Brain Research 768 Ž1997. 19–29 Research report Application of recombinant adenovirus for in vivo gene delivery to spinal cord Yi Liu a , B. Timothy Himes a a,b , Jon Moul a , Wenlin Huang a , Stella Y. Chow a , Alan Tessler Itzhak Fischer a, ) a,b , Department of Neurobiology and Anatomy, Allegheny UniÕersity of the Health Sciences, 3200 Henry AÕenue, Philadelphia, PA 19129, USA b Philadelphia Veterans Administration Hospital, Philadelphia, PA 19104, USA Accepted 15 April 1997 Abstract One strategy for treating spinal cord injury is to supply damaged neurons with the appropriate neurotrophins either by direct delivery or by transfer of the corresponding genes using viral vectors. Here we report the feasibility of using recombinant adenovirus for in vivo gene transfer in spinal cord. After injection of a recombinant adenovirus carrying a b-galactosidase Ž b-gal. reporter gene into the mid-thoracic spinal cord of adult rats, transgene expression occurred not only in several types of cells around the injection site but also in neurons whose axons project to this region from rostral or caudal to the injection site. Among labeled neurons were those of the red nucleus, the vestibular nuclei, reticular formation, locus coeruleus, and Clarke’s nucleus. A non-specific immune reaction, which could be blocked by immunosuppression with Cyclosporin A, reduced the number of transduced cells surviving at the injection site by 1 month. In neurons away from the injection site, where the immune response was minimal, transgene expression lasted for at least 2 months. These results support the idea that recombinant adenovirus can be used in the spinal cord for in vivo delivery of therapeutic genes important for supporting neuron survival and axon regeneration. q 1997 Elsevier Science B.V. Keywords: Adenovirus; b-Galactosidase; Gene transfer; Spinal cord; Immune response; Cyclosporin A; Clarke’s nucleus; Red nucleus 1. Introduction Compared to other commonly used gene transfer vectors, recombinant adenovirus has several advantages, including the capability of obtaining highly purified virus with a titer of up to 10 13 pfurml; the capacity of carrying a transgene of up to 8 kb in size; the ability to infect almost any type of cell including postmitotic cells, such as neurons and muscle cells; and the possibility of producing a high level of foreign gene expression due to multiple infection and the use of strong promoters w9,25,30x. Various recombinant adenoviruses have been employed successfully to deliver genes into lung w34x, muscle w32x, liver w22x and brain w1,11,12,25x. Several strategies have been used to transfer virus into regions of the brain, including direct stereotaxic injection w1,12,25x, osmotic disruption of the blood brain barrier followed by infusion of the virus through the internal carotid artery w14x, and injection into the cerebrospinal fluid ŽCSF. Žto infect ependymal cells. w2x. The feasibility of using adenovirus to deliver genes of therapeutic importance, such as the neurotrophin genes, to the injured spinal cord has, however, received little attention w16,26x. We used a recombinant adenovirus carrying the b-gal reporter gene to study the potential of adenovirus as an efficient vector for in vivo gene transfer in the spinal cord. A reduction of transgene expression following adenovirus mediated gene transfer is well known w25,41–44x, and has been attributed to an immunological reaction by the host and to promoter down-regulation w15x. To investigate the role of the host immune reaction, we examined transgene expression in animals that had been chronically immunosuppressed w5,17x. 2. Materials and methods 2.1. Recombinant adenoÕirus ) Corresponding author. Fax: q1 Ž215. 8439082; E-mail: fischer@auhs.edu 0006-8993r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. Recombinant adenovirus containing the lacZ gene under the control of cytomegalovirus ŽCMV. promoter ŽAd- 20 Y. Liu et al.r Brain Research 768 (1997) 19–29 CMVlacZ. was obtained from Dr. James M. Wilson at the University of Pennsylvania. The virus was constructed with deletions of E1a and part of E1b regions Žmu 1–9., to make it replication defective w27x. 2.2. Immunosuppression with cyclosporin A Cyclosporin A oral solution ŽSandoz Pharma. was diluted to a final concentration of 50 mgrml and administered via the drinking water beginning 3 days before surgery and continuing throughout the survival period. 2.3. Surgical procedures Twenty-two female Sprague-Dawley rats ŽTaconic Farms. weighing 200–250 g were studied. All procedures were approved by the institutional animal welfare commit- tee in accordance with the Public Health Service Guide for the Care and Use of Laboratory Animals. The rats were divided into the following groups. 