ARTICLE doi: /mthe , available online at on IDEAL several animal models of arthritis [15 19]. In fact, both

Transcription

1 doi: /mthe , available online at on IDEAL ARTICLE Ex Vivo Gene Delivery of IL-1Ra and Soluble TNF Receptor Confers a Distal Synergistic Therapeutic Effect in Antigen-Induced Arthritis Seon Hee Kim, 1 Eric R. Lechman, 1 Sunyoung Kim, 2 Joan Nash, 1 Thomas J. Oligino, 1 and Paul D. Robbins 1,* 1 Department of Molecular Genetics and Biochemistry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania 15261, USA 2 Seoul National University, Seoul, Korea * To whom correspondence and reprint requests should be addressed. Fax: (412) probb@pitt.edu. Intra-articular expression of antagonists of interleukin-1 (IL-1 ) and tumor necrosis factor- (TNF- ) in arthritic rabbit knee and mouse ankle joints by direct adenoviral-mediated intraarticular delivery results in amelioration of disease pathology in both the treated and contralateral untreated joints. Previous experiments suggest that direct adenoviral infection of resident antigen-presenting cells (APCs) and subsequent traveling of these cells to other sites of inflammation and lymph nodes might be responsible for this contralateral effect. To determine whether genetic modification of APCs is required for the contralateral effect, we have used an ex vivo approach utilizing genetically modified fibroblasts to express IL-1 receptor antagonist protein (IL-1Ra) and soluble TNF- receptor (stnfr) locally in arthritic joints. Retroviral vectors carrying IL-1Ra, stnfr-ig, or both genes together were used to stably infect autologous rabbit fibroblasts that were then injected intra-articularly into arthritic rabbit knee joints. The intra-articular delivery of either IL-1Ra- or stnfr-ig-expressing fibroblasts was antiinflammatory and chondro-protective in both the injected and noninjected contralateral joints. In addition, we demonstrate that the co-delivery of both antagonists in combination results in a synergistic effect in disease amelioration in both the treated and nontreated joints. These ex vivo results suggest that trafficking of vector-modified inflammatory cells is not the main mechanism responsible for the observed distal spread of the therapeutic effect. Moreover, the results demonstrate that local, ex vivo gene therapy for arthritis could be effective in blocking pathologies within untreated, distant arthritic joints. Key Words: arthritis, IL-1Ra, stnfr, inflammation, gene therapy, retroviral vectors INTRODUCTION Rheumatoid arthritis (RA) is a chronic systemic autoimmune disease characterized by joint inflammation and progressive destruction of cartilage and bone. Even though the etiology of RA remains unclear, there is now considerable evidence to suggest that interleukin-1 (IL-1 ) and tumor necrosis factor- (TNF- ) are the main mediators in the pathogenesis of RA [1]. Both IL-1 and TNF- are able to elicit the activation and proliferation of cultured synovial cells. In addition, both cytokines play a direct role in altering normal cartilage and bone metabolism because each can induce collagenase expression by synovial cells, inhibit proteoglycan synthesis in articular chondrocytes, and stimulate bone resorption in vitro [2,3]. Furthermore, IL-1 and TNF- can induce the expression of other pro-inflammatory cytokines such as IL-6, IL-8, and granulocyte-macrophage colony-stimulating factor (GM-CSF) [4 6]. IL-1Ra and stnfr are naturally occurring proteins whose levels are elevated in the serum of RA patients [7 10]. Soluble TNF receptors act to block the proinflammatory activities of TNF- by binding directly to the cytokine, preventing its interaction with the membranebound TNFR-I and TNFR-II receptors [11,12]. Similarly, the IL-1Ra protein inhibits the actions of IL-1 by competitively binding to the type I IL-1 receptor, but is unable to induce receptor signaling [13,14]. When delivered systemically as recombinant proteins, both stnfr and IL-1Ra were able either to prevent the onset of experimental arthritis or to reduce the severity of established disease in /02 $

2 ARTICLE doi: /mthe , available online at on IDEAL several animal models of arthritis [15 19]. In fact, both IL-1Ra and stnfr have exhibited efficacy in large controlled clinical studies in RA patients [20 26]. Moreover, a fusion of the p75 stnfr and the Fc immunoglobulin domain is now available commercially. However, neither IL-1 nor TNF antagonist alone can completely control disease. An open question still exists as to whether anti-stnfr therapy is mainly anti-inflammatory or whether it can B also prevent joint destruction. Conversely, while IL-1Ra has shown efficacy in blocking the radiological progression of disease, its anti-inflammatory properties are suspect. For these reasons, targeting a single mediator may not be sufficient to block the complex inflammatory response in RA. Our group earlier reported a phenomenon associated with direct adenoviral-mediated intra-articular delivery of stnfr and IL-1Ra to an arthritic rabbit knee, in which suppression of disease in the treated joint resulted in a therapeutic effect in the untreated, contralateral joint [17]. To date, this contralateral effect has been observed only following direct, intra-articular injection of gene transfer vectors, but has been observed for a variety of different therapeutic genes in mouse, rat, and rabbit models of arthritis. For example, we reported that the intra-articular administration of adenovirus expressing viral IL-10, IL-1Ra, or stnfr within rabbit joints with antigeninduced arthritis (AIA) considerably reduced disease indices in untreated contralateral joints [17,27]. Furthermore, periarticular administration of adenovirus expressing vil-10 into one paw of mice with collagen-induced arthritis (CIA) blocked arthritis onset in both the injected paws as well as within nontreated ipsilateral and contralateral paws [28]. Similarly, a therapeutic effect was observed in contralateral joints subsequent to direct administration of NF-κB deoxyoligonucleotide liposome complexes into inflamed rat knee joints [29]. A distal spreading of an anti-inflammatory effects also was noted after direct injection of either high-titer retrovirus or adenoviral vector expressing murine IL-4 into inflamed rat and mouse ankle joints, respectively [30,31]. Despite the observation that local, intra-articular gene transfer can confer a therapeutic effect in contralateral, untreated joints, the mechanism by which local gene transfer can disseminate the antiarthritic effects is currently poorly understood. One possible explanation for A FIG. 1. Retroviral vectors. (A) Constructs of modified retroviral vectors, MOIR, MOIT, and