|Home | About | Journals | Submit | Contact Us | Français|
For many years it has been virtually impossible to transfer genes into brain cells either to study or manipulate molecular, cellular, or, in vivo, behavioral processes. In addition to the physical barriers that protect the brain (bone and three layers of meninges), the earliest gene delivery systems, the retroviral vectors, require cell division to integrate the transgenes into their genomes to express any transgenes. Thus they limited transduction to dividing cells of the nervous system, e.g., astrocytes and oligodendrocytes, neurons in the neonatal brain undergoing cell division, or non-neural cells such as fibroblasts that could then be transplanted to the brain. Because neurons in the adult brain do not divide, retroviruses were of limited use to neuro-biologists wanting to manipulate the molecular makeup of neurons in vivo (Fisher and Gage, 1993; Suhr and Gage, 1999; Lowenstein and Castro, 2002; Hsich et al., 2002; see Kageyama et al., this volume, for further discussion).
Far greater transduction efficiencies in the nervous system can now be achieved with the recently developed lentiviral vectors. Lentiviruses are a subclass of retroviruses whose genomes contain additional viral proteins. Two of these virion proteins, matrix (MA) and virion protein R (VPR), are used to transport the cDNA/integrase complex (also called preintegration complex) across the nuclear membrane in the absence of mitosis. This allows lentiviruses not only to infect, but (at least in theory) also to express transgenes in both proliferating, e.g., astrocytes, and nonproliferating cells, such as neurons. In vivo experiments with lentiviral vectors, however, have demonstrated that only neurons express trans-genes, presumably due to a not yet characterized selectivity of lentiviral vectors to infect neuronal cells (Kay et al., 2001; Consiglio et al., 2001; Brooks et al., 2002).
The first lentiviral vectors were derived from human immunodeficiency virus-1 (HIV-1) but were pseudotyped by using the envelope glycoproteins from other viruses such as the vesicular stomatitis virus G protein (VSV-G), a fusion protein used to improve infection efficiency. The cell line 293 is used to generate these vectors using a three-plasmid cotransfection system. Three separate plasmids encode for the pseudotyped env gene, the transgene cassette, and a packaging construct respectively supplying the structural and regulatory genes in trans (Naldini et al., 1996). Currently other lentivurses, e.g., feline immune deficiency virus and equine infectious anemia virus, have also been engineered as gene transfer vectors. An important attraction of these latter systems is the fact that they are derived from animal, rather than human lentiviruses (Brooks et al., 2002; Curran and Nolan, 2002).
Lentivirus vectors have all the desirable properties of the early Moloney murine leukemia virus (Mo-MLV)-based vectors, but, in addition, can infect both dividing and quiescent cells and induce a limited inflammatory response even in the CNS (Blomer et al., 1997; unpublished observations). Lentiviruses have been used very successfully infecting and transducing the central nervous system (CNS), with almost negligible inflammatory responses (Deglon and Aebischer 2002; Hsich et al., 2002; Tate et al., 2002). Lentiviral vectors are currently being evaluated for safety, with a view to removing all nonessential regulatory genes to facilitate and accelerate approval for clinical trials. If the issues of increased production and safety can be overcome, higher titers are achieved, and public fears of some lentiviral vectors being HIV-1 derived are addressed, these vectors hold great promise for clinical applications within the nervous system (Ailles and Naldini, 2002; De Palma and Naldini, 2002; Galimi and Verma, 2002; Huentelman et al., 2002). Novel and/or improved vector systems are constantly being developed to provide a highly effective technology with which to explore the molecular basis of neuronal gene therapy (see Jakobsson et al., for further discussion).
Having briefly reviewed the current status of vectors available for gene transfer into the brain, we will now explore the potential application of replication-deficient adenovirus vectors as vehicles for gene delivery into the CNS (Table I). A significant feature of adenovirus as a potential vector for DNA delivery in human gene therapy protocols has been that a live (replication-competent) vaccine has been safely administered to many million human beings (U.S. military recruits) over several decades to provide protection against natural adenovirus infections (Rubin and Rorke, 1994). More recently, similar replication-competent adenovirus recombinants encoding defined immunogens have also been used in human vaccine trials, or as replication-competent vectors with or without additional anticancer genes to induce tumor cell killing. Although most adenovirus vaccine development has been based on exploiting replication-competent systems, replication-deficient adenovirus vectors have also proved to be highly effective agents with which to generate both humoral and cell-mediated immune responses to expressed transgenes. These vectors have definite attributes as agents for immunization. There is therefore considerable experience in using live adenovirus isolates, replication-competent adenovirus recombinants, and replication-deficient recombinant adenovirus as immunizing agents (Top et al., 1971a, b; Schwartz et al., 1974; Morin et al., 1987; Prevec et al., 1989; Eloit et al., 1990; Jacobs et al., 1992; Gallichan et al., 1993; Rubin, 1994; Wilkinson, 1994a,b; Wilkinson and Borysiewicz, 1995; Amalfitano and Parks, 2002; Nicklin and Baker, 2002; Wickham, 2000).
Furthermore, human gene therapy programs have been launched by a number of groups in the United States and Europe, resulting in the application of replication-deficient adenovirus recombinants in the treatment of patients suffering cystic fibrosis, coronary artery restenosis postpercutaneous angioplasty, brain tumors, head and neck carcinomas, ovarian carcinomas, melanoma, and metabolic disorders among others (Rosenfeld et al., 1992; Engelhardt et al., 1993; Mastrangeli et al., 1993; Simon et al., 1993; Boucher et al., 1994; Bout et al., 1994a,b; Brody et al., 1994a,b; Mittereder et al., 1994; Welsh et al., 1994; Wilson et al., 1994; Yei et al., 1994a,b; Zabner et al., 1994; Kirn et al., 2001; Rutanen et al., 2002; Bauknecht and Meinhold-Heerlein, 2002; Raper et al., 2002). Thus, a wealth of data has accumulated and continues to grow concerning the therapeutic administration of adenovirus and their effectiveness as well as their side effects in human patients.
