Gene Therapy (2003) 10, 657–667 & 2003 Nature Publishing Group All rights reserved 0969-7128/03 $25.00 www.nature.com/gt RESEARCH ARTICLE Selective gene expression in mediated via adeno-associated virus type 2 and type 5 vectors

M Cucchiarini1,w, XL Ren1, G Perides2 and EF Terwilliger1 1Division of Experimental Medicine, Harvard Institutes of Medicine and Beth Israel Deaconess Medical Center, Boston, MA, USA; and 2Department of Surgery, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA, USA

Microglia represent a crucial cell population in the central examined, an element derived from the gene for the murine nervous system, participating in the regulation and surveil- marker F4/80 was the most discriminating for lance of physiological processes as well as playing key roles microglia. Gene expression from vectors controlled by this in the etiologies of several major brain disorders. The ability element was highly selective for microglia, both in vitro and in to target gene transfer vehicles selectively to microglia would vivo. To our knowledge, this is the first demonstration of provide a powerful new approach to investigations of selective expression of transferred genes in microglia using mechanisms regulating brain pathologies, as well as enable AAV-derived vectors, as well as the first utilization of the development of novel therapeutic strategies. In this recombinant AAV-5 vectors in any macrophage lineage. study, we evaluate the feasibility of specifically and efficiently These results provide strong encouragement for the applica- targeting microglia relative to other brain cells, using vectors tion of these vectors and this approach for delivering based on two different serotypes of adeno-associated virus therapeutic and other genes selectively to microglia. (AAV) carrying cell-type-specific transcriptional elements to Gene Therapy (2003) 10, 657–667. doi:10.1038/sj.gt.3301925 regulate gene expression. Among a set of promoter choices

Keywords: CNS; microglia; gene transfer; rAAV; tissue-specificity

Introduction basic brain research, as well as for advancing the goal of gene therapy for CNS disorders. Microglia, the brain’s resident , are a crucial Attempts to genetically manipulate primary hemato- if minor component of the (CNS), poietic cells have principally focused upon transferring representing up to 10% of adult brain cells. In the CNS, genes into pluripotent stem cell populations. Relatively microglia participate in the regulation of diverse physio- little emphasis has been given to targeting more mature logical processes by controlling the development, myeloid lineages, in large part because they are strongly growth, and function of other brain cells. In addition, refractory to most types of gene transfer vehicles, even in microglia constitute a key immune surveillance system vitro. Similarly, gene transfer into brain cell populations responsible for destroying infectious agents, removing has focused largely upon the . Nevertheless, cell debris, and promoting brain tissue repair. As a some meaningful if tentative steps in the direction of consequence of their surveillance functions, microglia engineering microglia have been taken. Implantation of become activated in response to many different patho- ex vivo engineered microglial cell lines, or precursor cells, logic processes in the CNS, including AIDS dementia, has been successfully performed in several animal Alzheimer’s disease, multiple sclerosis, ischemia, and models.1,2 In another novel approach, engraftment into Creutzfeldt–Jakob disease. In turn, inappropriate or mouse brain of a neural stem cell line engineered to shed chronic over-activation of microglia is believed to play a retroviral vector resulted in gene transfer into the a key role in advancing the pathologies of several of microglia compartment.3 these diseases. In addition, microglia represent the major To approach this task, we elected to use recombinant cell type in the brain susceptible to infection by HIV. vectors based upon adeno-associated virus (rAAV) as Owing to their central roles in controlling brain injury gene transfer vehicles. AAV represents a small family of and disease, and their susceptibility to subversion in a replication-defective human parvoviruses not associated set of diverse, clinically important pathogenic processes, with any pathologies. Typical AAV vectors retain none of microglia constitute a pivotal population for study in the viral coding sequences,4 are nontoxic, and exhibit low immunogenicity,5,6 in marked contrast with other classes Correspondence: Dr EF Terwilliger, Division of Experimental Medicine, of viral vectors (adenoviruses, herpes simplex virus).7–9 Harvard Institutes of Medicine, 4 Blackfan Circle, Boston, MA 02115, The packaging limits of AAV-derived vectors are USA w comparatively small (about 5.5 kb), making inclusion of Current address: Laboratory of Experimental Orthopaedics, Department of Orthopaedic Surgery, Saarland University Medical Center, 66421 very long genomic DNA sequences impractical, but this Homburg, Germany is still fully adequate for many applications. rAAV can be Received 14 January 2002; accepted 13 September 2002 employed to transfer genes into cells of human origin as Targeting transgene expression to microglia with AAV M Cucchiarini et al 658 well as those of other primates and rodents. In human elements derived from genes encoding several different cells, the wild-type AAV genome integrates specifically myeloid-specific antigens: human CD11b and CD68, as within a short region of chromosome 19. However, well as murine F4/80. The properties of these vectors vectors deleted for the viral coding sequences integrate were compared, and contrasted against control vectors slowly and nonspecifically by a different mechanism.10 driven by the high level, nondiscriminatory CMV-IE Therefore, rAAV-delivered transgenes persist in target