2.3.1. Injection group without immunosuppression (n s 12) The surgical procedure has been described in detail elsewhere w20x. Briefly, rats were anesthetized with an intraperitoneal Ži.p.. injection of a cocktail containing xylazine Ž10 mgrkg., acepromazine maleate Ž0.7 mgrkg. and ketamine Ž95 mgrkg., and underwent laminectomy at the T7–8 level to expose one segment of the midthoracic spinal cord. A total of 5 m l of AdCMVlacZ viruses Ž10 11 pfurml in adenovirus stock buffer which contains 140 mM NaCl, 5 mM KCl, 0.1 mM Na 2 HPO4 , 5.5 mM glucose and 25 mM Tris; pH 7.4. was pressure injected into the right side of the spinal cord at three sites using a 10-ml Hamilton syringe with an attached glass pipet Žtip diameter 50 Fig. 1. Transgene expression at the injection site one week following surgery. The injection site shows intense staining for lacZ transgene expression. A, B: X-gal histochemical staining. C, D: immunocytochemical staining obtained with an anti-b-galactosidase Ž b-gal. antibody and DAB as the chromagen. Staining is localized to the injected side with little staining on the contralateral side. At higher magnification ŽB and D. both staining methods demonstrate detailed morphology of cell bodies and processes resembling a Golgi staining pattern Žarrows.. Scale bars: A, C s 100 mm; B, D s 40 mm. Y. Liu et al.r Brain Research 768 (1997) 19–29 m m., leaving the left side of the spinal cord intact. The injection site was marked by a 10-O silk suture ŽEthicon. through the dura. The animals were closely observed as they recovered from anesthesia and until the time of sacrifice. 2.3.2. Immunosuppressed injection group (n s 6) Animals that received Cyclosporin A according to the protocol described above underwent the same surgical procedures as the group that received injections of virus but were not treated with Cyclosporin A. 2.3.3. Transection–injection group Three operated control animals were anesthetized and underwent laminectomy and spinal cord transection at T7–8. Two weeks later the rats had a second operation in 21 which virus was injected into the spinal cord at T13 using the protocol described above. This group provided a control for retrograde transport and specificity of labeling. 2.3.4. Sham operation One sham-operated animal received 5 ml of adenovirus stock buffer alone injected at three sites into the right side of the spinal cord at T7–8, and was euthanized at 1 week. This animal served as a control for specificity of labeling. 2.4. Tissue preparation One week, 1 month and 2 months after surgery, seven animals Žfour non-immunosuppressed injection animals, two immunosuppressed injection animals and one transection–injection animal at each time point. were euthanized. Fig. 2. Transgene expression distant from the injection site. Transgene expression was detected by X-gal histochemistry ŽA–H. or b-gal immunocytochemistry ŽI.. One week following injection, expression level is high, resulting in a Golgi-like staining pattern in neurons of the red nucleus ŽA and B., locus coeruleus ŽD and E., reticular formation ŽG. and propriospinal neurons ŽH.. The staining is mostly localized to ipsilateral locus coeruleus and Clarke’s nucleus and contralateral red nucleus. In A, the crossing rubrospinal fibers are clearly visible Žarrowheads.. Two months following injection expression level is greatly reduced and staining is localized only to cell bodies, as shown for red nucleus ŽC., locus coeruleus ŽF. and Clarke’s nucleus ŽI.. Scale bars: A, D s 250 mm; B, C, E, F, H s 100 mm; G s 40 mm; I s 125 mm. 22 Y. Liu et al.r Brain Research 768 (1997) 19–29 Rats were anesthetized with an i.p. injection of sodium pentobarbital Ž100 mgrkg. and perfused transcardially with 200 ml of normal saline followed by 500 ml of ice-cold paraformaldehyde Ž4% in 0.1 M phosphate buffer, pH 7.4. over 30 min. The brain and spinal cord were dissected out and immersed in 0.1 M phosphate buffer ŽPB. at 48C overnight followed by cryoprotection for 2 days in 30% sucrose in 0.1 M PB containing 0.5 mM Thimerosal. Tissue Žwhole brain and spinal cord. was then cut into 0.5-cm blocks and processed in two ways. At each time point, whole mount specimens from two non-immunosuppressed injection animals, one immunosuppressed injection animal and one transection–injection animal were processed for X-gal histochemical staining Žsee below., and then embedded in OCT compound ŽFisher Scientific.. Tissue from the other two non-immunosuppressed injection animals and one immunosuppressed injection animal was directly embedded in OCT without whole mount X-gal staining. The frozen tissues were kept at y808C until cut into 20–40-m m thick sections on a cryostat. Coronal sections of brain and cross or longitudinal sections of spinal cord were mounted onto either charged or gelatin coated slides. 2.5. X-gal histochemical staining X-gal histochemical staining has been described in detail elsewhere w10x. Briefly, slides were rinsed three times Ž5 min each time. with PBS Ž137 mM NaCl, 2.7 mM KCl, 8 mM Na 2 HPO4 , 2.6 mM KH 2 PO4 , pH 7.2. followed by incubation in X-gal reagent ŽMolecular Probes, 1 mgrml final concentration. with X-gal mixer Ž35 mM K 3 FeŽCN. 6 , 35 mM K 4 FeŽCN. 6 , 2 mM MgCl 2 in PBS. at 378C overnight. Before coverslipping with DPX ŽFluka Chemical Co.., some sections were counterstained with cresyl violet. 