Targeting of the Transcription Factor STAT4 by Antisense Phosphorothioate Oligonucleotides Suppresses Collagen-Induced Arthritis¹

Rheumatoid arthritis (RA)³ is a chronic inflammatory systemic disease mainly affecting the synovial tissue (1, 2). Although its pathogenesis is still incompletely understood, RA is widely accepted to be an immune-mediated disorder. The significant contribution of immune cells to the pathogenesis could be clearly shown by the successful suppression of joint inflammation by anticytokine agents such as neutralizing Abs against TNF-α, soluble TNF receptor fusion proteins, or rIL-1β receptor antagonists (3, 4, 5). In fact, such agents targeting cytokine-driven immune processes play a major role in clinical disease management of RA patients (4, 5, 6).

Although macrophage-derived cytokines such as TNF-α and IL-1β appear to have a key role in mediating inflammatory synovitis, synovial T cells have also been suggested as being key players in both disease induction and promotion in RA. It is well established that the synovial tissue in RA is significantly infiltrated by T lymphocytes (7). However, the precise contribution of T cells to the pathogenesis of RA has not been entirely clarified. Synovial T cells are mainly of the CD4⁺ CD45 RO memory type (8, 9) and show a proinflammatory Th1 phenotype, although the classical Th1 T cell-derived cytokine IFN-γ is sparsely detectable in rheumatoid joints in contrast to TNF-α and IL-1β (10, 11). However, the recently described proinflammatory cytokine IL-17, which is mainly produced by T cells, is thought to be involved in the development of RA. IL-17 is abundant in the synovial fluids of RA patients, and T cell clones derived from RA patients frequently produce IL-17 (12, 13). Finally, T cell-targeted medications turned out to be efficient in the treatment of RA. Altogether, these data suggest that T cells play a crucial though still incompletely understood role in RA pathogenesis.

Important contributions to the understanding of the immune-mediated events in RA have originated from animal disease models such as the murine collagen-induced arthritis (CIA) model resembling various features of human RA. The development of arthritis in genetically susceptible mice after immunization with heterologous chicken collagen type II (CII) in CFA is dependent on both T and B cells (14). Similarly as synovial T cells in RA, T cells involved in the pathogenesis of CIA show a Th1-like phenotype. However, the molecular processes underlying the development and effector function of the T cell population in CIA are still poorly understood.

The heterodimeric cytokine IL-12 is known to be an important factor for the differentiation of naive T cells into Th1 effector cells (15, 16). After IL-12 binding to the high-affinity IL-12R, the Jak/STAT cascade is activated, leading to homodimerization of the transcription factor STAT4. Subsequently, STAT4 translocates to the nucleus where it binds to specific binding sites that augment IFN-γ gene expression and mediate the development of a Th1 T cell subset (15). However, in contrast to the role of IL-12 in promoting Th1 differentiation, the significance of IL-12-mediated processes in the pathogenesis of CIA is less clear because there are conflicting data with regard to this issue. The analysis of the contribution to CIA of either of the two IL-12-forming subunits, p40 and p35, using IL-12 p40^−/− and IL-12 p35^−/− deficient mice, respectively, revealed that p35^−/− mice are highly susceptible whereas p40^−/− mice show a strikingly different phenotype with the suppression of disease activity (17). Similar results were obtained when p40^−/− and p35^−/−mice were compared after induction of the T cell-mediated murine experimental autoimmune encephalitis model (EAE) (18). These data implied that another factor different from the IL-12 p70 heterodimer shares the p40 subunit and contributes to proinflammatory disease activity mediated by Th1-like T cells. Subsequently, Oppmann et al. (19) described a new cytokine, IL-23, which consists of a heterodimer of p40 and the newly identified subunit p19. IL-23 is mainly produced by dendritic cells and macrophages and preferably acts on memory T cells promoting their proliferation and the production of IL-17 (20, 21). In analogy to IL-12 signaling, Parham et al. (22) could demonstrate that IL-23 binding to the IL-23R complex leads to activation of the Jak/STAT cascade as well as to activation of STAT4.

Considering the conflicting data in terms of the role of IL-12 in the pathogenesis of CIA, recent reports suggest the possibility that the existence of intact IL-23 rather than IL-12 is crucial for the development of T cell-mediated proinflammatory disease stages (17, 23) and that both IL-12 and IL-23 mediate the activation of STAT4. We therefore addressed the role of STAT4 in CIA by exploiting two approaches. First, we studied the consequences of STAT4 deficiency on the outcome of CIA in vivo and observed a significantly reduced incidence and severity of disease manifestation in mice lacking STAT4. Second, we investigated the in vivo effects of an antisense (AS) DNA strategy against the translational start site of STAT4 on the clinical course of CIA. These data implied that cytokine-induced STAT4 activation exerts its arthritis-promoting effects especially during the phase of arthritis manifestation. Hence, targeting STAT4 emerges as an interesting target for the development of novel RA therapeutics in the future.

