Abstract
Gene therapy is a promising new approach in the treatment of rheumatoid arthritis. Gene delivery to diseased joints offers the prospect of achieving high, local concentrations of a therapeutic gene product in a sustained manner, while minimizing exposure of nontarget organs. We report that a single administration of a modified adenovirus encoding the Epstein-Barr-derived homologue of IL-10 can suppress the development of disease for extended periods of time when injected locally within the periarticular tissue surrounding the ankle joints of mice with collagen type II-induced arthritis. Furthermore, we show that injection of an adenoviral vector carrying the IL-10 gene into a single paw can suppress development of arthritis in other, noninjected paws of the same individual. The systemic protection resulting from local gene therapy occurred in the absence of detectable levels of viral IL-10 in the serum. Circulating Ab levels to heterologous collagen were unaffected; however, treatment with viral IL-10 significantly suppressed the development of Abs to autologous mouse type II collagen. Thus, the treatment of a single joint by local delivery of the vIL-10 gene may protect multiple joints of the same individual while avoiding deleterious side effects often associated with systemic therapy.
Rheumatoid arthritis (RA)3 is a chronic, systemic autoimmune disorder of unknown etiology characterized by synovial inflammation and erosion of articular cartilage and subchondral bone. Conventional pharmacologic therapies used in the treatment of RA have included nonsteroidal antiinflammatory drugs, immunosuppressive agents, corticosteroids, and disease-modifying antirheumatic drugs 1, 2, 3, 4, 5 . These therapies often reduce joint inflammation and provide relief from pain, but are ineffective in preventing the destruction of bone and cartilage or restoring joint function. Moreover, antiinflammatory drugs and potent immunosuppressive molecules can lead to undesirable side effects. Recently, however, new therapeutic strategies have been developed using proteins that mediate the pathogenesis of RA, including antagonists of proinflammatory cytokines such as IL-1 and TNF-α 6, 7, 8, 9, 10 .
Biological agents with demonstrated efficacy in animal modelsof arthritis include the IL-1R antagonist (IL-1Ra), Abs to IL-1 and TNF-α, and soluble TNF-α receptors 11, 12, 13, 14, 15, 16, 17 . Unfortunately, protein-based antiarthritic therapies must usually be given by frequent parenteral administration, and have often been associated with adverse side effects 18, 19, 20 . New techniques that deliver genes encoding immunomodulatory cytokines enable their expression in the body and offer the potential to overcome some of these problems. Specifically, gene transfer may facilitate the continuous release of therapeutic proteins following direct delivery of genes to tissues (in vivo delivery) or by the injection of modified cells engineered to secrete antiarthritic cytokines (ex vivo delivery) 21, 22 . Indeed, in vivo and ex vivo gene therapy has shown efficacy in ameliorating joint pathology in a number of animal models of inflammatory arthritis 23, 24, 25, 26, 27, 28, 29, 30, 31 . Both viral and nonviral vectors have been developed that are capable of efficient gene transfer; however, if administered systemically, these vectors generally achieve only transient expression. More importantly, host responses to viral components and diffuse expression of immunoregulatory molecules can lead to tissue inflammation, widespread immunosuppression, and toxicity if present at high concentrations 32, 33, 34, 35 . An alternative strategy is the delivery of genes locally to the joint. This approach would lead to the accumulation of antiarthritic proteins in and around articular tissues, where they are needed the most, and reduce exposure of extraarticular tissues and organs, thereby reducing potential side effects associated with systemic therapy 20 . For example, gene transfer to synovium and intraarticular expression of the IL-1Ra gene, via viral vectors, have been shown to be efficacious in reducing inflammation and cartilage loss in Ag-induced arthritis in the rabbit 23 , streptococcal cell wall arthritis in rats 29 , and collagen-induced arthritis (CIA) in mice 28 .