2.6. Immunocytochemical staining Immunocytochemical procedures were modified from those described by Milligan et al. w29x. Briefly, reactions that were to be visualized with diaminobenzidine ŽDAB. staining were carried out in 1% goat serum in PBS Ž140 mM NaCl, 2.6 mM KCl, 8 mM Na 2 HPO4 , 1.4 mM KH 2 PO4 , 0.2 mM Thimerosol; pH 7.4., and reactions using fluorescent secondary antibodies were carried out in 4% nonfat dry milk in PBS. The specificity of both staining procedures was controlled by omitting primary or secondary antibodies. Primary antibodies included a polyclonal rabbit anti-b-gal antibody Ž5X ™ 3X Inc.., diluted 1:1000; a monoclonal mouse anti-b-gal antibody ŽDevelopmental Studies Hybridoma Bank., diluted 1:50; a mouse monoclonal antibody which recognizes microtubule associated protein 2 ŽMAP2. of neurons w8x, diluted 1:50; a rabbit polyclonal antibody which identifies glial fibrillary acidic protein ŽGFAP. of astrocytes ŽBiomedical Technologies., diluted 1:500; a mouse monoclonal OX-42 antibody which recognizes the iC3b receptors on macrophages, microglia and activated CD 8 q T cells ŽHarlan Products for Bioscience., diluted 1:500. The DAB reactions were carried out using the Vectastain ABC kit ŽVector Labs.. The fluorescent secondary antibodies, including fluorescein ŽFITC.-conjugate donkey anti-rabbit IgGŽH q L., Texas red-conjugate donkey anti-rabbit IgGŽH q L., fluorescein ŽFITC.-conjugate goat anti-mouse IgG q IgM and Texas red-conjugate goat anti-mouse IgG ŽH q L., were purchased from Jackson Immunochemicals. 2.7. Image analysis Stained sections were observed with a Leica DMRBE microscope and images were captured using a Photometric Sensys KAF-1400 CCD camera ŽPhotometric. and pro- Fig. 3. Transgene expression in cerebellar Purkinje cells. Sections of cerebellum stained by X-gal histochemistry and counterstained with cresyl violet show lacZ expression localized in the Purkinje cell layer Žarrows. in a 1-month survival animal. Scale bars: A s 500 mm; B s 50 mm. Y. Liu et al.r Brain Research 768 (1997) 19–29 cessed on a Macintosh Power PC 8500 with IP Lab ŽScanalytics. and NIH image analysis software packages. 3. Results 3.1. Transgene expression at the injection site in non-immunosuppressed animals Adult rats received injections of 5 m l of recombinant adenovirus ŽAdCMVlacZ, 10 11 pfurml. containing the lacZ gene into the right side of the spinal cord at T7–8 level. One week following injection, the transgene expression level was very high as demonstrated by either X-gal histochemical ŽFig. 1A,B. or DAB immunocytochemical staining ŽFig. 1C,D.. At this and at later times the two methods provided comparable results. Intense staining ex- 23 tended over several segments rostral and caudal to the injection site. This spread was probably due to diffusion of the virus, as reported previously w25x. Diffusion was limited to the side of the spinal cord which had been injected, and the contralateral side remained unlabeled ŽFig. 1.; greater intersegmental than intrasegmental labeling could reflect the selectively low resistance to diffusion offered by the axonal tracts, which run longitudinally within the white matter. By 1 and 2 months, the number of cells at the injection site that expressed the transgene had greatly decreased, and only a small number of X-gal-labeled cells remained ŽFig. 6A.. 3.2. Transgene expression distant from the injection site in non-immunosuppressed animals lacZ gene expression occurred in neurons of brain stem nuclei, including those of the red nucleus ŽFig. 2A–C., Fig. 4. Characterization of cell types infected by the adenoviral vector in the injection site. Several types of cells expressing the transgene are recognized by specific markers, stained by immunofluorescent double labeling ŽA, B, D, E, G, H., and by their characteristic morphology as revealed by X-gal histochemistry ŽC, F, I.. Neurons Žarrows. are double-labeled by anti-b-gal antibody ŽA. and anti-MAP2 antibody ŽB. and identified by their morphology ŽC.. Astrocytes Žarrows. are double-labeled by anti-b-gal antibody ŽD. and anti-GFAP antibody ŽE. and identified by their morphology ŽF.. Clusters of lymphocytes Žarrows. are double-labeled by anti-b-gal antibody ŽG. and OX-42 antibody ŽH. and recognized by their morphology ŽI.. Scale bars: A, B, D, E, G, H s 20 mm; C s 50 mm; F, I s 25 mm. 24 Y. Liu et al.r Brain Research 768 (1997) 19–29 locus coeruleus ŽFig. 2D–F. and pontine reticular formation neurons ŽFig. 2G.. Spinal cord neurons caudal to the injection site, such as those of Clarke’s nucleus ŽFig. 2I., and rostral and caudal propriospinal neurons several segments away from the injection site ŽFig. 2H., also expressed the transgene. A common feature shared by these neurons expressing the transgene was that they all have axons that terminate in or pass through the injection site w38x. The pattern of expression therefore apparently resulted from uptake of the virus by axons with subsequent retrograde transport back to the cell bodies w16x. We did not, however, find expression in corticospinal neurons or in regions of the hypothalamus and tectum that are known to send fibers into spinal cord