Modulation of collagen-induced arthritis by adenovirus-mediated intra-articular expression of modified collagen type II

Native CII was solubilized from fetal calf articular cartilage by limited pepsin-digestion and purified as described earlier [9]. The purified collagen was dissolved in cold 0.01 M acetic acid at 4 mg/ml and stored frozen at -70°C until used.

CII was solubilized in 0.01 M acetic acid at a concentration of 4 mg/ml and emulsified with an equal volume of complete Freund's adjuvant (CFA) containing 4 mg/ml of Mycobacterium tuberculosis strain H37Ra (Difco Microbiology Products, Becton Dickinson, NJ, USA) [10]. Each mouse received 100 μg of CII emulsified in CFA intradermally at the base of the tail.

Recombinant adenovirus carrying cDNA for rCB11-A9 was generated using a BD Adeno-X Expression System (BD Biosciences Clontech (San Jose, California, USA)), which incorporates the rCB11-A9 expression cassette into a replication-incompetent (ΔE1/ΔE3) human adenoviral type 5 (Ad5) genome. All work was conducted in accordance with approved Institutional Biosafety Committee (IBC) protocols. In brief, an 834 bp of full-length murine CB11 gene was PCR-amplified from murine spleen cDNA and cloned into the PCR2 vector (Invitrogen, Carlsbad, California, USA). We introduced three point mutations (I260A, A261B, and F263N) within the immunodominant T cell determinant of CB11 (CII_124-402) to generate an rCB11-A9 construct. The rCB11-A9 cDNA was then excised with BamHI/EcoR I and subcloned into the same sites of the pShuttle2 vector to construct an rCB11-A9 specific expression cassette. For in vivo bioimaging analysis, a cDNA encoding the luciferase gene was also subcloned into the pShuttle2 to establish the Adeno-X-luciferase expression cassette. To produce recombinant adenoviral DNA containing rCB11-A9 or luciferase, we excised the expression cassettes from recombinant pShuttle2 plasmid DNA by digesting with I-Ceu I and PI-Sce I and ligated the expression cassettes with prelinearized BD Adeno-X Viral DNA (I-Ceu I and PI-Sce I digested). Low passage HEK293 cells were transfected with the resultant recombinant adenoviral DNA using the calcium phosphate method [11]. The recombinant adenoviral particles were harvested by lysing transfected cells. The resultant AdenoX-rCB11-A9 is a replication-incompetent recombinant adenovirus. High titer viral stocks (about 10⁸ to 10⁹ plaque forming units (pfu)/ml) were obtained by amplifying recombinant adenovirus in HEK 293 cells. A construct (pShuttle2-lacZ) was included in the BD Adeno-X Expression System and recombinant AdenoX-lacZ was generated as described above and used as a control. The recombinant adenoviral titers were determined by BD Adeno-X Rapid Titer Kit [11, 12].

In some experiments, a baculoviral expression system was used to produce rCB11 (CII_124-402^bac) in insect cells essentially as described earlier [13]. The cDNA for both recombinant CB11 and CB11-A9 (rCB11 and rCB11-A9) were subcloned into a Gateway entry vector (Invitrogen, Carlsbad, California, USA) and validated. The resultant Gateway entry vectors containing either rCB11 or rCB11-A9 were ligated with BaculoDirect Linear DNA (Invitrogen, Carlsbad, California, USA) and transfected into Sf9 insect cells. Supernatants from lysed insect cells were collected and screened for expression by performing SDS-PAGE and western blot analysis. After validated, high titers of recombinant baculovirus were obtained by re-infecting Sf9 cells twice and supernatants collected from lysed cells. To express the recombinant proteins Hi5 cells was infected with high titer of baculovirus. Supernatants from cultured Hi5 cells were harvested by centrifugation and the recombinant proteins purified by gel filtration and cation exchange chromatography, and dialyzed in dilute acetic acid.

The hind ankle joints of DBA/1 mice were injected intra-articularly with 10 ul of adenoviral vector 1 × 10⁷ pfu of adenovirus, containing the DNA for either luciferase, rCB11-A9, or Lac-Z. In some experiments, selected mice were injected intraperitoneally with luciferin, and the expression of the transgene (luciferase) was detected by bioluminescent imaging using a liquid nitrogen cooled CCD camera (Photometric Chemipro, Roper Scientific, NJ, USA) mounted on a dark box one hour later. Images were acquired and analyzed using Metamorph software (Universal Imaging Co., Dowlington, PA, USA).

