cultured cells is accomplished using cell lines that complement any gene product that is removed from the vector genome. Further immediate early gene deletions have been made in the HSV vector resulting in relatively nontoxic vectors (12). Like Ad, the HSV vector genome remains as an episome in the nucleus of the transduced cell. A unique aspect of HSV vectors, which are neurotrophic, is that HSV can establish latency in neurons. Therefore, the majority of gene transfer applications of HSV vectors has been directed toward the nervous system (13,14).
|
Application of Gene Transfer to the Nervous System |
|---|
|
|
Parkinson's Disease |
|---|
Widely used animal models of PD faithfully reproduce the biochemical phenotype of the disease. Chemical induction of lesions to the SNc of rodents or primates, using the toxins 6-hydroxydopamine (6-OHDA) or 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) leads to a pathogenic cascade resulting in apoptotic cell death of dopamine (DA) neurons, and have been used to study the therapeutic possibilities of gene transfer. Several different approaches have been employed. Gene transfer to provide the local expression of inhibitors of apoptosis in the SNc prevents both the loss of DA neurons and the development of a PD-like phenotype following the chemical insult (15–17). These peptide inhibitors of apoptosis must be expressed intracellularly in order to block the apoptotic cascade, so that gene transfer is uniquely suited to this approach. However, the pathogenic trigger for PD in humans is unknown and, while the biochemical and behavioral phenotype of the animal models is faithful to the human disease, the role of apoptosis in dopaminergic cell death in naturally occurring PD is controversial. In addition, the effects of prolonged expression of antiapoptotic factors in the brain have not been fully explored; some of these proteins are proto-oncogene products, and there might be important issues regarding their safety.
Another approach to prevent cell death in SNc employs the glial cell line-derived neurotrophic factor (GDNF), a peptide that was originally isolated by virtue of its trophic effects on dopaminergic cells in culture. GDNF gene transfer with any one of a number of different vector systems have been successfully used to affect GDNF gene transfer in experimental models, including LV (18–20), Ad (21–24), AAV (25–27), and HSV (28). Several points emerge from these studies, which differ mainly in the details of the experimental paradigms used. First, robust GDNF expression can be seen after gene transfer into the striatum or substantia nigra, and anterograde transport of GDNF to nerve terminals after transduction of the neuronal soma seems to be a property of GDNF rather than the vector system used. Second, GDNF appears to provide trophic support preventing degeneration of dopaminergic cells and loss of dopaminergic nerve terminals in both the 6-OHDA and MPTP models. This protection correlates with behavioral measures of nigrostriatal integrity and neurochemical assays examining DA production. Finally, in many circumstances, the application of GDNF is protective or restorative even after the toxic insult has taken place. A human trial in which the GDNF peptide was administered intracerebroventricularly (29) failed because of adverse effects from the intracerebroventricular administration, and concern about the penetration of the peptide into brain. Direct instillation of GDNF into brain parenchyma is being studied, but gene transfer may offer an alternative approach.
Inhibition of overactive neurons of the subthalamic nucleus (STN) by stereotactic ablation or deep brain stimulation has been shown to ameliorate motor signs in late-stage PD. A gene transfer strategy based on this approach has recently been reported (30). Transduction of STN neurons with glutamic acid decarboxylase (GAD), the rate-limiting enzyme for synthesis of the inhibitory neurotransmitter gamma-amino butyric acid (GABA) using an AAV vector resulted in synthesis and activity-dependent release of GABA from STN nerve terminals. Microelectrode studies in control animals showed that stimulation of the STN resulted in excitation of the majority of substantia nigra pars reticulata (SNr) neurons from which recordings were obtained, consistent with the known glutamatergic neurochemical phenotype of STN neurons. However, stimulation of GAD-transduced STN neurons produced a preponderance of inhibitory responses in the SNr neuron pool, suggesting that expression of GAD and consequent modification of the neurochemical phenotype had altered the physiological properties of the STN–SNc projection. Intriguingly, GAD transduction of the STN appeared to protect SNc dopaminergic neurons from a neurotoxic insult following administration of 6-OHDA (30). The protective effect seemed to be dependent on the induction of an inhibitory phenotype in the STN neurons, as destruction of the STN using ibotenic acid did not protect the SNc DA neurons from 6-OHDA. Combining neuroprotection with functional compensation is attractive; a phase I/II clinical trial has been approved by the U.S. Food and Drug Administration and is slated to begin in the near future.