w38x. In contrast to the sharp decrease in number of cells expressing lacZ at the injection site at longer post-injection intervals, the number of cells expressing the transgene in distant nuclei was stable for 2 months, the longest time-point studied ŽFig. 2C,F,I.. While the staining intensity remained high, at 1 week large amounts of b-galactosidase accumulated in the neurons and revealed a Golgi-like staining of cell bodies and processes ŽFig. 1 and Fig. 2., but at 1 and 2 months the staining was confined to cell bodies ŽFig. 2., suggesting down-regulation of transgene expression. Endothelial cells in cerebral vessels and choroid plexus have endogenous galactosidase activity. lacZ gene expression was also detected in cerebral endothelial cells, ependymal cells, and choroid plexus Ždata not shown.. Because endogenous b-gal activity was not seen when the X-gal histochemical reaction time was less than 4 h w24x, short incubation times allowed us to distinguish transgene expression from non-specific endogenous galactosidase ac- Fig. 5. Immune responses in the injection site following adenovirus injection in non-immunosuppressed animals. At 1 week, a dense accumulation of OX-42-positive cells surrounds the infected cells ŽA, B, C.. A and B are double-labeled by b-gal ŽA. and OX-42 ŽB. antibodies, and show several neurons Žarrows. surrounded by clusters of OX-42-positive cells Žarrowheads. resembling macrophages and microglia. In C, an adjacent section is stained by X-gal histochemistry and counterstained with cresyl violet and also shows infected neurons surrounded by small cells resembling microglia and lymphocytes Žthick arrows.. By 1 month most of the infected cells have disappeared from the injection site, but OX-42-positive cells remain along the injection tract ŽD, arrows. and are absent on the contralateral side of the spinal cord Žarrowheads.. Scale bars: A–C s 20 mm; D s 500 mm. Y. Liu et al.r Brain Research 768 (1997) 19–29 tivities. In three animals out of the 15 who received injections of virus, we also detected lacZ gene expression in the Purkinje cell layer of the cerebellum, where staining was intense and uniform ŽFig. 3.. Immunocytochemical staining using anti-b-gal antibody confirmed that the labeling of Purkinje cells represented specific staining. 25 staining in ascending projection neurons caudal to the transection, including those of Clarke’s nucleus Ždata not shown.. In the sham animal, which had received an injection of 5 ml of adenovirus stock buffer alone, there was no specific staining Ždata not shown.. 3.4. Characterization of cell types infected 3.3. Transgene expression in transection–injection and sham-operated animals Adult rats were transected at T7–8, and 2 weeks later 5 m l of AdCMVlacZ Ž10 11 pfurml. were injected into three sites at T13, caudal to the transection site. In the region of injection, we observed intense staining of X-gal at 1 week, followed by sharp decrease in the number of transduced cells at 1 and 2 months. We found no staining rostral to the transection site in either spinal cord or brain, but there was In order to characterize the types of cells expressing the transgene, we used double-labeling immunofluorescent staining with three specific markers. Neurons were identified by antibody against MAP-2 ŽFig. 4A,B., astrocytes by antibody against GFAP ŽFig. 4D,E., and macrophages, microglia and activated CD 8 q T cells by OX-42 antibody ŽFig. 4G,H.. In the injection site, we found that many of the cells that expressed the transgene were OX-42-positive, including lymphocytes ŽFig. 4G,H., microglia and Fig. 6. Transgene expression in the injection site in animals with or without immunosuppression. Animals receiving Cyclosporin A ŽB, C, D. display a temporal pattern of transgene expression that differs from that of animals without immunosuppression ŽA.. At 2 months, without immunosuppression, only a few cells are labeled by X-gal histochemistry. In contrast, animals receiving Cyclosporin A exhibit no obvious loss of transduced cells at 1 ŽB. and 2 months ŽC, D. in the injection site, as demonstrated by X-gal histochemistry. Note in C, at 2 months, the X-gal staining Žarrows. is limited to the side of spinal cord that had been injected. Scale bars: A, B, D s 40 mm; C s 500 mm. 26 Y. Liu et al.r Brain Research 768 (1997) 19–29 macrophages; relatively fewer cells were MAP-2- or GFAP-positive ŽFig. 4.. The identities of cells could also be confirmed by their characteristic morphologies ŽFig. 4C,F,I.. In distant nuclei only neurons expressed the transgene, and there was a sharp demarcation between the stained nuclei and their surrounding parenchyma where no staining occurred ŽFig. 2.. In the spinal cord several segments away from the injection site and in the brain no cells showed double labeling of b-galrOX-42 or b-galrGFAP Ždata not shown.. 