promoter. Transgene expression was monitored in each cells as a mixture of stable episomes and genuine case using two distinct and complementary indicator integrants, the relative proportions of which vary genes, lacZ and red fluorescent protein (RFP). Expression depending upon the cell type. However, unlike retro- from each vector was evaluated in vitro in microglia vis-a`- viruses, the episomal forms are also actively tran- vis other brain cells, as well as in a different macrophage scribed.11 lineage. While the efficiency of rAAV-mediated gene transfer Among the regulatory elements evaluated, the element varies widely between different cell types, this class of derived from the F4/80 sequence yielded the highest vector has been used successfully to transduce several level of specificity for microglia in primary cultures in primary cell types and tissues which are refractory to vitro, and provided a similar level of selectivity in the most viral vectors, including striated muscle, liver intact rat brain. Combined with the other unique hepatocytes, and vessel endothelium.5,6,12,13 Of particular properties of AAV, these results provide a strong relevance for this study, rAAV have been employed with foundation for a workable approach to effecting micro- good success by many groups to deliver genes into -specific gene transfer in living animals. populations in vitro and in cultured brain sections,14 as well as in the intact mammalian CNS.15–18 Also, while there has been little exploration of the utility Results of rAAV to effect gene transfer into differentiated macrophage populations, such as microglia, we pre- Vector design and construction viously employed an rAAV successfully on mature To attempt to restrict the expression of rAAV transgenes alveolar macrophages, as well as peripheral blood to microglia, several candidate transcription elements monocytes.12,19 were selected for the study. The tissue specificities and Most rAAV generated to date have been derived from expression patterns of the human CD11b and CD68 serotype 2 of the virus. In vivo, transduction with rAAV-2 antigens, as well as the murine F4/80, have been vectors in the brain tends to occur primarily into the analyzed in different types of myeloid cells, and each neurons,15,20 in contrast to the in vitro situation, where has been identified as a specific marker of particular and are also readily trans- myeloid lineages including, to varying degrees, micro- duced.21,22 However, successful targeting of an rAAV-2 to glia.25–28 The regulatory as well as coding sequences of oligodendrocytes in intact brain has also been reported, each gene have also been cloned.28–30 using the vector in concert with a strongly cell-type- The first candidate, CD11b, the a subunit of the specific promoter.22 CD11b/CD18 integrin adhesion molecule, is not present More recently, as additional serotypes of AAV have on the surfaces of immature myeloid precursors, but been cloned and characterized, vectors derived from expression increases during early cell differentiation and several other types of AAV have begun to be exploited. is subsequently restricted to mature monocytes, macro- Sequence alignments and other analyses indicate that phages, neutrophils, natural killer cells, and micro- AAV-5 is the most divergent of the known serotypes of glia.26,29 CD11b is also expressed at low levels in a few AAV.23,24 Of particular note, rAAV-5 vectors appear able other cell types, including fetal liver stem cells, acti- to deliver genes not only into neurons, but other major vated/memory CD8+ T cells, and B1 B cells. It has been brain cell populations in vivo, such as astrocytes, to a confirmed that this specificity of CD11b expression is degree more comparable to their capabilities observed in exerted at the transcriptional level, using a model of vitro. undifferentiated human myeloid leukemic cell lines. Although other strategies are under active develop- The second candidate, CD68, a lysosome-associated ment, presently, the most reliable way to restrict membrane glycoprotein, is present during early granu- expression of a vector transgene to particular cell types lomonopoietic differentiation and then becomes mark- is at the transcriptional level. In addition to cell-type edly upregulated in cells of several monocytic lineages, selectivity, the inclusion of cell-type-specific regulatory including tissue macrophages, blood monocytes, and, to elements in rAAV vectors can provide additional a lesser extent, microglia.27 CD68 expression has also advantages for gene delivery purposes.22 In contrast to been noted in several types of nonmyeloid cells, highly active viral promoters such as CMV-IE, the use of including lymphocytes, neutrophils, and megakaryo- cell-type-specific mammalian promoters can permit the cytes, although at much lower levels. synthesis of transgene proteins in concentrations closer The third promoter selected, derived from the murine to physiologically meaningful levels. Also, the phenom- F4/80 gene, is the least characterized of the set. F4/80, a enon of promoter silencing, characteristic of promoters transmembrane spanning cell surface molecule, is a such as CMV-IE in vivo,15 can often be avoided by using highly specific marker for macrophages.28 F4/80 is more cell-restricted mammalian elements, as for example expressed early during development28 although later in by employing the neuron-specific enolase (NSE) promo- myeloid differentiation than CD11b, but is more tightly ter to induce expression in neurons.16 restricted to myeloid lineages. Anti-F4/80 antibodies To generate microglia-specific AAV-derived vectors, react specifically with mouse tissue macrophages (from we constructed isogenic vectors from serotypes 2 and 5 the peritoneum, liver, spleen, kidneys, skin epidermis, in which expression was controlled by regulatory and thymus), macrophages derived from cultured bone

