Outline:
1. Possible roles of gene therapy in the CNS
2. Vector systems for delivery of genetic material to the CNS Retroviruses Recombinant herpesvirus Amplicon type defective herpesvirus vectors Possible drawbacks to HSV-1 viral vectors Other vectors for CNS gene therapy
3. Delivery of genetic material to the CNS The problem of the BBB Previous studies of virus delivery to brain Delivery of agents across the BBB by osmotic BBB disruption BBB disruption in the feline
4. Gene therapy of lysosomal storage disease A feline model of Gm2-gangliosidosis (Sandhoff disease) Requirements of a model system for gene therapy Gene therapy of animal models
1. POSSIBLE ROLES FOR GENE THERAPY IN THE CNS:
Recent advances in molecular biology have made gene therapy of the CNS a realistic goal. This review will focus on the potential of gene replacement therapy for correction of global neurodegenerative inherited enzyme deficiencies. However, gene therapy may also play a role in restoration of function in localized neurodegenerative metabolic disorders, and in treatment of malignant tumors.
Gene replacement therapy would correct genetic enzyme or protein deficiencies at the subcellular level. Although specific recessive inherited enzyme deficiencies are very rare, collectively they account for a significant proportion of neurodegenerative disease (Neuwelt et al. 1985; Rosenberg 1984). The lysosomal storage disorders are a major subset of genetic enzyme deficiencies characterized by accumulation of enzyme substrate in the lysosome, which can be particularly devastating in the CNS. Attempts to ameliorate the neurodegeneration resulting from lysosomal storage disorders have included enzyme replacement therapy by direct infusion of purified active enzyme, and bone marrow transplantation with donor cells producing the enzyme. Treatment of genetic enzyme deficiencies with somatic cell gene replacement bypasses many of the problems inherent in transplantation and enzyme replacement techniques and instead focuses on the source of the problem - a mutation resulting in a defective gene. The protein affected and the genetic defect responsible are known for many metabolic disorders. For gene replacement therapy, a normal gene is introduced into the patient in the appropriate tissue (CNS), in the appropriate cells (neurons and/or glia), thereby correcting the consequences of the defective gene (Fleischman 1991; Freese et al. 1990; Rosenberg 1984). Systemic gene therapy for genetically inherited errors of metabolism is already undergoing clinical trials in humans (Fleischman 1991). For CNS therapy, global delivery of corrective genetic material will be necessary.
The goal with CNS gene therapy would be to intervene as early as possible in the course of the disease in order to not only halt the progression of neurodegeneration, but to actually cure the disease. Infantile onset forms of neurodegenerative disorders of metabolism are characterized by motor and mental deterioration leading to an early death, and potentially irreversible changes are found in utero (Adachi et al. 1974). However, except for parents with a previous affected child, or known carrier populations, such as the Ashkenazi Jews and Tay-Sachs disease, in utero testing is not performed. For many of these diseases, there is a range of disease phenotypes related to the total level of enzyme activity (Leinekugel et al. 1992). The less severe juvenile and adult variants that have been characterized may be more amenable to gene therapy.
In addition to inborn errors of metabolism, gene therapy has the potential for restoration of neuronal function in late-onset localized neurodegenerative disorders such as Parkinson disease or Huntington's disease, or even Alzheimer's disease. The complexity of such illnesses will increase the difficulty of gene therapy. For example, the etiology of Parkinson disease is not known, but it is characterized by degeneration of the dopaminergic neurons of the nigrostriatal pathway. Release of dopamine from grafted fibroblasts (Wolff et al. 1989), or microencapsulated PC12 cells (Winn et al. 1991) reverses the disease phenotype in the rat model of Parkinson disease. Therefore, gene replacement with tyrosine hydroxylase (TH), the rate limiting enzyme in dopamine biosynthesis may be a plausible therapy (Horellou et al. 1989; Wolff et al. 1989). However, in order to bring about a sustained cure it may be necessary to introduce two (or more) genes since both TH and aromatic amino acid decarboxylase may be required to increase dopamine production. In disorders in which the underlying genetic defect is unknown, gene therapy with an apparently therapeutic gene may be palliative rather than curative. In addition to delivery and viral vector issues, gene therapy of diseases such as Parkinson disease will require gene transfer to a subset of neurons at specific locations within the brain, as well as tight regulation of gene transcription and protein expression.
Gene therapy may be a mechanism for the treatment of brain tumors. Reversal of the transformed phenotype of individual tumor cells can be accomplished by replacement or augmentation of a tumor suppressor gene such as P53 or the retinoblastoma gene (Huang et al. 1988). Inactivation of growth promoting activities such as oncogenes or autocrine growth factors can be accomplished by introduction of antisense genes (Rivera et al. 1989). These approaches require that a single gene defect be determined so that therapy with that specific gene will be efficacious, a goal that is not likely to be realized for neoplastic brain tumors. An approach that would be independent of specific neoplastic mutations is induction of a toxic phenotype in the tumor cells by expression of genes which confer a growth disadvantage or drug susceptibility. Transfection of murine sarcoma and lymphoma cells with the HSV-1 thymidine kinase (tk) gene in a retroviral vector induced susceptibility to antiviral drugs such as ganciclovir, which is activated by HSV-tk (Culver et al. 1992; Moolten et al. 1990; Ram et al. 1993). Alternatively, infection with the HSV-1 virus lacking the tk gene can specifically target tumor cells in the brain because normal brain tissues are poorly infected with tk negative HSV-1 (Martuza et al. 1991; Markert, et al. 1993). Development of tumor-specific agents with minimal neuronal toxicity, and delivery of these agents to tumor, especially those invasive neoplastic cells distant from the main tumor mass, will be major impediments to gene therapy of brain tumors.
2. VECTOR SYSTEMS FOR DELIVERY OF GENETIC MATERIAL TO THE CNS.
RETROVIRUSES: The most widely used vector systems in gene therapy are replication- defective murine and avian retroviruses packaged into infectious virions with helper virus. Retroviruses are capable of infecting a wide variety of cell types, causing integration of the recombinant viral material into random sites within the genome (Varmus 1988). Retroviruses may be used to deliver therapeutic genes into cells which are then grafted into the brain for treatment of localized CNS disease (Horellou et al. 1989; Wolff et al. 1989), or toxic genes for the treatment of tumors (Huang et al. 1988; Moolten et al. 1990). However, for a variety of reasons, retroviruses are unlikely to be useful in global CNS gene replacement (Table 1). Even in situations where a retrovirus may be effective there are limitations to this vector system that have yet to be overcome. For example, the efficiency of virus delivery and gene transfer in vivo limits the use of retroviruses in tumor therapy (Ram et al. 1993; Short et al. 1990). Long term sustained gene expression in retrovirus transfected cells in tissue grafts in brain may be problematic. The efficiency and persistence of retrovirus-mediated gene expression in the absence of selective pressure decreases in vivo (Friedmann 1989). Palmer et al. (1991) found that while transplanted transgenic skin fibroblasts persisted at constant levels for months, transgene expression rapidly decreased by three orders of magnitude.