3.5. Immune responses in the injection site and distant structures in non-immunosuppressed animals The immune response was prominent in the area of the injection site in the early post-injection period. One week following injection there was a dense accumulation of OX-42-positive cells ŽFig. 4H and Fig. 5B. and clusters of lymphocytes, which were also prominent within blood vessels. Some of these cells expressed the transgene ŽFig. 4G,H.. Large numbers of activated microglia surrounded b-gal-positive cells, suggesting an immunological reaction ŽFig. 5A–C.. By 1 month, most of the cells expressing X-gal had disappeared, and only a few weakly stained cells remained along the needle tract, which was overlapped by dense OX-42 immunoreactivity ŽFig. 5D.. By 2 months, the immune reaction was no longer detectable and only a few b-gal-positive cells were seen at the injection site. In contrast, distant nuclei whose neurons were labeled by retrograde transport of the virus did not show prominent OX-42 staining at any time Ždata not shown. and lacked accumulations of lymphocytes, macrophages and activated microglia, even though transgene expression continued throughout the survival period ŽFig. 2C,F,I.. 3.6. Transgene expression in immunosuppressed animals In the injection site, animals receiving Cyclosporin A displayed a temporal pattern of transgene expression that differed from that of animals without immunosuppression, although at 1 week staining patterns were similarly robust Ždata not shown.. In contrast to animals without immunosuppression, which had only residual X-gal staining at 2 months ŽFig. 6A., animals receiving Cyclosporin A demonstrated intense X-gal staining at 1 and 2 months ŽFig. 6B–D.. The pattern in distant projection regions was similar to that of animals without immunosuppression: transgene expression level but not the number of transduced cells was reduced at 1 and 2 months. 4. Discussion 4.1. Localization of transgene expression One week following injection of the recombinant adenovirus into the spinal cord, the viral infection extended for several segments rostral and caudal to the injection site. This was likely due to the diffusion of the virus, as previously observed following injection of adenovirus into the hippocampus w25x. In the injection site, different types of cells expressed the reporter gene, including neurons, astrocytes, macrophages and microglia. We attribute the limited localization of infection to the injected side of the spinal cord to a greater diffusion barrier provided by cells and tracts within a segment than along the axonal tracts that run longitudinally within the white matter. The number of cells expressing the transgene at the site decreased greatly with time, and by 2 months virtually no stained cells remained at the injection site. In contrast, expression of the b-gal reporter gene persisted for at least 2 months in cells at distances from the injection site, where transgene expression appeared to be limited to neurons that projected to the injected region. Three possible mechanisms may account for the expression seen in distant structures. ŽA. For neurons whose axons are located in the vicinity of the injection site, the virus could be picked up by their axons and retrogradely transported toward their cell bodies, as reported previously w16,18,33x. Expression of the reporter gene in these cells resulted in accumulation and subsequent diffusion of the gene products along the cellular processes, accounting for the Golgi-like staining pattern that we observed. This mechanism is likely to explain the expression that we found in neurons of the red nucleus, locus coeruleus, vestibular nuclei, nuclei of the reticular formation and Clarke’s nucleus. Indeed, lacZ expression remained caudal to the transection site in animals who were first transected and later received viral injections. The lack of staining in corticospinal neurons was surprising, and suggests that the virus can only be picked up andror retrogradely transported by selected populations of neurons whose axons are exposed to the injection. Previous studies have shown that cortical neurons can be infected directly by adenovirus injected into the adjacent brain w11x, or after retrograde transport of adenovirus injected into the striatum w6x. However, there have been no reports demonstrating retrograde infection of corticospinal neurons, suggesting that these neurons lack components important for uptake andror retrograde transport of the virus w40x. The same explanation can account for the lack of staining in neurons of hypothalamus and tectum that are known to project to the spinal cord. ŽB. Spread of the virus through the CSF circulation could account for the staining that we observed in the ependymal cells and choroid plexus, because the injection, consisting of a large number of viral particles, penetrated the meninges and allowed virus access to the CSF. Similar findings were reported following intraventricular administration of adenovirus w2x. This seems unlikely to account for neuronal staining since, except for Purkinje cells in some cases, all labeled neurons projected to the site of injection. Y. Liu et al.r Brain Research 768 (1997) 19–29 ŽC. The staining found in cerebral endothelial cells can be explained by the presence of virus in the cerebral