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 659 25,28,31 marrow precursors, and microglia. F4/80 is also CMV-IE SVpA expressed at low levels on blood monocytes,31 but is 2/CMV-IE-lacZ lacZ absent from the surfaces of other related cell types, (AAVβ-gal) AAV-2 AAV-2 including dendritic cells, polymorphonuclear leukocytes, 5’ ITR 3’ ITR lymphocytes, osteoclasts, and fibroblasts. To monitor the specificities of gene expression from *RFP rAAV bearing these different promoters, two distinct and complementary reporters were employed, the conven- tional lacZ marker encoding b-galactosidase (b-gal) and CD11b the more recently developed RFP.32 The use of two 2/CD11b-lacZ lacZ different genes was included to avoid over-reliance upon a single indicator, and control for complications in interpretation that might arise from endogenous back- ground contributions to either transgene signal. Unlike CD68 the case with b-gal, the fluorescent signal of RFP is also 2/CD68-lacZ lacZ readily detectable in living cells. Each myeloid-specific promoter was inserted up- stream of the lacZ gene in AAVb-gal (2/CMV-IE-lacZ), an AAV-2-derived vector plasmid,19 replacing the CMV- F4/80

IE promoter in that construct. The resulting vectors, 2/ 2/F4/80-lacZ lacZ CD11b-lacZ, 2/CD68-lacZ, and 2/F4/80-lacZ, are dia- grammed in Figure 1. The corresponding AAV-5-deriva- Figure 1 Structure of the rAAV vectors used in the study. A tives 5/CMV-IE-lacZ, 5/CD11b-lacZ, 5/CD68-lacZ, and representative set of recombinant vectors, based on type 2 AAV, is 5/F4/80-lacZ were constructed by substituting the Rous illustrated, containing either of the two marker genes chosen for the sarcoma virus (RSV) promoter present in vector plasmid analyses (lacZ or RFP) under the control of the CMV-IE promoter (2/ CMV-IE-lacZ, 2/CMV-IE-*RFP) or a myeloid-specific promoter (2/ pAAV5LacZ24 with each of the different myeloid- CD11b-lacZ or 2/CD11b-*RFP, 2/CD68-lacZ or 2/CD68-*RFP, 2/F4/80- specific regulatory elements, or a CMV-IE promoter. lacZ or 2/F4/80-*RFP). Alternatively, the promoter-transgene cassettes An RFP expression cassette (*RFP) was then inserted were cloned into vectors based on type 5 AAV (5/CMV-IE-lacZ or 5/CMV- into each of these vector constructs in place of the lacZ IE-*RFP, 5/CD11b-lacZ or 5/CD11b-*RFP, 5/CD68-lacZ or 5/CD68- marker, to produce matching sets of AAV-2 (2/CMV-IE- *RFP, 5/F4/80-lacZ or 5/F4/80-*RFP). *RFP, 2/CD11b-*RFP, 2/CD68-*RFP, and 2/F4/80-*RFP) and AAV-5 derived vector plasmids (5/CMV-IE-*RFP, 5/ CD11b-*RFP, 5/CD68-*RFP, and 5/F4/80-*RFP) carrying the RFP coding sequence. vectors. In most instances, the numbers of positive cells Packaging of each vector plasmid into virions was trended slightly higher in cultures exposed to rAAV-5 performed as described in Materials and methods. vectors than their rAAV-2 counterparts, despite the equivalent titers of the vector preparations, but the differences were not statistically significant. Prior activa- rAAV transduction of microglia and other brain cell tion of the cultures with rIFN-g, a proinflammatory populations cytokine and a prime macrophage activator, did not alter To test the feasibility of delivering and expressing rAAV- the relative rankings or extent of the gene expression (not delivered transgenes in microglia, while restricting their shown).33 expression within other brain cell populations, primary Similar results were obtained in experiments per- cultures of microglia were prepared from developing rat formed with sets of rAAV expressing the RFP indicator, . Purities of the cell preparations were at least 90%, as evaluated by confocal microscopy of treated microglia as assayed by immunocytochemistry for cell-type-spe- cultures. Intense fluorescence was observed using either cific markers. The cultures were then transduced with 2/CMV-IE-*RFP or 5/CMV-IE-*RFP (Figure 3c and g), each set of AAV-2 and AAV-5-derived vectors. Figure 2 compared to untreated cells, while the proportions of shows the results of X-Gal staining performed on fluorescent microglia observed after exposure to rAAV- cultured microglia 5 days after transduction with each *RFP bearing the myeloid-specific promoters were very vector. In cells exposed to either 2/CMV-IE-lacZ (Figure similar in each case to the numbers of stained cells seen 2b) or 5/CMV-IE-lacZ (Figure 2f), very high levels of following treatment with the corresponding lacZ vectors transgene expression were detected, in contrast to the (see representative results from day 5 post-transduction untreated control cells (Figure 2a), in which the back- in Figure 3). Together these results demonstrated ground staining was very low to undetectable. b-gal efficient rAAV-mediated gene transfer into microglia, activity was also detectable in cells transduced with the and confirmed that the patterns of expression seen with otherwise isogenic rAAV in which expression was driven different promoter elements were not dependent upon by one of the three myeloid-specific promoters. How- the gene being transferred. ever, major differences in the levels of expression were To determine the specificity of expression from rAAV noted between the three promoters. In particular, the F4/ bearing each of these promoter elements in other brain 80 element induced detectable transgene expression in cell types, cultures of primary rat neurons or astrocytes approximately 25% of the treated microglia, the CD68 were prepared from embryonic striatum, and exposed to element in a maximum of 10% of the cells, and the CD11b rAAV under conditions similar to those employed for the element in one cell or less per field, when the transgene microglia. Representative results in neurons, using cassettes were delivered in the context of AAV-5-derived vectors expressing RFP, are shown in Figure 4. Addition

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 660

Figure 2 X-Gal staining of rat primary microglia cultures transduced by AAV-2- and AAV-5-derived lacZ vectors. Transduced cells were fixed 5 days after the addition of 30 ml of each vector, and stained as described in Materials and methods ( Â 20 magnification): (a) No vector; (b) 2/CMV-IE-lacZ vector; (c) 2/ CD11b-lacZ vector; (d) 2/CD68-lacZ vector; (e) 2/F4/80-lacZ vector; (f) 5/CMV-IE-lacZ; (g) 5/CD11b-lacZ vector; (h) 5/CD68-lacZ vector; (i) 5/F4/80- lacZ vector.

Figure 3 Confocal microscopy analysis of rat primary microglia cultures transduced by AAV-2- and AAV-5-derived RFP-expressing vectors. Transduced cells were fixed 5 days after the addition of 30 ml of each vector, and mounted on microscope slides for confocal analysis (Carl Zeiss LSM 410) using a 568- nm filter ( Â 40 magnification): (a) No vector, (b) same field as (a) under transmitted light; (c) 2/CMV-IE-*RFP vector; (d) 2/CD11b-*RFP vector; (e) 2/ CD68-*RFP vector; (f) 2/F4/80-*RFP vector; (g) 5/CMV-IE-*RFP vector; (h) 5/CD11b-*RFP vector; (i) 5/CD68-*RFP vector; (j) 5/F4/80-*RFP vector.

of vectors of either serotype to neurons resulted in in neurons was undetectable after treatment with vectors intense fluorescence in most of the cells when expression driven by any of the myeloid-specific promoters. Use of was driven by the CMV-IE element, even at a low input lacZ instead of RFP as the indicator yielded the same dose. No differences in expression were noted between outcome. Closely parallel findings were obtained when rAAV-2 and rAAV-5 vectors. However, in marked primary astrocytes were exposed to a set of the vectors. contrast to the observations in microglia, RFP expression Expression from a vector was detected in astrocytes only

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 661

Figure 4 Confocal microscopy analysis of rat primary neuronal cultures transduced by the AAV-2- and AAV-5-derived RFP-expressing vectors. Transduced cells were fixed 2 days after the addition of 30 ml of each vector, and mounted on microscope slides for confocal analysis (Carl Zeiss LSM 410) using a 568-nm filter ( Â 10 magnification). (a) No vector, (b) same field as (a) under transmitted light; (c) 2/CMV-IE-*RFP vector; (d) 2/CD11b-*RFP vector; (e) 2/CD68-*RFP vector; (f) 2/F4/80-*RFP vector; (g) 5/CMV-IE-*RFP vector; (h) 5/CD11b-*RFP vector; (i) 5/CD68-*RFP vector; (j) 5/F4/80- *RFP vector.