MTIT. All vectors used in this study are derived from MOI or MTI improved retroviral vectors [35], which were designed to contain minimal viral sequences for safety against generation of replication-competent virus. The genes of interest encoding IL-1Ra and/or stnfr were inserted, with all gene transcription driven by the upstream Moloney murine leukemia virus (MoMLV) long terminal repeat (LTR) promoter. Neo-selectable marker in all of constructs was translated with encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES) from the same transcript. Also, the foot-and-mouth disease virus (FMDV) IRES is responsible for translation of stnfr in MTIT. (B) Transgene expression from the transduced cells. Fibroblast cells from rabbit were transduced with viral supernatant, selected with G418, and grown for days. Cell culture supernatants were collected at 48 hours after plating 1 million of the selected cells and levels of transgene product estimated by cytokine ELISA for human IL-1Ra and human stnfr. Values shown represent the mean ± SD of at least three different batches per each group. the distal spreading of anti-inflammatory effects is that sufficient levels of transgene product were present in the contralateral joints. However, with few exceptions, we are not able to measure detectable levels of transgene product in either serum or contralateral joints after local gene transfer. An alternative theory was that virus leakage from the joint space could result in a global immunosuppression of the animal. However, in marker gene studies, the intra-articular administration of adenoviruses encoding the marker gene luciferase into one knee joint of a rabbit bilaterally inflamed with AIA led to luciferase expression in both the injected knee joint and the contralateral knee joint, with no other major organs expressing luciferase activity [17]. These data suggested that systemic virus dissemination was not the source of the distal effect. However, the use of an Ad.eGFP marker virus suggested that a morphologically distinct subset of cells, identified 592

3 doi: /mthe , available online at on IDEAL ARTICLE A B FIG. 2. Effects of transgene expression on leukocytosis. One million genetically engineered fibroblast cells were transplanted to the knee joints of rabbits 24 hours post induction of AIA. Joints were lavaged on day 3 (A) and day 7 (B) post cell transplantation. Levels of infiltrating leukocytes were counted by hemocytometer. Values shown represent the mean ± SD. Asterisks denote values that differ at P < 0.05 compared with the leukocytes of joint fluid from the control group, which was treated with egfp-expressing cells. The black bars (T) represent the joints from each set of rabbits that received genetically modified cells expressing a therapeutic transgene. The open bars represent contralateral control joints (C) from the same rabbits that were treated with only a marker gene. as APCs, infected in the injected joint could traffic to the contralateral joint ([27]; E.R.L. et al., unpublished data). Furthermore, we recently demonstrated that in vivo delivery of vil-10 gene could generate APCs that, when adoptively transferred to recipient mice, could inhibit local and distal delayed-type hypersensitivity (DTH) reactions [32]. Finally, we have observed that intravenous injection of dendritic cells (DCs) genetically modified to express either IL-4 [33] or FasL [34] was able to reverse established murine CIA. Together, these data suggest that the direct genetic modification of cells, possibly APCs, and subsequent trafficking of these cells to other sites of inflammation might be responsible for the dissemination of a local therapeutic effect. To determine if the observed contralateral effect associated with local gene therapy for arthritis is restricted to direct in vivo genetic modification of cells able to traffic from the joint, such as APCs, we have examined the therapeutic effects of intra-articular injection of genetically modified synovial fibroblasts. In previous experiments using synovial cells genetically modified to express marker genes, no substantial trafficking of the modified cells has been observed following intraarticular injection [22]. We have generated retroviral vectors that express high levels of IL-1Ra, stnfr-ig, or both together in the same vector. These vectors were used to genetically modify autologous rabbit fibroblasts that were subsequently transplanted intra-articularly into arthritic rabbit knee joints. Here we report that intra-articular ex vivo delivery of either IL-1Ra- or stnfr-ig-expressing fibroblasts were able to block intra-articular leukocytosis and reduce cartilage degradation in treated joints. In addition, the co-delivery of both antagonists displayed a substantial synergistic effect in all measured indices of disease within treated joints. Most interestingly, ex vivo delivery of IL-1Ra and stnfr-ig, either alone or in combination, was able to confer anti-inflammatory and chondro-protective therapeutic effects on distal untreated joints. These results demonstrate that ex vivo, intra-articular gene delivery is effective in conferring the contralateral effect, similar to results generated with adenoviral-mediated in vivo gene delivery. These data also demonstrate that local intra-articular gene therapy could be a viable treatment for systemic polyarticular arthritis. RESULTS Gene Expression in Transduced Autologous Fibroblasts The cdna sequences encoding human IL-1Ra and stnfr- Ig were inserted into both the MOIN and MTIN vectors [35] to generate MOIR (IL-1Ra), MOTR (stnfr-ig), and MTIT (IL-1Ra and stnfr-ig), respectively (Fig. 1A). We used supernatants from stable CRIP producer lines to transduce autologous rabbit fibroblasts. Following stable selection of modified cells in G418, we measured expression of IL-1Ra and stnfr-ig in recovered tissue culture supernatants (24 hour) by cytokine ELISA (Fig. 1B). An average of 39 ng/ml of IL-1Ra was produced from the rabbit fibroblasts transduced with the MOIN (IL-1Ra) vector, while cells modified with MTIT (both IL-1Ra and stnfr-ig) expressed 40 ng/ml of IL-1Ra per million cells. Autologous rabbit fibroblasts infected with MOTR (stnfr-ig) and MTIT (IL-1Ra and stnfr-ig) produced an average of 7.3 ng/ml and 12.7 ng/ml per million cells over 24 hours, respectively. 593