Traditional pharmacologic therapy of PD involves correction of the neurochemical deficit by systemic delivery of the DA precursor, L-DOPA, or by use of agents that act directly on striatal DA receptors. In the first gene therapy approaches to PD, gene transfer was employed to deliver the gene coding for tyrosine hydroxylase (TH), the rate-limiting enzyme for DA formation to the striatum (31,32). Gene transfer resulted in enhanced DA production and observable behavioral benefit in a rodent model. More recently, it has been demonstrated that simultaneous delivery of multiple genes encoding enzymes that drive DA synthesis more effectively corrects the DA-deficient phenotype than single-enzyme replacement. Synthesis of DA from tyrosine depends on two reactions, catalyzed by TH and aromatic acid decarboxylase (AADC). The first requires a cofactor that is synthesized by glutamyl transpeptidase (GTP)-glutamyl cyclohydrolase I (GCH1). Various vector systems have been used in recent preclinical studies to deliver different combinations of these enzymes. These include multicistronic LVs simultaneously encoding GCH1, TH, and AADC (33), and combinations of AAV vectors separately encoding GCH1 and TH (34) or GCH1, TH, and AADC (35,36). In all cases, coexpression of the enzymes and functional recovery of the experimentally lesioned animals was observed. A further refinement to this strategy has used coexpression of the vesicular monoamine transporter with AADC to enhance the uptake of nascent DA into synaptic vesicles after its synthesis (37). While transgene-mediated enhanced DA expression has been demonstrated to effectively correct the motor phenotype in lesioned rodent and primate models, it is unclear at present whether the nonphysiological sustained delivery of DA in the striatum by these kinds of approaches will alleviate or exacerbate the problem of adverse effects. Because the therapeutic and toxic doses of L-DOPA alter in individual patients over the course of the disease, it is likely that, as a minimum, it will be necessary to design vectors with inducible enzyme expression or enzymes with controllable activity before clinical application of this kind of gene therapy can be contemplated.
|
Gene Therapy for Diffuse Brain Disease |
|---|
An alternative approach to treating AD is to attempt to prevent the accumulation of amyloid protein in the brain, for example, by producing an enzyme such as neprilysin that would degrade amyloid peptides in vivo. The principle of using gene transfer to affect release of an enzyme diffusely through the brain has, to date, been explored principally in studies of lysosomal storage diseases such as MPS. For example, injection of a recombinant Ad vector expressing beta-glucuronidase directly into the lateral ventricles of mutant mice increased the beta-glucuronidase activity in crude brain homogenates to 30% of heterozygote activity. Histochemical demonstration of beta-glucuronidase activity in the brain revealed that the enzymatic activity was found principally in ependymal cells and choroids plexus (39). An Ad vector expressing aspartylglucosaminidase (AGA) injected intraventricularly into the brain mice with aspartylglucosaminuria resulted in AGA expression in the ependymal cells lining the ventricles and diffusion of AGA into the neighboring neurons. One month after administration of the wild-type Ad-AGA, a total correction of lysosomal storage in the liver and a partial correction in brain tissue surrounding the ventricles was observed (40). Similar results have been obtained in the MPS VII mouse injected with an Ad vector expressing beta-glucuronidase, with the distribution of enzyme activity and phenotypic correction increased by mannitol-induced disruption of the brain–cerebrospinal fluid barrier (41), or with AAV vectors expressing beta-glucuronidase injected directly into brain parenchyma (42,43). One advantage of using gene transfer to express an enzyme is that the secreted enzyme may disseminate along the neuraxis, resulting in widespread reversal of the hallmark pathology, and suggesting that a limited number of appropriately spaced sites of gene transfer may provide overlapping spheres of enzyme diffusion to cover a large volume of brain tissue (42,44,45). AAV-mediated correction has also been reported to improve cognitive function in the murine model of MPS VII (46), and injection of a feline immunodeficiency virus-based vector expressing beta-glucuronidase into striatum unilaterally resulted in bihemispheric correction of the characteristic cellular pathology, and correction of established impairments in spatial learning and memory resulted in a dramatic recovery of behavioral function (47).