when the transgene was controlled by the CMV-IE were noted following treatment of the cells with rIFN-g promoter, and not by any of the myeloid-specific (not shown). elements. These dramatic differences between vector However, in marked contrast to the results in micro- expression in primary microglia, as opposed to neurons glia, no expression was detected from either rAAV-2 or or astrocytes, were reproducible between different cell rAAV-5 vectors when the indicator was under the control preparations, and when different time points were of the F4/80 promoter. Similar findings were obtained evaluated. Although each cell type was demonstrably using rAAV-2 and rAAV-5 vectors expressing lacZ. Also susceptible to AAV transduction, expression of the gene striking was the failure of the F4/80 regulatory element cassette could be induced in microglia even as it was to induce any transgene expression in this cell popula- blocked in neurons and astrocytes, by careful selection of tion, either in the context of AAV-2 (Figure 5f) or AAV-5 the promoter driving the transgene. (Figure 5j), despite inducing the highest levels of transgene expression in microglia of any of the mye- rAAV transduction of other tissue macrophages loid-specific promoters. The F4/80 promoter therefore The vector sets were next tested upon primary macro- represented the most specific element evaluated for phages from a very different lineage, to further evaluate transduction of resident brain macrophages, as opposed the myeloid-specific character of each promoter in the to either other brain cell populations or other macro- context of rAAV. Mature human lung alveolar macro- phages. phages purified from broncholavage fluids were selected because we have previously established their suscept- rAAV-mediated gene transfer in vivo ibility to transduction with rAAV-2.12,17 Figure 5 shows The ability to restrict expression of an AAV-delivered the results of confocal analyses performed on alveolar transgene to microglia vis-a`-vis other brain cell types in macrophages 3 days following transduction with the vivo was evaluated next, employing one of the vectors different vectors expressing RFP. High levels of fluores- characterized in vitro as exhibiting the highest specificity cence were detected in cells transduced either with 2/ for microglia. 5/F4/80-*RFP was delivered by stereotaxic CMV-IE-*RFP (Figure 5c) or with 5/CMV-IE-*RFP injection into the striatum of male Sprague–Dawley rats. (Figure 5g), compared to untreated cells (Figure 5a). Brain sections were subsequently collected at two time Strong transgene expression was also observed when points, and processed for histological analysis. Double the RFP sequence was controlled by either the CD11b immunostaining was performed using specific antibo- promoter (Figure 5d and h) or by the CD68 promoter dies against RFP and the F4/80 surface marker (Figure (Figure 5e and i). As was noted in the microglia 6). In brains collected 3 days after injection, little or no studies, expression trended higher after exposure of RFP expression was detected. However, in brains macrophages to rAAV-5 vectors compared to their rAAV- collected 3 weeks postinjection, the RFP marker was 2 counterparts, but the differences noted were small. As easily visualized (Figure 6b) and was seen to colocalize with the microglia, no changes in expression patterns with the F4/80 antigen in cells around the injection site

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 662

Figure 5 Confocal microscopy analysis of human primary lung alveolar macrophages transduced by the AAV-2- and AAV-5-derived RFP-expressing vectors. Transduced cells were fixed 3 days after addition of 50 ml of each vector, and mounted on microscope slides for confocal analysis (Carl Zeiss LSM 410) using a 568-nm filter ( Â 40 magnification): (a) No vector, (b) same field as (a) under transmitted light; (c) 2/CMV-IE-*RFP vector; (d) 2/CD11b-*RFP vector; (e) 2/CD68-*RFP vector; (f) 2/F4/80-*RFP vector; (g) 5/CMV-IE-*RFP vector; (h) 5/CD11b-*RFP vector; (i) 5/CD68-*RFP vector; (j) 5/F4/80- *RFP vector.

ab

100 µm 100 µm

F4/80 RFP Figure 6 Double immunofluorescence analysis of rat brain sections following stereotaxic injection of the 5/F4/80-*RFP vector. Male Sprague–Dawley rats received 2 ml in one hemisphere, injected as described in Materials and Methods. At 3 weeks after vector administration, sections were fixed and analyzed by indirect immunofluorescence for colocalization of the RFP marker and the F4/80 antigen. Sections were mounted on microscope slides for confocal analysis and viewed under a Leica DMIL inverted microscope and photographed with a Kodak DC290 camera: (a) Anti-F4/80 reactivity (fluorescein), (b) same field as (a) showing anti-RFP reactivity (rhodamine).

(Figure 6a). The RFP-expressing cells were ramified cells injection site in treated animals, at either early or late with a morphology and size characteristic of microglia. time points, and there was no evidence of increased While not all cells in the region exhibiting F4/80 infiltration by microglia into the region. This observation immunoreactivity expressed RFP, most of the RFP- is consistent with previous findings regarding injections positive cells also showed a strong signal for F4/80. of rAAV into brain tissue.20,34 It also stands in contrast to The results, clearly indicating the coincidence of the results following intracerebral injections with some vector transgene product with the microglia marker, classes of viral vector, or genetically engineered cell confirmed that expression could be achieved selectively lines, into immunocompetent animals, which provoked in microglia with a vector such as 5/F4/80-*RFP. No vigorous and destructive immune responses leading to overt signs of inflammation were noted around the rapid clearance of all the genetically modified cells.7–9