Table 1. Gene therapy with retroviruses _______________________________________________________ Target limited to mitotic cells (fibroblasts, tumor cells, bone marrow). Limited size of DNA insert (4-8 kb). Viral gene insertion at random sites in host genome
Regulation of gene expression depends on insertion site.
Small possibility that viral integration may disturb cell function. Low titers and low levels of viral integration. Transient gene expression in the absence of selective pressure. ________________________________________________________ Problems associated with gene therapy using retroviruses that render it inappropriate for the treatment of CNS disorders.
Retroviruses have additional limitations which restrict their use in gene therapy of the CNS (Table 1). Most importantly, retroviral integration and gene expression requires target cells which undergo mitosis and cell division (Varmus 1988). Thus, while tumor cells or cell grafts can be subjected to retrovirus-mediated gene transfer, neurons and other post-mitotic cells cannot be transformed with retroviral vectors. Since glial cells divide at a low rate it may be possible to transfect glia with recombinant retroviruses. In one study (Short et al. 1990), injection of a retrovirus into the CNS resulted in little or no gene expression in parenchymal cells of the adult rat brain. Even in the presence of a retrovirus packaging cell line graft, few if any normal brain cells were seen to produce a recombinant gene product after 1 week.
Another drawback to retroviral-mediated gene transfer is the small possibility that the retroviral insertion will disrupt normal cell function. For example, insertion in the midst of normal gene sequences could turn off a necessary enzyme or disrupt normal regulation. In contrast, insertion in promoter sequences could lead to inappropriate gene expression, leading to oncogene activation and cell transformation. Finally, retroviruses can accommodate insertion of a relatively small amount (4-8 kb) of foreign genetic material, which effectively limits the insert to cDNA rather than genomic clones (Varmus 1988). The size limitation increases the difficulty of multiple gene replacement and cell-type specific control through promoter and enhancer sequences.
RECOMBINANT HERPESVIRUS: Vectors derived from HSV-1 have a number of advantages which make them suitable for CNS gene therapy (Table 2). HSV-1 can infect a wide range of cell types, including mature, post-mitotic neurons in adult CNS. This neurotropic virus establishes long-term residence in the neuronal cell body as a circular or concatameric extrachromosomal element, averting the problem of inappropriate insertion into the host genome. HSV-1 can enter a latent state in which only a small number of transcripts are expressed, and remain benign for the life time of the host. The virus genome contains about 150 kb of double-stranded DNA that encodes approximately 75 genes (McGeoch et al. 1988). For a detailed review of the current status of recombinant HSV-1 viral vectors, see Breakefield and DeLuca (1991).
Table 2. CNS gene therapy with recombinant HSV-1 _______________________________________________________ Advantages:
Infection of neurons and other post-mitotic cells
No DNA insertion into host genome
High titers (some replication compromised viruses)
Virus can enter latency Disadvantages:
Lytic infection (replication compromised)
Neurotoxicity
Recombination to wild type or lytic virus
Transient gene expression
Low titer (some replication defective viruses; amplicon) ____________________________________________________________ Advantages and disadvantages to the use of recombinant HSV-1 viral vectors for the treatment of CNS disorders.
Several investigators are developing attenuated strains of HSV-1 in which a gene involved in HSV-1 replication, such as the immediate early viral genes ICP0 or ICP4, or the gene for the early viral protein tk is replaced with lacZ or other transgene. Mutants in which the ICP4 gene (also known as IE3) encoding the viral transcription factor Vmw175 (Davidson and Stow 1985) is disabled, such as lacZ-expressing viral vectors Gal4 (Chiocca et al. 1990; Huang et al. 1992), Cgal3 (Johnson et al. 1992) and 8117/43 (Dobson et al. 1990), are completely replication defective and do not go beyond primary infection. Although replication defective vectors seem ideal for neuronal gene transfer, these particular mutants can still remain toxic in vitro (Cgal3, Johnson et al. 1992), or have low infectivity and transient gene expression in vivo (Gal4, Chiocca et al. 1990).
Replication compromised recombinant HSV-1 mutants grow at a reduced rate, such as ICP0 negative mutant 7134 (Chiocca et al. 1990; Huang et al. 1992), or replicate only in cells that undergo mitosis, such as tk negative mutant RH105 (Chiocca et al. 1990; Huang et al. 1992). The high levels of endogenous tk in mitotic cells can substitute for viral tk, which theoretically allows one to target lytic growth to brain tumor cells (Martuza et al. 1991). Other mutants retain growth competence but have reduced neurotoxicity, such as the R3616 mutant (Markert et al. 1993) which lacks the 34.5 gene involved in preempting neuronal protein synthesis. These mutant recombinant HSV-1 have drastically reduced neurotoxicity compared to wild type HSV-1 and can be produced in high titers. Use of virus mutants that retain the ability to replicate has the severe disadvantage of neuronal cytotoxicity. It is possible that this may be overcome by generating mutants which enter latency and do not reactivate. In the latent state of infection, only the latency associated transcripts (LAT) genes are expressed (Sawtell and Thompson 1992; Wolfe et al. 1992). Recombinant genes under control of the latency specific promoters could thus be constitutively expressed, without further neurotoxicity.
Several investigators have examined in vivo infection, toxicity and gene expression from recombinant HSV-1 expressing marker genes (Huang et al. 1992; Chiocca et al. 1990; Fink et al. 1992; Breakefield and DeLuca 1991). Long term (24 weeks) non-neurotoxic expression of the lacZ transgene was observed in mouse sensory neurons after mutant HSV-1 infection (Dobson et al. 1990). Pilot studies of neuronal gene replacement have also begun with recombinant HSV-1. A replication compromised tk-negative HSV-1 expressing the hypoxanthine-guanine phosphoribosyltransferase (HPRT) gene (Palella et al. 1988, 1989). This viral vector can infect rat neuroma cells in culture and express human HPRT mRNA (Palella et al. 1988), and can also transfer the HPRT gene to mouse brain (Palella et al. 1989). In another in vivo study, Wolfe et al. (1992) used a recombinant HSV-1 to successfully transfer -glucuronidase to the CNS of a murine model of mucopolysaccharidosis VII.
AMPLICON TYPE DEFECTIVE HERPESVIRUS VECTORS. A promising vector system derived from HSV-1 is the amplicon cloning-amplifying vector originally developed by Spaete and Frenkel (1982). We have modified this system to produce the replication defective pHSV constructs shown in figure 1 (Geller and Breakefield 1988; Geller et al. 1990). The prototype vector pHSVlac contains the bacterial lacZ gene as a marker for gene expression, under the control of the HSV-1 IE 4/5 promoter. The pHSV vector construct lacks most of the HSV-1 genome and contains only the HSV-1 origin of DNA replication, the HSV-1 a(1) packaging site and the IE 4/5 promoter, required for DNA replication, packaging and recombinant gene expression respectively. This vector can be maintained as a plasmid in bacteria and series of head to tail concatamers can be packaged into infectious virions with helper virus. A packaging system that utilizes helper HSV-1 with an ICP4 (IE3) gene deletion, in combination with a cell line that provides the ICP4 gene product (Davidson and Stow 1985), yields product virions with a titer of 2 x 107 (Geller et al. 1990) as diagrammed in figure 2. Low levels of recombination to wild type are observed with this system. Reversion frequencies of 1-9 x 10-5 may be reduced further in the future by using double mutant helper virus.