In the mouse model of MPS IIIB resulting from a defect in alpha-N-acetylglucosaminidase (NaGlu), a NaGlu-expressing AAV vector injected into brain resulted in 6 months of expression of rNaGlu in multiple brain regions of adult MPS IIIB mice. The vector transduced an area of 400–500 microns surrounding the infusion sites, but after 6 months the correction of glycose amino glycan storage involved neurons of a much larger area (48). In a mouse model of metachromatic leucodystrophy, Naldini and colleagues demonstrated that an LV vector encoding a functional arylsulfatase A gene injected into the brain of adult mice with germ-line inactivation of the mouse gene-encoding arylsulfatase A resulted in sustained expression of active enzyme throughout a large portion of the brain, with long-term protection from development of neuropathology and hippocampal-related learning impairments (49).
Correction of phenotypic deficits in both histology and behavior in MPS mice using gene transfer has been impressive, and the reversal of established deficits (47) represents an important clinical feature in consideration of the development of a practical treatment. Several features of this model should be kept in mind. The relevant gene product is taken up by cells throughout the brain by binding to mannose-6-phosphate receptors. Thus, global correction of these diseases can be achieved by transduction of a fraction of cells within the brain as long as the gene product released from the cells is adequately distributed through the brain. In other models using enzyme replacement, it has been noted that replacement of as little as 10% of the normal enzyme activity may be sufficient to correct the phenotype. Regarding the application to human disease, issues of volume of distribution need to be explored. Even though correction of an animal model has not yet been demonstrated, a human trial of gene transfer to treat Canavan disease using liposomes to transfer aspartoacylase has been reported (50), and the same group has now begun a similar study in children using an AAV vector to express the enzyme in the brain.
|
Diseases of Peripheral Nervous System: Polyneuropathy and MND |
|---|
 TopAbstractTechniques of Gene TransferApplication of Gene Transfer...Parkinson's DiseaseGene Therapy for Diffuse... Diseases of Peripheral Nervous... Gene Transfer for Treatment...SummaryReferences |
|---|
Using transduction of DRG neurons by peripheral inoculation of an HSV vector, we have demonstrated a protective effect against the development of neuropathy in 3 different models of polyneuropathy. Selective large fiber nerve degeneration caused by overdose of pyridoxine can be prevented by subcutaneous inoculation of an HSV-based vector containing the coding sequence for NT-3, measured by the amplitude and conduction velocity of the evoked sensory response, as well as preservation of H-wave amplitude (54). Treated animals show preservation of a population of large myelinated fibers that otherwise degenerate in this condition, and the preservation of electrophysiologic and histologic parameters is reflected in behavioral testing of treated animals (54). Inoculation of an HSV-based vector expressing NGF under the control of the human cytomegalovirus promoter prior to the start of pyridoxine intoxication provides a similar protective effect (55). Injection of an replication-incompetent HSV vector expressing NGF under the control of the human cytomegalovirus promoter 2 weeks after the induction of diabetes (by injection of streptozotocin) prevents the development of neuropathy measured by reduction in evoked sensory nerve amplitude, and also increases expression of neuropeptides in the DRG (56). A protective effect has also been observed by transfer of VEGF using a plasmid injected into muscle in models of ischemic and diabetic neuropathy (57,58), although one must assume that the protective effect in those models results from circulating levels of VEGF achieved by muscle transduction and thus may not avoid the potential for systemic side effects. MND is a serious and fatal affliction without currently effective treatment. Like polyneuropathies, administration of trophic factors appears to slow the progression of the disease in rodent models, but a human trial of ciliary neuronotrophic factor in MND had to be abandoned because of the cytokine-like side effects of the