Interest in exploiting the new capabilities offered by adenoviruses as gene therapy vectors has been very high. The key features that make adenovirus potentially a credible vector for neurological gene therapy specifically are that (1) sufficiently high titers can be easily produced as to allow their administration in vivo, (2) they can transduce many different differentiated cell types, including postmitotic neurons, (3) expression in the target cell can be restricted to the transgene only (reviewed in Legrand et al., 2002; Nicklin and Baker, 2002), and (4) they can be scaled up to very high titres, i.e., 1012–1013 IU/ml. So far adenovirus recombinants have been used to transduce cells of the lung (see above for CFTR transfer; Gilardi et al., 1990; Rosenfeld et al., 1991; Yei et al., 1994a,b), liver (Herz and Gerard, 1993; Ishibashi et al., 1993; Engelhardt et al., 1994; Hayashi et al., 1994; Kozarsky et al., 1994), arteries and other blood vessels (Lemarchand et al., 1992; Roessler et al., 1993; Kingston et al., 2001), joints, bone marrow cells, and various differentiated circulating cells of the immune system (Haddada et al., 1993), heart (Stratford-Perricaudet et al., 1992; Kass-Eisler et al., 1993), skeletal muscle (Quantin et al., 1992; Stratford-Perricaudet et al., 1990, 1992; Acsadi et al., 1994a,b), brain (Akli et al., 1993; Bajocchi et al., 1993; Davidson et al., 1993; Le Gal La Salle et al., 1993; Ambar et al., 1999; Dewey et al., 1999; Thomas et al., 2000a,b, 2001a,b), and spinal cord (Lisovoski et al., 1994), and to release factor IX into the bloodstream (Smith et al., 1993). The widespread potential applications of adenovirus systems has promoted further developments of the adenovirus vectors, in particular, efficient strategies have been devised for the insertion of transgenes into recombinant adenovirus. Also, the further developments have concentrated on two main limitations of the adenoviral vectors, namely, the limited transgene capacity of first-generation adenoviral vectors and the inflammatory and immune responses caused by the adenoviral virions and its genome. Inserts of up to 8 kbp are feasible in replication-deficient recombinants, which can be grown without helper virus on trans-complementing 293 cells (Graham et al., 1977; Berkner, 1988; Wilkinson, 1994a,b), gutless vectors, and the new helper-dependent high-capacity vectors have a theoretical capacity of up to 30–36 kbp (Lowenstein et al., 2002; Hartigan-O’Connor et al., 2002; Zhou et al., 2002).
Recent reviews have provided detailed protocols and methodologies involved in constructing and utilizing adenovirus vectors for gene transfer to cells of the brain both in vitro and in vivo (Lowenstein, 1995; Southgate et al., 2001; Thomas et al., 2001a,b; see Phillips, 2002). In this chapter, we will consequently concentrate on progress that has been made in the application of such vectors within the context for their exploitation for neurobiology and discuss potential future developments in this field.
Adenoviruses have the ability to infect a wide range of different tissues and cell types. It has been shown convincingly that adenovirus type 5 serotype initiates infection by binding to the coxsackievirus and adenovirus receptor (CAR) through the knob domain located at the distal extreme of the fiber protein (Bergelson et al., 1997; Roelvink et al., 1999). CAR is a 46-kDa integral membrane protein with a typical transmembrane region and an extracellular region composed of two immunoglobulin (Ig)-like domains (Tomko et al., 1997). The C-terminal knob of the fiber protein confers the specificity of the cellular receptor recognition. Following its binding to CAR, adenovirus is subsequently internalized through an interaction with integrins on the cell surface of the target cells, especially αvβ5 (Wickham et al., 1993). Entry into the cell is by receptor-mediated endocytosis (Wickham et al., 1994), after which the viral nucleocapsid is released from endosomes into the cytoplasm, and transported to the nucleus (Greber et al., 1993; Trotman et al., 2001). The double-stranded DNA genome (~36 kb; Fig. 1) is then released into the nucleus of the infected cell (Trotman et al., 2001). Upon infection of a permissive cell, the replication cycle can be divided into three phases of gene expression: immediate early (E1 genes), early (E1–E4 genes), and finally the late events, which follow the onset of viral DNA replication. The very first region of the genome to become active is the left-hand region of the left strand, E1, which encodes a series of transcriptional activators, which will promote progression into the early phase of gene expression. These phases of virus replication are associated with host cell cycle progression to S1, onset of viral DNA replication, as well as the production of gene products associated with evasion of host antivirus defenses. Five late transcriptionally active regions are found at the right of the left strand, and are produced from a single major late promoter by alternative splicing (L1, L2, L3, L4, and L5). Expression of late viral genes becomes activated after the start of viral DNA replication and will eventually lead to the assembly of progeny virions, in excess of 105 new virions per infected cell (Shenk, 1995).