4 ARTICLE doi: /mthe , available online at on IDEAL A B FIG. 3. Effects of transgene expression on cartilage matrix metabolism. (A) Breakdown of proteoglycan in cartilage. We measured concentration of GAGs released into joint fluids spectrophotometrically. (B) Synthesis of proteoglycan in articular cartilage. Fragments of articular cartilage were shaved from the femoral condyles, and their in vitro rate of 35 SO 4 2 incorporation into GAGs measured by scintillation counting. Values shown represent the mean ± SD. Asterisks denote values that differ at P < 0.05 compared with value from joints of control group, which was delivered with egfp-expressing cells. The black bars (T) represent the joints from each set of rabbits that received genetically modified cells expressing a therapeutic transgene(s). The open bars represent contralateral control joints (C) from the same rabbits that were treated with only a marker gene. Therapeutic Efficacy of IL-1Ra and stnfrc in the Rabbit AIA Model We have demonstrated that the administration of replication-defective adenovirus expressing IL-1Ra and stnfr- Ig into bilaterally inflamed AIA rabbit joints resulted in a therapeutic effect in both treated and untreated contralateral joints. We investigated whether the contralateral effect is restricted to direct in vivo delivery of therapeutic genes by examining the therapeutic efficacy of ex vivo gene transfer of genetically modified autologous fibroblasts expressing IL-1Ra, stnfr-ig, as well as both genes in combination. We induced experimental arthritis by injection of chicken ovalbumin (OVA) into the rear knee joints of 26 rabbits previously immunized to antigen. At 24 hours post disease induction, 10 6 genetically engineered and selected autologous rabbit fibroblast cells were injected into each inflamed knee joint. The number of cells injected was selected on the basis of earlier experiments examining efficacy in rabbit AIA models following injection of IL-1Ra-expressing synovial fibroblasts [10,36]. Cells expressing the therapeutic genes were injected into the left knee of all experimental group rabbits, while autologous fibroblasts carrying the enhanced green fluorescent protein (egfp) marker were injected into the right knees. As a quantitative index of inflammation in the rabbit knee, we counted the numbers of leukocytes in the joint fluid on days 3 and 7 (Fig. 2). On day 3, the control rabbit group, treated with autologous fibroblasts expressing the marker gene egfp in both knees, were highly inflamed with mean levels of infiltrating leukocytes exceeding cells per ml of recovered lavage fluid. In the IL-1Ra- or stnfr-ig-treated knees, a small therapeutic effect on leukocytic infiltration was observed on day 3. No considerable reduction in inflammation was observed in contralateral knees for either single-therapy group on day 3. However, when both genes were administered in combination, a synergistic therapeutic effect was observed on day 3 in both the treated and untreated contralateral joints, with mean leukocytic infiltration levels of and , respectively. By day 7, the control egfp joints were still highly inflamed with a mean infiltration of cells per ml of recovered lavage fluid. However, at this time point, all three experimental treatment groups showed considerable reduction in infiltrating leukocyte levels in the treated joints. Joints that received combination therapy again displayed a mild synergistic effect compared with each therapeutic gene alone. Interestingly, all three groups showed a substantial reduction of inflammation in the untreated contralateral joints that received only a marker gene. Inhibition of Breakdown and Synthesis of Cartilage Matrix Typically, there exists in normal joints a homeostatic balance between cartilage matrix synthesis and breakdown that is skewed toward a loss of cartilage within inflamed RA joints. The levels of glycosaminoglycans (GAGs) released into synovial fluid were measured by 1,9- dimethylmethylene blue (DMMB) dye binding assay at day 7 (Fig. 3A) to examine the effect of IL-1Ra and stnfr- Ig on cartilage degradation in rabbit AIA knees. The experimental control rabbit group that received egfp marker gene-transduced fibroblasts had high levels of GAGs within the lavage fluid, with an average of 41.0 g/ml compared with 9.0 g/ml in naive control joints. When compared to egfp control rabbits, the IL-1Ra-treated rabbits showed a 41% reduction in the amount of released GAGs in the treated joint, with only a moderate reduction in the contralateral joint (31.9 g/ml). Rabbit joints 594

5 doi: /mthe , available online at on IDEAL ARTICLE FIG. 4. Histological analysis of recovered rabbit synovial tissue. Antigen-induced arthritis was induced in both knees of four groups of rabbits immunized with ovalbumin. At 24 hours after induction, all rabbits were injected in the right knee with 1 million genetically engineered fibroblast cells expressing egfp and in the left knee with autologous fibroblasts producing either therapeutic gene for the treatment group or egfp for control. A week after injection, synovial tissues from each group were harvested, sectioned, and stained with hematoxylin and eosin. Magnification: 100. (A) Synovium from a normal naive rabbit knee. (B) Synovium recovered from rabbit knee receiving cells expressing egfp in both knees after arthritis induction. (C, D) Synovial tissue recovered from rabbit knee injected with IL-1Ra-expressing fibroblasts in left knee (C) and egfp-expressing cells (D). (E, F) Tissue recovered from rabbit knees injected with stnfr-producing cells in the left knee (E) and egfp-producing cells in the right (F). (G, H) Synovial tissue recovered from rabbit knees injected with fibroblasts secreting both of IL-1Ra and stnfr in the left knee (G) and egfp in the right (F). A C E B D F G H expressing stnfr-ig showed only a modest reduction in the treated joint (33.7 g/ml), and no therapeutic effect was observed in the untreated contralateral joint. Most interestingly, rabbit knees that were injected with autologous fibroblasts expressing both IL-1Ra and stnfr-ig displayed a 70.7% reduction in mean GAG levels within treated knees, while the contralateral knees showed a 52% reduction in GAG release (Fig. 3A). We measured 35 S incorporation into proteoglycans in articular cartilage shavings from femoral chondyles on day 7 (Fig. 3B), to examine whether overexpression of IL-1Ra and stnfr-ig could block the inhibition of new matrix synthesis associated with arthritis. The control group of rabbits receiving egfp-transduced fibroblasts in both knee joints showed an inhibition of cartilage synthesis, producing only 15.6% the amount of GAGs as that of naive control joints. Rabbits joints injected with IL-1Ra-expressing cells showed > 60% of normal cartilage synthesis rates in both treated and contralateral knees. In contrast, rabbits treated with stnfr-ig alone displayed no protection of matrix synthesis rates. However, even in the absence of any observable therapeutic efficacy of stnfr-ig, rabbit knees treated with the combination therapy showed a moderate additive effect in the treated joints, maintaining 84% of control levels, with a striking spread of this effect to the untreated contralateral joints (66%). Histological Analysis The therapeutic effects following ex vivo gene transfer were examined histologically. Synovial tissue isolated from the rabbits killed at day 7 was sectioned and stained with hematoxylin and eosin (Fig. 4). When compared with synovium from nondiseased control rabbit knees (Fig. 4A), sections from the egfp-treated control group show a severe synovitis typical of that seen with AIA (Fig. 4B). The synovium is considerably expanded and hypercellular, with increased 595