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 663 Discussion CD11b or CD68 noted in human alveolar macrophages. The results in alveolar macrophages confirmed the basic The capability to genetically manipulate microglia functionality of the CD11b and CD68 promoter frag- populations with ease would lead directly to new ments employed, as well as affirming the high specificity avenues for exploration into the etiologies of multiple of the F4/80 element for microglia. The latter result is brain pathologies, as well as afford opportunities for also in accord with previous studies showing very low or gene therapy approaches to some of these disorders. undetectable levels of F4/80 in alveolar macrophages.28 Unfortunately, most classes of viral vector have proven It is possible that the low expression levels achieved in to be poor agents to transfer genes into microglia and rat microglia using the CD11b and CD68 elements were other myeloid lineages. One possible exception are because of deficiencies of the human promoter sequences vectors derived from complex lentiviruses, such as in rodent cells. However, this seems unlikely. Strong HIV,35,36 which naturally target this cell type. However, transgene expression from the human CD68 promoter most such vectors also carry significant stretches of has been described in murine macrophages,39 and native viral coding sequence, because their packaging transgenic mice bearing gene cassettes under the control signals are embedded within viral open reading frames. of the human CD11b promoter exhibited vigorous For clinical applications, they are also freighted by expression of the transgene product in their macro- additional safety considerations associated with using phages.43–45 agents derived from pathogenic viruses. The disparities between expression of native gene Technologies based on rAAV have proven to be products and production from a heterologous transgene powerful tools for delivering genes into several organ driven by the isolated promoter may relate to additional systems, and most prominently for targeting brain sequences that contribute to maximal expression in cells.15–18 The susceptibility of microglia to infection by microglia. For example, Greaves et al30 demonstrated AAV-2 was originally suggested by Bartlett et al37 using that the human CD68 sequence lies directly downstream wild-type virus. The presence of the AAV was detected of the eIF4A1 gene. Some additional far-upstream in microglia following stereotaxic injection of Cy3- regulatory sequences may thereby contribute to the labelled virions into rat brain, although preferential cell-type specificity of the native CD68 product. Simi- binding and uptake by neurons occurred to a much larly, the macrophage specificity of the native CD11b greater extent and at much earlier times (minutes instead promoter has been reported to be fully contained within of hours), and they were unable to detect AAV gene a fragment extending from bp À412 to bp +83,29 but expression in the microglia. Using an in vitro system, we expression of CD11b-driven transgenes has not been confirmed the permissivity of primary microglia to rAAV detected in microglia of transgenic animals, even when transduction and were able to detect high levels of longer gene fragments were employed. transgene expression in these cells, using two different These results represent the first demonstration of gene reporter genes placed under the control of the strong transfer into primary microglia using an rAAV, as well as heterologous CMV-IE promoter, instead of the much the first successful use of rAAV-5 in any macrophage weaker AAV p5 promoter, which controls expression of lineage, and show that it is practical to potently suppress the major viral proteins in wild-type AAV. transcription from vector-delivered transgenes in neu- We then employed two strategies to circumvent the rons, while permitting it in another brain cell type. Direct preference for neurons characteristic of classic rAAV-2 injection of an rAAV vector carrying the F4/80 element vectors. First, we constructed AAV-5-derived vectors, also demonstrated the feasibility of genetically manip- which transduce the major brain cell populations in vivo ulating microglia in vivo. In this study, sustained and with less selectivity compared to rAAV-2.15,20 Using restricted transgene expression was observed in micro- rAAV-5 vectors, we observed transduction efficiencies glia 3 weeks postinjection, but not after 3 days. This time in microglia in vitro at least as high as those achieved lag in the appearance of RFP is in good agreement with with equivalent titers of rAAV-2. Second, we generated reports by other investigators characterizing AAV- vectors carrying myeloid-specific promoters in order to mediated gene transfer into neurons and other cell types restrict gene expression to these cells, a strategy in vivo.16,22,46 The stringency of the restriction to micro- previously employed in retroviral38 and adenoviral glia imposed by the F4/80 element, observed both in vectors.39 Among the promoters tested, the element vitro and in vivo, is particularly impressive considering controlling expression of the murine F4/80 resulted in the exquisite susceptibility of neurons to uptake of AAV. the highest levels of transgene activity in the microglia, Transduction efficiencies observed with the different while simultaneously exhibiting the maximum discrimi- rAAV on microglia were similar to those of the alveolar nation against expression in another type of macrophage. macrophages, with both myeloid cell types requiring Several other studies have suggested that F4/80 is the about 10 Â higher vector input doses, compared with most reproducible and specific antigen for detection of neurons, to achieve similar effects. this cell type,26,28,40 compared with either CD11b,26,41 or Taken together, these results provide strong encour- CD68.41,42 agement for the further examination of recombinant Unexpectedly, gene expression driven by the CD11b AAV vectors for targeting therapeutic genes to microglia, and CD68 promoters in transduced microglia was and should lead to valuable applications for experimen- notably muted, particularly in the case of CD11b. The tal brain research as well as for gene therapy. Several absence of more robust transgene expression from these attempts have also been made recently to redirect the elements in microglia was not due to a lack of sequences host cell tropism of AAV-2 by inserting short peptide necessary for promoter activity per se, or the presence of ligands into the sequence of the capsid protein VP3, the poison sequences, as demonstrated by the much higher major constituent of the virion.47–50 These served both to levels of expression from transgenes driven by either inactivate the natural broad tropism of the virus, and

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 664 confer new binding affinities for particular cell-surface plasmid containing the AAV-2 replication and encapsi- receptors. As X-ray crystallography data about the dation functions.4,55 The AAV-5-based plasmids were structure of the AAV-2 capsid is becoming available, derived from pAAV5LacZ, provided by JA Chiorini, as the outcomes of such manipulations should become was 5RCB, a helper plasmid containing the AAV-5 more predictable in the very near future. As the knowl- replication and encapsidation functions.24 2/CMV-IE- edge base about the capsid structures of different types lacZ is an AAV-2 vector plasmid containing the lacZ gene of AAV continues to expand, such approaches, combined under the control of the cytomegalovirus immediate- with careful choice of regulatory elements in the early promoter (CMV-IE) and the simian virus 40 (SV40) delivered transgenes, should lead in the near future to small t antigen intron/polyadenylation signal (SVpA), bona fide ‘designer vectors’ exhibiting both high-affinity which has been described previously as AAVb-gal.19 and very tight selectivity for their cellular targets. A 496-bp segment of the CD11b promoter sequence, extending from position À412 to +83 of the human CD11b gene was provided by DG Tenen.29 A CD68 Materials and methods regulatory element, consisting of the 460-bp sequence upstream of the start codon of CD68, followed by the first Cells intron of the gene (83 bp) that acts as a macrophage- The 293 line, an adenovirus-transformed human em- specific enhancer, was provided by DR Greaves.30 The bryonic kidney cell line was maintained in Eagle’s F4/80 promoter fragment, comprised of the 667 bp minimal essential medium (Mediatech cellgro, Herndon, immediately upstream of the start site of translation, VA, USA) containing 10% fetal bovine serum (FBS) was provided by A McKnight.28 (Gibco BRL Life Technologies, Grand Island, NY, USA) The Red Fluorescent Protein RFP expression cassette and 100 U/ml-100 mg/ml penicillin–streptomycin (pen– (*RFP) was derived by PCR amplification from plasmid strep). pDsRed1-N1, encoding the Discosoma sp. red fluorescent Primary rat microglia cultures were prepared from the (Clontech Laboratories, Palo Alto, CA, USA; cat. no. cortices of 1 to 2-day-old neonatal Wistar rats as 6921.1). *RFP incorporated (1) a 22-bp leader/Kozak previously described51,52 and maintained in DMEM/F- element, (2) a synthetic 48-bp in-frame sequence contain-