circulation w31x. The lack of staining in the adjacent parenchyma indicated that the virus was confined within the blood vessels by the blood–brain barrier. Animals receiving only buffer injection showed staining in endothelial cells of cerebral blood vessels and choroid plexus. Two observations suggest that this staining was non-specific: first, no other structures were stained either at the injection site or elsewhere in the spinal cord or brain. Second, the staining in animals receiving buffer injection disappeared when the staining time was reduced to less than 4 h whereas staining in animals receiving viral injections remained even when the staining time was reduced to 2–3 h. The mechanisms that account for the apparently specific labeling of cerebellar Purkinje cells in some, but not all animals are not clear. 4.2. Effects of the immune response Two possible mechanisms could account for the differences between transgene expression at the injection site and distant structures. One explanation for the decreased number of cells expressing the transgene at the injection site at 1 and 2 months is the strong immune reaction evidenced by intense OX-42 staining at the injection site. In contrast, distant structures showed no increase in OX-42 staining in animals that received intraspinal injection of virus compared to control groups. In support of this interpretation we have demonstrated that animals immunosuppressed with Cyclosporin A showed no obvious reduction in the number of transduced cells in the injection site even at prolonged survival times. The localized immune reaction is probably a nonspecific process in response to the injury w6,7,40x, rather than a specific antiviral reaction, since there was no comparable loss of transduced cells in distant projection regions. Another possible explanation for the extensive elimination of transduced cells at the injection site was the large number of injected viruses that resulted in multiple infection of individual cells and very high levels of b-galactosidase expression. Indeed, we found that 1 week following injection the expression level at the injection site greatly exceeded that of the other regions w15,30x. It is also possible that excessive production and accumulation of the transgene product, the bacterial bgalactosidase, was detrimental to the host cells w28x. 4.3. Downregulation of transgene expression We also found evidence for downregulation of transgene expression. One week after injection, large amounts of b-galactosidase accumulated in the soma and diffused along the processes of transduced cells in both the injection site and various projection regions resulting in a Golgi-like staining. However, 2 months after injection, 27 staining was present only in the soma, indicating a decrease in b-galactosidase expression. This effect was independent of the immune response since it was also observed in immunosuppressed animals. The most likely explanation for the decreased level of transgene expression over time is the downregulation of the viral promoter in vivo that affected both the injection site and distant projection regions w15x. However, in the injection site there was also a large loss of transduced cells due to a strong immune response, whereas in distant projection regions the immune response did not appear to play a major role. 4.4. A potential Õector for gene therapy in spinal cord injury Several approaches have proven effective in preventing axotomy-induced neuron cell death: fetal CNS tissue transplantation w3,4,20x, transplantation of genetically modified cell lines w21,36x, and administration of exogenous neurotrophic factors either to the injury site w13,35x or cell bodies w19,23,37x. One common feature of these strategies is the administration of neurotrophic factors that are capable of rescuing axotomized neurons which would otherwise die. The differences lie in the sources of neurotrophins and the routes of delivery. The results of the present studies suggest that introducing recombinant adenovirus containing neurotrophin genes into a site of spinal cord injury could deliver therapeutic neurotrophins in two ways. Ž1. The virus will transduce cells surrounding the lesion site and induce them to synthesize and subsequently release neurotrophins to the axons and terminals of axotomized neurons. Provision of neurotrophins to injured axons has prevented axotomy-induced neuronal death in several studies of experimental spinal cord injury w13,20,21x, consistent with the classic target derived role of neurotrophins. Ž2. Axotomized and intact neurons that pick up the virus via their axons will also be transduced and therefore become capable of synthesizing and releasing neurotrophins adjacent to their cell bodies. This mechanism might allow transduced neurons to rescue themselves Žif they are axotomized. and adjacent neurons in an autocrine or paracrine fashion. 