The incidence and severity of arthritis were determined by visually examining each forepaw and hindpaw and scoring them on a scale of 0 to 4 as described previously [10]. Scoring was conducted by two examiners, one of whom was unaware of the identity of the treatment groups. Each mouse was scored thrice weekly beginning three weeks post immunization and continuing for eight weeks. The incidence of arthritis (number of animals with one or more arthritic limbs) and mean severity score (sum of the severity scores/total number of animals in the group) was recorded at each time point.

In a prevention protocol, four groups of 10 DBA/1 mice each were administered intra-articularly in the ankles, either adenoX-rCB11-A9 or adenoX-LacZ. The mice were immunized with CII/CFA either three or seven days after the injection.

In a treatment protocol, groups of three DBA/1 mice were immunized with CII/CFA and at the time arthritis reached a severity score of two or greater, the mice were administered intra-articularly in the hind ankles either adenoX-rCB11-A9 or adenoX-LacZ.

Mice were bled at six weeks after immunization and sera were analyzed for antibodies reactive with native CII using a modification of an ELISA previously described [10]. Serial dilutions of a standard serum were added to each plate. From these values, a standard curve was derived by computer analysis using a four-parameter logistic curve. Results are reported as units of activity, derived by comparison of test sera with the curve derived from the standard serum which was arbitrarily defined as having 50 units of activity. Reactivity to CII was not detected in sera obtained from normal mice.

Groups of three DBA/1 mice were administered intra-articularly either adenoX-rCB11-A9 or adenoX-LacZ and the mice were immunized with CII/CFA three days after the injection. Draining lymph node cells were harvested 14 days after the immunization and cultured (5 × 10⁶ cells/ml) with 100 μg/ml of either the mouse collagen immunodominant peptide, Ova (negative control), or purified protein derivative (PPD) (positive control). Supernatants were collected 72 hours later and analyzed for the presence of multiple cytokines (IL-4, IL-5, IL-10, IL-2, interferon (IFN)γ, and IL-17 by a Bio-plex mouse cytokine assay (Bio-Rad, Hercules, CA, USA) according to the manufacturer's protocol. Values are expressed as picograms per ml and represent the mean values for each group.

The incidence of arthritis in various groups of mice was compared using Fisher's Exact Test. Mean severity scores and antibody levels were compared using Student's t test.

Using a replication-defective, recombinant adenovirus, we incorporated the cDNA for lac-Z, and assessed the transfection efficiency of the recombinant adenovirus delivered into arthritic ankles of DBA/1 mice previously immunized with CII/CFA. Each hind ankle was injected intra-articularly with 10⁷ pfu of the adenoviral particles. Forty-eight hours later, the animals were sacrificed and histology was performed on the involved joints. As shown in Figure 1, staining with β-galactosidase clearly demonstrated that the adenovirus-expressed gene product was present in synovial cells, lining the surface of the synovium near the cartilage surface (Figure 1). Although most of the transfected cells are fibroblast-like synoviocytes, a smaller number of monocyte-like synovial cells were also transfected. These data confirm that arthritic synovial cells can be readily transfected with adenoviral constructs and that an adenovirus carrying gene can be efficiently expressed.

To develop a unique collagen-based therapy, we built upon our previous work demonstrating that a synthetic peptide of CII, which contained three amino acid substitutions (A9), could effectively suppress arthritis. We used PCR to introduce three point mutations within the CB11 portion of recombinant type II collagen (CII_{124-402,260A, 261B, 263D}), (rCB11-A9) so that the resulting molecule contained the A9 sequence at the exact site of the wild-type sequence. To test for safety, we developed a baculovirus construct and expressed the rCB11-A9 protein in drosophila cells, because insect cells express a modest activation of lysine hydroxylase and hydroxylysine glycosyltransferase, allowing partial glycosylation of the product. This system closely mimics the post-translational system of mammalian cells. The baculovirus-expressed collagen was purified, emulsified with CFA, and used to immunize DBA/1 mice to observe for the development of arthritis. We found that rCB11-A9 was unable to induce either arthritis or antibodies to CII (Table 1). On the other hand, the unmodified control rCB11-induced arthritis at its expected incidence of 40% as well as inducing a significant antibody response to murine CII (Table 1). These data suggest that rCB11-A9 will be safer than many conventional therapies, if used to treat arthritis because it is non-immunogenic and non-arthritogenic.