12 (Gibco BRL Life Technologies), 10% FBS, pen–strep at ing a (gly)4 motif followed by a nucleolar localization 56 371C under 5% CO2. Under the conditions of prepara- domain derived from the HIV tat gene, and (3) the RFP tion, the cultures were 95–99% pure, as characterized by coding sequence. immunocytochemistry for the microglia marker ED-1. Prior to transduction, the microglia were cultured for 2 rAAV vector packaging, purification, and titration days in 24-well plates on microscope coverglasses (Fisher Recombinant AAV-2 and AAV-5 vectors were packaged Scientific, Pittsburgh, PA, USA; cat. no. 12-545-80) using adenovirus 5 (Ad5) to provide helper functions, in precoated with 0.2 mg/ml poly-D-lysine (Sigma, St combination with the trans-acting AAV factors supplied Louis, MO, USA; cat. no. P6407) at a density of between by pAd8 and 5RCB, respectively, as previously de- 1 and 2 Â 105 microglia well. For activation, the cells were scribed.4,12,17,19,24,46,55 Briefly, semiconfluent 293 cells were incubated with 100 U/ml of recombinant human IFN-g infected by Ad5 at a multiplicity of infection (MOI) of 3–5 (rIFN-g) (Sigma; cat. no. I3265) for 48 h at 371C 3 days and then cotransfected with 18 mg of AAV vector plasmid following rAAV transduction.27 and 6 mg of helper plasmid, using a calcium phosphate Rat primary neuronal cultures were prepared from precipitation method. At 3 days post-transfection, cells embryonic day 17–18 rat striatum as previously de- were harvested and subjected to four freeze–thaw cycles scribed53 and maintained in DMEM/F-12, with 1 Â B27 to release the cell-associated virions. Residual Ad5 was supplement (Gibco BRL Life Technologies), pen–strep at inactivated by heating at 561C for 30 min, and any 6 371C under 5% CO2. Cells (2 Â 10 ) were grown in 12- remaining plasmid DNA was removed by treatment with well plates on microcoverglasses (VWR Scientific Pro- 25 U of RNase-free DNase (Promega, Madison, WI, USA; ducts, West Chester, PA, USA; cat. no. 483–046) precoated cat. no. M6101) per milliliter at 371C for 60 min. Viral with 0.2% polyethylenimine (Sigma; cat. no. P3143) in vector preparations were purified by overnight dialysis 0.05 M Na2B4O7 Á 10H2O pH 8.4 for 2 days prior to rAAV at 41C against Dulbecco’s modified Eagle’s medium (Bio- transduction. The ratio of cells staining positive histo- Whittaker, Walkersville, MD, USA; cat. no. 12-604F), chemically for NSE versus glial fibrillary acid protein using Slide-A-Lyzer 10 K MWCO dialysis cassettes from (GFAP) was greater than 10 : 1. Pierce (Rockford, IL, USA; cat. no. 66425), when the Human primary lung alveolar macrophages were reporter gene was lacZ and Spectra/Por CE DispoDialy- prepared from bronchoalveolar lavages as previously zers 50K MWCO from Spectrum (Laguna Hills, CA, described12 and cultured in RPMI 1640 (Mediatech USA; cat. no. 135 522) when the reporter gene was RFP. cellgro), 10% FBS, pen–strep at 371C under 5% CO2. Titers of the viral vector preparations were screened by Cells (0.4 Â 106) were grown in 12-well plates on micro real-time PCR as previously described57 using a Prism coverglasses precoated with 0.2 mg/ml poly-D-lysine for 7700 Sequence Detector System from Perkin-Elmer (PE)- 2 days prior to rAAV transduction. For activation, the Applied Biosystems (Foster City, CA, USA) and the cells were incubated with 1000 U/ml of rIFN-g for 24 h at TaqMan Universal PCR Master Mix (PE-Applied Biosys- 371C 2 days following rAAV transduction.54 tems; cat. no. 4304437). Titers estimated by this method were between 1010 and 1011 transgene copies per Plasmids milliliter, with no consistent significant differences The AAV-2-derived vector plasmids used in this study between the titers of rAAV-2 and rAAV-5 preparations. were derived from a parental genomic clone, pSSV9, The primer sets and fluorescent probes were purchased obtained from RJ Samulski, as was pAd8, a helper from PE-Applied Biosystems and Metabion (Martinsried,

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 665 Germany), respectively. Titer estimations obtained by Hill, NC, USA) and JA Chiorini (Gene Therapy and this methodology were consistently at least one log Therapeutics Branch, National Institute of Dental and higher than functional titers obtained by serial dilution of Craniofacial Research, Bethesda, MD, USA) for provid- vector preparations on 293 cells. It has been estimated ing AAV precursor plasmids, DG Tenen (Hematology/ that the ratio of virus particles to functional vectors is Oncology Division, Harvard Institutes of Medicine, about 500–1.58 Boston, MA, USA) for the CD11b promoter element, D In vitro vector transductions were performed by Greaves and A McKnight (Sir William Dunn School of incubating the target cells with rAAV in a limiting Pathology, Oxford, UK) for the CD68 and F4/80 amount of serum-free medium for 90 min at 371C. Cells promoter sequences, J de Vellis and R Cole (UCLA were then cultured in appropriate medium with serum. Mental Retardation Research Center, Los Angeles, CA, USA) for primary microglia cultures, H Koziel (Pulmon- Analyses of transgene expression ary Laboratory, Beth Israel Deaconess Medical Center, Detection of b-galactosidase (b-gal) activity in cells Boston, MA, USA) for primary lung alveolar macro- transduced with lacZ-expressing vectors was performed phage preparations, and JW Francis (Department of by X-Gal staining using a standard method.59 For the Neurology, Massachusetts General Hospital, Boston, evaluation of fluorescence in cells exposed to RFP- MA, USA) for rat primary neuronal cultures. We also expressing vectors, cells were first fixed in 2% parafor- thank CJ Cahill and DA Brown (Joslin Diabetes Center maldehyde, 23 mM NaH2PO4 Á H2O, 77 mM Na2HPO4,pH and Beth Israel Deaconess Medical Center, Boston, MA, 7.4 for 30 min at room temperature and mounted on USA) for help with confocal imaging, V Petkova microscope slides (Fisher Scientific; cat. no. 12-550-15) (Hematology/Oncology Division, Harvard Institutes of using Aqua Poly/Mount mounting medium (Poly- Medicine, Boston, MA, USA) for assistance with real- sciences, Inc., Warrington, PA, USA; cat. no. 18606). time PCR, H Madry (Laboratory of Experimental Fluorescence was then detected using a confocal micro- Orthopaedics, Department of Orthopaedic Surgery, scope (Carl Zeiss LSM 410, Thornwood, NY, USA) with a Saarland University Medical Center) for helpful discus- 568-nm filter. sions, and J Delahanty for assistance with graphics preparation and editing. CNS injections and histology rAAVs were injected into the brains of male Sprague– Dawley rats (150–200 g). Briefly, rats were anesthetized References with 90 mg/kg ketamine, 10 mg/kg xylazine, and mounted on a Kopf stereotactic frame. A burr hole was 1 Sawada M et al. Brain-specific gene expression by immortalized drilled 3 mm left of the Bregma, and 2 ml of rAAV were microglial cell-mediated gene transfer in the mammalian brain. injected into the striatum at a depth of 3 mm at a rate of FEBS Lett 1998; 433: 37–40. 1 ml/min using a glass micropipette air pressure injection 2 Benninger Y et al. Differentiation and histological analysis of system. The wounds were closed with wound clips, the embryonic stem cell-derived neural transplants in mice. Brain animals were allowed to recover, and were then returned Pathol 2000; 10: 330–341. to their cages with unlimited access to food and water. 3 Lynch WP, Sharpe AH, Snyder EY. Neural stem cells as The rats were killed at different times post-treatment by engraftable packaging lines can mediate gene delivery to