A number of studies suggest that pHSVlac can stably express a gene in both peripheral and CNS neurons, in culture or in the adult mammalian brain. Expression of the lacZ gene from pHSVlac was observed for 2 weeks to several months in cultured neurons from throughout the nervous system (Geller and Freese 1990; Geller et al. 1991). Following stereotactic injection of HSV-1 vectors into the brain of adult rats stable expression of several recombinant genes has been observed in neurons around the injection site as well as in neurons projecting to the injection site (Freese et al. 1990; Geller et al. 1991). Initial attempts to develop gene therapy approaches to neurodegenerative diseases with localized sites of action have been encouraging. A HSV-1 viral vector that expresses nerve growth factor can prevent effects of peripheral nerve axotomy (Federoff et al. 1992) and is currently under study in the CNS. Apart from gene therapy, amplicon-mediated gene transfer can also be used to examine neuronal physiology. In one study, expression of a constitutively active adenylate cyclase from an amplicon vector in cultured sympathetic neurons leads to increases in neurotransmitter release that can be measured for at least one week (Geller et al. 1993). Transfer of the glucose transporter gene to adult rat hippocampus has been achieved using an amplicon type HSV-1 vector, as a first step in altering CNS physiology to counter the neurotoxic effects of metabolic disorders (Ho et al. 1993). We have begun studies with both the normal human and normal feline HEXB cDNAs inserted into a defective pHSV-1 vector.
POSSIBLE DRAWBACKS OF HSV-1 VIRAL VECTORS. Both recombinant HSV-1 and our replication defective amplicon HSV-1 vector systems have potential for use in global neurodegenerative diseases such as the lysosomal storage diseases, but many questions remain to be addressed (Table 2). Disadvantages of the use of recombinant HSV-1 include the relative difficulty of gene manipulation in the viral genome because of its large size and because many important genes are duplicated. This can be overcome by utilizing the amplicon system or by using plasmid cloning cassettes and recombination in culture. Viral titer remains a problem. The titers of defective HSV-1 vectors currently obtained may be adequate for therapy of localized neurodegeneration (Federoff et al. 1992) and for local alteration of CNS physiology (Ho et al. 1993). However, higher titers are likely to be required for application to global neurodegenerative diseases since the defective HSV-1 vectors do not replicate in vivo. In the current generation of viral vectors there appears to be a trade-off in that viruses with low neurotoxicity also have low infectivity, gene expression and titer. Like retroviruses, HSV-1 transgenes under control of heterologous promoters, even strong viral promoters such as the CMV promoter, may decrease expression in vivo (Dobson et al. 1990). In addition, the viral immediate early promoters used in the amplicon system, as well as many recombinant HSV-1 vectors, have only transient activity, as they normally function only in the initial stage of infection. Although the HSV-1 constructs have been shown to infect brain cells and express transgene for weeks to several months, further studies are required to determine how persistent such gene expression will prove to be in vivo.
The major problem with using HSV-1 as a gene transfer vector is neurotoxicity. Productive HSV-1 infection is lytic, resulting in necrosis. Viral encephalitis is a possible complication of any of the replication competent HSV-1 mutants, and is not an acceptable alternative to neurodegenerative disease. The defective HSV-1 viral vectors as well as the deletion mutant helper virus used to package the amplicon DNA can revert to wild type at a low but finite frequency due to homologous recombination with the complementing gene present in the packaging cell line; therefore, improved packaging systems will be required before HSV-1 gene vectors can be used in humans. In vivo, the presence of latent wild type virus is possible, with the resulting possibility of recombination and spread of mutant virus. It may be impossible to treat patients who are seropositive for anti-herpes antibodies, or pretreatment with anti- herpes drugs may be necessary. A more difficult problem will be to counter the neurotoxic attributes of viral genes and viral proteins. Even the most replication defective viruses may maintain cytotoxicity induced by expression of the immediate early genes, or by the viral capsid proteins internalized with benign viral genome. The ICP4 negative Cgal3 virus was found to have cytopathic effects in cultured cells even though it was incapable of lytic growth (Johnson et al. 1992). The authors found that many replication defective and replication compromised HSV-1 were cytotoxic within three days of infection (Johnson et al. 1992). However, other laboratories have demonstrated few cytopathic effects in vivo using similar virus mutants (Chiocca et al. 1990; Dobson et al. 1990; Federoff et al. 1992; Geller et al. 1991; Huang et al. 1992).
ADENOVIRUS AND ADENO-ASSOCIATED VIRUS (AAV): Recently, adenovirus has been shown to be a promising vector for gene transfer to quiescent cells, including neurons (Le Gal La Salle et al. 1993; Davidson et al. 1993; Akli et al. 1993; Bajocchi et al. 1993). Adenovirus has a number of similarities to HSV-1 that make it suitable for neuronal gene therapy, without the disadvantages of lytic infection and acute neuronal toxicity (Table 3). Adenovirus has a large genome which can accommodate at least 7.5 kb, with a theoretical capability of 35 kb, of foreign genetic material, and it can be produced in the high titers (1010-1011 pfu/ml) necessary for global gene therapy (Horwitz 1991). Like HSV-1, adenovirus genes are expressed without DNA integration, and persistent latent infections can be established, although the mechanisms for infection and latency of adenovirus have not been as well characterized as for HSV-1 (Horwitz 1991). Adenovirus can be made replication defective by deleting immediate early genes such as the E1 gene, involved in immortalization and transformation, or the E3 gene, which is involved in host gene turn off and is important in latency (Horwitz 1991; Bajocchi 1993).
Table 3. CNS gene therapy with adenovirus vectors _______________________________________________________ Advantages:
Infection of neurons and other quiescent cell types
Gene expression without DNA integration in host genome
Theoretical insert capacity of 35 kb
No oncogenicity in humans or animal models
Low levels of neuronal toxicity Disadvantages:
Possible inflammatory response
Transient gene expression ________________________________________________________ Advantages and disadvantages to the use of adenovirus vectors for the treatment of CNS disorders.
Following direct inoculation of lacZ -expressing replication defective adenovirus into the caudate putamen, extensive regions of the brain parenchyma were positive for Gal activity, which continued to be detected in some neurons for up to 8 weeks (Davidson et al. 1993). In another study (Akli et al. 1993), transgene expression was detected in neurons, astrocytes and other cells within 0.5-1 mm of the injections site for up to 45 days, with no signs of cytopathic effects except at very high titers of infection. Adenovirus thus appears to be able to transfect a variety of CNS cells types, and gene expression lasts, in some cases, as long or longer than any of the HSV studies to date. However, when defective adenoviruses were used to transfer the gene for 1- antitrypsin to ependymal cells in the CNS, protein secretion was only observed for one week (Bajocchi et al. 1993). This may be the result of the different promoters used in these vectors, or cell type specific gene inactivation. The neurotoxicity that was observed with concentrations of adenovirus vector above 109 pfu/ml indicates that characterization and abrogation of the mechanisms governing adenovirus toxicity will be necessary before these vectors can be used in global gene replacement therapy. An additional concern is the pervasive presence of anti-adenovirus antibodies in adults (Straus 1984), which could result in an inflammatory response or reduced efficacy of adenovirus vectors.