6 ARTICLE doi: /mthe , available online at on IDEAL A B C D E F FIG. 5. In vivo expression of IL-1Ra and stnfr within AIA rabbit joints and serum. (A) IL-1Ra expression in AIA rabbit joint fluid on day 3 and (B) on day 7. (C) Quantitation of IL-1Ra in bloodstream of AIA rabbit on day 7. (D) In vivo expression of stnfr-ig in AIA rabbit joints on day 3 and (E) on day 7. (F) stnfr-ig expression within serum on day 7. At days 3 and 7 after transplantation of genetically modified fibroblasts, the knees were lavaged and the blood was collected. The recovered fluid and serum were analyzed by ELISA specific for human IL-1Ra and human TNFR. Values given are the mean ± SD. The black bars (T) represent the joints from each set of rabbits that received genetically modified cells expressing a therapeutic transgene(s). The open bars represent contralateral control joints (C) from the same rabbits that were treated with only a marker gene. numbers of resident synovial cells and infiltrating mononuclear cells. In contrast, joints treated with IL-1Ra showed an observable reduction in the severity of synovitis (Fig. 4C). Joints receiving only the marker gene egfp, contralateral to those that received IL-1Ra, also showed a considerable reduction in pathology (Fig. 4D). A similar pattern was observed within the group of rabbits treated with stnfr-ig. Joints receiving stnfr-ig showed a detectable reduction in synovitis, with a moderate reduction in the untreated contralateral joints (Figs. 4E and 4F). Joints receiving combination therapy appeared to have a similar or greater reduction in disease pathology both in the treated and untreated contralateral joints (Figs. 4G and 4H). Transgene Expression in the Joint and Serum We quantified levels of IL-1Ra and stnfr-ig in both serum and recovered lavage fluids from inflamed rabbit knee joints by cytokine ELISA (Fig. 5). In joints treated with autologous cells expressing IL-1Ra, an average of pg/ml and 320 pg/ml of IL-1Ra were expressed at days 3 (Fig. 5A) and 7 (Fig. 5B), respectively. Levels of IL-1Ra in combination-treated joints averaged pg/ml at day 3 (Fig. 5A) and increased to 140 pg/ml by day 7 (Fig 5B). Importantly, IL-1Ra could not be detected in any contralateral joint or in the serum (Fig. 5C). In contrast to IL- 1Ra, stnfr-ig expression could be measured in transplanted joints with levels of receptor averaging pg/ml on day 3 (Fig. 5D) and increasing to pg/ml by day 7 (Fig. 5E). The concentration of stnfr-ig in joints that received the dual treatment was 430 pg/ml on day 3 (Fig. 5D), and these levels climbed to 4 ng/ml by day 7 (Fig. 5E). Joints contralateral to those expressing stnfr-ig had very low, yet detectable, levels of receptor. Furthermore, stnfr-ig was detected in the serum of the stnfr-ig-treated rabbits, averaging 300 pg/ml at day 7 (Fig. 5F). However, within the combination-treated animals much more substantial levels of the receptor were evident in the serum at day 7 (Fig. 5E), with mean levels reaching 800 pg/ml. DISCUSSION In this study, we investigated whether in vivo genetic modification of cells is a requirement for the distal spreading of locally induced anti-inflammatory therapeutic effects, termed the contralateral effect. So far, all previous reports of a contralateral effect have been observed with direct, intra-articular injection of adenoviral, retroviral, or liposome-based gene transfer vectors. Thus, we have generated retroviral vectors that expressed high levels of IL-1Ra, soluble TNF- receptor, or both proteins together. The vectors were used to infect autologous rabbit fibroblasts, a cell type that we have shown is unable to traffic efficiently 596

7 doi: /mthe , available online at on IDEAL ARTICLE FIG. 6. Model for the contralateral effect. following intra-articular injection ([22]; unpublished data). The genetically modified synovial cells were subsequently transplanted into rabbit knees with experimental AIA. The administration of 10 6 fibroblasts expressing either IL-1Ra or stnfr-ig along was both anti-inflammatory and chondro-protective in treated joints. In addition, delivery of both antagonists in combination showed a considerable additive and possibly even synergistic effect in disease amelioration in treated joints, especially in regard to effects on cartilage metabolism. Most interestingly, when the agonists were delivered either alone or simultaneously, a considerable contralateral effect was observed in joints receiving fibroblasts transduced with a retrovirus expressing only a marker gene. Moreover, the extent of the contralateral effect was comparable to our previous results using direct intra-articular injection of adenoviral vectors carrying similar therapeutic genes. These results strongly suggest that polyarticular therapeutic effects seen with local gene therapy are not attributable to direct genetic modification of resident cells within the joint. Delivery of IL-1Ra and stnfr-ig to inflamed rabbit knees was able to inhibit several pathophysiological responses of established AIA. When IL-1Ra and stnfr-ig were administered alone, an equally potent suppression of leukocytic infiltration was observed in treated joints. Interestingly, coadministration of both transgenes resulted in a synergistic downregulation of inflammation. These results suggest that both TNF-α and IL-1 may play an equal role in the recruitment of immune cells to the joint space. A similar synergistic effect on inflammation and cartilage degradation was also observed after the intraarticular co-injection of adenoviral vectors carrying the genes for soluble IL-1 and TNF in rabbit AIA [17]. These data also concur with earlier results suggesting that IL-1 and TNF- work synergistically in the pathology of RA [37]. Thus these data suggest that targeting of multiple inflammatory mediators may be the most beneficial approach for RA treatment. Our results suggest that intra-articular IL-1Ra expression was considerably more efficacious than stnfr-ig in both the prevention of cartilage matrix