Most available recombinant adenoviruses for use as vectors for gene transfer are rendered replication defective through deletions in the E1, E2, E3, or E4 regions, or through combinations thereof. These vectors, have been termed “first (and sometimes second and third)-generation” adenovirus vectors, depending on the authors, and the particular deletions to increase the cloning capacity (Gilardi et al., 1990; Davidson et al., 1993; Brody et al., 1994b; Wilkinson, 1994a,b; see Philips, 2002). The regions deleted comprise some genes that are necessary for replication (e.g., those located in the E2 region), to others that do not appear to be necessary for replication (e.g., the E3 and E4 region) and that encode genes associated with modulating the host immune response and transcriptional control. No deletions are neutral, and they will have varying effects on the production, replication, expression, and toxicity of the particular viral vectors. Replication-deficient adenovirus recombinants deleted in various genes or genomic regions are commonly grown on trans-complementing 293 cells that express an integrated copy of the Ad5 E1 gene, complemented by other regions in the case of deletions in the E2 region (Graham et al., 1977). Transcomplementation with the E3 or E4 region has usually not been necessary.
Efficient systems to construct adenovirus recombinants are now available both from a number of different laboratories, and more recently also from various commercial sources (e.g., Microbix Biosystems Inc., Toronto, Canada; QBio Gene, Montreal, Canada). These allow the construction of both replication-competent and replication-deficient recombinant adenoviruses for gene transfer (Prevec et al., 1989), using conventional and well-established DNA cloning and transfection procedures (Sambrook, 1989). Most published reports utilize adenovirus (Ad) recombinant derived from either serotypes 5 or 2 (see Table II), although there are some 47 human serotypes identified and many more animal isolates, some of which have also been used within the context of gene transfer (Cotten et al., 1993). The exploration of other adenoviral serotypes has substantially increased since 1995 as some of their distinct properties concerning the receptors used for entry, the consequent cell types transduced, have been discovered to differ in important ways from adenovirus type 2 and 5 (Zabner et al., 1999; Segerman et al., 2000; Shayakhmetov et al., 2000). In Ad E1-deleted vectors, transgenes are usually inserted into the E1 region under the control of exogenous viral promoters (e.g., RSV-LTR, IE-hCMV). Cell specific alone, or combined with inducible promoters, or other elements that increase expression levels can also be used effectively to target expression to specific cell types, to regulate its expression, and to increase expression levels (Babiss et al., 1986; Bessereau et al., 1994; Grubb et al., 1994; Smith-Arica et al., 2000, 2001; Southgate et al., 2001; Gerdes et al., 2000; Harding et al., 1998; Ralph et al., 2000; Glover et al., 2002). Most recently the use of replication-competent adenoviruses for the treatment of turmors has also begun to be explored (Nanda et al., 2001; Doronin et al., 2000, 2001).
The deletion of the Ad E1 gene region results in a failure of the replication-deficient Ad vector to activate both early and late phase transcription from the viral genome. Consequently expression of only a transgene under the control of a constitutive or cell-type-specific or inducible promoter is achieved in the infected target cell. Although this genetic barrier to adenovirus gene expression is efficient, it is not absolute. Even limited breakthrough of Ad gene expression does occur in many if not most tissues, and can thus be problematic, it may either by itself affect the physiology of the target cell, induce cytotoxic responses, a cellular immune response, as well as provide antigenic epitopes recognized by the adaptive arm of the immune response (Engelhardt et al., 1994; Yang et al., 1994a,b; Thomas et al., 2000b, 2001b; Lowenstein and Castro, 2002; Mahr and Gooding, 1999; Wold et al., 1994, 1999; Hackett et al., 2000; Wood et al., 1996a). Although infections at low multiplicity with E1-deleted vectors are unlikely to lead to high level adenoviral gene expression, the same is not true at high multiplicity of infections such as used in gene therapy applications. This is likely to remain the case, even if deletion of Ela, Elb, and other regions of the adenovirus genomes provides a better block than simply the Ela deletion (Imperiale et al., 1984; Spergel and Chen-Kiang, 1991). Nevertheless extremely high input multiplicities of infection (in excess of 100 or even higher infectious units per cell regularly achieved in in vivo experimental designs) can result in breakthrough of low level expression of the Ad genome (Engelhardt et al., 1994; Kass-Eisler et al., 1994; Yang et al., 1994a,b). A further modification to current vectors is therefore of interest, i.e., the supplementation of deletions in E1, E3, with an additional conditional mutation in E2a, the so-called “second-generation vectors” (Engelhardt et al., 1994; Yang et al., 1994a,b). The E2a region plays a role in activating late phase gene expression. Recombinants carrying this additional mutation have proved to be less reactogenic and permit prolonged expression of the transgene (Engelhardt et al., 1994; Yang et al., 1994a,b). Expression of late genes from the adenoviral genome that are normally expressed only following adenoviral DNA replication suggests that limited replication of the adenoviral genome may occur in some cells. This remains to be proven. In view of the fact that the liver contains proteins that can substitute for E1a function, this topic ought to be investigated carefully.
Further, forskolin can increase the breakthrough to late gene expression. Forskolin results in enhanced levels of intracellular calcium and activation of the transcription factor cAMP responsive element binding protein (CREB or ATF), a factor that is known to bind and stimulate transcription from the Ad early promoter (Muchardt et al., 1990). It is therefore not surprising that forskolin treatment of cells infected with a replication-deficient Ad recombinant can enhance breakthrough to late phase gene expression. Both a high input multiplicity of infection (around the site of virus inoculation or instillation) and elevated calcium levels are likely to occur in transduced cells following in vivo gene transfer. Both these phenomena are very likely to coincide in neurons surrounding the site of adenovirus direct injection into the brain. Physiological conditions at the site of gene transfer could thus further regulate transgene expression from viral vectors.