Adeno-associated virus is a small vector (5 kb maximum DNA insert) that integrates into the host genome as a provirus, by an unknown mechanism (Muzyczka 1992). Replication defective mutants require both wild-type AAV and adenovirus for replication (Muzyczka 1992). While this decreases the possibility of a productive infection, it increases the technical difficulty of producing AAV mutants, and limits the titers that can be attained. A number of basic questions remain to be determined about AAV: Will it infect neurons? Is cell division necessary? Is there any chronic toxicity associated with this vector? Nevertheless, AAV can transduce a variety of cells with high frequency, and it is not associated with any human disease (Muzyczka 1992). Recently, Walsh et al. (1992) were able to obtain stable human gamma-globin gene insertion in erythroleukemia cells using a replication negative AAV vector, with regulated high levels of gene expression. Thus AAV shows promise as a gene therapy vector, but it remains to be seen if it will be an effective vector for treatment of global neurodegenerative disease.
OTHER VECTORS FOR CNS GENE THERAPY: Encapsulation of plasmid DNA in cationic liposomes provides a vector system that may circumvent the toxicity inherent in replicative or nonreplicative viruses, while still protecting the DNA from the extracellular environment and providing a mechanism (lipid fusion) for DNA transfer to target cells. Inoculation of liposome:DNA complexes into rodent brain results in DNA incorporation and gene expression by brain cells, albeit in a very transient manner (Jiao et al. 1992; Ono et al. 1990). More recently, i.v. infusion of liposome:DNA complexes resulted in transgene expression in a number of tissues for at least 9 weeks, with no sign of toxicity (Zhu et al. 1993). Plasmid DNA vectors have the advantage that they are easy to clone, can be produced in quantity, and probably will not integrate into the host genome. Transfection efficiency of such vectors may be low, and cell type specificity may be difficult to attain. In addition, it remains to be seen if sustained gene expression can be obtained in the CNS using these vectors. As with all gene transfer vectors, development of promoters to attain long term, high levels of gene expression in a cell-type specific manner will be a major emphasis of future research.
3. DELIVERY OF GENETIC MATERIAL TO THE CNS:
The problem of the BBB. A major impediment to gene replacement therapy of a disease affecting the CNS, either locally or globally, is the delivery of recombinant genetic material to the brain. The BBB is a capillary barrier that results from a continuous layer of endothelial cells bound together with tight junctions that allow very little transcellular or pericellular transport of blood borne molecules (Rapoport 1986; Neuwelt 1989). The barrier excludes molecules from the brain based on electric charge, lipid solubility and molecular weight, and normally excludes molecules of greater than 180 Mr. Large biological molecules such as antibodies and complexes such as viruses have very little access across the normal BBB (Rapoport 1986; Neuwelt 1989; Neuwelt et al. 1991), and even the leaky barrier present in diseased brain or tumor has significantly reduced permeability to large particles. Circumventing the BBB to gain access of therapeutic agents to brain and brain tumor has been a major area of emphasis of this laboratory (Neuwelt 1989; Neuwelt et al. 1981-1993). Because of the impermeability of the BBB, systemically administered enzyme, enzyme producing cells, or genetic vectors are unable to correct neurodegenerative diseases.
Previous studies of virus delivery to brain: Infection of the CNS with wild type HSV-1 generally results from neurotropic spread from the periphery via the olfactory bulb, leading to viral encephalitis in 2.3 cases per million population per year in the USA (Corey and Spear 1986). In experimental animals, corneal inoculation with wild type HSV-1 can lead to latent infection of defined areas of the CNS, during which a transgene is constitutively expressed (Sawtell and Thompson 1992; Wolfe et al. 1992). This in vivo mechanism requires productive infection, with secondary or tertiary spread to the neurons of final residence, and cannot be used with the replication defective viral vectors that are being studied for gene replacement therapy. Furthermore, even if global delivery of recombinant virus could be achieved by this mechanism, widespread viral encephalitis is not a reasonable alternative to neurological disease.
A straight-forward mechanism to gain access of replication defective recombinant material to the brain is direct injection of mutant HSV-1, adenovirus, or other genetic vector into the brain parenchyma (Chiocca et al. 1990; Freese et al. 1990; Davidson et al. 1993). Focal inoculation of viral vectors has potential for gene therapy of neurodegenerative diseases that affect a specific area of the CNS(e.g. Parkinson disease affects the nigrostriatal dopamine system). We have demonstrated that injection of recombinant HSV-1 constructs into rat brains can lead to specific infection of neurons and expression of recombinant protein (Federoff et al. 1992). The volume of delivery following direct injection of small volumes (1-2 l) of the replication defective viruses into brain parenchyma is limited by diffusion to a small area around the inoculation site, although defective adenovirus have been reported to spread as far a 1 mm from the needle tract (Akli et al. 1993). Compromised virus can spread further by secondary infection, but this is accompanied by unacceptable neurotoxicity. Inoculation of larger volumes (10-20 l) can cause pathology at the injection site from tissue displacement, and also results in loss out the needle tract. Alternatively, slow infusion (convection) is a relatively new method that may increase the volume of distribution of virus in the brain. Bulk flow through brain interstitium is generated by maintaining a pressure gradient during interstitial infusion (Bobo et al. 1992). The lack of access of inoculated virus to the vast majority of the brain points out that delivery by focal inoculation is not a suitable method for treatment of global neurodegenerative diseases or enzyme deficiencies that affect the entire CNS (e.g. metabolic enzyme deficiencies).
Intracerebral transplantation of recombinant cells which produce virus is a mechanism for delivery of genetic material to the brain that circumvents the BBB. In studies leading to tumor therapy, transplantation of a cell line capable of packaging and releasing retrovirus bearing a marker gene resulted in labeling of up to 10% of cells in a C6 glioma (Short et al. 1990) and 10-70% transduction in 9L gliosarcoma (Ram et al. 1993), which were much higher success rates than could be obtained by inoculation, although little or no glial infection was observed. Delivery of retrovirus- packaging cell lines into brain tumors may have therapeutic applicability, although tumor responses seen in small animals may not be transferable to large animals with large or infiltrative tumors. As discussed above, retroviral vectors are unlikely to be useful in CNS gene replacement. However, if the system devised for packaging of retrovirus could be tailored for a neurotropic virus, such as HSV-1, amplicon, adenovirus or AAV, then transplantation of virus packaging cells might be useful for neuronal gene therapy.