Noninvasive bioimaging is an exciting new development that can be applied to clinical diseases to monitor the duration of gene expression and determine the extent of the therapeutic effect. As adenoviral constructs typically can be transcribed but not replicated with cell division, it was important to predict the length of time the introduced therapy might be effective. We developed a system for in vivo bioimaging in which the firefly luciferase reporter gene was incorporated into our adenoviral vector and this construct was injected into murine ankle joints. Serial bioluminescence imaging of gene expression was performed on days 1, 3, 12 and 18 following intraperitoneal injection of luciferin, the substrate of luciferase. As shown in Figure 2, the injected sites of joints clearly showed the expression of luciferase, as indicated by the green luminescent color and the expression of luciferase could be detected as late as 18 days post injection. By day 21 the green color was no longer detectable. Taken together, these data confirm that a transgene carried by the adenoviral vector gene can be efficiently transferred into the joints and a sustained release of expression can be successfully achieved.

All the previous data suggest that local expression of rCB11-A9 in arthritic joints will be able to effectively modulate CIA. To test this hypothesis the rCB11-A9 was incorporated into the adenoviral genome and the resulting construct (AdenoX-rCB11A9) tested. To evaluate potency in the treatment of arthritis, DBA/1 mice were injected intra-articularly (in the hind ankles) with the adenoX-rCB11-A9 either three or seven days prior to immunization with CII/CFA and were observed for the development of arthritis. Control mice were injected with adenoX-lacZ. As predicted, the mice treated with the adenoX-rCB11-A9 demonstrated a significant decrease in the severity of arthritis as manifested by the severity scores and visual inspection (Figure 3, panels a and b). The control adenoX-Lac-Z construct had no effect. Concordant with a decrease in the incidence and severity of arthritis, antibody production to CII was significantly decreased (Table 2). The hindpaws injected with AdenoX-rCB11A9 were profoundly affected when compared with adenoX-lacZ injected control hindpaws, (severity scores of 0 vs 2.8 ± 2.7, P ≤ 0.025 if injected three days prior to immunization and 0 vs 2.2 ± 1.8, P ≤ 0.01 if injected seven days prior to immunization). The non-injected forepaws developed arthritis with attenuated severity (severity scores of 1.3 ± 1.5 vs 4.8 ± 2.1, P ≤ 0.01 when treated three days prior, to imunization or 1.8 ± 2 vs 4.8 ± 1.5, P ≤ 0.01 when treated seven days prior to immunization). These data indicate that therapy with adenoX-rCB11-A9 significantly down regulated the immune responses to CII in vivo and attenuated the development of arthritis.

We have reported that a major component of the mechanism of action for the synthetic peptide analog A9 is its ability to cause T cells to secrete a suppressive cytokine profile. Its effectiveness is based on its ability to transform potential inflammatory T cells and/or bystander T cells into therapeutic (regulatory-like) T cells [8, 14]. Based on our previous observation that the activated antigen-specific T cells found in draining lymph nodes can accurately reflect the T cell responses of arthritic joints [15], we examined the secretion of a panel of cytokines IFN-γ and IL-2 (Th1), IL-10 and IL-4 (Th2) and IL-17 (Th17), by testing supernatants from draining lymph node cells of mice cultured with the murine immunodominant determinant. We found that following treatment with adenoX-CB11-A9, the Th1 and Th17 cytokine responses to murine CII were significantly decreased compared with those induced following treatment with adenoX-lacZ. Similarly, the Th2 cytokines IL-5 and IL-10 were decreased. On the other hand, treatment with the adenoX-rCB11-A9 induced a significantly greater IL-4 response to murine CII when compared with the lac-Z control (Table 3). These data are consistent with the concept that IL-4 has a unique role in the suppression of arthritis that is only partially duplicated by other Th2-type cytokines in the absence of IL-4 [16]. Taken together these data suggest that the adenoX-rCB11-A9 therapy may work by inducing a population of T cells to redirect their cytokine response to secrete predominantly IL-4, a cytokine known to ameliorate arthritis. The ability to induce a population of regulatory-like T cells to secrete suppressive cytokines in the presence of murine CII as well as the ability to redirect inflammatory cells toward a more suppressive phenotype may explain the profound downregulatory effects adenoX-rCB11-A9 has on CIA.

In clinical situations, gene therapy is more likely to be used therapeutically rather than to prevent disease. To determine the effectiveness of this treatment on well-established arthritis, we immunized mice with CII/CFA and at the onset of arthritis (severity score greater than two), we introduced the adenoX-rCB11-A9 into the joints. As shown in Figure 3, the mice injected with the adenoX-rCB11-A9 had a reversal of arthritis, reaching severity scores of 0 within five days and the modulation of arthritis lasted approximately three weeks after the injection. Control mice injected with the adenoX-lacZ control did not improve and progressed to develop a more severe arthritis (Figure 4). The time course for the modulation of arthritis fits quite accurately with the time course predicted by the bioimaging data.