The induction of an immune response to elements of the Ad vector is now perceived as a significant limitation of the vector in long-term gene therapy protocols, especially when systemic administration is needed. Several groups have characterized the effects of the virions on brain inflammation, and the effects of peripheral injections on the stimulation of adaptive immune responses (Byrnes et al., 1995, 1996a,b; Matyszak and Perry, 1996; Wood et al., 1996a; Wood, 1996; Morral et al., 1997; Matyszak 1998; Perry, 1998; Gerdes et al., 2000; Thomas et al., 2000b, 2001a,b, 2002). The effects of each of these on transgene expression are reviewed in detail elsewhere (Lowenstein and Castro, 2002). Although gene transduction remains possible in seropositive animal models and thus possibly in human clinical trials, repeated exposure to viral antigens will be expected to reduce the efficiency of gene transfer. The lack of need to utilize adjuvant when immunizing against adenovirus indicates that even small amounts of vectors leaking into the peripheral immune compartments could activate such immune responses. In the nervous system elements of the adaptive immune response, but not innate inflammatory processes and cells, are able to completely shut-off transgene expression, apparently in the absence of cytotoxicity (unpublished observations). Whether similar phenomena occur in other organs remains to be determined.
Utilizing different Ad serotypes as vectors in sequential treatment administrations could to some extent circumvent this problem. Breakthrough to early phase transcription was demonstrated by the identification of a humoral immune response to early phase proteins. In a mouse model breakthrough has been strongly correlated with the induction of CD8+ cytotoxic T-lymphocyte (CTL) response to adenovirus proteins, which is responsible both for an inflammatory response and an elimination of target cells expressing a transgene (Yang et al., 1994a,b). It has, however, been difficult to conclusively demonstrate that such CTLs are effectively cytotoxic in vivo, and thus some of these results and their implications for immune-mediated killing of target cells have been challenged (Wadsworth et al., 1997). Many experiments have been performed using extremely high doses of input virus. If lower doses of adenovirus recombinant are inoculated into a mouse, major histocompatibility complex-I (MHC-I) responses specific for a viral transgene in the absence of a detectable response to the vector can be induced (Schadeck et al., 1999). However, high input doses of recombinant may be essential for efficient in vivo gene transfer.
It is also important to consider that many experiments are performed in non-permissive or semipermissive species (Ginsberg et al., 1991; Ross and Ziff, 1992), e.g., mice, rats, and nonhuman primates. In such systems, Ad E1 recombinants are unable to replicate due to a “species block.” It would thus be useful to test recombinants in a more relevant species, such as “permissive” cotton rat (Sigmoidon hispidus), to determine whether cell-specific factors in addition to species-specific factors regulate the capacity of adenovirus to replicate in particular target organs (Oualikene et al., 1994).
Many researchers see the current cloning capacity of ~7–8 kb as a limitation toward the more general applicability of adenoviral vectors. There is a stringent limitation exerted by the adenovirus particle, which will package only an additional 3% ( ~1 kb) in addition to the full-length viral genome (Shenk, 1995). Thus, a new generation of high-capacity helper-dependent adenoviral vectors (also known as “gutless” or “gutted” vectors; HD-Ad) has been developed that is devoid of all viral coding sequences (Fig. 2) (Kochanek et al., 1996; Mitani et al., 1995; Chen et al., 1997, 1999; Morral et al., 1998; Morsy et al., 1998; Schiedner et al., 1998; Burcin et al., 1999; Sandig et al., 2000; Akagi et al., 1997; Maione et al., 2001; Parks et al., 1996, 1999a,b; Cregan et al., 2000; Ng et al., 2001, 2002; Umana et al., 2001; Lowenstein et al., 2002). These vectors have a minimum requirement for the extreme termini of the linear adenovirus genome, containing only those cis-acting elements for viral DNA replication and packaging, mainly the inverted terminal repeat (ITR) sequences and packaging signal. Because these elements are contained ~500 bp from the ends of the genome (Grable and Hearing, 1992), helper-dependent vectors have the potential encode range of few hundred base pairs to up to ~30 kb of foreign DNA, which is close to the size of the native genome.
HD-Ad are copropagated with an E1-deleted helper virus, which provides in trans all of the proteins required for packaging the vector. Up to now, several systems have been developed to prevent packaging of the helper virus. The Cre/loxP-based system for the generation of HD-Ad involves the use of a first-generation helper virus, where the packaging signal is flanked by loxP recognition sites (Hardy et al., 1997). Infection of Cre-expressing 293 cells with the helper virus results in excision of the viral packaging signal, rendering the helper virus DNA unpackagable but still able to replicate and provide helper functions for HD-Ad vector propagation (Chen et al., 1996). Purification by cesium chloride centrifugation is necessary to reduce the titer of the helper virus to negligible levels, typically ranging from 0.1 to 0.01% of the HD-Ad titer (Morsy et al., 1998). Recently, another Flp/frt-base system has been developed. The Flp recombinase was used in place of Cre, and shown to excise the frt-flanked packaging signal in helper virus efficiently (Lowenstein and Castro, 2001; Ng et al., 2001; Umana et al., 2001). The most recent improvement to this system is the development a new Cre-expressing cell line based on E2T, an E1 and E2a complementary cell line (Zhou et al., 2001). Thus an E1 and E2a double-deleted helper virus can be used with the new cell line to produce HD-Ad vector with low helper contamination, further improving HD vector safety (Zhou et al., 2001). Compared with first-generation adenovirus vectors, the HD-Ad vector can efficiently transduce a wide variety of cell types from numerous species in a cell cycle-independent manner as first generation, but HD-Ad vector has the added advantage of increased cloning capacity, reduced toxicity and immune responses, and prolonged stable transgene expression in vivo (Schiedner et al., 1998; Thomas et al., 2000a,b; Thomas et al., 2001a,b). This system is essentially analogous to herpes simplex virus (HSV1) amplicons and shares some of their limitations (Lowenstein, 1994, 1995), e.g., low titers and the presence of variable amounts of helper virus in viral stocks, although novel systems that produce vectors in a helper-free fashion are now available (Wade-Martins et al., 2001; Logvinoff and Epstein, 2001; Wang et al., 2002). Eventually, it should be possible to develop this system so that the recombinants can be produced in the absence of helper virus to generate a safer vector with markedly reduced potential to be reactogenic.