systemically administered trophic factor (52). An AAV-based vector expressing GDNF has been demonstrated to protect a motor neuron-like cell line from apoptotic cell death in vitro (59). After intramuscular injection of the NT-3 adenoviral vector, pmn mice (a model of MND) showed a 50% increase in life span, reduced loss of motor axons, and improved neuromuscular function as assessed by electromyography. These results were further improved by co-injecting Ad vector coding for ciliary neuronotrophic factor (60). Administration of an Ad vector expressing cardiotrophin 1 (CT-1) to newborn pmn mice led to sustained CT-1 expression in the injected muscles and bloodstream, prolonged survival of animals, and improved motor functions. CT-1-treated mice showed a significantly reduced degeneration of facial motor neurons and phrenic nerve myelinated axons. The terminal innervation of skeletal muscle, grossly disturbed in untreated pmn mice, was almost completely preserved in CT-1-treated pmn mice (61). This approach relies on systemic release from injected muscle, and thus may not avoid the problems of systemic administration. Achieving adequate systemic levels from muscle transduction in larger animals may prove difficult. To date, no vectors have been created from viruses that would naturally target motor neurons in a manner similar to the targeting of DRG neurons by HSV-based vectors, and efforts to construct vectors that would target to motor neurons have been unsuccessful to date.
|
Gene Transfer for Treatment of Pain |
|---|
Iadarola and coworkers demonstrated that a recombinant Ad encoding a secreted form beta-endorphin injected intrathecally into lumbar cerebrospinal fluid transduced meningeal cells, and that beta-endorphin secretion attenuated inflammatory hyperalgesia without affecting basal nociceptive response (62). HSV-mediated gene transfer to deliver and express opioid peptides to be released from primary afferent terminals may be used to alter the physiology of postsynaptic neurons, affecting nociceptive transmission in the spinal dorsal horn. An HSV vector containing the human proenkephalin gene injected subcutaneously in the foot produces an antihyperalgesic effect in rodents (63), and a 50% reduction in the spontaneous pain behavior during the delayed phase of the formalin test of inflammatory pain (64). The naltrexone-reversible analgesic effect in inflammatory pain is maximal 1 week after vector inoculation, and can be reestablished by reinoculation of the vector after the initial effect has waned (64). In a spinal nerve ligation model of neuropathic pain, injection of the vector 1 week after spinal nerve ligation produced a naloxone-reversible antiallodynic effect that was continuous, persisted for several weeks, and could also be reestablished by reinoculation of the vector after the original effect had waned. In the neuropathic pain model, vector-mediated enkephalin expression enhances the effect of morphine, reducing the ED50 of morphine from 1.8 mg/kg to 0.15 mg/kg, and the vector continues to provide an antiallodynic effect in the face of tolerance to morphine induced by repeated injection of the drug (65). A similar analgesic effect for HSV-mediated expression of proenkephalin has been demonstrated in a model of polyarthritis (66) and in a rodent model of pain caused by cancer in bone (67). The latter study represents a model that would be an appropriate first test of this approach in humans, and we presented a proposal for a phase I human trial of the proenkephalin-expressing vector in the treatment of pain resulting from cancer metastatic to bone to the National Institutes of Health's Recombinant DNA Advisory Committee in June 2002.
|
Summary |
|---|
For diseases that commonly afflict the aging nervous system, it seems likely that gene transfer will be used to target the delivery of therapeutic proteins of peptides to specific regions of the nervous system to achieve therapeutic effects to treat conditions that are not principally genetic in origin. Human trials for the treatment of AD have begun, and other trials targeting PD and intractable pain are poised to begin. The results of these studies will serve as an important milestone in the development of this field.
|
Acknowledgments |
|---|
Received February 5, 2003
Accepted May 9, 2003
|
References |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