DELIVERY OF AGENTS ACROSS THE BBB BY OSMOTIC BBB disruption: Delivery of protein, particles, or virus across the BBB can be accomplished by the hypertonic mannitol blood-brain-barrier disruption technique as reported by our laboratory (Neuwelt 1989; Neuwelt et al. 1981-1993). This reversible disruption appears to result from shrinkage of cerebrovascular endothelial cells and subsequent increased permeability of the tight junctions (Rapoport 1986). Osmotic disruption requires that vascular endothelium be perfused with a high osmolarity solution (e.g. 25% mannitol) for a short period of time (30 sec), and is thus restricted to a first pass through either the left or right cerebral circulation or the vertebral circulation. Vascular permeability to small molecules (sugars, amino acids, chemotherapeutic drugs such as methotrexate) as well as large molecules (e.g. antibodies), is increased maximally at 15 min after carotid infusion of hypertonic mannitol, after which BBB vascular permeability decreases rapidly, and returns to pre-infusion levels within two hours. This method has been well characterized in animals and is used clinically for the delivery of chemotherapeutic agents to intracerebral tumor and brain around tumor in humans (Neuwelt et al. 1991).
In early studies, we examined the utility of osmotic BBB disruption for enzyme replacement therapy. We demonstrated passage of -hexosaminidase A across the BBB of the rat following hyperosmotic disruption, as well as neuronal uptake of exogenously applied enzyme and correct intracellular localization (Neuwelt et al. 1981, 1984). Although this method was effective, repeated disruption would be necessary due to the transitory nature of protein supplementation.
Studies were undertaken to determine if viral particles could be delivered across the BBB of the rat (Neuwelt et al. 1991). Osmotic BBB modification with intracarotid mannitol (25%) was immediately followed by bolus intracarotid administration of 0.5 ml of purified, UV-inactivated KOS strain HSV-1 endogenously labeled with [35S]-methionine (2.0 x 106 cpm, approximately 5 x 108 plaque forming units per ml). A fourfold increase in radioactivity in the disrupted hemisphere was found compared to saline-infused controls, suggesting that HSV-1 virions can pass the BBB. The distribution of viral particles within the disrupted hemisphere was not assessed, and will require the use of mutants bearing a marker gene to reveal the extent of virus delivery following BBB disruption. Such experiments are currently underway in this laboratory.
It remains controversial whether viral particles can pass through the vascular endothelium or traverse beyond the basement membrane. In order to address this question, we have examined the delivery of virus sized iron oxide particles (hydrodynamic diameter 20 + 4 nm) to brain parenchyma following osmotic BBB disruption (Neuwelt et al. 1993). Global delivery of these particles to the interstitial spaces and uptake by neurons of the disrupted cerebral hemisphere was assessed by magnetic resonance imaging, histochemistry for iron, and electron microscopy. These results are suggestive that recombinant virus or other genetic vectors such as liposomes (Zhu et al. 1993) can be delivered to individual neurons by means of osmotic BBB disruption.
BBB DISRUPTION IN THE CAT. As described below, we are currently attempting CNS gene replacement in the feline using the osmotic BBB disruption technique. The disadvantage of using a feline model has to do with the vascular anatomy of the cat brain. Unlike the rodent, canine, or human, the arterial blood supply to cat cerebrum is not supplied by an internal carotid artery that connects directly to the Circle of Willis. Instead, the common carotids of the cat lead directly into a carotid rete which is a large collection of anastomotic arteries that are bathed by the cavernous sinus. This system provides a cooling mechanism for arterial blood to cat brain, but also inhibits infusion rates and decreases mannitol concentrations due to arterial blood dilution. Our preliminary experiments utilizing hypertonic mannitol via the common carotid artery have been unsatisfactory, due to the dilution of mannitol to a subthreshold osmotic environment for BBB opening.
An alternative is to use the vertebral arteries, which join to form the basilar artery and supply the medulla and pons. We have successfully performed this procedure in the canine by percutaneous catheterization of the vertebral artery with the Seldinger technique (Neuwelt et al., 1981). With the recent development of tracker catheters, use of 1 mm vessels, which are routinely catheterized at our institution, has allowed us to enter the vertebral artery in the cat. Tests of mannitol infusion into the vertebral artery has yielded BBB disruption in the feline, but the results are less than in the rodent or canine. The use of higher osmolarity solutions (2 M arabinose) coupled with external compression of the carotid arteries may be sufficient for reproducible BBB disruptions.
4. GENE THERAPY OF LYSOSOMAL STORAGE DISEASE:
A FELINE MODEL FOR GM2-GANGLIOSIDOSIS (Sandhoff Disease): The Gm2- gangliosidoses are lysosomal storage diseases resulting from a deficiency of - Hexosaminidase enzyme activity. The two major isozymes of -hexosaminidase in normal tissues are -hexosaminidase A, composed of an alpha and a beta subunit, and -hexosaminidase B, a dimer of two beta subunits (Mahuran 1991). Type I Gm2- gangliosidosis, or Tay-Sachs disease, is caused by a defect in the alpha chain of - hexosaminidase A, encoded by the HEXA gene. Type II, or O-variant Gm2- gangliosidosis, known as Sandhoff disease, results from a defect in the common beta subunit, coded by the HEXB gene (Mahuran 1991). Defects in HEXB gene expression thus result in lack of active beta chain protein, and loss of both isozymes. Sandhoff disease is characterized by storage of Gm2-gangliosidoses and other related asialoglycolipids in the brain and other tissues. Patients with the infantile form of Sandhoff disease display an early onset of disease symptoms and die by 4 years of age. The juvenile onset Sandhoff disease phenotype is less severe and affected patients may survive to adulthood.
Few naturally occurring animal models of Gm2-gangliosidosis have been discovered, perhaps due to differences in -hexosaminidase enzyme activity or substrate specificity between species. A feline model of Gm2-gangliosidosis has been described (Cork 1977) which demonstrates growth of abnormal neurites from cortical neurons (Walkley et al. 1990). We have characterized a second feline model of infantile onset Sandhoff disease in the Korat cat (Muldoon et al. 1993; Neuwelt et al. 1985). Affected offspring of these highly inbred Korat cats produced by our colony of obligate carrier animals show clinical symptoms of neurodegeneration within 4-6 weeks of age. The molecular defect appears to be the deletion of a cytosine residue at position +39 in exon 1 of the gene, resulting in a frame shift and a stop codon at base 180 (Muldoon et al.1993). The genotype of newborn animals can be determined by PCR analysis for the presence of this mutation. The results of the PCR test are suggestive that a second early onset mutation may exist in our colony, and a third mutation resulting in late onset disease may exist in the Korat cat population at large. However, even with a second mutation, homozygous affected animals can be identified within days of birth for studies of gene therapy of the CNS.
REQUIREMENTS OF A MODEL SYSTEM FOR GENE THERAPY: In order for gene therapy to be successful, several condition must be met (Fleischman 1991). The Korat cat model for Sandhoff disease meets these conditions and has several advantages which make it amenable to gene therapy. First, the defective gene is known and characterized, a necessary prerequisite for replacement. The normal human HEXB gene and the Korat HEXB gene have been cloned and sequenced, and both are available for use in expression vectors. Recombinant gene expression can be detected in vivo by PCR analysis of insert and HSV-1 vector sequences, or by addition of a flag sequence to distinguish the transgene from the normally encoded mRNA. Second, the defect in Sandhoff disease results from a recessive mutation in a single gene. Gene replacement with a single normal allele should be sufficient to abrogate all of the disease phenotype, as opposed to a disease with a multi-gene locus or an unknown basic underlying defect, or one caused by a dominant mutation.