4.5. A proÕiso Our observation of rapid reduction of number of transduced cells around the injection site in animals without immunosuppression suggests that a non-specific inflammatory response damages the host tissue at the site of viral introduction. However, several strategies are now available to counteract this problem: immunosuppression by Cyclosporin A or other agents Žas shown in the present study.; second-generation adenovirus vectors that are designed to elicit less immune response w39x; and replacement of the lacZ gene with neurotrophin genes that would avoid overproduction and accumulation of foreign proteins 28 Y. Liu et al.r Brain Research 768 (1997) 19–29 while providing the cells in the injury site with necessary neurotrophic support. w15x Acknowledgements We are grateful to Dr. Marion Murray for critical reading of the manuscript. This work was supported by NIH Grants NS24707, NS24725 and NIH training Grants NS10090 and HD07467 and the Research Service of the Veterans Administration. w16x w17x w18x References w1x S. Akli, C. Caillaud, E. Vigne, L.D. Stratford-Perricaudet, L. Poenaru, M. Perricaudet, A. Kahn, M.R. Peschanski, Transfer of a foreign gene into the brain using adenovirus vectors, Nature Genet. 3 Ž1993. 224–228. w2x G. Bajocchi, S.H. Feldman, R.G. Crystal, A. Mastrangeli, Direct in vivo gene transfer to ependymal cells in the central nervous system using recombinant adenovirus vectors, Nature Genet. 3 Ž1993. 229– 234. w3x H. Bernstein-Goral, B.S. Bregman, Spinal cord transplants support the regeneration of axotomized neurons after spinal cord lesions at birth: a quantitative double-labeling study, Exp. Neurol. 123 Ž1993. 118–132. w4x B.S. Bregman, E. Kunkel-Bagden, P.J. Reier, H.N. Dai, M. McAtee, D. Gao, Recovery of function after spinal cord injury: mechanisms underlying transplant-mediated recovery of function differ after spinal cord injury in newborn and adult rats, Exp. Neurol. 123 Ž1993. 3–16. w5x P. Brundin, H. Widner, O.G. Nilsson, R.E. Strecker, A. Bjorklund, Intracerebral xenografts of dopamine neurons: the role of immunosuppression and the blood–brain barrier, Exp. Brain Res. 75 Ž1989. 195–207. w6x A.P. Byrnes, R.E. MacLaren, H.M. Charlton, Immunological instability of persistent adenovirus vectors in the brain: peripheral exposure to vector leads to renewed inflammation, reduced gene expression, and demyelination, J. Neurosci. 16 Ž1996. 3045–3055. w7x A.P. Byrnes, J.E. Rusby, M.J. Wood, H.M. Charlton, Adenovirus gene transfer causes inflammation in the brain, Neuroscience 66 Ž1995. 1015–1024. w8x J.E. Crandall, S.A. Tobet, I. Fischer, T.O. Fox, Age-dependent expression of microtubule-associated protein 2 in the ventromedial nucleus of the hypothalamus, Brain Res. Bull. 22 Ž1989. 571–574. w9x R.G. Crystal, Transfer of genes to humans: early lessons and obstacles to success. wReviewx, Science 270 Ž1995. 404–410. w10x A.M. Dannenberg Jr., M. Suga, J.E. Garcia-Gonzales, Macrophages in granulomas: histochemical evidence suggesting local control of heterogeneous functions, Hamatologie und Bluttransfusion 27 Ž1981. 109–119. w11x B.L. Davidson, E.D. Allen, K.F. Kozarsky, J.M. Wilson, B.J. Roessler, A model system for in vivo gene transfer into the central nervous system using an adenoviral vector, Nature Genet. 3 Ž1993. 219–223. w12x B.L. Davidson, S.E. Doran, D.S. Shewach, J.M. Latta, J.W. Hartman, B.J. Roessler, Expression of Escherichia coli beta-galactosidase and rat HPRT in the CNS of Macaca mulatta following adenoviral mediated gene transfer, Exp. Neurol. 125 Ž1994. 258–267. w13x P.S. Diener, B.S. Bregman, Neurotrophic factors prevent the death of CNS neurons after spinal cord lesions in newborn rats, Neuroreport 5 Ž1994. 1913–1917. w14x S.E. Doran, X.D. Ren, A.L. Betz, M.A. Pagel, E.A. Neuwelt, B.J. Roessler, B.L. Davidson, Gene expression from recombinant viral w19x w20x w21x w22x w23x w24x w25x w26x w27x w28x w29x w30x w31x w32x vectors in the central nervous system after blood–brain barrier disruption, Neurosurgery 36 Ž1995. 790–965. H.D. Durham, H. Lochmuller, A. Jani, G. Acsadi, B. Massie, G. Karpati, Toxicity of replication-defective adenoviral recombinants in dissociated cultures of nervous tissue, Exp. Neurol. 140 Ž1996. 14–20. F. Finiels, M. Gimenez y Ribotta, M. Barkats, M.L. Samolyk, J.J. Robert, A. Privat, F. Revah, J. Mallet, Specific and efficient gene transfer strategy offers new potentialities for the treatment of motor neurone diseases, NeuroReport 7 Ž1995. 373–378. B. Finsen, P.H. Poulsen, J. Zimmer, Xenografting of fetal mouse hippocampal tissue to the brain of adult rats: effects of cyclosporin A treatment, Exp. Brain Res. 70 Ž1988. 117–133. G.D. Ghadge, R.P. Roos, U.J. Kang, R. Wollmann, P.S. Fishman, A.M. Kalynych, E. Barr, J.M. Leiden, CNS gene delivery by retrograde transport of recombinant replication-defective adenoviruses, Gene Ther. 2 Ž1995. 132–137. K. Giehl, P. Mestres, W. Tetzlaff, BDNF and NT-3 exert different and overlapping effects on GAP-43 and T-alpha-1-tubulin expression in axotomized corticospinal neurons of the rat, Soc. Neurosci. Abstr. 21 Ž1995. 1056. B.T. Himes, M.E. Goldberger, A. Tessler, Grafts of fetal central nervous system tissue rescue axotomized Clarke’s nucleus neurons in adult and neonatal operates, J. Comp. Neurol. 339 Ž1994. 117–131. B.T. Himes, J. Solowska-Baird, L. Boyne, E.Y. Snyder, A. Tessler, I. Fischer, Grafting of genetically modified cells that produce neurotrophins in order to rescue axotomized neurons in rat spinal cord, Soc. Neurosci. Abstr. 21 Ž1995. 537. H.A. Jaffe, C. Danel, G. Longenecker, M. Metzger, Y. Setoguchi, M.A. Rosenfeld, T.W. Gant, S.S. Thorgeirsson, L.D. Stratford-Perricaudet, M. Perricaudet, Adenovirus-mediated in vivo gene transfer and expression in normal rat liver, Nature Genet. 1 Ž1992. 372–378. N.R. Kobayashi, A.M. Bedared, W. Tetzlaff, BDNF reverses the atrophy of chronically injured rat rubrospinal neurons after cervical ŽC3. axotomy, Soc. Neurosci. Abstr. 21 Ž1995. 1056. B. Lal, M.A. Cahan, P.O. Couraud, G.W. Goldstein, J. Laterra, Development of endogenous beta-galactosidase and autofluorescence in rat brain microvessels: implications for cell tracking and gene transfer studies, J. Histochem. Cytochem. 42 Ž1994. 953–956. G. Le Gal La Salle, J.J. Robert, S. Berrard, V. Ridoux, L.D. Stratford-Perricaudet, M. Perricaudet, J. Mallet, An adenovirus vector for gene transfer into neurons and glia in the brain, Science 259 Ž1993. 988–990. F. Lisovoski, J. Cadusseau, S. Akli, C. Caillaud, E. Vigne, L. Poenaru, L. Stratford-Perricaudet, M. Pericaudet, A. Kahn, M. Peschanski, In vivo transfer of a marker gene to study motoneuronal development, Neuroreport 5 Ž1994. 1069–1072. P.R. Lowenstein, G.W.G. Wilkinson, M.G. Castro, A.F. Shering, A.R. Fooks, D. Bain, Non-neurotropic adenovirus: a vector for gene transfer to the brain and possible gene therapy of neurological disorders, in: D. Latchman ŽEd.., Genetic Manipulation of the Nervous System, Academic Press, New York, 1996, pp. 11–39. M. Michelle, Foreign protein not viral proteins seem responsible for immune response, Gene Ther. 19 Ž1996. 3–4. C.E. Milligan, T.J. Cunningham, P. Levitt, Differential immunochemical markers reveal the normal distribution of brain macrophages and microglia in the developing rat brain, J. Comp. Neurol. 314 Ž1991. 125–135. R.L. Neve, Adenovirus vectors enter the brain. wReviewx, Trends Neurosci. 16 Ž1993. 251–253. H. Ooboshi, M.J. Welsh, C.D. Rios, B.L. Davidson, D.D. Heistad, Adenovirus-mediated gene transfer in vivo to cerebral blood vessels and perivascular tissue, Circ. Res. 77 Ž1995. 7–13. T. Ragot, N. Vincent, P. Chafey, E. Vigne, H. Gilgenkrantz, D. Couton, J. Cartaud, P. Briand, J.C. Kaplan, M. Perricaudet, A. Kahn, Efficient adenovirus-mediated transfer of a human minidystrophin gene to skeletal muscle of mdx mice, Nature 361 Ž1993. 647–650. Y. Liu et al.r Brain Research 768 (1997) 19–29 w33x V. Ridoux, J.J. Robert, X. Zhang, M. Perricaudet, J. Mallet, G. Le Gal La Salle, Adenoviral vectors as functional retrograde neuronal tracers, Brain Res. 648 Ž1994. 171–175. w34x M.A. Rosenfeld, W. Siegfried, K. Yoshimura, K. Yoneyama, M. Fukayama, L.E. Stier, P.K. Paakko, P. Gilardi, L.D. Stratford-Perricaudet, M. Perricaudet, S. Jallat, A. Pavirani, J.-P. Lecocq, R.G. Crystal, Adenovirus-mediated transfer of a recombinant alpha 1-antitrypsin gene to the lung epithelium in vivo, Science 252 Ž1991. 431–434. w35x M. Sendtner, B. Holtmann, R. Kolbeck, H. Thoenen, Y.A. Barde, Brain-derived neurotrophic factor prevents the death of motoneurons in newborn rats after nerve section, Nature 360 Ž1992. 757–759. w36x M.C. Senut, M.H. Tuszynski, H.K. Raymon, N.H. Liou, K.R. Jones, L.F. Reichardt, Regional differences in responsiveness of adult CNS axons to grafts of cells expressing human neurotrophin 3, Exp. Neurol. 134 Ž1995. 36–55. w37x W. Tetzlaff, N.R. Kobayashi, K.M. Giehl, B.J. Tsui, S.L. Cassar, A.M. Bedard, Response of rubrospinal and corticospinal neurons to injury and neurotrophins, in: F.J. Sell ŽEd.., Prog. Brain Res., Vol. 103, Elsevier, Amsterdam, 1994, pp. 271–286. w38x D.J. Tracey, Ascending and descending pathways in the spinal cord, in: G. Paxinos ŽEd.., The Rat Nervous System, 2nd ed., Academic Press, New York, 1995, pp. 67–76. 29 w39x Q. Wang, M.H. Finer, Second-generation adenovirus vectors. wReviewx, Nature Med. 2 Ž1996. 714–716. w40x M.J.A. Wood, H.M. Charlton, K.J. Wood, K. Kajiwara, A. Byrnes, Immune responses to adenovirus vectors in the nervous system, Trends Neurosci. 19 Ž1996. 497–501. w41x Y. Yang, Q. Li, H.C. Ertl, J.M. Wilson, Cellular and humoral immune responses to viral antigens create barriers to lung-directed gene therapy with recombinant adenoviruses, J. Virol. 69 Ž1995. 2004–2015. w42x Y. Yang, F.A. Nunes, K. Berencsi, E.E. Furth, E. Gonczol, J.M. Wilson, Cellular immunity to viral antigens limits E1-deleted adenoviruses for gene therapy, Proc. Natl. Acad. Sci. USA 91 Ž1994. 4407–4411. w43x Y. Yang, J.M. Wilson, Clearance of adenovirus-infected hepatocytes by MHC class I-restricted CD4q CTLs in vivo, J. Immunol. 155 Ž1995. 2564–2570. w44x Y. Yang, Z. Xiang, H.C. Ertl, J.M. Wilson, Upregulation of class I major histocompatibility complex antigens by interferon gamma is necessary for T-cell-mediated elimination of recombinant adenovirus-infected hepatocytes in vivo, Proc. Natl. Acad. Sci. USA 92 Ž1995. 7257–7261.