STAT4-deficient BALB/c mice (24) were purchased from The Jackson Laboratory and backcrossed five times in the DBA/1 background for arthritis experiments. Genotyping was performed by PCR using ear biopsy-derived DNA with the following primer sets: STAT4 wild-type (WT) sense, 5′-CCAACTCAGACAGCAACT-3′; STAT4 WT AS, 5′-GCTCTTTGAGCAGGGATGG-3′; Neo sense, 5′-CTTGGGTGGAGAGGCTATTC-3′; and Neo AS, 5′-AGGTGAGATGACAGGAGATC-3′. Furthermore, mice used in CIA experiments were homozygous for I-A^q as determined by FACS analysis of peripheral mononuclear blood cells (biotinylated anti-mouse I-A^q and I-A^d from BD Biosciences; PE-conjugated streptavidin from Dianova). Male DBA/1 mice were derived from the animal facility of the University of Mainz. In all experiments only male STAT4^−/−, STAT4^+/−, and STAT4^+/+ mice were used at the ages of 8–12 wk during the time of immunization. Animals were housed in the animal facility of the University of Mainz (Mainz, Germany) and maintained under standard laboratory conditions with chow and water ad libitum.

For arthritis induction, mice were immunized intradermally (i.d.) in the ear and at the base of the tail with 100 μg of CII (Sigma-Aldrich) emulsified in CFA (Difco) at day 0. An i.p. booster injection with 50 μg of collagen alone in PBS followed at day 21.

The mice were visually checked for the appearance of arthritis and were considered to be arthritic when either redness or swelling of one or more digits or other parts of the paw were macroscopically detectable. The severity of arthritis was graded in a semiquantitative manner on a scale from 0 (no signs) to 4 (maximal signs of inflammation) for each paw separately: 0, no symptoms; 0.5, redness or swelling of one digit; 1, swelling and erythema of two digits; 2, swelling of more than two digits or mild swelling of the limb; 3, significant but not maximal swelling of the complete or almost complete paw; 4, maximal swelling and erythema of the entire limb and later merging into ankylosis. The scores of four paws per mouse were added so that the maximal possible score per mouse was 16.

Mice were killed at indicated time points followed by the removal of hind paws and fixation of the samples in 10% formic acid for at least 12 h. After decalcification in EDTA, the specimens were embedded in paraffin and stained with H&E. The histopathological grading of joint lesions was performed semiquantitatively on a scale from 0 to 3. The following criteria were examined: inflammatory infiltrate (0, none; 1, mild; 2, moderate; 3, severe infiltrate); synovial lesion (0, no lesion; 1, mild alteration; 2, moderate alteration; 3, severe/complete destruction of the synovia); cartilage destruction (0, none; 1, mild; 2, moderate; 3, severe destruction with loss or complete fragmentation of cartilage); and bone destruction (0, none; 1, mild destruction of subchondral bone; 2, moderate destruction; 3, severe destruction with loss of large areas of bone).

The histological grading was done in a blinded fashion by the same pathologist (P.S.) (25). The total scores represent the mean values of all single scores (mean ± SD) of histologically diseased individual mice within the indicated group.

IFN-γ, TNF-α, and IL-6 production by bulk splenocytes from STAT4^−/− and STAT4^+/+ mice was measured ex vivo at the indicated time points (i.d. on days 10, 40, and 56) after immunization. Briefly, spleens were removed and a single-cell suspension was generated by pushing the spleen through a sterile 40-μm cell strainer. After removal of erythrocytes from the cell suspension by hypotonic lysis in ACK (ammonium chloride potassium phosphate) buffer, spleen cells were washed in PBS and consecutively cultured in 24-well plates with the indicated amounts of CII, 10 U/ml murine IL-12 (PeproTech) or remained unstimulated in RPMI 1640 medium (BioWhittaker) supplemented with 10% heat-inactivated FCS, 3 mM l-glutamine, 10 mM HEPES buffer, and 100U/ml penicillin and streptomycin (BioWhittaker). Where indicated, 24-well plates were coated with 10 μg/ml hamster anti-mouse CD3 Ab (clone 145-2C11) (BD Pharmingen) overnight at 4°C, and unbound Abs were removed by washing with PBS before use. After 24 h, supernatants were collected and either cytokine ELISA analysis was performed immediately as described below or the supernatants were stored at −20°C for later analysis.

Cytokine concentrations in supernatants were measured by sandwich ELISA. Briefly, 96-well plates (Maxisorb; Nunc) were coated overnight at 4°C with 3 μg/ml (IFN-γ), 2 μg/ml (IL-6), or 800 ng/ml (TNF-α) of monoclonal anti-mouse Abs in PBS (100 μl/well). After washing the plates three times with washing buffer (PBS with 0.05% Tween 20; Sigma-Aldrich), wells were blocked with 200 μl/well blocking buffer (PBS, 1% BSA, 5% sucrose, and 0.05% NaN₃) for 1 h at 37°C followed by three washes with washing buffer. Serial dilutions of samples and standards in PBS and 1% BSA (in the case of TNF-α, TBS with 0.1% BSA) were transferred to ELISA plates (100 μl/well) and incubated overnight at 4°C. Plates were washed three times and diluted polyclonal biotinylated Abs for each cytokine were added (400 ng/ml for IFN-γ; 200 ng/ml for IL-6; 300 ng/ml for TNF-α) in 100 μl/well TBS and 0.1% BSA for 2 h at room temperature. After three washes with PBS and 0.05% Tween 20, 100 μl of streptavidin-peroxidase (DakoCytomation) in a dilution of 1/1000 in PBS and 1% BSA was added to each well and incubated for 30 min at room temperature. Plates were then washed an additional three times followed by the addition of 100 μl/well substrate buffer containing 3,3′,5,5′-tetramethyl-benzidine. The color reaction was abrogated by adding 50 μl/well sulfuric acid. The colored product was measured at a wave length of 450 nm using an EMAX precision microplate reader (Molecular Devices). All Abs and standard proteins were purchased from R&D Systems. The IL-23 ELISA was purchased from NatuTec and performed according to the manufacturer’s directions.