In this study, we have chosen to express the gene encoding the Epstein-Barr viral homologue of IL-10, known as viral IL-10, within the synovium and peripheral tissues of mice with CIA, in a preventative protocol. CIA shares many immunologic and pathologic features with RA and has been used extensively in the discovery and development of antiarthritic therapeutic agents 36, 37, 38, 39 . IL-10 inhibits the production of proinflammatory molecules such as IL-1, IL-6, IL-8, IL-12, and TNF-α by activated monocytes and Th1-type lymphocytes, and down-regulates MHC II expression on APCs 40, 41, 42, 43, 44, 45 . Additionally, IL-10 up-regulates the production of antiinflammatory molecules such as IL-1Ra and soluble TNF-α receptors 40, 46 . Systemic administration of rIL-10 has been reported to suppress the incidence, delay the onset, and reduce the severity of disease in CIA in mice and rats 47, 48, 49, 50 . Furthermore, neutralization of endogenous murine IL-10 with mAbs leads to more severe disease in these animal models 47, 51 . Viral IL-10 shares many of the immunosuppressive and antiinflammatory properties of cellular IL-10. However, unlike cellular IL-10, viral IL-10 does not induce MHC class II expression on B cells and does not act as a costimulatory molecule in T cell activation 52, 53 . Therefore, viral IL-10 may serve not only as a superior antiinflammatory agent, but also as a stealth molecule that reduces the body’s immune reaction to the adenoviral vector used in gene transfer of the vIL-10 gene to the joint 52, 54 .
In the studies described below, we report that direct transfer of the viral IL-10 gene to the paws of mice can prevent the development of CIA. Apparailly et al. 31 have recently reported that systemic delivery of an adenoviral vector encoding vIL-10 also inhibits murine CIA. Our adenoviral research, in contrast, indicates that systemic delivery in this fashion exerts only a weak and transient antiarthritic effect at nontoxic doses. Local delivery, in contrast, resulted in a strong and persistent suppression of CIA, without evidence of adverse side effects. Furthermore, we describe the unexpected and potentially important observation that when only one or two paws were injected with the adenoviral vector, all four paws were protected from developing disease.
Materials and Methods
Induction of CIA
Native bovine collagen type II, dissolved in 0.05 M acetic acid at a concentration of 2 mg/ml, was emulsified in an equal volume of CFA (Difco, Detroit, MI). Mice were injected once intradermally into the base of the tail with 100 μl of the emulsion (100 μg CII/mouse). Joint count was defined as the number of paws per animal with definitive edema and erythema of the metatarsal or metacarpal regions; paws with only one or two swollen digits were noted, but were not included in the joint count. The degree of paw swelling was determined using a spring-loaded caliper (Dyer, Lancaster, PA), and the clinical severity of arthritis was measured according to a graded scale: 0, normal; 1, significant edema and erythema; 2, joint distortion; and 3, joint ankylosis. Each paw was graded, resulting in a possible maximal score of 12 for each mouse. All mice used in these experiments were female DBA/1 lac J mice, 8–10 wk of age, obtained from The Jackson Laboratory (Bar Harbor, ME), housed and treated at the University of Pittsburgh central animal facility following National Institutes of Health guidelines.
Cell culture
Draining lymph nodes were removed and single cell suspensions were prepared in RPMI 1640 supplemented with heat-inactivated 10% FBS and antibiotics. Lymphocytes were cultured for 72 h, 5 × 106 cells/ml, and supernatants were collected, centrifuged, and stored at −80°C until assayed for viral IL-10 by ELISA, as described below. Paws were cut off at the hairline, stripped of hair and skin, and minced finely with sterile forceps and scalpel. Tissues were pushed through a sieve, and cellular exudates were cultured as above for 72 h. Supernatants were assayed for the expression of viral IL-10 by ELISA.
Adenoviral constructs
The adenoviral vectors used in these studies carry deletions in the E1 and E3 regions that allow infection, but block the ability of the virus to replicate in nonpermissive cells. The cDNAs encoding viral IL-10, the β-galactosidase (LacZ), and luciferase (Luc) are driven by the human CMV early promoter. Infectious adenovirus was produced in the human embryonic kidney 293 cell line (American Type Culture Collection, Bethesda, MD). The adenovirus encoding the firefly luciferase gene was constructed and generously provided by Dr. Jay Kolls (Louisiana State University, New Orleans, LA). Viral titers were determined by OD at 260 nm, in which 1 U = 1012 viral particles.