CO2 inhalation, and their brains collected. Brains were microglia: evidence from studying retroviral env-related neuro- fixed overnight in freshly prepared 10% paraformalde- degeneration. J Virol 1999; 73: 6841–6851. hyde in phosphate buffer, washed with Tris buffer, and 4 Samulski RJ, Chang LS, Shenk T. Helper-free stocks of cryoprotected by immersion in 30% sucrose prior to recombinant adeno-associated viruses: normal integration does being embedded in O.C.T. compound (Sakura/Tissue- not require viral gene expression. J Virol 1989; 63: 3822–3828. Tek, Torrence, CA, USA; cat. no. 4583). The tissues were 5 Flotte TR et al. Stable in vivo expression of the cystic fibrosis sectioned at 6 mm on a cryo-microtome (Leica CM1850; transmembrane conductance regulator with an adeno-associated Northvale, NJ, USA) and placed on microscope slides. virus vector. Proc Natl Acad Sci U S A 1993; 90: 10613–10617. RFP and F4/80 were detected by staining the sections 6 Xiao X, Li J, Samulski RJ. Efficient long-term gene transfer into with an anti-RFP monoclonal antibody (Clontech La- muscle tissue of immunocompetent mice by adeno-associated virus vector. J Virol 1996; 70: 8098–8108. boratories; cat. no. 8374-2) at a 1 : 50 dilution, and a rat 7 Wood MJ et al. Inflammatory effects of gene transfer into the anti-F4/80 antibody (Serotec, Raleigh, NC, USA; cat. no. CNS with defective HSV-1 vectors. Gene Ther 1994; 1: 283–291. MCAP497) at a 1 : 100 dilution, respectively. Antigen 8 Byrnes AP, Rusby JE, Wood MJ, Charlton HM. Adenovirus gene reactivities were visualized using a rhodamine-conju- transfer causes inflammation in the brain. Neuroscience 1995; 66: gated goat anti-mouse (Caltag, Burlingame, CA, USA; 1015–1024. cat. no. M30002) at 1 : 100, and a biotinylated rabbit anti- 9 Lawrence MS et al. Inflammatory responses and their impact on rat (Vector Laboratories, Burlingame, CA, USA; cat. no. beta-galactosidase transgene expression following adenovirus BA-4000) at 1 : 150 together with FITC-conjugated strep- vector delivery to the primate caudate . Gene Ther 1999; 6: tavidin (Vector Laboratories; cat. no. SA-5001) at 1 : 200. 1368–1379. 10 Yang CC et al. Cellular recombination pathways and viral terminal repeat hairpin structures are sufficient for adeno- Acknowledgements associated virus integration in vivo and in vitro. J Virol 1997; 71: 9231–9247. This research was funded by an award to EF Terwilliger 11 Flotte TR, Afione SA, Zeitlin PL. Adeno-associated virus vector from the Pediatric AIDS Foundation and by NIH grant gene expression occurs in nondividing cells in the absence of NS43986. We are grateful to RJ Samulski (The Gene vector DNA integration. Am J Respir Cell Mol Biol 1994; 11: 517– Therapy Center, University of North Carolina, Chapel 521.