The technology for gene transfer to the brain has progressed substantially during the past 5 years, and new improved systems are being rapidly developed. Table I compares the major systems for gene transfer into neurons in vitro and into the brain in vivo. Vectors vary dramatically in their efficiency of gene transfer, longevity of expression, associate toxicity, and size of the transgene they can harbor. Some vectors mainly direct short-term expression of transgenes, either because of their intrinsic toxicity [e.g., Semliki Forest virus (SFV), vaccinia], or promoter shut-off, a poorly understood phenomenon by which promoters within viral vectors eventually stop being active in spite of the vector genomes being present. Adenoviruses mediate long-term transgene expression in the central nervous system (up to over 1 year) in naive (nonimmunized) animals. Immunization completely eliminates expression from first-generation vectors but not high-capacity helper-dependent adenoviral vectors, whereas in preimmunized animals, transgene expression mediated by first-generation adenoviruses declines within 2–4 weeks postinjection, but expression from the high-capacity helper-dependent adenoviral vectors is reduced only to approximately 50% and stays stable thereafter (Thomas et al., 2000a,b; Thomas et al., 2001b).
In 1993 four groups independently reported for the first time in vivo gene transfer into several types of brain cells using adenovirus recombinant encoding β-galactosidase (Akli et al., 1993; Bajocchi et al., 1993; Davidson et al., 1993; Le Gal La Salle et al., 1993). Since these initial breakthroughs, there has been a logarithmic increase of original papers using adenovirus recombinants to transduce brain cells (see Table II). Although there is some specificity in the capacity of different adenovirus serotypes to infect/express in different target cells (Acsadi et al., 1994a,b; Bessereau et al., 1994; Grubb et al., 1994; Millecamps et al., 1999) a recombinant adenovirus-derived vector can indeed express trans-genes in all brain cells, e.g., neurons, astrocytes, oligodendrocytes, ependymal cells, fibroblasts, macrophages, endothelial blood vessel cells, retinal pigment epithelium, photoreceptors, as well as retinal neurons proper, and peripheral nerve Schwann cells either in vivo or in vitro (Table II; Akli et al., 1993; Bajocchi et al., 1993; Caillaud et al., 1993; Davidson et al., 1993; Le Gal La Salle et al., 1993; Bain et al., 1994; Bennett et al., 1994; Jomary et al., 1994; Li et al., 1994; Byrnes et al., 1995; Lowenstein, 1995; Shering et al., 1997; Morelli et al., 1999; Smith-Arica et al., 2000; Thomas et al., 2000a,b; Umana et al., 2001; Thomas et al., 2000b, 2001a,b, 2002).
While initial experiments were performed using recombinant adenovirus to transfer marker proteins into the brains of both rodents and primates, recombinant vectors encoding therapeutic genes for treatment of neurological diseases have now been generated (Table III) (Chen et al., 1994; Davidson et al., 1994; Perez-Cruet et al., 1994; Shewach et al., 1994; Lowenstein, 1995; Geddes et al., 1997; Ambar et al., 1999; Dewey et al., 1999; Gupta et al., 2001; Bohn et al., 1999; Lawrence et al., 1999; Bohn, 2000; Connor et al., 1999, 2001; Kozlowski et al., 2000, 2001; Amalfitano and Parks, 2002). Expression of transgenes after in vivo administration of vectors to the brain was detected up to 5–6 months and even after 18 months postinoculation (Geddes et al., 1997; Thomas et al., 2000a,b; Thomas et al., 2000b, 2001a,b; Zermansky et al., 2001).
Other examples of adenoviruses encoding therapeutic transgenes for preclinical applications include recombinant adenoviral vectors encoding hypoxanthine phosphoribosyltransferase (HPRT), which could be utilized in the treatment of the Lesch–Nyhan syndrome. Although these vectors have mainly been utilized to transduce HPRT in primate (Davidson et al., 1994) and rodent brains (Davidson et al., 1994; Lowenstein, 1995; Southgate et al., 2001), there is yet no published account of their capacity to complement an HPRT deficiency in animal models for the Lesch–Nyhan syndrome. Adenovirus vectors expressing tyrosine hyroxylase, and their therapeutic efficacy in a rat model of Parkinson’s disease, have been reported (Horellou et al., 1994; Corti et al., 1999a,b; Hida et al., 1999). Recombinant adenoviruses encoding neuroprotrective factors such as glial cell line-derived neurotrophic factor (GDNF), aimed to prevent dopaminergic neuron degeneration in a rat model of Parkinson disease, have also been successfully developed (Bilang-Bleuel et al., 1997; Choi-Lundberg et al., 1997, 1998; Bohn et al., 1999; Lawrence et al., 1999; Bohn, 2000; Connor et al., 1999, 2001; Kozlowski et al., 2000, 2001).