In situ detection of β-galactosidase transgene expression
Paws of mice injected with Ad-LacZ were split longitudinally and fixed in 10% buffered Formalin overnight at 4°C. After washing in PBS, tissues were reacted overnight at 37°C with 5-bromo-4-chloro-3-indolyl-β-galactopyranoside solution (X-gal; Sigma, St. Louis, MO). Some tissues were then decalcified in 20% EDTA, embedded in paraffin, and sectioned at 5–8 μm for histologic analysis.
Luciferase assay
Tissues were dissected and stored at −80°C. Approximately 0.7 g of each tissue was mixed with 2 ml 0.25 M Tris-HCl (pH 7.5), and the mixture was homogenized by hand with a tightly fitting dounce homogenizer. The homogenate was collected, put through three freeze-thaw cycles, and centrifuged 15 min at low speed in a table-top clinical centrifuge. Luciferase activity in 100 μl of supernatant was measured in a luminometer, as directed, using a luciferase assay system according to the manufacturer (Promega, Madison, WI).
Quantitation of viral IL-10 by ELISA
Viral IL-10 concentrations in serum and tissue culture supernatants were quantitated using a sandwich ELISA with specific mAbs purchased from PharMingen (San Diego, CA). These Abs do not cross-react to murine IL-10. In brief, 96-well microtiter plates were coated with primary anti-human/anti-viral IL-10 mAb overnight and washed with PBS/0.05% Tween 20, and nonspecific protein-binding sites were blocked with 20% FBS in PBS. Plates were washed, and samples and standards were incubated in duplicate for 4 h at room temperature. Plates were washed to remove unbound proteins, and a biotin-conjugated secondary mAb specific to viral IL-10 was added. After a 1-h incubation at room temperature, plates were washed and incubated at room temperature with an avidin-alkaline phosphatase conjugate. After washing, enzyme activity was determined with a colorimetric assay by adding p-nitrophenylphosphate substrate (Sigma) and measurement of the OD at 405 nm in a microplate reader.
Anti-collagen Ab ELISA
Microtiter plates were coated overnight with 1 μg/well native bovine or murine collagen type II (M. Griffiths, Salt Lake City, UT) and blocked with 10% nonfat milk in PBS. Plates were incubated with serial dilutions of serum obtained from treatment and control mice in these studies, or anti-CII-reactive Abs purified from CII-immunized mice. Plates were washed and incubated with goat anti-mouse IgG, or with anti-mouse IgG1, IgG2a, or IgG2b conjugated to alkaline phosphatase to determine total Ig and Ig subclass anti-CII levels. The amount of bound Ab was estimated after incubation with substrate, and measurement of enzyme activity was quantitated at 405 nm.
Statistical analysis
Ab levels were analyzed by group means with the Student’s t test. Incidence of arthritis was analyzed by χ2 test, disease onset, and the significance of clinical severity by the Mann-Whitney U-test.
Results
Detection of marker gene and vIL-10 expression in tissues after periarticular or systemic injection of adenovirus
The transduction of murine tissues by adenoviral vectors was confirmed by detecting the expression of marker gene products (β-galactosidase or luciferase) or viral IL-10 following local injection of 1 × 109 adenoviral particles into the joints of each paw in naive mice. Dissection and histologic analysis of paws injected with an adenovirus encoding the bacterial β-galactosidase gene (LacZ) revealed a diffuse distribution of the expressed gene product, detected by blue staining, in and around the articulating joints and other connective tissues of the paw (data not shown). For this reason, the site of local injection is described in this communication as periarticular.