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 666 12 Inouye RT et al. Potent inhibition of human immunodeficiency 34 During MJ et al. In vivo expression of therapeutic human virus type 1 in primary T cells and alveolar macrophages by a genes for dopamine production in the caudates of MPTP- combination anti-Rev strategy delivered in an adeno-associated treated monkeys using an AAV vector. Gene Ther 1998; 5: virus vector. J Virol 1997; 71: 4071–4078. 820–827. 13 Snyder RO et al. Correction of hemophilia B in canine and 35 Wrzesinski S et al. HTLV type 1 Tax transduction in microglial murine models using recombinant adeno-associated viral cells and astrocytes by lentiviral vectors. AIDS Res Hum vectors. Nat Med 1999; 5: 64–70. Retroviruses 2000; 16: 1771–1776. 14 Keir SD et al. Gene transfer into hypothalamic organotypic 36 Mordelet E et al. Brain engraftment of autologous macrophages cultures using an adeno-associated virus vector. Exp Neurol transduced with a lentiviral flap vector: an approach to 1999; 160: 313–316. complement brain dysfunctions. Gene Ther 2002; 9: 46–52. 15 Kaplitt MG et al. Long-term gene expression and phenotypic 37 Bartlett JS, Samulski RJ, McCown TJ. Selective and rapid uptake correction using adeno-associated virus vectors in the mamma- of adeno-associated virus type 2 in brain. Hum Gene Ther 1998; lian brain. Nat Genet 1994; 8: 148–154. 9: 1181–1186. 16 Klein RL et al. Neuron-specific transduction in the rat septo- 38 Malik P et al. Retroviral-mediated gene expression in human hippocampal or nigrostriatal pathway by recombinant adeno- myelomonocytic cells: a comparison of hematopoietic cell associated virus vectors. Exp Neurol 1998; 150: 183–194. promoters to viral promoters. Blood 1995; 86: 2993–3005. 17 Wu P, Phillips MI, Bui J, Terwilliger EF. Adeno-associated virus 39 Laukkanen J et al. Adenovirus-mediated gene transfer of a vector-mediated transgene integration into neurons and other secreted form of human macrophage scavenger receptor inhibits nondividing cell targets. J Virol 1998; 72: 5919–5926. modified low-density lipoprotein degradation and foam-cell 18 Pitkow LJ et al. Facilitation of affiliation and pair-bond formation formation in macrophages. Circulation 2000; 101: 1091–1096. by vasopressin receptor gene transfer into the ventral forebrain 40 Lawson LJ, Perry VH, Gordon S. Turnover of resident of a monogamous vole. J Neurosci 2001; 21: 7392–7396. microglia in the normal adult mouse brain. Neuroscience 1992; 19 Du B, Wu P, Boldt-Houle DM, Terwilliger EF. Efficient 48: 405–415. transduction of human neurons with an adeno-associated virus 41 Rezaie P, Patel K, Male DK. Microglia in the human fetal spinal vector. Gene Ther 1996; 3: 254–261. cord – patterns of distribution, morphology and phenotype. 20 Chamberlin NL, Du B, de Lacalle S, Saper CB. Recombinant Brain Res Dev Brain Res 1999; 115: 71–81. adeno-associated virus vector: use for transgene expression and 42 Hutchins KD, Dickson DW, Rashbaum WK, Lyman WD. anterograde tract tracing in the CNS. Brain Res 1998; 793: 169– Localization of microglia in the human fetal cervical spinal 175. cord. Brain Res Dev Brain Res 1992; 66: 270–273. 21 Doll RF et al. Comparison of promoter strengths on gene 43 Back A, East K, Hickstein D. Leukocyte integrin CD11b promoter delivery into mammalian brain cells using AAV vectors. Gene directs expression in lymphocytes and granulocytes in trans- Ther 1996; 3: 437–447. genic mice. Blood 1995; 85: 1017–1024. 22 Chen H, McCarty DM, Bruce AT, Suzuki K. Gene transfer and 44 Dziennis S et al. The CD11b promoter directs high-level expression in oligodendrocytes under the control of basic expression of reporter genes in macrophages in transgenic mice. protein transcriptional control region mediated by adeno- Blood 1995; 85: 319–329. associated virus. Gene Ther 1998; 5: 50–58. 45 Early E et al. Transgenic expression of PML/RARalpha 23 Bantel-Schaal U, Delius H, Schmidt R, zur Hausen H. Human impairs myelopoiesis. Proc Natl Acad Sci U S A 1996; 93: adeno-associated virus type 5 is only distantly related to other 7900–7904. known primate helper-dependent parvoviruses. J Virol 1999; 73: 939–947. 46 Davidson BL et al. Recombinant adeno-associated virus type 2, 4, 24 Chiorini JA, Kim F, Yang L, Kotin RM. Cloning and character- and 5 vectors: transduction of variant cell types and regions in Proc Natl Acad Sci USA ization of adeno-associated virus type 5. J Virol 1999; 73: 1309– the mammalian central nervous system. 97 1319. 2000; : 3428–3432. 25 Austyn JM, Gordon S. F4/80, a monoclonal antibody directed 47 Girod A et al. Genetic capsid modifications allow efficient re- specifically against the mouse macrophage. Eur J Immunol 1981; targeting of adeno-associated virus type 2. Nat Med 1999; 5: 11: 805–815. 1052–1056. 26 Perry VH, Hume DA, Gordon S. Immunohistochemical localiza- 48 Rabinowitz JE, Xiao W, Samulski RJ. Insertional mutagenesis of tion of macrophages and microglia in the adult and developing AAV2 capsid and the production of recombinant virus. Virology mouse brain. Neuroscience 1985; 15: 313–326. 1999; 265: 274–285. 27 Peudenier S, Hery C, Montagnier L, Tardieu M. Human 49 Wu P et al. Mutational analysis of the adeno-associated virus microglial cells: characterization in cerebral tissue and in type 2 (AAV2) capsid gene and construction of AAV2 vectors primary culture, and study of their susceptibility to HIV-1 with altered tropism. J Virol 2000; 74: 8635–8647. infection. Ann Neurol 1991; 29: 152–161. 50 Shi W, Arnold GS, Bartlett JS. Insertional mutagenesis of the 28 McKnight AJ, Gordon S. The EGF-TM7 family: unusual adeno-associated virus type 2 (AAV2) capsid gene and genera- structures at the leukocyte surface. J Leukoc Biol 1998; 63: 271– tion of AAV2 vectors targeted to alternative cell-surface 280. receptors. Hum Gene Ther 2001; 12: 1697–1711. 29 Pahl HL, Rosmarin AG, Tenen DG. Characterization of the 51 McCarthy KD, de Vellis J. Preparation of separate astroglial and myeloid-specific CD11b promoter. Blood 1992; 79: 865–870. oligodendroglial cell cultures from rat cerebral tissue. J Cell 30 Greaves DR, Quinn CM, Seldin MF, Gordon S. Functional Biol 1980; 85: 890–902. comparison of the murine macrosialin and human CD68 52 Cole R, de Vellis J. Preparation of and promoters in macrophage and nonmacrophage cell lines. cultures from primary rat glial cultures. In: Shahar A, de Vellis J, Genomics 1998; 54: 165–168. Vernadakis A, Haber B (eds). A Dissection and Tissue Culture 31 Gordon S et al. Antigen markers of macrophage differentiation in Manual of the Nervous System. Liss AR: New York, 1989, pp 121– murine tissues. Curr Top Microbiol Immunol 1992; 181: 1–37. 133. 32 Matz MV et al. Fluorescent proteins from nonbioluminescent 53 Francis JW et al. Enhancement of diphtheria toxin potency by Anthozoa species. Nat Biotechnol 1999; 17: 969–973. replacement of the receptor binding domain with tetanus toxin 33 Young HA, Hardy KJ. Role of interferon-gamma in immune cell C-fragment: a potential vector for delivering heterologous regulation. J Leukot Biol 1995; 58: 373–381. proteins to neurons. J Neurochem 2000; 74: 2528–2536.

Gene Therapy Targeting transgene expression to microglia with AAV M Cucchiarini et al 667 54 Isler P et al. Interleukin-12 production by human alveolar 57 Clark KR, Liu X, McGrath JP, Johnson PR. Highly purified macrophages is controlled by the autocrine production of recombinant adeno-associated virus vectors are biologically interleukin-10. Am J Respir Cell Mol Biol 1999; 20: 270–278. active and free of detectable helper and wild-type viruses. 55 Samulski RJ, Chang LS, Shenk T. A recombinant plasmid from Hum Gene Ther 1999; 10: 1031–1039. which an infectious adeno-associated virus genome can be 58 Carter BJ. The growth cycle of adeno-associated virus. In: Tijssen excised in vitro and its use to study viral replication. J Virol 1987; P (ed). Handbook of Parvoviruses, Vol. 1. CRC Press: Boca Raton, 61: 3096–3101. FL, 1990, pp 155–168. 56 Schwarze SR, Ho A, Vocero-Akbani A, Dowdy SF. In vivo 59 Nguyen JT et al. Adeno-associated virus-mediated delivery of protein transduction: delivery of a biologically active protein antiangiogenic factors as an antitumor strategy. Cancer Res 1998; into the mouse. Science 1999; 285: 1569–1572. 58: 5673–5677.

Gene Therapy