breakdown and the protection of GAG synthesis rates. On day 7, GAG levels in the lavage fluids of IL-1Ra-treated joints were near half the levels of those joints treated with stnfr-ig (Fig. 3A). In fact, the chondro-protective effect in the stnfr-igtreated joints was so weak that no therapeutic effect was observed in untreated contralateral joints. Combination therapy yielded a synergistic chondro-protective effect, however, with treated joints and untreated contralateral joints releasing 71% and 52% less cartilage matrix than untreated controls, respectively. In addition, stnfr-igtreated joints showed no rescue of cartilage matrix synthesis rates. In fact, both treated and untreated control joints exhibited matrix synthesis rates statistically similar to those of untreated control rabbit joints. Conversely, IL- 1Ra-treated joints protected nearly 65% of control values of GAG synthesis rates, while also generating a strong protective effect in the untreated joint. Once more, a considerable synergistic effect was observed with combinationtreated joints, preserving nearly 84% of control levels while the contralateral joint maintained 65% of control values. These observations suggest that IL-1 is the major factor responsible for suppressing the synthesis of cartilage GAGs and effecting degradation of matrix during AIA, at least in the rabbit, with TNF playing a much smaller role. These results also correlate with earlier published results in which IL-1 was implicated in cartilage breakdown and leukocytosis in a murine model of AIA [38]. Taken together, these results raise questions about the efficacy of existing commercial therapies that target only TNF-. While TNF- antagonists may suppress joint inflammation, ongoing cartilage destruction within the joint may not be affected. When expressed alone, the levels of IL-1Ra peaked at 268 pg/ml on day 3 and fell to an average of 140 pg/ml by day 7. This decrease in expression may be due, in part, to promoter shutdown or cytotoxic T-cell-mediated killing of the genetically modified fibroblasts expressing human IL-1Ra. In contrast, when IL-1Ra was expressed in tandem with stnfr-ig, expression increased from 254 pg/ml at day 3 to 320 pg/ml at day 7. It is unclear why co-expression of both antagonists allowed for the preservation of IL- 1Ra expression levels. Whereas absolute levels of IL-1Ra were superior to stnfr-ig production in transduced cells in vitro (Fig. 1B), greater levels of stnfr-ig were detected in joint fluid in vivo at day 3 and day 7 (Fig. 5). In fact, in treated joints, stnfr-ig accumulated to very high levels, with measurable levels detectable in the serum and contralateral joint. The presence of detectable amounts of transgene product in both the serum and contralateral joints after local gene delivery to rabbit joints has not been 597

8 ARTICLE doi: /mthe , available online at on IDEAL observed previously in this particular model of disease. The soluble TNF- receptor construct used in this study contains the C H 2-C H 3 domain of a mouse IgG1 heavy chain portion fused with the extracellular domain of the 55-kDa TNFα receptor [17]. The attachment of Ig heavy chains stabilizes this construct, thereby extending its half-life, which may account for the high levels of transgene product within injected joints and the subsequent detection within serum and contralateral joints. We already have demonstrated that direct gene delivery to joints had therapeutic effects that extended to untreated contralateral joints [17]. Possible mechanisms that could be involved in this effect include dissemination of virus following injection, systemic levels of protein, trafficking of genetically or functionally modified cells, or possibly a neurogenic component. We have used marker genes to demonstrate that there is not widespread viral dissemination following adenoviral injection [17,27]. Moreover, the marker gene trafficking studies suggested that a distinct subset of virally transduced cells, possibly APCs, were able to traffic from infected joints to contralateral joints [33]. These results and other data supported an earlier proposed mechanism for the distal transfer of a local therapeutic effect [27,32]. We have hypothesized that the direct modification of infiltrating cells and their subsequent trafficking to other sites of inflammation was integral to the systemic dissemination of local therapeutic effects. However, as we clearly have demonstrated in this report, ex vivo delivery of retrovirally transduced synovial fibroblasts was able to confer a strong contralateral effect in distal untreated joints as well as in the treated joints. These results suggest that direct genetic modification of cells is not a requirement for the distal spreading of a local therapeutic effect to other joints. Considering that cells need not be directly modified to confer a distal therapeutic effect in untreated joints, there still exists considerable evidence to suggest that APCs participate in the systemic dissemination of the therapeutic effect. We have shown earlier that DCs, genetically modified in vitro to express the appropriate cytokine, have the ability to suppress established CIA in mice [33]. Specifically, bone marrow derived DCs transduced with adenoviruses expressing either FasL [34] or IL-4 were able to suppress CIA when injected either locally or systemically. In addition, we demonstrated that in vivo delivery of vil-10 at sites of inflammation could generate APCs that could inhibit both local and distal DTH reactions in recipient mice by adoptive transfer [32]. Furthermore, adoptive transfer of bone marrow derived DCs cultured in the presence of recombinant IL-10 or infected ex vivo with Ad.vIL-10 and pulsed with antigen were shown to suppress DTH reactions in an antigen-specific manner [32]. To examine further the mechanism underlying the contralateral effect, we also have developed a double-antigen arthritis model in the rabbit where disease in each of the hind knee joints is induced by injection of different antigens, either OVA or keyhole limpet hemocyanin (KLH) (E.R.L., et al., manuscript submitted). Using this doubleantigen model, we have shown that the contralateral effect observed following intra-articular injection of Ad.vIL-10 is antigen-specific, able to suppress disease in the contralateral joint only if the antigen used to induce disease is the same as in the injected joint. These results, in addition to demonstrating the antigen specificity of the contralateral effects, also strongly suggest that the effect is not due to systemic levels of protein that would confer system, nonspecific immunosuppression. These data are consistent with results generated in adoptive