A time-course study determined that expression of the marker gene luciferase could be detected for at least 4 wk in naive joint tissues and lymph nodes draining the paws (popliteal, maxillary) following periarticular injection of Ad-Luc into each footpad (Table I⇓). Luciferase activity was at or below background in other tissues, organs, and lymph nodes not draining the injected paws (<250 U). In contrast, luciferase was detectable in lung and liver for 3 days and in spleen for 5 days following a single systemic (i.v. or i.p., respectively) injection of equivalent concentrations of Ad-Luc (Table II⇓).
Luciferase activity in organs of mice injected with Ad-Luc periarticularlya
Luciferase activity in organs of mice injected with Ad-Luc systemically (i.v. or i.p.)a
In agreement with luciferase expression, viral IL-10 protein could be detected in conditioned media of lymph node cells obtained from injected paws, up to 3 wk after periarticular injection of high doses of Ad-vIL-10 (1 × 108 to 1 × 109 particles/paw). Measurable quantities of viral IL-10 protein could also be detected in culture medium of cells obtained directly from paws injected with Ad-vIL-10 at all concentrations tested (1 × 107 to 1 × 109 particles/paw) (Fig. 1⇓). Viral IL-10 protein could not be detected in cell cultures of nondraining lymph nodes, spleen, liver, or lung.
Production of viral IL-10 by paws and lymph node cells following in vivo injection of virus. Groups of three mice were injected periarticularly with Ad-vIL-10 at different concentrations (1 × 107 to 1 × 109 particles/paw). Three weeks after injection of Ad-vIL-10 virus, lymph node cells draining the paw were cultured as single cell suspensions (5 × 106 cells/ml) for 72 h. Paws injected with Ad-vIL-10 were cut off at the hairline, skin was removed, and remaining tissues were finely minced. Cell suspensions obtained from paw exudates were washed once in PBS and cultured in media for 72 h. Supernatants of lymph node and paw exudate cultures were assayed for secreted viral IL-10 by ELISA, as described in Materials and Methods.
No viral IL-10 protein could be detected in the serum of mice injected periarticularly with Ad-vIL-10 at any time point. Conversely, viral IL-10 could be detected for 72 h in the serum of mice following either i.v. or i.p. injection of equivalent concentrations of the Ad-vIL-10 vector, but was undetectable thereafter (Table III⇓). However, conspicuous inflammation was detected in lung and liver tissues following systemic administration of greater than 1 × 109 particles (data not shown). In preliminary studies, inflammation was also seen in naive paws after periarticular injections, most likely due to an immune response to the adenoviral vector. It was interesting, however, that high concentrations of Ad-vIL-10 caused less paw inflammation than equivalent concentrations of either Ad-Luc or Ad-LacZ, although the differences in paw swelling did not reach statistical significance (data not shown). Nonetheless, experiments using concentrations of adenoviral vectors higher than 1 × 109 adenoviral particles/paw or systemic concentrations greater than 4 × 109 particles were not pursued.
Viral IL-10 protein expression in sera after injection of Ad-vIL-10a
Effect of Ad-vIL-10 on the development of CIA
Experiments were designed to investigate the influence of viral IL-10 expression on the development of CIA. Twenty-eight days after immunization with CII, groups of mice were injected systemically (i.p. or i.v.) with 1 × 109 particles Ad-vIL-10. Additional groups of mice received periarticular injections consisting of 1 × 107 to 1 × 109 particles of Ad-vIL-10 into each individual paw. Control mice received equal concentrations of Ad-Luc or an equal volume of PBS systemically or periarticularly into each paw. Mice receiving Ad-vIL-10 systemically, whether the route of injection was i.v. or i.p., failed to gain any beneficial effect from viral IL-10 transfer and expression (Table IV⇓). However, gene transfer and expression of 1 × 109 particles of Ad-vIL-10 to each paw completely protected mice from developing CIA for the duration of the study (10 wk postimmunization with CII) (Fig. 2⇓). In contrast, 75% (p < 0.01) of mice injected with Ad-Luc and 87.5% (p < 0.001) of mice injected with PBS alone developed arthritis. Suppression of CIA by periarticular delivery of Ad-vIL-10 was dose dependent. Even though some mice in both treatment groups receiving lower than optimal concentrations of Ad-vIL-10 (<1 × 109 particles/paw) eventually developed arthritis, incidence was significantly less in mice receiving 1 × 108 particles/paw compared with mice in either of the control groups (p < 0.05 versus Ad-Luc- or PBS-treated mice). In addition, treatment with both 1 × 108 and 1 × 107 particles/paw significantly delayed the onset of arthritis (p < 0.02 versus Ad-Luc-treated group). However, the administration of Ad-vIL-10 did not significantly reduce the severity or mean joint count in those mice that developed arthritis compared with arthritic mice in either control groups (data not shown).