Because adenovirus recombinants can be used to transduce essentially all brain cells, they have also been postulated to be applicable to gene therapy protocols for the treatment of brain tumors, by utilizing them to deliver cytotoxic genes directly into the tumors. Thus, a number of groups have utilized several adenovirus recombinants, e.g., expressing HSV1 thymidine kinase or Fas ligand, Fas receptor, or p53, under the control of viral promoters. In experimental paradigms the use of these viruses appears to be promising (Badie et al., 1994; Boviatsis et al., 1994; Brody et al., 1994a; Chen et al., 1994; Perez-Cruet et al., 1994; Shewach et al., 1994; Ambar et al., 1999; Dewey et al., 1999; Morelli et al., 1999; Shinoura et al., 2000a,b; Shono et al., 2002; see George et al., this volume).
Recombinant retroviruses have been used in the treatment of brain tumors (Ram et al., 1993; Palu et al., 1999; Tamura et al., 2001), and replication-competent and conditional HSV1 vectors as well as HSV1 amplicon vectors have also been tried (Martuza et al., 1991; Markert et al., 1993; Kramm et al., 1997; Pechan et al., 1999; Herrlinger et al., 2000; Papanastassiou et al., 2002; see Hu and Coffin, this volume). In a recent phase I clinical trial the efficacy and safety profiles of retrovirus and adenovirus expressing HSVI TK have been compared (Sandmair et al., 2000). Adenoviruses showed very promising results, both in terms of their safety and also their efficacy in glioma growth control (Sandmair et al., 2000). However, the only vector system that has progressed to a phase III clinical trial has been a retroviral vector expressing HSV-1 thymidine kinase. The trial was halted because of the absence of positive effects in the gene therapy arm of patients treated (Rainov, 2000; Rainov and Kramm, 2001). With Phase I and II being essentially toxicity trials, Phase III efficacy clinical trials are crucial in order to determine whether any of the new therapies are indeed effective, because such results can be obtained only rarely from the much smaller Phase I and II trials.
Two groups have also made interesting observations on the use of adenovirus recombinants that should be of interest to neuroanatomists, namely the use of adenovirus for pathway tracing and cell labeling. Ridoux and colleagues (1994a) observed retrograde labeling of substantia nigra neurons after injections of recombinants expressing β-galactosidase into the striatum; similar results were also observed by Byrnes et al. (1995). Thus, replication-defective adenovirus could be used for retrograde labeling of neuronal pathways. Interestingly, HSV1-derived recombinants can also be used to trace neuronal pathways (Ugolini et al., 1989), with some recombinants being specific for anterograde transport and others for retrograde transport (Zemanick et al., 1991), while certain pseudorabies recombinants have been shown to be specific markers for individual neuronal circuits and capable of retrograde axonal transport (Card et al., 1991, 1992; Mazarakis et al., 2001; Enquist et al., 1998; DeFalco et al., 2001). Lisovoski et al. (1994) utilized adenovirus recombinants to label neurons of the spinal cord at different stages of development to study their morphological development. The staining they obtained after in vivo administration filled the dendritic arbors of neurons, thus allowing for their morphological and morphometric examination during spinal cord development.
An effective way of delivering gene products to the brain, rather than by introducing it into constituent cells, is by transplanting genetically engineered cells directly into the CNS. So far, most genetically engineered cells expressing trangenes for transplantation have been transduced with recombinant retroviral vectors. Ridoux et al., (1994b) have used adenovirus to tranduce primary cultures of rat astrocytes in vitro and then transplanted these into the CNS of host rats. Expression of transgene was detected for at least 5 months. Thus, adenovirus might constitute an alternative to retrovirus for transducing cells in preparation for transplantation into the brain. In many cases expression from retroviral vectors ceases after a few days to a few weeks; thus, it will be interesting to compare both systems vis-à-vis length of transgene expression.
Expression of transgenes after the administration of adenovirus recombinants into adult brains has been seen by some groups to last from 6 to 18 months. It also was reported that injection of recombinants into neonates has allowed expression to be sustained over periods up to a year. Long-term modulation of the immune response and the elucidation of its exact role in the regulation of long-term expression from adenoviral recombinants administered into the CNS will be crucial to achieve stable expression over a long period of time, which will be needed to implement gene therapy for chronic neurological disorders in humans. (Stratford-Perricaudet et al., 1990; Kass-Eisler et al., 1994; Geddes et al., 1997; Thomas et al., 2000a,b; Thomas et al., 2000b, 2001a,b; Zermansky et al., 2001).
Scant information is available on the interactions of adenovirus with the brain. Thus it will be of great importance to explore this field, e.g., adenovirus entry into brain cells, transport to the nucleus, and viral replication in neurons of permissive species. In addition, the effect of adenovirus recombinants on the electrophysiology of specific neuronal populations will have to be investigated. It also will be critical to assess the effects of adenoviral delivery into different brain regions on animal behavior. It is surprising that not much work on these topics has been published in the past 5 years.
Essentially there are two types of applications for which vectors for gene transfer into the brain could be used. Short-term expression could be used, for example, to express cytotoxic products to kill tumor cells, or to provide drugs to block ischemia-induced neurotoxicity. Ideally, the area of brain tissue to be targeted should be focal, and therapeutic benefit should be predicted after short-term transgene expression (Table III). Long term expression would be needed to treat chronic neurodegenerative disorders, such as Parkinson’s disease, Alzheimer’s disease, or amyotrophic lateral dystrophy. In this case, the area of brain tissue will be larger, and the results of the therapeutic intervention may not be seen until after several years.