A low level of lysosomal enzyme activity is required for normal cellular function, on the order of 5-10% of the level found in normal individuals (Mahuran 1991; Leinekugel et al. 1992). Therefore, gene replacement with a low overall level of gene expression would be sufficient for cure. This would be especially important in treatment of late-onset disease, in which a low level of enzyme is present and may require only a small increase for normal function (Leinekugel et al. 1992). The - hexosaminidases, like many other lysosomal enzymes, can be secreted by expressing cells and taken up by non-expressing cells (Hickman et al. 1974; Leinekugel et al. 1992) or transferred by direct cell contact (Olsen et al. 1983). Thus one cell, producing a large quantity of enzyme, could cure a number of surrounding cells. Access to target cells could be a problem with other desirable proteins, due to insolubility or to lack of secretion or uptake mechanisms.
Finally, expression of the HEXB gene, and other lysosomal enzyme genes, is constitutive in normal cells of all tissues, so the complication of highly regulated expression is avoided. Attempting to obtain correct expression of cell type specific, cell cycle specific or hormonally regulated genes is impractical, at present, although the elucidation of specific promoter elements may increase our ability to direct transcription. In the Korat model, an expression vector with a constitutive promoter could be used, and existing cellular machinery should be sufficient to process even a large amount of recombinant protein. No known disorder has been characterized that results from the presence of an obligatory enzyme in excess quantities.
GENE THERAPY OF ANIMAL MODELS. In addition to our plans for gene delivery to the feline model of Gm2-gangliosidosis, other laboratories are investigating HSV-1 viral vectors as vehicles for gene replacement in global neurodegenerative disorders. Palella et al. (1988) have transferred the HPRT gene to a rat neuroma cell model of Lesch-Nyhan syndrome. In vivo studies with this model system demonstrated that the defective virus could transfer the HPRT gene to mouse brain (Palella et al., 1989). Wolfe et al recently demonstrated the transfer of the GUSB gene encoding - glucuronidase to the CNS of a murine model of mucopolysaccharidosis VII (MPS VII) using a HSV-1 system (Wolfe et al. 1992). Like Sandhoff disease, MPS VII is a lysosomal storage disorder in which substrate accumulates in the lysosomes of a number of cell types including neurons. The HSV-1 viral vector in this case was a fully replication competent virus with the GUSB gene under control of the viral LAT promoter (Wolfe et al. 1992). Transgene expression was seen only in cells in which the virus established latency, in contrast to the amplicon HSV-1 constructs which will express transgene in all primary infected cells. Viral delivery to CNS was accomplished by productive infection of the cornea followed by secondary and tertiary infection of brain stem and trigeminal ganglia. This method of delivery is not available to replication defective virus which will not spread past the primary infection. It is unclear what results would be obtained with this virus following focal intracerebral inoculation or BBB disruption. We hypothesize that in future we will be able to deliver the cat HEXB gene in a neurotropic vector to affected cat brain using BBB disruption of the cat vertebral artery. Even if delivery is limited to the vertebral circulation in our feline model, it will demonstrate virus delivery to brain, infection of neurons, gene and protein expression, and substrate mobilization.
CONCLUSION:
Genetic enzyme deficiencies account collectively for a significant proportion of the neurodegenerative disorders. No treatment is available for many of these disorders other than prenatal diagnosis and elective abortion. A proposed strategy for treatment involves replacement of the deficient enzyme with a normal one from some exogenous source. The three major alternative sources of enzyme are direct replacement of purified enzyme, tissue transplantation with enzymatically competent donor tissue, and insertion of a new normally functioning gene. To be effective in dealing with a neurodegenerative disorder, the administered enzyme, regardless of source, must be made accessible to CNS tissue by circumvention of the BBB. In this regard, one means of altering this barrier is to reversibly disrupt the BBB by osmotic manipulation. The vectors, techniques and experimental approaches may have widespread application in several relatively common neurodegenerative disorders. However, major refinements in viral vectors and delivery mechanisms, as well as rigorous attention to safety measures will be required before CNS gene replacement therapy becomes a reality in humans.
FIGURE LEGENDS:
Figure 1. Structure of the defective HSV-1 vector.
The vector pHSV-1 contains the pBR322 sequences necessary for the replication of the DNA construct as a bacterial plasmid, including the E. Coli origin of replication and the ampicillin resistance gene (cross hatched area). The HSV-1 oris sequence (bricked circle) and the a region sequences comprise the sequences required for packaging of the construct into virions. The HSV-1 IE 4/5 promoter drives expression of the recombinant gene (shaded region) with the SV40 polyadenylation signal (spotted region) for production of recombinant protein in vivo. The prototype vector pHSVlac contains the bacterial lacZ gene (Geller and Breakefield 1988; Geller et al. 1990). Other HSV-1 recombinant vectors might contain the HEXB gene for treatment of Gm2-gangliosidosis or the tyrosine hydroxylase gene for treatment of Parkinson disease.
Figure 2. Production of recombinant replication defective HSV-1
The recombinant HSV-1 vector (pHSVHEXB or pHSVlac, etc.) is delivered by CaPO4 transfection into the packaging cell line M64A, which contains the HSV-1 IE3 gene. The transfected cells are infected with the HSV-1 IE3 deletion mutant helper virus. The cells then contain all the elements necessary for packaging of more replication defective helper virus (H) as well as the recombinant HSV-1 vector (R) (Geller et al. 1990). The virus stock is purified and can then be utilized in gene therapy experiments such as the correction of the enzyme deficiency in our Korat cat model of Gm2-gangliosidosis. Recombinant virus could be delivered to brain by the blood brain barrier disruption technique (Neuwelt et al. 1991).
REFERENCES:
Adachi, M., Schneck, L., and Volk, B. W. (1974) Ultrastructural studies of eight cases of fetal Tay-Sachs disease. Laboratory Investigation 30:102.
Akli, S., Caillaud, C., Vigne, E., Stratford-Perricaudet, L. D., Poenaru, L., Perricaudet, M., Kahn, A., and Peschanski, M. R. (1993) Transfer of a foreigh gene into the brain using adenovirus vectors. Nature Genetics 3:224-228.
Bajocchi, G., Feldman, S. H., Crystal, R. G., and Mastrangeli, A. (1993) Direct in vivo gene transfer to ependymal cells in the central nervous system using recombinant adenovirus vectors. Nature Genetics 3:229-234.
Bobo, R. H., Laske, D., Akbasak, A., and Oldfield, E. H. (1992) Convection-enhanced drug distribution: a new method of drug delivery to brain (abstract) in 60th Annual Meeting of the American Axxociation of Neurological Surgeons, p310.
Breakefield, X. O. and DeLuca, N. A. (1991) Herpes simplex virus for gene delivery to neurons. New Biologist. 3:203-218.
Chiocca, E. A., Chio, B. B., Cai, W., DeLuca, N. A., Schaffer, P. A., DiFiglia, M., Breakefield, X. O. and Martuza, F. L. (1990) Transfer and expression of the lacZ gene in rat brain neurons mediated by herpes symplex virus mutants. New Biologist. 2:739- 746.