Effect of Ad-vIL-10, injected periarticularly in all four paws, on the development of CIA. Groups of eight mice were immunized with 100 μg bovine collagen type II on day 0. On day 28, treatment groups received separate periarticular injections of either Ad-vIL-10 (1 × 107 to 1 × 109 particles), Ad-Luc (1 × 109 particles), or equal volumes of PBS into each paw. Mice were assessed three times per week for the onset of arthritis. Values are the percentage of mice in each group with a macroscopic score of >1.
Effect of systemic gene delivery on the incidence and severity of collagen-induced arthritisa
Importantly, local periarticular gene transfer was able to protect uninjected paws from developing CIA (Fig. 3⇓). Mice were injected periarticularly on day 28 with 1 × 109 particles of Ad-vIL-10 into only two paws, while the remaining two paws received equal volumes of PBS. Incidence of arthritis was confined to just 28.6% (two of seven mice) of mice injected into two paws with Ad-vIL-10, whereas 100% (seven of seven mice) of mice injected with the control virus (Ad-Luc; p < 0.01) and 85.7% of mice injected with PBS only (p < 0.05) developed disease. Protection from the development of arthritis was long lasting, up to 70 days (10 wk) after primary immunization with CII. In the two mice treated with Ad-vIL-10 that did develop disease, arthritis only developed in paws receiving injections of PBS (4/14), whereas all paws receiving direct periarticular injections of Ad-vIL-10 were protected (0/14) (p < 0.05). Severity of arthritis, however, was not affected by treatment with Ad-vIL-10. These results were confirmed in a separate study by injecting only one paw per mouse periarticularly with 1 × 109 particles of either Ad-vIL-10 or Ad-Luc. Again, all mice injected with Ad-Luc developed arthritis (10 of 10); however, significantly fewer mice became arthritic in the Ad-vIL-10-treated group (3 of 10, p < 0.01), although one mouse developed arthritis in the Ad-vIL-10-injected paw on this occasion (Fig. 3⇓). Although there was no significant difference between mice injected periarticularly in two paws versus injection of one paw, these data may reflect the variation between experiments. Once again, severity of arthritis in the Ad-vIL-10 group that developed disease was not reduced compared with the control group (data not shown). Nonetheless, these findings indicated a dramatic and unexpected protective, systemic effect of local vIL-10 gene therapy.
Ability of Ad-vIL-10 injected periarticularly in two or one paws to protect uninjected paws from developing CIA. Groups of 7 or 10 mice immunized with 100 μg bovine collagen type II on day 0. On day 28, treatment groups received periarticular injections of either Ad-vIL-10 (1 × 109 particles/paw) or Ad-Luc (1 × 109 particles/paw). Control mice received equal volumes of PBS. Mice were injected with virus either in two paws, selected at random (7 mice per group), or in just one rear paw (10 mice per group). Values are the percentage of mice in each group with a macroscopic score of >1.