transfer experiments whereby the anti-inflammatory effects of local Ad.vIL-10 administration could be transferred to recipient mice via APCs [32], but only if the same inciting antigen was used to induce disease. These results, taken together with the results observed using ex vivo gene transfer, suggest that DCs functionally modified in vivo or in culture, play a major role in the suppression of both local and distal joint inflammation. The results using genetically modified synovial cells to deliver IL-1Ra and stnfr-ig intra-articularly do not rule out the possible contribution of elevated systemic protein levels in conferring the contralateral effect. However, that no systemic levels of human IL-1Ra could be detected in the serum or in the contralateral knee, even though a contralateral effect was observed following intra-articular delivery of IL-1Ra-expressing synovial cells, suggests that an alternative mechanism is involved. In contrast to IL-1Ra, it is still possible the elevated levels of stnfr-ig observed in the serum could contribute to the observed contralateral effect. It is important to note that previous unpublished studies from our laboratory have demonstrated that the steadystate intra-articular level of stnfr-ig needed to confer a therapeutic effect in the injected joint is at least 1 ng/ml. Substantially lower levels of stnfr-ig were observed in the contralateral knees following ex vivo gene transfer. Moreover, as described earlier, we have demonstrated, using a double-antigen model of arthritis, that the contralateral effect in the rabbit is antigen-specific, strongly suggesting that these effects are not mediated by systemic levels of proteins that confer general immunosuppression. In consideration of our earlier results as well as those reported in this paper, we propose the mechanism for the contralateral effect, outlined in Fig. 6. Intra-articular expression of IL-1 and TNF inhibitors such as IL-1Ra, sil-1r, and stnfr, as well as immunosuppressive cytokines such as vil- 10 that downregulate IL-1 and TNF expression, are able to modulate the activity of intra-articular APCs, including both DCs and macrophages. These modified APCs then are able to traffic to sites of immunomodulation such as the lymph nodes and possibly even distant sites of inflammation to regulate the immune response in an antigen-dependent manner. This model is consistent with observations demonstrating a contralateral effect following in vivo, adenoviralmediated delivery of a variety of therapeutic agents to joints as well as with intra-articular injection of NF-κB decoys. NFκB inhibition in APCs in vivo can result in downregulation 598

9 doi: /mthe , available online at on IDEAL ARTICLE of co-stimulatory molecules, blocking maturation of APCs that renders them more tolerogenic. It is important to note that the proposed model is not incompatible with other mechanisms mediating the observed contralateral effects. It is still possible that low systemic levels of the therapeutic proteins can contribute to the observed effect. In addition, there exists considerable evidence for a role of the nervous system in the hallmark bilateral symmetry of RA [39]. In fact, several animal studies have shown that induction of monoarthritis in one joint can lead to pathology in the contralateral untreated joint [40,41]. However, despite possible roles for other pathways in conferring the observed contralateral effect, our data strongly suggest that local intra-articular gene expression has a major function in conferring a contralateral effect in part through modulation of function of resident APCs. A phase I clinical trial to determine the safety and efficacy of local, ex vivo gene transfer to arthritic joints has been recently completed [22]. In this clinical study, autologous synovial fibroblasts were genetically modified by retroviral infection to express IL-1Ra and then injected intraarticularly into diseased MCP joints. In all patients treated, no observable adverse effects occurred, with expression of exogenous IL-1Ra noted in all treated knuckle joints. Despite the extensive evidence that ex vivo gene therapy is therapeutic, it has been suggested that local gene therapy for a systemic disorder such as RA is unrealistic. However, here we demonstrate that local, intra-articular ex vivo gene transfer is not only therapeutic in the injected joint, but also therapeutic in untreated joints. Thus our results suggest that local, ex vivo gene therapy could be effective in treating pathologies associated with a systemic autoimmune disease like RA. MATERIALS AND METHODS Construction of plasmids and production of retrovirus. IL-1Ra cdna [16] was amplified from MFG-IL-1Ra plasmid using the following primers: IL-1Ra5, 5 -GGATCCTTATGGAAATCTGCAGAGG-3 ; IL-1Ra3, 5 -GGATC- CTCGAGCTACTCGTCCTCCTG-3. Soluble TNF receptor gene (stnfr-ig) was amplified from stnfr-ig plasmid [17] using TNFsR5 and Ig3 oligomers: TNFsR5, 5 -ATCGATGCCATGGGCCTCTCCACCGTG-3 ; Ig3, 5 -ATC- GATTCATTTACCAGGAGAGTG-3. The IL-1Ra fragment amplified by PCR was inserted into the BamHI site of MOIN, COIN, and MTIN [35], resulting in MOIN-IL-1Ra, COIN-IL-1Ra, and MTIN-IL-1Ra. Also, stnfr-ig was inserted into the BamHI site of MOIN and inserted into the SalI site of MTIN-IL-1Ra as a blunt fragment resulting in MOIN-sTNF-Rc-Ig and MTIT (Fig. 1). High-titer amphotropic producer cell lines were generated for each virus. Briefly, 10 mg of each plasmid was individually transfected into both CRIP and PA317 packaging cell lines by calcium phosphate precipitation. After cloning by limiting dilution, we selected G418-resistant clones. These individual clones were screened for virus production by transduction of NIH3T3 and HIG-82 cells with 1 ml virus supernatant. We assayed culture media from retrovirally transduced NIH3T3 and HIG-82 cells for IL-1Ra and stnfr-ig production by ELISA (R&D Systems, Minneapolis, MN). Viral supernatant from each producer line was harvested 48 hours after a 1:5 passage and filtered with a m-pore syringe filter. The viral supernatants were either used directly or stored at 80 C. All producer lines were maintained in DMEM containing 10% FBS and antibiotics. Animals. Female New Zealand White (NZW) rabbits weighing kg were obtained from Myrtles Rabbitry (Thomson Station, TN) and housed at the University of Pittsburgh Central Animal Facility. Animals were allowed to acclimate 3 days before experimentation and were fed water and chow ad libitum. Acquisition, culture, and transplantation of autologous fibroblasts. Dermal tissue was recovered from the abdominal region of each rabbit. Animals were first anesthetized by i.m. injection containing 5 mg xylazine and 50 mg