The availability of viral vectors that express short term at high levels of expression could lead to important new treatments for brain diseases. A case in point are brain tumors, which are now being treated experimentally with adenovirus recombinants expressing conditional cytotoxic gene products (Badie et al., 1994; Boviatsis et al., 1994; Brody et al., 1994a,b; Chen et al., 1994; Perez-Cruet et al., 1994; Shewach et al., 1994; Takamiya et al., 1992; Barba et al., 1994; Benedetti et al., 1997; Ram et al., 1997; Bansal and Engelhard, 2000; Ikeda et al., 2000; Trask et al., 2000; Jacobs et al., 2001; Rainov and Kramm, 2001). Brain tumors constitute a good target disease for gene therapy for the following reasons: (1) the disease is focal; (2) the therapeutic objective—the destruction of tumor cells— needs to be achieved within the short term, and this can be done using cytotoxic or conditionally cytotoxic gene products; (3) the disease is life threatening within 6–12 months of diagnosis; (4) no effective treatments are available; and (5) adenovirus-particle-induced inflammation could be a beneficial adjunct to tumor cell elimination. Similarly, replication-deficient and replication-competent HSV1-derived vectors have also been also developed (Martuza et al., 1991; Markert et al., 1993) and these vectors have already been tested in Phase I–III clinical trials (Martuza et al., 1991; Markert et al., 1993, 2000; Rampling et al., 1998, 2000; Rainov, 2000).
Achieving long-term expression in brain following the administration of recombinant vectors will open up the development of gene therapy clinical trials for human neurological diseases where long-term expression is paramount, such as Alzheimer’s disease, Parkinson’s disease, Huntington’s disease, and the metabolic and degenerative brain disorders (Amalfitano and Parks, 2002; Hsich et al., 2002; Lowenstein, 2002; Lowenstein and Castro, 2002). What makes it difficult to treat such diseases is that they affect large areas of the brain, many extending throughout large anatomical regions of the human CNS, and they progress relatively slowly over years or decades. For Parkinson’s diseases there are as yet no presymptomatic diagnostic methods available, so treatment can be instituted only when symptoms appear. In families suffering from Huntington’s disease, presymptomatic diagnosis is now possible (Holloway et al., 1994) and this might also soon become possible in Alzheimer’s disease (Saunders et al., 1993; Nalbantoglu et al., 1994; Scinto et al., 1994). The diffuse nature of these diseases has led to the development of other therapeutic interventions that could target genes and their products to different areas located throughout the brain, such as transplantation of genetically engineered cells (Fisher, 1993), the direct intracranial administration of neuronal growth factors (NGF), (Seiger et al., 1993), or more recently the development of embryonic and or neural stem cells. HSV1 (Lokensgard et al., 1994; Burton et al., 2001; Kay et al., 2001; Latchman and Coffin, 2001; Lilley et al., 2001) and adeno-associated virus vectors (Kaplitt et al., 1994; Sanlioglu et al., 2001; Fu et al., 2002; Muramatsu et al., 2002) might also provide alternative vectors to achieve long-term transgene expression in brain for clinical applications in humans.
Although human brain infections caused by adenovirus are extremely rare, it has been reported that adenovirus can cause central nervous system disease, especially in children and immuncompromised individuals, although rare cases have also been described in nonimmunocompromised patients (Ginsberg and Prince, 1994; Engel, 1995; Chirmule et al., 1999; Ginsberg, 1999). Meningoencephalitis, encephalitis, and cerebellar ataxia have been reported in children in association with adenovirus infection of the CNS, mostly adenovirus type 7, 1, 2, or 32 (Chou et al., 1973; Kelsey, 1978; Davis et al., 1988; Osamura et al., 1993). Direct inoculation of adenovirus into the brain of rhesus monkeys has demonstrated that several adenovirus serotypes can cause neuropathology, although adenovirus did not appear to replicate to a great extent in the primate brain (Ginsberg and Prince, 1994; Engel, 1995; Chirmule et al., 1999; Ginsberg, 1999). Also, neuropathology due to an anamnestic reponse of the immune system following the injection of the adenovirus type 5 into the primate brain was recently described (Davidson et al., 1994).
The possibility that latent adenovirus infection might be localized to the human brain, and that adenovirus could enter the brain through infected macrophages from the periphery, has not yet been examined in enough detail, although widespread neuronal inclusions diagnostic of adenovirus particles have been seen in some fatal encephalitis cases (Chou et al., 1973). However, all these data taken together with our observations that wild-type adenovirus can express endogenous proteins in neurons, and brain glial cells, strongly suggest that adenovirus could replicate, or at last express many of its genes after infecting human neurons or glial cells and that expression might proceed for a longer time than predicted, even in the absence of symptoms of encephalitis. If there is a risk of adenoviral encephalitis it will have to be carefully considered at a time adenovirus vector is directly administered into the brain. It is clear that many more studies will be required to clarify the origins of neurotoxicity after direct injections of adenovirus virions into the brain. The use of the completely deleted high-capacity helper-dependent adenoviral vectors will completely avoid these potential complications.