Expression of vIL-10 in cells obtained from injected and uninjected paws
Cells were recovered from the paws of arthritic mice injected with higher, antiarthritic (1 × 109 particles/paw) or lower, ineffective (1 × 107 particles/paw) concentrations of Ad-vIL-10 virus. Viral IL-10 protein could be detected by ELISA in culture media conditioned by cells recovered from the joint exudates of nonarthritic, Ad-vIL-10 (1 × 109 particles/paw)-treated paws for up to 3 wk after periarticular injections of virus (weeks 4–7 after primary immunization with CII) (Fig. 4⇓). At later time points, however, viral IL-10 protein could not be detected in injected paws (data not shown), even though mice remained free from the development of CIA. These data implied that expression of Ad-vIL-10 at higher concentrations imparted a long-lasting protective effect from the development of CIA, which did not require the continued expression of the viral IL-10 protein. In contrast, viral IL-10 could not be detected in culture supernatants of paws receiving suboptimal doses of Ad-vIL-10 (1 × 107 particles/paw) that had developed arthritis during this same time period. These results implied that the delay in onset of arthritis observed in groups of mice receiving lower doses of Ad-vIL-10 correlated with the level or persistence of local viral IL-10 expression.
Expression of vIL-10 in cells obtained from injected and noninjected paws. Cells were obtained from the paws of arthritic mice and nonarthritic mice treated periarticularly 3 wk prior in one paw with Ad-vIL-10 (1 × 107 or 1 × 109 particles/paw). Cells were cultured for 72 h, and conditioned medium was assayed for the presence of viral IL-10 protein by ELISA, as described in Materials and Methods. Values represent the mean viral IL-10 protein in ng/ml of triplicates of culture supernatant.
Effect of Ad-vIL-10 treatment on anti-collagen type II Abs
Serum Ab titers in mice injected periarticularly with Ad-vIL-10 were compared with Ad-Luc or PBS only-treated controls. Viral IL-10 treatment did not reduce the total anti-bovine CII Ab titers (Fig. 5⇓) and did not induce any change in the ratios of isotype-specific Igs to collagen type II (IgG1, IgG2a, IgG2b, or IgG3) (data not shown). However, serum Abs to mouse type II collagen were reduced significantly in mice treated locally with Ad-vIL-10 (Fig. 6⇓).
Total Ig to bovine collagen type II. Groups of 10 mice were immunized with 100 μg bovine collagen type II on day 0. On day 42, mice sera were obtained and assayed for anti-bovine CII Abs by ELISA. Values represent OD at 405 nm of serial dilutions of pooled sera from each group.
Total Ig to autologous mouse collagen type II. Groups of 10 mice were immunized with 100 μg bovine collagen type II on day 0. On day 42, mice sera were obtained and assayed for anti-murine CII Abs by ELISA. Values represent OD at 405 nm of serial dilutions of pooled sera from each group.
Discussion
These studies demonstrate that direct injection of adenovirus encoding viral IL-10 in the proximity of the ankle joints prevents the onset of collagen type II-induced arthritis in mice. Importantly, we show for the first time that gene transfer and expression of optimal concentrations of viral IL-10 (1 × 109 particles/paw) in one or more joints confer protection from the subsequent development of arthritis in nontreated joints. Furthermore, it is striking that this effect persists for up to 70 days after immunization with CII, even though viral IL-10 protein was detectable in cultures of transduced cells for only 3 wk following injection with the adenoviral vector. Protection from arthritis by local viral IL-10 gene transfer and expression was dose dependent. Mice injected with suboptimal Ad-vIL-10 (1 × 107 to 1 × 108 particles/paw) developed disease; however, the incidence of arthritis was significantly diminished and time of onset was delayed compared with mice receiving control adenoviral vectors or PBS alone. Therefore, our data are in strong agreement with the known antiinflammatory and immunosuppressive properties of IL-10. In contrast, systemic administration (i.v. or i.p.) of 1 × 109 particles of Ad-vIL-10 could only achieve short-term expression (1–3 days) of viral IL-10 protein and did not affect the incidence, time of onset, or severity of CIA compared with control mice. However, a recent study by Apparailly et al. 31 has demonstrated that systemic administration of 1 × 109 plaque-forming units of an adenoviral vector expressing viral IL-10 suppressed development and reduced the severity of CIA, and that viral IL-10 protein could be detected for up to 7 days after injection. The reasons for the discrepancy between our results and Apparailly et al. may be as follows. First, the dose of adenoviral vIL-10 vector used by their studies most likely represents a 100-fold greater viral dose than in our studies (in which 1 plaque-forming unit is approximately 1 × 102 particles of virus). This may explain why serum levels of viral IL-10 were still present 7 days after injection in their studies, whereas we could only detect viral IL-10 in the serum for 3 days following i.v. injection. In our experiments, such high viral loads produced hepatoxicity. Second, we might also speculate that a large, systemic dose of adenovirus could in fact divert Ag-nonspecific immune effector cells, thereby reducing the actual incidence and severity of arthritis in these mice.