As indicated in some experiments, cytokine levels in cell culture supernatants were measured using the FlowCytomix mouse Th1/Th2 cytokine multiplex kit (Bender MedSystems; catalog no. BMS820FF) according to the manufacturer’s directions.

Proliferation of splenocytes was determined by measuring [³H]TdR incorporation. Briefly, 1 × 10⁵ splenocytes derived from naive mice or mice 10 days after immunization were coincubated in a round-bottom 96-well plate (Greiner Bioscience) with 10, 50, or 100 μg/ml CII or remained unstimulated. After 54 h. 1 μCi of [³H]thymidine was added to the cell culture for additional 18 h. After 72 h, the [³H]TdR incorporation was measured in a scintillation counter (LS2800; Beckman Coulter). Each proliferation assay was done in triplicate.

To increase resistance to nucleases, only oligonucleotides with a phosphorothioate backbone modification were used (Carl Roth). The STAT4 AS oligonucleotide consisted of a 21-mer analog to the 5′ end of the coding region of murine STAT4, which spans the translation initiation site. The STAT4 mismatched (MM) control oligonucleotide was a 21-mer analog as well that contained a total of six nucleotide exchanges compared with the sequence of the AS oligonucleotide.

The genomic sequence chosen for targeting was 5′-GAGAGAGCCAGCATGTCTCAGTGGAATC-3′ from murine STAT4 (gene identifier 6755669). The sequences of the phosphorothioate oligonucleotides were 5′-TTCCACTGAGACATGCTGGCT-3′ (STAT4 AS) and 5′-TTCTACTAAGATAAGCGAGCT-3′ (STAT4 MM).

Basic local alignment search tool analysis of either sequence revealed some sequence similarities to several murine bacterial artificial chromosome clones. However, extensive analysis of these genomic DNA sequences ruled out the binding of the oligonucleotides used to a proven or even merely predicted open reading frame other than murine STAT4 mRNA.

Splenocytes from naive healthy DBA/1 mice were isolated as described above. The transient transfection of splenocytes with STAT4 AS and MM oligonucleotides was performed using a cationic lipid formulation (Lipofectamine; Invitrogen Life Technologies). Briefly, 15 μl of Lipofectamine was incubated with 16 μM oligonucleotides in 200 μl serum-free RPMI 1640 medium for 45 min at room temperature. Then, 2 × 10⁶ splenocytes were slowly added to the DNA/Lipofectamine complex and coincubated in the presence of RPMI 1640 medium containing 5% FCS at 37°C. After 2 h the medium was changed to RPMI 1640 medium containing 10% FCS. The cells were incubated in 48-well plates and stimulated with 10 U/ml recombinant murine IL-12 (PeproTech) for 6 h. Untransfected splenocytes that were either stimulated with IL-12 or left unstimulated served as controls. After 6 h, total RNA of the cells was isolated (RNeasy mini kit; Qiagen). The RNA was stored at −80°C if not used immediately.

Unseparated splenocytes from naive DBA/1 WT mice were stimulated with IL-12 for 6 h after being transiently transfected with phosphorothioate-modified oligonucleotides (STAT4 AS or MM) or untransfected to serve as controls as described above. After isolation of total RNA, a real-time RT-PCR analysis of STAT4 expression was performed using a one-step RT-PCR kit (LightCycler RNA Master SYBR Green I; Roche) according to the manufacturer’s directions. The PCR was done on a LightCycler instrument (Roche). The results were internally normalized to the expression of GAPDH. A standard curve for both of the PCR products was generated by PCR amplifying serial dilutions with known concentrations of STAT4 and GAPDH cDNAs and then the ratio of STAT4 to GAPDH expression was calculated. The value of untransfected, IL-12 stimulated splenocytes was set as 100%. The primer sequences used were as follows: STAT4 sense primer, 5′-GAGAAGTGAGATTCCACTCTG-3′; STAT4 AS primer, 5′-CACTCTCCAGTTTCATCTGC-3′; GAPDH sense primer, 5′-ACCACAGTCCATGCCATCAC-3′; and GAPDH AS primer, 5′-TCCACCACCCTGTTGCTGTA-3′.