As stated above, recombinant adenoviruses are very attractive vectors for gene transfer and therapy within the CNS, however, they still cause acute inflammation, and any of their expressed genes can be the target of adaptive immune responses. Local inflammation is elicited in the brain in response to the injection of either a first-generation, or a high-capacity helper-dependent adenoviral vector. This acute inflammation is dose dependent, self-limited (disappears within 30 days), adenoviral-capsid dependent, and does not affect long-term adenoviral-mediated transgene expression (Byrnes et al., 1995, 1996a; Kajiwara et al., 1997; Gerdes et al., 2000; Thomas et al., 2000b, 2001a,b, 2002). Injection of adenoviral vectors into the brain leads to persistent expression (e.g., up to 18 months). This is thought to be due to the failure to stimulate an effective antiadenoviral T cell response following the careful injection of these vectors into the brain parenchyma (Byrnes et al., 1996b; Wood et al., 1996a; Kajiwara et al., 1997; Gerdes et al., 2000; Thomas et al. 2000b, 2001a,b, 2002). Activation of antiadenoviral immune response by peripheral immunization with adenovirus leads to massive lymphocyte infiltration of the brain parenchyma, macrophage/microglial activation, up-regulation of MHCI and II, and loss of vector-mediated transgene expression (Byrnes et al., 1996a; Gerdes et al., 2000; Thomas et al., 2000b, 2001a,b, 2002). Mechanisms of loss of transgene expression have not been completely elucidated. Either cytotoxicity of transduced cells or downregulation of mRNA expression could explain it. Thus, loss of transgene expression from first-generation adenoviral vectors in the brain following systemic immunization against adenovirus limits their utility for long-term neurological gene therapy applications (Lowenstein and Castro, 2002).
Recent work from our laboratory indicates that high doses of E1/E3-deleted viral vectors (>108 IU) cause direct acute cytotoxicity, and chronic inflammation, which lead to reduced transgene expression and substantial tissue damage (Thomas et al., 2001a,b). We thus reasoned that an absolute reduction in the dose of vectors would be necessary to eliminate both the acute cytotoxicity and chronic inflammation. The use of the powerful mCMV promoter in adenoviral vectors has recently uncovered the capacity to transfer transgenes into the brain, and achieving similar results in the levels of transgene expression while using 2–3 logs lower doses of vectors when compared to vectors employing the hCMV promoter. The important reduction in total viral dose needed thus allows transduction in the absence of acute cytotoxicity or chronic brain inflammation (Gerdes et al., 2000).
Additionally, E1/E3-deleted adenoviruses do express adenoviral proteins encoded within their genomes (Yang et al., 1994a,b). These can either generate specific immune responses or provide target epitopes recognized by the activated adaptive immune system. We thus studied transgene expression and inflammatory responses elicited by the intrastriatal injection of E1–E3-deleted or HC-Adv vectors in four different paradigms: (1) acutely in naive rats, (2) long term, in naive animals, (3) long term, in naive animals, followed by a peripheral immunization against adenovirus type 5, and (4) medium term, in animals immunized against adenovirus type 5 preceding intracranial vector injection (Thomas et al., 2000b, 2001a,b, 2002).
These studies have shown the following. (1) Injection of 1 × 107 IU of either E1/E3-deleted vectors or HC-Adv into the brain causes an acute, transitory inflammatory response, with both cellular (e.g., activation of microglia) and molecular components (e.g., upregulation of MHCI-I). This acute inflammation does not affect long-term transgene expression (see Figs. Figs.33–6). (2) Within 12 months the expression of a marker transgene from first generation vectors decreases. This decrease is less when expression is directed by an HC-Adv. (3) Following peripheral immunization against adenovirus type 5, transgene expression from an E1/E3-deleted adenovirus vector is completely abolished within 45 days, whereas expression from a HC-Adv remains unaffected (Fig. 4). (4) Down-regulation of transgene expression appears to be mediated at the transcriptional level (specific reduction of transgene mRNA levels), rather than cytotoxicity (unpublished data). (5) Expression from an E1/E3-deleted adenovirus in brains of animals previously immunized against adenovirus type 5 is almost completely eliminated by 14 days postinjection, whereas expression from an HC-Adv is reduced to only 50%, and remains stable for at least 2 months (Thomas et al., 2000b, 2001a,b, 2002; Lowenstein and Castro, 2002). The complexities, causes, and consequences of using viral vectors as gene transfer vectors in the brain have been explored elsewhere (Lowenstein, 2002).
The early generations of replication-deficient adenovirus vectors, as well as the most recently developed high-capacity helper-dependent adenovirus vectors are extremely versatile vehicles for gene transfer into the brain, and have thus become invaluable reagents to neuroscientists. Current applications cover wide areas from helping to address basic neurobiological problems to clinical trials for the treatment of human diseases. The main advantages of adenovirus-derived vectors are as follows:
Although the results so far have been highly encouraging (Tables (TablesIIII and III), adenovirus vector systems continue to be optimized for application to somatic cell gene therapy and specifically for neurological gene transfer. Currently, progress in clinical neurological gene therapy is both dependent on and limited by the technology for gene transfer. Selection of the most appropriate vector system for each application is crucial. Adenoviral vectors clearly have distinct advantages for efficient gene delivery into nervous system both in vitro and in vivo. Vector development will undoubtedly further enhance the utility of these vectors and allow their implementation for gene therapy applications to treat neurological diseases.
Work in the Gene Therapeutics Research Institute at Cedars Sinai Medical Center is funded by R01 NS42893.01 and R01 NS44556.01 grants from the National Institutes of Health, National Institute for Neurological Disorders and Stroke. We also thank the funding our Institute receives from the Board of Governors at Cedars Sinai Medical Center and the encouragement and support of every one of its members. We also wish to thank Dr. Shlomo Melmed for unparalleled support and academic leadership. We are grateful to Ms. Cheryl Cathcart for her superb administrative and organizational skills and to Mr. Danny Malaniak for his encouragement, support, and commitment. Ms. Juanita Gutierrez was responsible for the editing and preparation of this manuscript for publication and we are grateful for her dedication and superb skills.