Production of IgG2a anti-collagen Abs to heterologous type II collagen has been associated with the development of arthritis; however, the development of both humoral and cellular responses to autologous, murine collagen type II is thought to be crucial to the development of CIA in susceptible strains 39 . Gene transfer of viral IL-10 periarticularly or systemically did not affect the total Ig Ab response to heterologous bovine CII, nor was there any shift in the anti-CII isotype titers in the sera of treated mice. Importantly, in our studies, Ab titers to autologous type II collagen were significantly reduced. Taken together, these data suggest that viral IL-10 may be a potent immunosuppressive and antiarthritic agent without the potential to increase autoantibody production, and therefore may be an attractive therapeutic agent in the treatment of RA.
Gene transfer techniques have been successful in delivering antiarthritic proteins to synovium and ameliorating the inflammation and joint destruction typically observed in various animal models of arthritis 22, 23, 28, 30 . In several other studies, transgenes were delivered by i.v. or i.p. injection of viral vectors encoding antiinflammatory cytokines or by adoptive transfer of cells modified to secrete therapeutic proteins 24, 25, 26, 27, 28 . In agreement with our results in mice, systemic gene transfer with viral vectors often resulted in high but short-term serum levels of expressed protein, an approach that is not likely to produce significant clinical benefits in the treatment of a chronic disease like RA. In studies in which exogenous IL-10 had a suppressive effect on the development and severity of CIA in mice and rats, IL-10 was delivered continuously by either miniosmotic pumps 50 or i.p. injections, once or twice daily 49, 50, 51 . In contrast, a one-time intraarticular injection of cells modified to secrete IL-1Ra reduced the symptoms of experimental arthritis in mice, rats, and rabbits and demonstrated the potential of gene delivery via an ex vivo approach in the treatment of chronic disease 23, 28, 39 . Indeed, clinical trials have recently been initiated using autologous synovial cells transduced ex vivo with a retrovirus encoding human rIL-1Ra in RA 55 . However, ex vivo strategies are time consuming, impractical, and expensive, as they require in vitro infection, expansion of transduced cells in culture, and extensive safety testing before reimplantation.
An alternative approach is the direct transduction of therapeutic genes to tissues within the affected joint. This strategy has been used successfully to deliver genes encoding marker proteins and antiinflammatory agents to synovial tissue in the rabbit, by direct intraarticular injection of knee joints using adenoviral 56 or retroviral vectors 57 . Of course one concern is the potential for viral vectors to cause inflammation at the site of injection. In particular, adenoviral vectors are highly immunogenic and have been reported to cause inflammation and to accelerate the onset and increase the severity of arthritis when given in high dose 24 . In our own studies, adenoviral vectors encoding marker genes (LacZ or luciferase) and possessing titers of greater than 1 × 109 particles caused transient inflammation in and around transduced tissues following direct injection into the paw, limiting the potential effectiveness of in vivo gene therapy. A recent report, however, demonstrated the ability of an adenoviral vector encoding viral IL-10 to down-regulate the host immune response to adenoviral proteins while maintaining its potent antiinflammatory and immunosuppressive activity 54 . In our studies, high concentrations of Ad-vIL-10 in vivo consistently caused less inflammation than equivalent concentrations of control adenoviral vectors, and did not aggravate the onset of CIA. It is unknown whether specific responses to adenoviral Ags were reduced in these animals compared with controls.