To prove sufficient splenic uptake of AS DNA, healthy mice were injected i.v. with FITC-labeled, phosphorothioate-modified STAT4 AS DNA. Eight hours after administration of the drug, mice were killed and immunofluorescence staining on cryosections for CD3 (Caltag Laboratories) was performed. Briefly, the sections were fixed in 4% paraformaldehyde followed by rat anti-mouse CD3 staining (Caltag Laboratories) in a 1/100 dilution. After washing with TBS buffer, sections were stained with the secondary biotinylated goat anti-rat IgG (1/150 dilution), washed again, and finally incubated with streptavidin-Cy3 in a 1/150 dilution.

In the CIA experiments, 8- to 12-wk-old male DBA/1 mice were immunized with CII in CFA (i.d. on day 0 and boosted on day 21 with CII in PBS i.p.) according to the protocol described above. On day 26 after immunization, mice received an i.v. treatment with 800 μg of STAT4 AS oligonucleotides. Control animals were either treated with 800 μg of STAT4 MM oligonucleotides or with the solvent PBS alone (total volume, 200 μl). Treatment with 400 μg DNA (STAT4 AS or MM) or PBS injection i.v. was repeated on day 33. Arthritis development was checked by inspection at the indicated time points and scored as described above. At day 40 after immunization, the mice were killed and hind paws removed for were histological examination as described above.

In some experiments, splenocytes derived from immunized mice were sorted using magnetic beads (MACS) to enrich for CD4⁺ and to separate Thy1.2⁺ (CD90) from Thy1.2⁻ cells according to the manufacturer’s directions (Miltenyi Biotec).

Whole cell lysates from CD4⁺- and CD11b⁺-enriched splenocytes derived from immunized DBA/1 WT mice were generated and Western blot analysis was performed using a standard protocol. Briefly, 30 μg protein was used per lane and separated by electrophoresis on a 7.5% SDS polyacrylamide gel. After wet blotting on nitrocellulose, the membrane was blocked in PBS with 0.2% Tween 20 and 5% nonfat dry milk at 37°C for 1 h. Then the blots were incubated with rabbit anti-STAT4 (Santa Cruz Biotechnology) at 4°C overnight, washed at least three times in PBS with 0.2% Tween 20, and subsequently incubated with HRP-conjugated anti-rabbit IgG (code P0217; DakoCytomation) in PBS with 0.2% Tween 20 and 5% nonfat dry milk for 5 h at room temperature. After three washes with PBS and 0.3% Tween 20 the final reaction was developed using a chemiluminescent system (ECL; Amersham Biosciences). For reprobing, the blots were stripped in 0.2M glycine (pH 2.8). All blots were reprobed with anti-β-actin and HRP (Santa Cruz Biotechnology) for normalization and developed as described above.

CD4⁺ T cells were isolated from naive splenocytes using Dynal beads and transfected with STAT4 AS, STAT4 MM, or unrelated oligonucleotides preincubated with Lipofectamine. After stimulation with 10 U/ml rIL-12 (Genzyme) for 48 h the cells were harvested, nuclear proteins were extracted, and a gel retardation assay was performed using a STAT4-specific DNA binding site that has been previously shown to bind to STAT4 (26, 27). The binding reaction was performed at room temperature for 30 min. After electrophoresis, the gels were dried and exposed to Kodak MS films on intensifying screens at −80°C. The sequences of the oligonucleotides for EMSA were 5′-AGTCCTTGAATGGTGTGAAGTAAAAGTGCCTTCAAAGAATCCCC-3′ and 5′-TCAGGAACTTACCACACTTCATTTTC ACGGAAGTTTCTTAGGGG-3′.

For supershift assays, 5 μg of nuclear extracts from splenocytes were incubated with a hamster anti-mouse STAT4 Ab (Abcam) for 1 h on ice followed by a 30-min incubation with STAT4 probe at room temperature and electrophoresis as mentioned above.

Naive CD11b⁺ splenocytes from BALB/c WT mice were isolated by MACS and transfected for 3.5 h with fluorescein-labeled STAT4 AS oligonucleotides (16 μM) preincubated with Lipofectamine as described above. After culturing the cells for 18 h in the presence of 1 μg/ml LPS, cytospins of these cells were generated and analyzed by fluorescence microscopy. In brief, after 18 h of incubation the cells were washed twice in RPMI 1640 and then resuspended in medium at 10⁶ cells/ml. Cells (10⁵) were transferred on cytospin slides and spun for 5 min at 500 rpm. After air drying of the slides, the nuclei were counterstained with Hoechst dye (blue) diluted 1/1000 in PBS for 30 min at room temperature. The slides were washed several times in PBS and 0.1% Tween 20 and mounted in one drop of mounting medium (Vector Laboratories).

To gain insight into the contribution of STAT4 for the development of arthritis, we generated arthritis-susceptible STAT4-deficient mice by backcrossing STAT4^−/− BALB/c mice five times into the DBA/1 background. The induction of CIA and the monitoring for clinical signs of arthritis was performed as specified in Materials and Methods. In all arthritis experiments only male, 8- to 12-wk-old STAT4^−/−, STAT4^+/−, and STAT4^+/+ mice homozygous for MHC class II molecule H-2^q were used for immunization.

Interestingly, STAT4^−/− mice were almost completely resistant to arthritis induction. Fig. 1,A displays the pooled results of two independent, representative experiments in which STAT4-deficient mice had little or no signs of disease. As shown in Table I, only one of 23 collagen-immunized STAT4^−/− mice (4%) developed macroscopic signs of arthritis (onset day 27, maximum score 8). STAT4^+/− heterozygous mice had a delayed onset (mean day 45 ± 11.8 vs day 36 ± 9.6 in WT mice) and a reduced incidence of arthritis (44 vs 62% in WT mice) (Table I). Once afflicted, however, these mice showed clinical signs of arthritis comparable to those of WT mice (arthritis index 5.31 in STAT4^+/− vs 4.94 in WT mice). In addition, there was no difference between fifth generation STAT4^+/+ mice originating from backcrossing and DBA/1 inbred mice in terms of incidence and severity of arthritis development, indicating a complete reconstitution of arthritis susceptibility in backcrossed F5 mice (data not shown).

The above results were further supported by investigating the histopathological differences between collagen-immunized STAT4^+/+ and STAT4^−/− DBA/1 mice. Fifty-six days after immunization, STAT4^−/− mice displayed strikingly reduced histopathological activity of arthritis as compared with STAT4^+/− and STAT4^+/+ mice (Fig. 1,B). Only 4 of 18 analyzed STAT4^−/− mice showed a few signs of arthritis, reflecting an overall mild disease manifestation (Table II). The overall mean score of arthritic STAT4^−/− mice was ∼75% reduced as compared with STAT4^+/− and WT mice. The latter two groups showed no differences (Fig. 1 B), however.

In addition, we performed a detailed analysis of several important histopathological features of an arthritic joint, including the evaluation of the inflammatory infiltrate, synovial lesions, and cartilage and bone destruction, respectively. As shown in Table II, similarly to the clinical pattern the morphological signs of arthritis in STAT4^−/− mice were dramatically reduced in comparison to STAT4^+/− and WT mice. The range of the histological scores was 0.5–1 in arthritic STAT4^−/− mice vs 1–3 in heterozygous and STAT4^+/+ mice, respectively (Table II). In contrast, the histological analysis did not reveal any significant differences between immunized STAT4^+/− and STAT4^+/+ mice with regards to frequency (7 of 14 vs 11 of 21 mice) and histological criteria of arthritis (Table II). Taken together, these results imply that STAT4 is critical for the incidence of disease manifestation in CIA.

Proinflammatory cytokines could contribute in a crucial way to the diminished degree of arthritis in STAT4^−/− mice. To analyze potential differences in the collagen-specific response of splenocytes ex vivo after in vivo priming in STAT4^−/− and STAT4^+/+ mice, we analyzed and compared the cytokine production at day 56 after immunization. Splenocytes were restimulated ex vivo with either 50 μg/ml CII or 10 U/ml IL-12 or remained unstimulated for 24 h. Production of IFN-γ, TNF-α, and IL-6 was measured in the supernatants by ELISA. As shown in Fig. 1 C, STAT4^−/− splenocytes did not produce significant amounts of IFN-γ in response to either stimuli or under unstimulated conditions, consistent with a role for STAT4 in controlling IFN-γ production of T cells. In contrast, STAT4 WT T cells produced marked amounts of IFN-γ upon Ag-specific stimulation. In addition, STAT4 deficiency led to decreased levels of both TNF-α and IL-6 in comparison with the levels produced by STAT4 WT splenocytes after restimulation with collagen ex vivo at day 56.

To further discriminate between T cell- and non-T cell-derived cytokine expression profiles, we separated Thy1.2⁺ T cells and Thy 1.2⁻ splenocytes from collagen-immunized STAT4 WT and STAT4-deficient mice. Interestingly, anti-CD3/anti-CD28 Ab plus IL-12 stimulated Thy1.2⁺ T lymphocytes from collagen-immunized STAT4^−/− mice did not only produce reduced levels of IFN-γ, IL-6, and TNF-α but also significantly lower amounts of IL-17 as compared with WT control lymphocytes (Fig. 1,D). In contrast, IL-23 production of Thy1.2⁻ cells was indistinguishable between STAT4^−/− or STAT4^+/+ mice (Fig. 1 D). STAT4 thus regulates the production of a panel of T cell-derived proinflammatory cytokines with known pathogenic roles in CIA.