Corey, L. and Spear, P. G. (1986) Infections with herpes simplex viruses. The New England Journal of Medicine 314:749-756.
Cork, J. C., Munnell, J. F., Lorenz, M. D., Murphy, J. V., Baker, J. J., and Rattazzi, M. C. (1977) Science 196:1014-1017.
Culver, K. W., Ram, Z., Wallbridge, S., Ishii, H., Oldfield, E. H. and Blaese, R. M. (1992) In vivo gene transfer with retroviral vector-producer cells for treatment of experimental brain tumors. Science 256:1549-1552.
Davidson, B. L., Allen, E. D., Kozarsky, K. F., Wilson, J. M., and Roessler, B. J. (1993) A model system for in vivo gene transfer into the central nervous system using an adenoviral vector. Nature Genetics 3:219-223.
Davidson, I. and Stow, N., D. (1985) Expression of an immediate early polypeptide and activation of a viral origin of DNA replication in cells containing a fragment of herpes simplex virus DNA. Virology 141:77-88.
Dobson, A. T., Margolis, T. P., Sedarati, F., Stevens, J. G. and Feldman, L. T. (1990) A latent, nonpathogenic HSV-1 derived vector stably expresses -galactosidase in mouse neurons. Neuron. 5:353-360.
Federoff, H. J., Geschwind, M. D., Geller, A. I. and Kessler, J. A. (1992) Expression of NGF in vivo from a defective herpes simplex virus1 vector prevents effects of axotomy on sympathetic neurons. Procedings of the National Academy of Science USA 89:1636-1640.
Fink, D., Lawrence, R., Sternberg, L., Weber, P., Marina, M., Goins, W. and Glorioso, J. (1992) In vivo expression of -galactosidase in hippocampal neruons by HSV- mediated gene transfer. Human Gene Therapy 3:11-19.
Fleischman, R. A. (1991) Southwestern internal medicine conference: human gene therapy. American Journal of Medical Science 301:353-363.
Freese, A., Geller, A. I. and Neve, R. (1990) HSV-1 vector mediated neuronal gene delivery. Stragies for molecular neuroscience and neurology. Biochem. Pharmacol. 40:2189-2199.
Friedmann, T. (1989) Progress toward human gene therapy. Science 244:1275-1281.
Geller, A. and Freese, A. (1990) Infection of cultured central nervous system neurons with a defective herpes simplex virus I vector results in stable expression of Escherichia coli -galactosidase. Procedings of the National Academy of Science USA 87:1149-1153.
Geller, A. I. and Breakefield, X. O. (1988) A defective HSV-1 vector expresses Escherichia coli -galactosidase in cultured peripheral neurons. Science 241:1667- 1669.
Geller, A. I., During, M. J., Haycock, J. W., Freese, A., and Neve, R. (1993) Long-term increases in neurotransmitter release from neuronal cells expressing a constitutively active adenylate cyclase from a herpes simplex virus type 1 vector. Procedings of the National Academy of Science USA 90:7603-7607.
Geller, A. I., During, M. J. and Neve, R. L. (1991) Molecular analysis of neuronal physiology by gene transfer into neurons with Herpes Simplex Virus vectors. Trends in Neuroscience 14:428-432.
Geller, A. I., Keyomarsi, K., Bryan, J. and Pardee, A. B. (1990) An efficient deletion mutant packaging system for defective herpes simplex virus vectors: Potential applications to human gene therapy and neuronal physiology. Procedings of the National Academy of Science USA 87:8950-8954.
Hickman, S., Shapiro, L. J. and Neufeld, E. F. (1974) A recognition marker required for uptake of a lysosomal enzyme by cultured fibroblasts. Biochemical and Biophysical Research Communications 57:55-61.
Ho, D. Y., Mocarski, E. S.,and Sapolsky, R. M. (1993) Altering central nervous system physiology with a defective herpes simplex virus vector expressing the glucose transporter gene. Procedings of the National Academy of Science USA 90:3655- 3659.
Horellou, P., Guibert, B., Leviel, V. and Mallet, J. (1989) Retroviral transfer of a human tyrosine hydroxylase cDNA in various cell lines: regulated release of dopamine in mouse anterior pituitary AtT-20 cells. Procedings of the National Academy of Science USA 86:7233-7237.
Horwitz, M. S. ((1991) Adenoviridae and their replication. In: Fundamental Virology (eds Fields, B. N. and Knipe, D. M.) 1679-1721 (Raven Press, New York).
Huang, H.-J. S., Yee, H.-K., Shew, J.-Y., Chen, P.-L., Bookstein, R., Friedmann, T., Lee, E. Y.-H. P. and Lee, W.-H. (1988) Suppression of the neoplastic phenotype by replacement of the RB gene in human cancer cells. Science 242:1563-1566.
Huang, Q., Vonsattel, J. P., Schaffer, P. A., Martuza, R. L., Breakefield, X. O. and DiFiglia, M. (1992) Introduction of a foreign gene (escherichia coli lacZ) into rat neostriatal neurons using herpes simplex virux mutants: A light and electron microscopic study. Experimental Neurobiology 115:303-316.
Jiao, S., Acsadi, G., Jani, A., Felgner, P. L., and Wolff, J. A. (1992) Persistence of plasmid DNA and expression in rat brain cells in vivo. Experimental Neurology 115:400-413.
Johnson, P., A., Miyanohara, A., Levine, F., Cahill, T. and Friedmann, T. (1992) Cytotoxicity of a replication-defective mutant of herpes simplex virus type 1. Journal of Virology 66:2952-2965.
Le Gal La Salle, G., Robert, J. J., Berrard, S., Ridoux, V., Stratford-Perricaudet, L. D., Perricaudet, M., and Mallet, J. (1993) An adenovirus vector for gene transfer into neurons and glia in the brain. Science 259:988-990.
Leinekugel, P., Michel, S., Conzelmann, E. and Sandhoff, K. (1992) Quantitative correlation between the residual activity of -hexosaminidase A and arylsulfatase A and the severity of the resulting lysosomal storage disease. Human Genetics 88:513- 523.
Mahuran, D. J. (1991) The biochemistry of HEXA and HEXB gene mutations causing Gm2- gangliosidosis. Biochimicalet Biophysica Acta 1096:87-94.
Markert, J. M., Malick, A., Coen, D. M., and Martuza, R. L. (1993) Reduction and elimination of encephalitis in an experimental glioma therapy model with attenuated herpes simplex mutants that retain susceptibility to acyclovir. Neurosurgery 32:597- 603.
Martuza, R. L., Malick, A., Markert, J. M., Ruffner, K. L. and Coen, D. M. (1991) Experimental therapy of human glioma by means of a genetically engineered virus mutant. Science 252:854-856.
McGeoch, D., Dalrymple, M., Davison, A., Dolan, A., Frame, M., McNab, D., Perry, L., Scott, J. and Taylor, P. (1988) The complete DNA sequence of the long unique region in the genome of herpes simplex virus type 1. Journal of General Virology 69:1531- 1574.