More importantly, our results show for the first time that direct, in vivo periarticular gene therapy with Ad-vIL-10 to selected joints can prevent peripheral, nontreated joints from developing CIA. It thus may prove possible to combine the safety benefits of local gene delivery with the polyarticular effects of systemic gene delivery while treating only one joint. Recently, periarticular administration of an adenoviral vector encoding the FasL gene induced apoptosis of cell types involved in the pathogenesis of CIA and reduced the severity of established arthritis, and notably, was able to protect nontreated joints 30 . Our data imply that migration and peripheral infection of distant tissues by adenoviral vectors cannot fully account for the protective effect of vIL-10 gene transfer. It may be alternatively postulated that cells infected within the articular compartment migrate to peripheral joints and suppress proinflammatory events leading to CIA. Experimental evidence arguing against this hypothesis includes a report demonstrating that ex vivo delivery of fibroblasts expressing human IL-1Ra to knee joints of mice prevented CIA in the draining, ipsilateral paw; however, remaining uninjected paws developed arthritis equal to control animals 28 . Cell types other than fibroblasts may possess greater migratory abilities than synovial fibroblasts, and in conjunction with the potent immunomodulatory properties of vIL-10, may account for the observed protection of peripheral joints in our studies.
It is clearly important to understand the mechanisms underlying the protective effect of adenoviral-mediated vIL-10 therapy in CIA. IL-10 has been reported to diminish the MHC II expression in monocytes and inhibits the up-regulation of costimulatory molecules on macrophages and DCs, cell types present in the periarticular tissue of the mouse paw 58, 59, 60 . We suggest that down-regulation of MHC class II molecules on the cell surface may impair the ability to present antigenic self peptides, in this case murine type II collagen, to T cells. Alternatively, periarticular expression of IL-10 may impair APC function by down-regulating costimulatory molecule expression. Recent studies have shown that migration of DCs to lymphoid tissues causes maturation-induced changes, including up-regulation of accessory and costimulatory molecules 60 . Immature DCs incubated with IL-10 fail to express costimulatory molecules crucial to the activation of Th1 cells. Instead, it appears that these DCs induce Ag-specific anergy in Th1 cells, preventing their proliferation in response to Ag and reducing the secretion of IFN-γ 61 . In addition, immature DCs treated in vitro with IL-10 can lead to the induction of a Th2, rather than a Th1 response in vivo 62 . We suggest that IL-10 expression within the periarticular tissues of CII-immunized mice may also interfere with the maturation of local, immature DCs 63 , and thus affect the priming of autoreactive T cells.
Whether IL-10 down-regulates MHC class II expression periarticularly, promotes a Th2 rather than a Th1 response, and/or induces Th1 peripheral tolerance toward self Ags in CIA remains to be investigated. Regardless of the mechanisms underlying this protective effect, the data presented in this study and elsewhere clearly encourage the continuing development of gene-based treatments for arthritis and, in particular, the further investigation of viral IL-10 as a potential alternative to IL-10-based immunosuppressive approaches to inflammatory disease.
Acknowledgments
We thank Ms. Lori Miller and Dr. Christian Lattermann for their excellent technical assistance, and Dr. Paul H. Wooley for very helpful suggestions and advice involving statistical analysis and content.
Footnotes
↵1 This work was supported in part by National Institutes of Health Grant PO1 DK44935 and a grant from the Ferguson Foundation and 5 RO1 AR44526-02 NIH.
↵2 Address correspondence and reprint requests to Dr. Christopher H. Evans, Department of Orthopaedic Surgery, University of Pittsburgh, 200 Lothrop Street, Room C313, Presbyterian Hospital, Pittsburgh, PA 15213. E-mail address: cevans{at}vms.cis.pitt.edu
↵3 Abbreviations used in this paper: RA, rheumatoid arthritis; Ad-vIL-10, adenoviral vector carrying interleukin-10; CIA, collagen-induced arthritis; CII, type II collagen; DC, dendritic cell; IL-1Ra, interleukin-1 receptor antagonist; vIL-10, viral IL-10.
- Received July 24, 1998.
- Accepted December 10, 1998.
- Copyright © 1999 by The American Association of Immunologists