ketamine. Isoflurane at L/minute was used to maintain the anesthesia for the duration of the surgery. The abdominal region was shaved and prepped with betadine. A surgical skin biopsy was performed under sterile conditions, and the skin was closed with absorbable suture. We digested tissue sections with clostridial collagenase (2 mg/ml, 2 hours) at 37 C and the cells grown as monolayers. Then we grew rabbit primary fibroblasts to ~ 50% confluence in 100-mm dishes containing 10 ml of Ham s F12 medium supplemented with 10% FBS and antibiotics (GIBCO- BRL, Gaithersburg, MD). One million autologous cells were incubated with 2 ml viral supernatant in the presence of 8 g/ml Polybrene (Abbott Laboratories, Chicago, IL) for 5 hours at 37 C. Afterwards, we added 3 ml of fresh medium and returned the plates to the incubator for 20 hours. The viral supernatants were then replaced with fresh medium containing 0.8 mg/ml of G418 and grown for days. For intra-articular transplantation, transduced cultures were treated with trypsin, washed, and resuspended in Gey s balanced salt solution (GIBCO-BRL, Gaithersburg, MD) to a final concentration of cells/ml. A total of 0.2 ml of the cell suspensions was injected through the patellar tendon into the knee joints of recipient rabbits. Induction of AIA in rabbits. NZW rabbits were sensitized to chicken ovalbumin (OVA) by a series of two intradermal injections of 5 mg of OVA (Sigma, St. Louis, MO), emulsified in Freund s adjuvant, given 2 weeks apart. Two weeks after the second injection, acute monoarticular arthritis was initiated in the hind knee joints by the injection of 5 mg OVA dissolved in 1 ml saline. At 24 hours after the initiation of AIA, autologous fibroblasts were transplanted into inflamed knee joints by injection through the patellar tendon. A total of 22 rabbits were divided into three treatment groups. All three experimental rabbit groups received 1 million autologous fibroblasts expressing either stnfr-ig (6 rabbits), IL-1Ra (6 rabbits), or stnfr-ig and IL-1Ra in combination (10 rabbits), into the left knee joints, respectively. All three groups of rabbits received 1 million cells expressing the egfp marker gene into the right knee as a control. The inflamed control group of 6 rabbits received 1 million autologous fibroblasts expressing egfp into both joints. A noninflamed group of 3 rabbits served as naive controls. At days 3 and 7 post transplantation, the rabbit knees were lavaged by injection of 1 ml Gey s balanced salt solution (Life Technologies, Rockville, MD) through the patellar tendon. After manipulation of the joint to allow for ample mixing, the needle was reinserted and the fluid aspirated. The leukocytes in the recovered lavage fluids were measured by hemocytometer. Also, the levels of stnfr-ig and IL-1Ra expression in recovered lavage fluids and serum were measured by cytokine ELISA kit (R&D Systems, Minneapolis, MN). At day 7, the rabbits were euthanized by Beuthanasia overdose, and their knees dissected. Synovial capsules were immersion-fixed in 10% neutral buffered formalin for several days. The fixed tissues were then dehydrated in a gradient of alcohols, embedded in paraffin, sectioned at 5 mm, mounted on glass slides, and stained with hematoxylin and eosin. Sections were examined by light microscopy at 100 magnification. Cartilage metabolism. Glycosaminoglycans (GAGs) released into the joint space were measured in recovered lavage fluids. Joint aspirates were first centrifuged at 12,000g for 10 minutes to remove debris. We digested aliquots of 100 ml of the recovered lavage fluid with papain. Papain suspension (type III; 20 l, 19 units/mg protein; Sigma, St. Louis, MO) was added to 1 ml of buffer containing 10 mm EDTA and 0.4 M sodium acetate (ph 5.2). Equal volumes of lavage fluid and papain suspension were incubated overnight at 65 C, to degrade the proteoglycan core proteins. Papain was inactivated by the addition of iodoacetic acid to final concentration of 4 mm. After incubation of sample with 2 U of hyaluronate lyase at 37 C overnight, we measured the concentration of the sulfated GAGs by a DMMB binding assay [35]. 599

10 ARTICLE doi: /mthe , available online at on IDEAL The proteoglycan synthesis rates of articular cartilage chondrocytes were monitored by radiolabeled inorganic sulfate incorporation. Briefly, cartilage from the tibial plateau and femoral condyles was aseptically dissected from the joint and weighed. Approximately 30 mg of cartilage was then incubated in 1 ml of Neuman & Tytell (Life Technologies) serumless medium containing 40 mci of 35 2 SO 4 per each sample at 37 C. After 24 hours, proteoglycans were extracted from the cartilage shavings by incubation for 48 hours in 1 ml of 0.5 M NaOH at 4 C with gentle agitation. Following chromatographic separation of unincorporated counts, radiolabeled GAGs released into the culture medium and recovered by alkaline extraction from the cartilage were measured using a scintillation counter as earlier described. Statistical analysis. Data were pooled by experimental group and the mean ± SD were calculated using the Microsoft Excel graphing and statistical program. Both the Student s t-test and one-way analysis of variance (ANOVA) were conducted to determine the presence of any significant differences. P values less than 0.05 were considered statistically significant. ACKNOWLEDGMENTS We thank Jaculyn Waddel for technical assistance, Chang-Hwan Han for histological analysis, and Steve Ghivizzani and Chris Evans for helpful discussion. This work was supported in part by grant DK44935 and contract AR from the National Institutes of Health. RECEIVED FOR PUBLICATION APRIL 29; ACCEPTED AUGUST 30, REFERENCES 1. Arend, W. P., and Dayer, J. M. (1995). Inhibition of the production and effects of interleukin-1 and tumor necrosis factor- in rheumatoid arthritis. Arthritis Rheum. 38: Evans, C., and Robbins, P. D. (1994). Prospects for treating arthritis by gene therapy. J. Rheumatol. 21: Robbins, P. D., and Evans, C. H. (1996). Prospects for treating autoimmune and inflammatory diseases by gene therapy. Gene Ther. 3: Zoja, C., et al. (1991). Interleukin-1 and tumor necrosis factor- induce gene expression and production of leukocyte chemotactic factors, colony-stimulating factors, and interleukin-6 in human mesangial cells. Am. J. Pathol. 138: Aloisi, F., et al. (1991). Production of hemolymphopoietic cytokines (IL-6, IL-8, colony-stimulating factors) by normal human astrocytes in response to IL-1 and tumor necrosis factor-. J. Immunol. 149: Pang, G., Couch, L., Batey, R., Clancy, R., and Cripps, A. (1994). GM-CSF, IL-1, IL-1, IL- 6, IL-8, IL-10, ICAM-1 and VCAM-1 gene expression and cytokine production in human duodenal fibroblasts stimulated with lipopolysaccharide, IL-1 and TNF-. Clin. Exp. Immunol. 