The above data suggested that the reduced clinical signs of arthritis in STAT4^−/− mice might be mediated at least partially by a diminished supply of proinflammatory cytokines that are known to contribute to arthritis manifestation. However, it remained to be determined whether the disease-modifying effect of STAT4 deficiency is, in addition, due to an impaired Ag priming. To directly address this early step in the course of CIA, STAT4^−/− and STAT4^+/+ male mice were immunized i.d. with collagen in CFA on day 0. On day 10, spleens were removed and splenocytes were restimulated ex vivo with increasing dosages of collagen as indicated or remained unstimulated for 24 h. Supernatants were analyzed for the production of indicated cytokines by ELISA (Fig. 2). Similar to the above results, there was a marked impairment of IFN-γ production by STAT4^−/− splenocytes as compared with WT cells. However, there was no difference in the capacity of TNF-α or IL-6 production between WT and STAT4-deficient mice on day 10, suggesting that the differences observed above at later stages of disease are secondary to the STAT4 effects on IFN-γ production during early phases of CIA. In addition, we performed proliferation assays with splenocytes derived from the same experimental setting. After 54 h, cells were pulsed with [³H]thymidine for an additional 18 h. As shown in Fig. 2, both STAT4^−/− and STAT4^+/+ splenocytes from collagen-immunized mice proliferated in response to collagen to a similar degree and to a greater extent as compared with the control splenocytes derived from nonimmunized DBA/1 WT mice. However, there was no difference in proliferation between STAT4^−/− and STAT4^+/+ splenocytes. Taken together, these data suggested that there is no relevant STAT4-dependent impairment of the primary response to collagen in immunized mice in terms of the proliferative capacity and production of disease-promoting proinflammatory cytokines, with the important exception of a strikingly decreased IFN-γ production.

STAT4 is known to be critical for the differentiation of T cells in Th1 effector cells and is also expressed by both human dendritic cells and monocytes in response to stimulation (28). To further address the question of which types of cells express STAT4 during CIA, we analyzed STAT4 protein expression during the course of CIA in WT DBA/1 mice. At the indicated time points after the immunization of male DBA/1 WT mice, spleens were removed and CD11b⁺ and CD4⁺ splenocytes were isolated using immunomagnetic beads (Fig. 3). Whole cell protein extracts of these two cell populations were generated and then Western blotting with a STAT4-specific Ab was performed. As shown in Fig. 3, STAT4 protein expression was significantly increased starting on day 27 after arthritis induction preferentially in CD4⁺ T cells and to a lesser extent in CD11b⁺ splenocytes. Such an increase was followed by a reduction of STAT4 levels in T cells starting at days 30 and 32 of CIA. Moreover, we observed a second period of increased STAT4 protein expression in CD4⁺ splenocytes starting at day 37. Taken together, the data show that in the pathogenesis of arthritis, splenic T cells express STAT4 in response to Ag recall in a temporarily ordered, biphasic manner.

Because the above data suggested that the STAT4 protein is specifically expressed during the manifestation period of arthritis and might exert a disease-modulating function in CIA pathogenesis, we next aimed at specifically suppressing STAT4 expression in vivo. Accordingly, we designed several phosphorothioate-modified AS oligonucleotide sequences directed to the translational start site of STAT4 and screened these oligonucleotides in vitro for their ability to diminish STAT4 mRNA levels and STAT4 protein expression.

To demonstrate efficient transfection and appropriate localization of oligonucleotides, magnetically sorted CD11b⁺ splenocytes from naive DBA/1 mice were transiently transfected with fluorescein-labeled STAT4 AS oligonucleotides in complex with Lipofectamine. After 18 h of stimulation with LPS, cytospins of these cells were generated and analyzed by fluorescence microscopy. As shown in Fig. 4 A, the majority of the cells (>50%) were successfully transfected with the AS oligonucleotide as shown by nuclear colocalization of the FITC-labeled oligonucleotides.

We further investigated the functional capacity of these oligonucleotides to interfere with STAT4 mRNA levels. For this purpose, splenocytes were transiently transfected with STAT4 MM or AS oligonucleotides and stimulated with IL-12 for 6 h followed by RNA isolation and STAT4 mRNA real-time PCR analysis. A phosphorothioate-modified STAT4 AS oligonucleotide led to an ∼70% reduction of STAT4 mRNA levels in STAT4 AS oligonucleotide-transfected splenocytes in comparison with untreated controls (Fig. 4 B). In contrast, a STAT4 MM oligonucleotide (STAT4 MM) did not result in a marked reduction of STAT4 mRNA level in transfected splenocytes.

Next, we addressed the question whether STAT4 AS treatment would also influence the extent of STAT4 protein expression and especially reduce the STAT4 protein/DNA interaction. In these studies, we isolated CD4⁺ T cells and treated these cells with STAT4 AS, STAT4 MM, or completely unrelated oligonucleotides in the presence of rIL-12 for 48 h. After extracting nuclear proteins we performed gel-retardation assays (EMSA) using a consensus STAT4 DNA binding site. As shown in Fig. 4 C, the coincubation of CD4 T cells with specific STAT4 AS oligonucleotides led to a significant reduction of STAT4 binding to a STAT4 reference DNA binding site, indicating reduced amounts of STAT4 in the nucleus under STAT4 AS treatment. The addition of a STAT4-specific Ab to STAT4 consensus site resulted in a reduction of the protein/DNA complex, indicating the presence of STAT4 in this complex (data not shown).

In a subsequent series of studies, we aimed at the analysis of the effects of STAT4 oligonucleotides in vivo. We found that the i.v. injection of FITC-labeled STAT4 AS oligonucleotides resulted in an uptake of the DNA by CD3⁺ splenic T cells (Fig. 4,D). This system enabled us to evaluate the in vivo effects of the STAT4 AS oligonucleotides on the clinical and histological outcome in collagen immunized mice (CIA). Based on the kinetics of STAT4 protein expression in splenocytes of mice after CIA induction, we decided to administer the AS reagents on day 26 and to repeat the application on day 33 after immunization. Mice were injected i.v. with STAT4 AS oligonucleotides, STAT4 MM oligonucleotides, or the solvent PBS alone at the same time points. As shown in Fig. 5 A, treatment with STAT4 AS oligonucleotides resulted in a significantly delayed onset of arthritis manifestation and in an ∼70% reduction of the clinical end score at day 40 as compared with mice injected with STAT4 MM oligonucleotides or just PBS. These results indicate that interference with transcribed STAT4 mRNA by administering STAT4 AS oligonucleotides ameliorates the course of the disease in a specific manner.

In subsequent studies, we performed histopathological comparisons of paraffin sections from STAT4 AS-, STAT4 MM-, and PBS-treated DBA/1 mice. Mice were killed on day 40 after immunization and the hind paws were removed for histological studies. Consistent with the clinical assessment, all histological aspects of arthritis were strikingly reduced after STAT4 AS oligonucleotide treatment as compared with administration of identical amounts of STAT4 MM DNA and PBS (Fig. 5,B). Fig. 5 C shows representative histopathological evidence for significant reduction of joint inflammation and joint destruction after the administration of STAT4 AS oligonucleotides in comparison with STAT4 MM and PBS treatment. These results showed that, in addition to the reduced clinical manifestation, STAT4 AS oligonucleotide treatment also strikingly diminishes the joint damage in CIA.

In the present study, we have identified a crucial role of the transcription factor STAT4 in the pathogenesis of CIA. Furthermore, by using STAT4 AS oligonucleotides this study suggests that the disease-promoting action of STAT4 especially takes place during the manifestation phase of the disease. Our data imply that specific targeting of STAT4 might be a relevant future therapeutic approach in RA treatment.

STAT4 plays a multifaceted role in the immune system, as STAT4 is activated by a number of cytokines (e.g., IL-2, IL-12, IL-15, IL-21, IL-23, and IL-27) (16, 29, 30, 31). All of these cytokines are known to promote autoimmune disorders, which highlights STAT4 as an appropriate therapeutic target for the attenuation of disease progression by specifically intervening with its transcriptional effects. Presumably, the best characterized roles for STAT4 are the transcriptional mediation of IL-12-induced production of IFN-γ, the regulation of the IL-12Rβ2 chain expression, and the promotion of the differentiation of naive T cells into proinflammatory Th1 effector cells (15).

Several recent studies suggested a more complex picture of IL-12 action in murine CIA. First, although early Ab mediated neutralization of the IL-12 p40 subunit starting with immunization did not lower the incidence but significantly diminished the severity of arthritis, treatment of mice with established arthritis led to an exacerbation of clinical symptoms (32, 33). These results suggest that IL-12 might play a dual role in CIA. Secondly, IL-12Rβ2^−/−, IFN-γ^−/−, IFN-γR^−/−, and STAT1^−/− mice, although lacking critical hallmarks of proinflammatory Th1 T cells, somehow show unexpectedly enhanced susceptibility to various autoimmune disease models (34, 35, 36, 37). However, in contrast to IL-12 p35-deficient mice, mice lacking the p40 subunit showed a dramatically reduced susceptibility and severity of CIA, suggesting that IL-23 p19/p40 may play an important role in CIA (17, 38). Indeed, by using p19-deficient mice it could be demonstrated that, similarly as in EAE (39), arthritis susceptibility is dependent on the presence of the IL-23 p19 subunit (17). Hence, the shared use of the p40 subunit by IL-23 and IL-12 provided a possibility to explain the obvious dichotomy of the p40^−/− and p35^−/− phenotype in various classical Th1-mediated animal models (17, 19).

Beyond the shared use of the p40 subunit, IL-12 and IL-23 converge on the transcriptional level by activating the same intracellular transcription factor, STAT4. The significance of the STAT4 activation in CIA has not yet been analyzed. In our study we demonstrate that STAT4^−/− mice show a dramatically reduced incidence and severity of CIA development. This result provides further significant insight into the IL-12/IL-23/STAT4 balance in CIA for several reasons. First, it proves that STAT4-mediated pathways are crucial for the development of arthritis. These data are consistent with other reports investigating the contribution of STAT4 to different autoimmune disease models such as EAE, experimental autoimmune myocarditis, and proteoglycan-induced arthritis (40, 41, 42). Second, the clinical outcome of STAT4^−/− mice in CIA is similar to the weak or absent CIA induction in p40^−/− mice that lack both IL-23 and IL-12 (38). However, other cytokines such as IL-15 may activate STAT4 signaling and contribute to an arthritis manifestation that would be compromised in STAT4^−/− mice (43). Thus, it is possible that cytokine signaling in addition to IL-12 and IL-23 signaling is impaired in STAT4-deficient mice during CIA. However, the question remains as to how disrupted STAT4 signaling suppresses CIA. A defective priming of naive T cells could take place as a consequence of abrogated IL-12 signaling in STAT4^−/− mice, leading to a reduced disease manifestation. However, STAT4^−/− mice did not show any reduction in the primary proliferative response to Ag in comparison to WT mice, ruling out grossly inhibited priming. Furthermore, there was no evidence for a direct regulatory effect of STAT4 on IL-6 and TNF production during the early phases of CIA, indicating that the reduced IL-6 and TNF levels at late stages of the disease are secondary to the effects of STAT4 on IFN-γ production.

STAT4 expression in CD4⁺ T cells from WT mice was clearly up-regulated during CIA in a temporally orchestrated, biphasic manner. STAT4 expression was also up-regulated in CD11b monocytes, consistent with a recent report that monocytes derived from the synovial tissues of RA patients express STAT4 (28). However, the kinetics of STAT4 expression during the onset and early progression of arthritis might be correlated with the known dynamics of the disease-promoting cytokine expression in CIA induced by a specific cytokine milieu. Marinova-Mutafchieva et al. (44) analyzed the cytokine expression pattern of lymph node cells from days 1 to 10 after the onset of arthritis. Interestingly, these cells produced rapidly decreasing amounts of IFN-γ after collagen restimulation, whereas the amounts of IL-6 and TNF were progressively increasing (44). Thus, we suggest that the very distinct and organized expression profile of proinflammatory cytokines released by arthritic T cells is cytokine milieu dependent, e.g., an IL-12- and sequentially IL-23 driven and, at least in part, STAT4-mediated process during the early stages of arthritis that is reflected by a similarly orchestrated STAT4 expression pattern. In any case, STAT4 in T cells was found to regulate the expression of various proinflammatory cytokines, such as IFN-γ, IL-6, TNF, and IL-17, that are known to play a pivotal role in CIA (44).

The specific pattern of STAT4 expression in the course of disease manifestation led us to further analyze the functional role of STAT4 at this stage of arthritis. To address this issue, we developed an AS strategy using oligonucleotides that were directed against the translational start site of STAT4 mRNA. STAT4 AS oligonucleotides led to a significant reduction of STAT4 mRNA levels and the diminished binding to a STAT4 consensus binding site by STAT4 derived from nuclear extracts after AS DNA treatment. Importantly, transient transfection of CD11b⁺ or CD19⁺ splenocytes with STAT4 AS or MM DNA did not alter their cytokine expression profile (IL-1, TNF, and IFN-γ in CD11b⁺ cells and IL-10 and IFN-γ in CD19⁺ B cells) under stimulation, arguing against unspecific activating signals elicited by the oligonucleotide sequences used (data not shown). These data encouraged us to use this approach for CIA therapy in vivo. We treated WT mice with CIA by injecting STAT4 AS and MM oligonucleotides i.v. on day 26 after immunization and repeated such treatment 1 wk later. Strikingly, treatment with STAT4 AS DNA but not with MM DNA led to a significant reduction of both clinical and histopathological signs of arthritis. These findings have several implications for the pathogenesis of CIA. The data imply that the biphasic pattern of STAT4 protein expression in CIA is functionally relevant. Based on our results and the work of others (44), we speculate that the STAT4 expression kinetics may precede the expression and mimic the dynamics of the expression of crucial arthritis mediators as IFN-γ, IL-6, and TNF during the onset of disease. Our findings suggest that at this stage of CIA STAT4 is predominantly a proinflammatory transcriptional mediator whose targeting is beneficial for CIA. In conclusion, the observation of naturally occurring STAT4 expression during arthritis manifestation and successful interference with STAT4 action by an AS approach further support the above findings in knockout studies and suggest that STAT4 plays a pivotal role in CIA. Importantly, STAT4 plays a major role especially at later stages of CIA rather than exclusively regulating early T cell priming and thus emerges as a key novel therapeutic target in chronic autoimmune arthritis.

Taken together, it appears that STAT4 activation in CIA is likely to be mediated by various proinflammatory cytokines, such as IL-12, IL-23, and IL-15, that converge to induce STAT4 activation. Our data suggest a model in which STAT4 drives the production of IFN-γ and thereby Th1 T cell effector function in CIA. STAT4-induced Th1 T cell activation subsequently causes the production of key downstream effector molecules such as IL-6 and TNF, thereby promoting synovial inflammation, tissue destruction, and further recruitment of effector cells. Whether, in later stages of CIA, STAT4 is additionally central for IL-23-mediated expansion of synovial Th17 T cells—a recently identified T cell subset that secrets IL-17, IL-6, and TNF and critically contributes to arthritis manifestation—remains to be analyzed in future studies (20, 21, 45).

Thus, our data unequivocally identify STAT4 as an important transcriptional mediator of proinflammatory signals leading to arthritis manifestation. Specific targeting of STAT4 expression and activation by AS strategies thus emerges as a highly effective approach for the treatment of experimental arthritis with important implications for the design of novel therapies for RA in humans.

We thank Renate Eilers for excellent technical help with the preparation of the histological sections.

The authors have no financial conflict of interest.