Moolten, F. L., Wells, J. M., Heyman, R. A. and Evans, R. M. (1990) Lymphoma regression induced by ganciclovir in mice bearing a herpes thymidine kinase transgene. Human Gene Therapy 1:125-134.
Muldoon, L. L., Pagel, M. A., Neuwelt, E. A. and Weiss, D. L. (1993) Characterization of the molecular defect in a feline model for type II Gm2-gangliosidosis (Sandhoff disease). Submitted for publication.
Muzyczka, N. (1992) Use of Adeno-Associated Virus as a General Transduction Vector for Mammalian Cells. Current Topics in Microbiology and Immunology 158:97- 129.
Neuwelt, E. A. (1989) Implications of the blood-brain barrier and its manipulation. Volume 1, Basic science aspects. Volume 2, Clinical aspects. Plenum Publishing Co, New York.
Neuwelt, E. A., Barranger, J. A., Brady, R. O., Pagel, M., Furbish, F. S., Quirk, J. M., Mook, G. E. and Frenkel, E. P. (1981) Delivery of hexosaminidase-A to the cerebrum after osmotic modification of the blood-brain barrier. Procedings of the National Academy of Science USA 78:5838-5841.
Neuwelt, E. A., Barranger, J. A., Pagel, M. A., Brady, R. O. and Frenkel, E. P. (1984) Delivery of active hexosaminidase-A across the blood brain barrier in rats. Neurology 34:1012-1019.
Neuwelt, E. A., Glasberg, M., Diehl, J., Frenkel, E. P. and Barnett, P. (1981) Osmotic blood-brain barrier disruption in posterior fossa of the dog. Journal of Neurosurgery 55:742-748.
Neuwelt, E. A., Johnson, W. G., Blank, N. K., Pagel, M. A., Masien-McClure, C., McClure, M. J. and Wu, P. M. (1985) Characterization of a new model of GM2 gangliosidosis (Sandhoff's disease) in Korat cats. Journal of Clinical Investigation 76:482-490.
Neuwelt, E. A., Pagel, M. A. and Dix, R. D. (1991) Delivery of ultraviolet-inactivated 35S- herpesvirus across an osmotically modified blood-brain barrier. Journal of Neurosurgery 74:475-479.
Neuwelt, E. A., Goldman, D., Dahlborg, S. A., Crossen, J., Ramsey, F., Goldstein, S. M., Braziel, R., and Dana, B. (1991) Primary CNS lymphoma treated with osmotic blood- brain barrier disruption: prolonged survival and preservation of cognitive function. Journal of Clinical Oncology 9:1580-1590.
Neuwelt, E. A., Weissleder, R., Nilaver, G., Roman-Goldstein, S., Szumowski, J., Pagel, M. A., Kroll, R. A., Remsen, L. G., McCormick, C. I., Jones, R. S., Shannon, E. M., and Muldoon, L. L. Delivery of virus-sized iron oxide particles to rodent brain. Submitted for publication, 1993.
Olsen, I., Muir, H., Smith, R., Fensom, A. and Watt, D. J. (1983) Direct enzyme transfer from lymphocytes is specific. Nature 306:75-77.
Ono, T., Fujino, Y., Tsuchiya, T., and Tsuda, M. (1990) Plasmid DNAs directly injected into mouse brain with lipofectin can be incorporated and expressed by brain cells. Neuroscience Letters 117:259-263.
Palella, T. D., Hidaka, Y., Silverman, L. J., Levine, M., Glorioso, J., and Kelley, W. N. (1989) Expression human HPRT mRNA in brains of mice infected with a recombinant herpes simplex virus-1 vector. Gene 80:137.
Palella, T. D., Silverman, L. J., Schroll, C. T., Homa, F. L., Levine, M. and Kelley, W. N. (1988) Herpes simplex virus-mediated human hypoxanthine -guanine phosphoribosyltransferase gene transfer into neuronal cells. Molecular and Cellular Biology 8:457-460.
Palmer, T. D., Rosman, G. J., Osborne, W. R. A. and Miller, A. D. (1991) Genetically modified skin fibroblasts persist long after transplantation but gradually inactivate introduced genes. Procedings of the National Academy of Science USA 88:1330- 1334.
Ram, Z., Culver, K. W., Walbridge, S., Blaese, R. M., and Oldfield, E. H. (1993) In situ retroviral-mediated gene transfer for the treatment of brain tumors in rats. Cancer Research 53:83-88.
Rapoport, S. I. and Bobinson P. J. (1986) Tight-junctional modification as the basis of osmotic opening of the blood-brain barrier. Annals of the New York Academy of Sciences 481:250-267.
Rosenberg, R. N. (1984) Molecular genetics, recombinant DNA techniques, and genetic neurological disease. Annals of Neurology 15:511-520.
Sawtell, N. M. and Thompson, R. L. (1992) Herpes simplex virus type 1 latency- associated transcription unit promotes anatomical site-dependent establishment and reactivation from latency. Journal of Virology 66:2157-2169.
Short, M. P., Choi, B. C., Lee, J. K., Malick, A., Breakefield, X. O. and Martuza, R. L. (1990) Gene delivery to glioma cells in rat brain by grafting of a retrovirus packaging cell line. Journal of Neuroscience Research 27:427-433.
Spaete, R. R. and Frenkel, N. (1982) The herpes simplex virus amplicon: A new eucaryotic defective-virus cloning-amplifying vector. Cell 30:295-304.
Straus, S. E. (1984) Adenovirus infections in Humans. In: The Adenoviruses (ed. Ginsberg, H. S) 451-496 (Plenum Press, New York).
Varmus, H. (1988) Retroviruses. Science 240:1427-1435.
Walkley, S. U., Wurzelmann, S., Rattazzi, M. C. and Baker, H. J. (1990) Distribution of ectopic neurite growth and other geometrical distortions of CNS neurons in feline Gm2- gangliosidosis. Brain Research 510:63-73.
Walsh, C. I., Liu, J. M., Xiao, X., Young, N. S., and Nienhuis, A. W. (1992) Regulated high level expression of a human g-globin gene introduced into erythroid cells by an adeno-associated virus vector. Procedings of the National Academy of Science USA 89:7257-7261.
Winn, S. R., Tresco, P. A., Zielinski, B., Greene, L. A., Jaeger, C. B. and Aebischer, P. (1991) Behavioral recovery following intrastriatal implantations of microencapsulated PC12 cells. Experimental Neurology 113:322-329.
Wolfe, J. H., Deshmane, S., L. and Fraser, N. W. (1992) Herpesvirus vector gene transfer and expression of -glucuronidase in the central nervous system of MPS VII mice. Nature Genetics 1:379-384.
Wolff, J. A., Fisher, L. J., Xu, L., Jinnah, H. A., Langlais, P. J., Iuvone, P. M., O'Malley, K. L., Rosenberg, M. B., Shimohama, S., Friedmann, T. and Gage, F. H. (1989) Grafting fibroblasts genetically modified to produce L-dopa in a rat model of Parkinson disease. Procedings of the National Academy of Science USA 86:9011-9014.
Zhu, N., Liggitt, D., Liu, Y., and Debs, R. (1993) Systemic gene expression after intravenous DNA delivey into adult mice. Science 261:209-211.