96: al-janadi, M., al-balla, S., al-dalaan, A., and Raziuddin, S. (1993). Cytokine profile in systemic lupus erythematosus, rheumatoid arthritis, and other rheumatic diseases. J. Clin. Immunol. 13: Toussirot, E., et al. (1994). Serum levels of interleukin 1-, tumor necrosis factor-, soluble interleukin 2 receptor and soluble CD8 in seronegative spondylarthropathies. Rheumatol. Int. 13: Hidaka, T., et al. (2001). Dynamic changes in cytokine levels in serum and synovial fluid following filtration leukocytapheresis therapy in patients with rheumatoid arthritis. J. Clin. Apheresis 16: Klimiuk, P. A., et al. (2001). Serum cytokines in different histological variants of rheumatoid arthritis. J. Rheumatol. 28: Loetscher, H., Steinmetz, M., and Lesslauer, W. (1991). Tumor necrosis factor: receptors and inhibitors. Cancer Cells 3: Abe, Y., et al. (1993). Role of 55- and 75-kDa tumor necrosis factor membrane receptors in the regulation of intercellular adhesion molecules-1 expression by HL-60 human promyelocytic leukemia cells in vitro. J. Immunol. 150: McIntyre, K. W., et al. (1991). Inhibition of interleukin 1 (IL-1) binding and bioactivity in vitro and modulation of acute inflammation in vivo by IL-1 receptor antagonist and anti- IL-1 receptor monoclonal antibody. J. Exp. Med. 173: Svenson, M., Hansen, M. B., Heegaard, P., Abell, K., and Bendtzen, K. (1993). Specific binding of interleukin 1 (IL-1) and IL-1 receptor antagonist (IL-1ra) to human serum. High-affinity binding of IL-1ra to soluble IL-1 receptor type I. Cytokine 5: Arend, W. P. (1993). IL-1 antagonism in inflammatory arthritis. Lancet 341: Bandara, G., et al. (1993). Intraarticular expression of biologically active interleukin 1- receptor-antagonist protein by ex vivo gene transfer. Proc. Natl. Acad. Sci. USA 90: Ghivizzani, S. C., et al. (1998). Direct adenovirus-mediated gene transfer of interleukin 1 and tumor necrosis factor soluble receptors to rabbit knees with experimental arthritis has local and distal anti-arthritic effects. Proc. Natl. Acad. Sci. USA 95: Mori, L., Iselin, S., De Libero, G., and Lesslauer, W. (1996). Attenuation of collageninduced arthritis in 55-kDa TNF receptor type 1 (TNFR1)-IgG1-treated and TNFR1-deficient mice. J. Immunol. 157: Joosten, L. A., Helsen, M. M., van de Loo, F. A., and van den Berg, W. B. (1996). Anticytokine treatment of established type II collagen-induced arthritis in DBA/1 mice. A comparative study using anti-tnf-, anti-il-1 /, and IL-1Ra. Arthritis Rheum. 39: Elliott, M. J., et al. (1994). Repeated therapy with monoclonal antibody to tumour necrosis factor (ca2) in patients with rheumatoid arthritis. Lancet 344: Elliott, M. J., et al. (1994). Randomised double-blind comparison of chimeric monoclonal antibody to tumour necrosis factor (ca2) versus placebo in rheumatoid arthritis. Lancet 344: Evans, C. H., et al. (1996). Clinical trial to assess the safety, feasibility, and efficacy of transferring a potentially anti-arthritic cytokine gene to human joints with rheumatoid arthritis. Hum. Gene Ther. 7: Moreland, L. W., et al. (1997). Treatment of rheumatoid arthritis with a recombinant human tumor necrosis factor receptor (p75) Fc fusion protein. N. Engl. J. Med. 337: Bresnihan, B., et al. (1998). Treatment of rheumatoid arthritis with recombinant human interleukin-1 receptor antagonist. Arthritis Rheum. 41: Weinblatt, M. E., et al. (1999). A trial of etanercept, a recombinant tumor necrosis factor receptor Fc fusion protein, in patients with rheumatoid arthritis receiving methotrexate. N. Engl. J. Med. 340: Jiang, Y., et al. (2000). A multicenter, double-blind, dose-ranging, randomized, placebocontrolled study of recombinant human interleukin-1 receptor antagonist in patients with rheumatoid arthritis: radiologic progression and correlation of Genant and Larsen scores. Arthritis Rheum. 43: Lechman, E. R., et al. (1999). Direct adenoviral gene transfer of viral IL-10 to rabbit knees with experimental arthritis ameliorates disease in both injected and contralateral control knees. J. Immunol. 163: Whalen, J. D., et al. (1999). Adenoviral transfer of the viral IL-10 gene periarticularly to mouse paws suppresses development of collagen-induced arthritis in both injected and uninjected paws. J. Immunol. 162: Miagkov, A. V., et al. (1998). NF-(B activation provides the potential link between inflammation and hyperplasia in the arthritic joint. Proc. Natl. Acad. Sci. USA 95: Boyle, D. L., et al. (1999). Intra-articular IL-4 gene therapy in arthritis: anti-inflammatory effect and enhanced th2 activity. Gene Ther. 6: Kim, S.-H., et al. (2000). Gene therapy for established murine collagen-induced arthritis by local and systemic adenovirus-mediated delivery of interleukin-4. Arthritis Res. 2: Whalen, J. D, Thomson, A. W., Lu, L., Robbins, P. D., and Evans, C. H. (2001). Viral IL- 10 gene transfer inhibits DTH responses to soluble antigens: evidence for involvement of genetically modified dendritic cells and macrophages. Mol. Ther. 4: Kim, S.-H., et al. (2001). Effective treatment of established murine collagen-induced arthritis by systemic administration of dendritic cells genetically modified to express IL- 4. J. Immunol. 166: Kim, S.-H., Kim, S., Oligino, T. J., and Robbins, P. D. (2002). Effective treatment of established mouse collagen-induced arthritis by systemic administration of dendritic cells genetically modified to express FasL. Mol. Ther. (in press). 35. Kim, S.-H., et al. (1998). Construction of retroviral vectors with improved safety, gene expression, and versatility. J. Virol. 72: Hung, G. L., et al. (1994). Suppression of intra-articular responses to interleukin-1 by transfer of the interleukin-1 receptor antagonist gene to synovium. Gene Ther. 1: Henderson, B., and Pettipher, E. R. (1989). Arthritogenic actions of recombinant IL-1 and tumour necrosis factor- in the rabbit: evidence for synergistic interactions between cytokines in vivo. Clin. Exp. Immunol. 75: van de Loo, A. A., et al. (1995). Role of interleukin-1 in antigen-induced exacerbations of murine arthritis. Am. J. Pathol. 146: Mapp, P. I., et al. (1990). Substance P-, calcitonin gene-related peptide- and C-flanking peptide of neuropeptide Y-immunoreactive fibres are present in normal synovium but depleted in patients with rheumatoid arthritis. Neuroscience 37: Kidd, B. L., et al. (1995). Neurogenic influences on contralateral responses during experimental rat monoarthritis. Brain Res. 688: Decaris, E., et al. (1999). Evidence for neurogenic transmission inducing degenerative cartilage damage distant from local inflammation. Arthritis Rheum. 42: