Inflammation and Immune Response of Intra-Articular Serotype 2 Adeno-Associated Virus or Adenovirus Vectors in a Large Animal Model

Intra-articular gene therapy has potential for the treatment of osteoarthritis and rheumatoid arthritis. To quantify in vitro relative gene transduction, equine chondrocytes and synovial cells were treated with adenovirus vectors (Ad), serotype 2 adeno-associated virus vectors (rAAV2), or self-complementary (sc) AAV2 vectors carrying green fluorescent protein (GFP). Using 6 horses, bilateral metacarpophalangeal joints were injected with Ad, rAAV2, or scAAV2 vectors carrying GFP genes to assess the in vivo joint inflammation and neutralizing antibody (NAb) titer in serum and joint fluid. In vitro, the greater transduction efficiency and sustained gene expression were achieved by scAAV2 compared to rAAV2 in equine chondrocytes and synovial cells. In vivo, AAV2 demonstrated less joint inflammation than Ad, but similar NAb titer. The scAAV2 vectors can induce superior gene transduction than rAAV2 in articular cells, and both rAAV2 and scAAV2 vectors were showed to be safer for intra-articular use than Ad vectors.


Introduction
In elderly people, joint disorders including osteoarthritis and rheumatoid arthritis remain major causes of mobility loss [1]. Gene therapy has a great potential for the treatment of these conditions by using intra-articular gene delivery, such as previously described for interleukin-1 receptor antagonist (IL-1Ra) [2,3], transforming growth factor-β1 [4], or tumour necrosis factor alpha antagonist protein [5]. Direct intraarticular administration of gene delivery vectors is an attractive and effective strategy to transduce articular cartilage and synovium with therapeutic genes, because joints are discrete and accessible cavities that can be easily injected [6].
A choice of viral vectors has been used to introduce therapeutic genes to intra-and periarticular tissues including adenovirus (Ad) [7,8] and serotype 2 adeno-associated virus (AAV2) [9][10][11], and Ad vectors have been shown to induce rapid and successful transgene expression, but intra-articular usage has been restricted due to robust inflammatory and immune reactions, as well as limited by transient transgene expression [6][7][8]. In contrast, recombinant AAV2 (rAAV2) has gained much attention as a safer vector for gene delivery to joints, because it can induce sustained transgene expression in the absence of significant inflammation [6,[9][10][11]. In addition, rAAV2-mediated joint therapy may be advantageous by its greater penetration into articular cartilage, likely due to smaller particle size [12]. Moreover, AAV2 vectors have been modified to package self-complementary DNA, capable of bypassing the rate-limiting step of secondstrand synthesis in order to accelerate and improve the transduction efficiency [12,13]. Such self-complementary AAV2 (scAAV2) vectors have been shown to effectively transduce chondrocytes and synovial cells [14] and cartilage explants [12] in vitro, and induce superior transgene expression in cartilage and synovium tissues after injection of 5 × 10 11 particles in vivo in rodent models [3]. Taken together, intraarticular administration of scAAV2 vectors has potential for future clinical application, but the inflammation and 2 Arthritis immune reactions, cell tropism, and application to larger joints as in people have not been directly compared among rAAV2, scAAV2, or Ad vectors in vivo.
The objectives of our study were to assess the relative gene transduction and duration of expression among Ad, rAAV2, and scAAV2 vectors carrying green fluorescent protein (GFP) in chondrocytes and synovial cells in vitro and compare the inflammation and immune response by the intra-articular administration of Ad, rAAV2, and scAAV2 vectors in vivo in equine joints of similar size to human joints. We hypothesized that scAAV2 vectors would show accelerated and greater gene transduction in vitro compared to rAAV2 and induce less inflammation and immune response in vivo compared to Ad vectors.

Viral Vector Production.
Recombinant E-1 deleted human serotype 5 adenovirus preparations, single-stranded human serotype 2 AAV2, and scAAV2 containing GFP under the control of the cytomegalovirus (CMV) promoter were generated and purified by cesium chloride density gradient method [12,15]. Particle count of Ad vectors was determined by optical density at 260 nm, and particle titers of rAAV2 and scAAV2 vectors were determined by DNase-resistant particle (DRP) values using real-time PCR assay as described previously [16].

In Vivo Experimental Design and Vector Administration.
The in vivo experimental design is summarized in Figure 1. All procedures were approved by the Institutional Laboratory Animal Care and Use Committee at The Ohio State University. By using 4 healthy adult horses (3 Thoroughbreds and 1 Quarter horses) aged 5-12 years (median age: 7) and weight 418-536 kg (mean weight: 429 kg), 8 metacarpophalangeal joints (2 joints/horse) were treated by the intra-articular administration of GBSS, Ad-GFP (5 × 10 11 particles/joint), rAAV2-GFP (5 × 10 11 DRP/joint), scAAV2-GFP (5 × 10 11 DRP/joint), or scGFP (1 × 10 13 DRP/joint). In joints assigned randomly to be injected with Ad, GBSS was injected at day 0 and the Ad-GFP at day 14 after a normalization period, because Ad vectors were expected to induce rapid transgene expression. The GBSS-induced inflammation was assessed between day 0 and 14 prior to the Adinjection.
Intra-articular administration of viral vectors was performed while horses were standing and sedated with xylazine hydrochloride (i.v., 1 mg/kg). After the aseptic preparation of skin over the lateral aspect of the metacarpo/tarsophalangeal joints, the limb was held off the ground at approximately 120 degree flexion, and a 20 G 1.5 inch needle was inserted into the joint space between the proximal sesamoid bone and lateral distal condyle of the third metacarpal/tarsal bone by passing through the lateral collateral sesamoidean ligament. The intra-articular administrations of assigned treatments were performed only after the 2 mL joint fluid was collected to ensure the needle placement.

In Vivo Inflammatory and Immune Response.
Physical examinations were performed weekly after the joint injection by evaluating circumference of the injected joints and painfree range of joint motion. Circumference of the injected joint was recorded as the mean of 3 measurements obtained over the injection site by use of a cloth measuring tape [18]. Range of pain-free motion was recorded as the mean of 3 goniometer measurements of the flexed joint immediately Arthritis 3 before elicitation of an aversion response such as lifting of the head or movement of the limb forward/backward [18]. Lameness grades were assigned to each horse by an experienced equine clinician (ALB) using the American Association of Equine Practitioners' lameness with 0.5 increments [19], whereas a joint on the nonlame limb was assigned as 0 lameness grade.
Joint fluid samples were collected weekly by using the same technique described above and analyzed for cell count and total protein concentration. The interleukin 1 beta (IL-1β) and IL-1Ra protein concentration in the joint fluid was quantified by ELISA (Equine IL-1Ra DuoSet, R&D Systems, Minneapolis, MN). Serum samples were collected weekly, and neutralizing antibody (NAb) titer for Ad and AAV2-vectors were measured for serum and joint fluid samples from all joints. Titers were determined by analyzing the ability of serum antibody to inhibit Ad-GFP (1 × 10 4 particles/cell) or rAAV2/scAAV2-GFP infection (1 × 10 6 DRP/cell) on Human carcinoma cells (HeLa cells: 5 × 10 3 cells/well in 96-well plates) at 48 hours [15]. By applying twofold dilution series of sample serum, the NAb titer was calculated as the highest serum dilution inhibiting Ad or rAAV2 or scAAV2-GFP transduction by >50%.

Statistical Analysis.
Repeated-measure analysis of variance (ANOVA) (SAS Institute Inc., Cary, NC) was used to evaluate the effects of Ad or rAAV2 or scAAV2-mediated GFP gene delivery with the posttest multiple comparisons between the treatment groups at each time point using Proc Mixed statistical models for continuous outcomes and Genmod statistical models for categorical outcomes. Significance level was set at P < 0.05 for all analyses.

In Vitro Gene Transduction. Self-complementary AAV2
vectors produced greater and more rapid GFP gene transduction (Figures 2(a) and 2(b)) compared to rAAV2 vectors in equine chondrocytes and synovial cells. Also, scAAV2 vectors showed more sustained GFP gene transduction (Figures 2(a) and 2(b)) compared to Ad vectors when they showed comparable % transduction in equine chondrocytes and synovial cells, where the onset of GFP gene transduction (Figures 2(a) and 2(b)) was equally rapid to Ad vectors.

In Vivo Inflammatory and Immune
Response. The Ad vectors induced greater articular inflammation compared to rAAV2 and scAAV2 vectors with significantly greater joint fluid cell counts (P < 0.03: Figure 3(b)), joint circumference (P < 0.04: Figure 3(c)), and reduced range of joint motion (P < 0.02: Figure 3(d)), joint fluid IL-1β concentration (P < 0.001: Figure 3(e)), and lameness grade (P < 0.04: Figure 3(f)), although all parameters were recovered to normal values within 4-5 weeks. For the rAAV2-or scAAV2injected joints, inflammatory parameters were increased, but not significantly different than GBSS-injected joints, except a significantly greater joint fluid protein concentration at day 2 (Figure 3(a)). None of inflammatory parameters were significantly different between GBSS and rAAV2 or scAAV2 at low dose, and scAAV2 at high dose. None of joint fluid samples from Ad/AAV2-injected joints showed detectable concentration of IL-1Ra protein.
The NAb titers against Ad-or AAV2 were significantly greater in the Ad-or AAV2-injected joint fluid compared to the serum or uninjected contralateral joint fluid (P < 0.03: Figures 4(a) and 4(b)), and injected synovial fluid titers remained high (>100) to the end of study period. The NAb titers for both serum and uninjected contralateral joint fluid against Ad or AAV2 peaked earlier at 2 weeks after intraarticular administration compared to injected synovial fluid (Figures 4(a) and 4(b)).

Discussion
Self-complementary AAV2 vectors showed promising and superior efficacy in vitro. Equine chondrocytes and synovial cells showed greater, more rapid, and sustained gene transduction by scAAV2 vectors compared to rAAV2 (Figures 2(a) and 2(b)). As the result, even 100 times lower dosage of scAAV2-GFP vectors (1 × 10 4 DRP/cell) was able to induce the approximately equal % transduction to rAAV2-GFP vectors (1 × 10 6 DRP/cell) in equine chondrocytes (Figure 2(b)). Bypassing rate-limiting second-strand DNA synthesis is expected to be largely beneficial to transfect the certain types of cells with low DNA replication; therefore, use of scAAV2 vector may be greatly advantageous for an intra-articular administration due to the low cell division rate of chondrocytes [14]. Because scAAV vectors do not require the second-strand DNA synthesis step, more rapid and superior gene expression of scAAV over rAAV vectors is expected in human chondrocytes and synovial cells as well. Our study also showed that, for the first time, there were no significant differences in the inflammation and immune response between the intra-articular rAAV2 and scAAV2 vectors (Figures 3(a) to 3(f) and Figure 4(b)). This is an expected result because the rAAV2 and scAAV2 vectors used in this study were the same serotype and contained the same capsid proteins, where the innate and adaptive immune responses are primarily initiated by recognition of the capsid proteins [20,21]. An advantage of scAAV2 vectors may be detected in humans that have NAb titer against wild-type AAV2 (80% of human population) in that lower vector dosage may be effective and incite less immune reaction. Previous studies showed that, while the preexisting immunity by host exposure of the wild-type AAV2 is a significant limiting factor for rAAV2 gene transfer, the humoral immunity to the AAV2 capsid could be prevented by lowering the AAV2 particles administered [22].
Our study showed that AAV2 vectors may be a safer vector of choice for intra-articular gene administration, as they induced significantly less inflammatory response in the joints than Ad vector (Figure 3). The inflammation shown in the Ad-injected joints was expected and was transient. The signs of swelling, pain, and increased joint fluid protein and cell counts were significantly greater in Ad vectors compared to rAAV2 or scAAV2 vectors. Importantly, all parameters were returned to normal values within 4-5 weeks (Figure 3). Our results support that the adverse effects associated with an intra-articular Ad vector injection is temporary and in this study was clinically acceptable. The inflammation upon repeat injection was not studied in our report, but may be greater than first injection based on an amnestic immune response to the Ad vector.
To the authors' knowledge, this is the first study to report that intra-articularly administered Ad and AAV2 vectors induced antibody production in both serum and joint fluid. Previously, an intra-articular AAV2 injection has shown to increase serum antibody in human clinical trial [23]. The NAb production in the joint fluid may inhibit the repetitive injections of the gene delivery vectors in the same joint. A previous study reported that, after intra-articular IL-1Ra gene transduction by scAAV2 vectors, the reinjection of the vectors could not generate detectable levels of IL-1Ra expression [3]. In this regard, modification of the vectors might be necessary to prevent the immune responses that might interfere with successful repeat gene therapy application, including the capsid epitope alterations to decrease immunogenicity of the vector [24], cross-packaging of genomes into different AAV2 capsid serotype [25], or the application of transient immunosuppression at the time of the second vector administration [26]. Interestingly, the NAb titers in the Ad-or AAV2-injected joints were greater than serum titer and remained high (>100 NAb titer) until the end of the study period, even after the serum Ad or AAV2 titers were faded (Figures 4(a) and 4(b)). This may indicate the persistence of vectors in the injected joints.
Another important finding in our study is that, for both Ad and AAV2 vectors, the NAb titers against Ad/AAV2 vectors were increased in the synovial fluid from not only the Ad/AAV2-injected joint but also the contralateral joint where Ad/AAV2 vectors were not administered (Figures 4(a) and 4(b)). The increment and peak of the NAb titer in the contralateral joints corresponded to the serum NAb titer suggesting an infiltration of Ad antibodies from systemic circulation into the contralateral joint cavity. This may imply that treating one joint with viral vector may limit an efficacy of repeated application of intra-articular gene therapy in both the previously treated joints and the distant joints, at least for 6-8 weeks after the initial joint injection. A similar phenomenon has been noticed with intraocular administration of AAV2 vectors resulting in systemic antibody production and blocked transgene expression upon readministration in the contralateral eye [27].
Our study supported the safety of intra-articular gene therapy using AAV vectors in human joints. In this study, the direct intra-articular administration of relatively high dosage of rAAV/scAAV vectors (1 × 10 13 DRP/joint) did not cause any detrimental inflammatory responses in equine metacarpo/tarsophalangeal joints with an average normal synovial fluid volume of 4.4 mL [28]. Osteoarthritis and rheumatoid arthritis commonly occur in human knee or elbow joints with an average normal synovial fluid volume of 4.5 mL [29]. In the literature, successful IL-1Ra or marker gene transductions of the articular cartilage and synovium were reported by the intra-articular administration of 4.7 × 10 11 particles of rAAV in the mice knee joints [9], 5 × 10 11 particles of scAAV in the rabbit knee joints [3], and 1.5 × 10 12 particles of rAAV in the rabbit knee joints [11]. No significant inflammatory effects were evident in those reports, whereas the capacities of these rodent joints (0.5 mL) were estimated