Patients with psoriasis and psoriatic arthritis respond well to tumor necrosis factor α (TNFα) blockers in general; however, there is now mounting evidence that a small cohort of patients with rheumatoid arthritis who receive TNFα blockers develop psoriasis. This study was undertaken to explore the mechanisms underlying TNFα blockade–induced exacerbation of skin inflammation in murine psoriasis-like skin disease.
Skin inflammation was induced in BALB/c scid/scid mice after they received CD4+CD45RBhighCD25− (naive CD4) T cells from donor mice. These mice were treated with either anti–interleukin-12 (anti–IL-12)/23p40 antibody or murine TNFRII-Fc fusion protein and were examined for signs of disease, including histologic features, various cytokine levels in the serum, and cytokine or FoxP3 transcripts in the affected skin and draining lymph node (LN) cells. In a separate study, naive CD4+ T cells were differentiated into Th1 or Th17 lineages with anti-CD3/28 magnetic beads and appropriate cytokines in the presence or absence of TNFα. Cytokine gene expression from these differentiated cells was also determined.
Neutralization of TNFα exacerbated skin inflammation and markedly enhanced the expression of the proinflammatory cytokines IL-1β, IL-6, IL-17, IL-21, and IL-22 but suppressed FoxP3 expression in the skin and reduced the number of FoxP3-positive Treg cells in the draining LNs. TNFα also demonstrated a divergent role during priming and reactivation of naive T cells.
These results reveal a novel immunoregulatory role of TNFα on Th17 and Treg cells in some individuals, which may account for the exacerbation of skin inflammation in some patients who receive anti-TNF treatments.
Tumor necrosis factor α (TNFα) is a pleiotropic cytokine that has multiple proinflammatory and costimulatory effects on a broad range of cell types (1–6). TNF regulates cell trafficking, activation, maintenance of secondary lymphoid structures, and host defense against various pathogens, thus playing a major role in orchestrating inflammation and immunity (7). Recently, increasing evidence indicates that TNFα may have immunosuppressive effects, since long-term exposure to TNFα can directly prevent the activation of T cells (8, 9) or promote the expansion of Treg cells (10). Although the exact mechanism of TNFα in modulating immune function is not clear, its role in inflammatory bowel disease and rheumatic diseases, including rheumatoid arthritis (RA) and psoriasis, has been well established (11, 12). TNFα antagonists, including the anti-TNF monoclonal antibodies infliximab and adalimumab as well as the soluble TNF receptor (sTNFRII-Fc) etanercept, have demonstrated clinical efficacy in treating human RA (13), ankylosing spondylitis (14), psoriatic arthritis, and Crohn's disease (15). However, one unexpected side effect of TNF antagonism reported in the literature is the new onset or worsening of psoriatic skin lesions (16). It is paradoxical that TNFα antagonists are efficacious for treating psoriasis in some patients while inducing psoriasis in other patients (17–21).
We have developed a skin inflammation model by adoptively transferring CD4+CD45RBhighCD25− naive T cells into scid/scid recipient mice. Certain histologic and immunopathologic features in this model mimic those of human psoriasis (22). In particular, we have demonstrated that Th1 (TNFα, interferon-γ [IFNγ], and IL-12), Th17 (IL-17 and IL-22), and proinflammatory cytokines (IL-1 and IL-6), which are important in the pathogenesis of human psoriasis (23, 24), also play a significant role in this skin inflammation model. With this model, we previously demonstrated that the IL-12/23p40 pathway is critical in regulating the complex inflammation cascade that leads to the development of disease. An antibody neutralizing the p40 subunit shared by IL-12 and IL-23 has shown efficacy in psoriasis clinical trials (25, 26).
In the present study, we found that, consistent with the results of our previous study (22), antagonizing the p40 pathway almost completely ameliorated skin inflammation, whereas neutralization of TNFα exacerbated skin lesion development. We further demonstrated that TNFα neutralization enhances Th17 cytokine gene expression, correlating with a decrease in the percentage of Treg cells found the in the draining lymph node (LN).
MATERIALS AND METHODS
Female BALB/cBy donor mice and CB17/prkdcscid/J recipient mice were purchased from The Jackson Laboratory (Bar Harbor, ME). All mice were housed in a specific pathogen–free environment at a Wyeth animal facility, and were studied between 6 and 8 weeks of age. All protocols were approved by the Wyeth Animal Care and Use Committee.
Phycoerythrin (PE)–conjugated anti-CD4, fluorescein isothiocyanate (FITC)–conjugated anti-CD45RB, and allophycocyanin (APC)–conjugated anti-CD25 were purchased from BD PharMingen (San Diego, CA). Anti-mouse IL-12/23p40 (clone C17.8) and rat IgG2a (isotype control antibody) were purchased from BioLegend (San Diego, CA). Human TNFα was purchased from R&D Systems (Minneapolis, MN). Human and mouse soluble receptor for TNFα (TNFRII-Fc) were generated in house (Wyeth Research, Cambridge, MA). Anti–Eimeria tenella (mouse IgG2a, HB 8389 2.03.7, isotype control) was from American Type Culture Collection (Rockville, MD). Mouse IgG2a was used as isotype control for TNFRII-Fc because both proteins share the same Fc region. This isotype antibody was previously used in studies of collagen-induced arthritis (CIA) and did not demonstrate any biologic effects (27).
Cell purification, adoptive transfer, and treatment of mice.
Cells for adoptive transfer were prepared as previously described (22), with slight modification. CD4+ T cells were enriched from BALB/cBy splenocytes using a mouse CD4 enrichment kit (R&D Systems). The cells were then labeled with PE-conjugated anti-CD4, FITC-conjugated anti-CD45RB, and APC-conjugated anti-CD25 antibodies. Cells were subsequently sorted using a MoFlo cell sorter (Dako, Fort Collins, CO). CD4+CD45RBhighCD25− cells were collected and were >95% pure. Cells were resuspended in saline, and 4 × 105 cells per mouse were injected intraperitoneally into CB17/prkdcscid/J mice.
Mice were treated with antibodies/soluble receptors as indicated above and were monitored for external signs of skin lesions twice each week. At the end of the study, mouse ears and LNs were collected for further ex vivo studies.
Mice were evaluated twice a week starting 10 days after adoptive transfer, by investigators who were blinded with regard to treatment. To record disease progression, mice were scored for disease severity based on external physical appearance using a semiquantitative scale ranging from 0 to 6, where 0 = no skin or ear abnormalities, 0.5 = slight erythema on either the ears or eyelids, 1 = mild to moderate erythema on the ears or eyelids, with mild thickening of the ear (<2% of the body surface), 2 = moderate to severe erythema on 2–10% of the body surface and mild scaling, 3 = severe erythema and scaling on 10–20% of the body surface, 4 = very severe and extensive erythema and scaling on 20–40% of the body surface, 5 = very severe and extensive erythema and scaling on 40–60% of the body surface, and 6 = very severe and extensive erythema and scaling on >60% of the body surface (22). Specific observations were noted based on fur condition, ear manifestations, eyelid appearance, and presence of abnormalities on the limbs and tail.
Tissue specimens were processed into paraffin tissue blocks using routine methods and sectioned. Sections were stained with hematoxylin and eosin.
Quantitation of cytokine transcripts.
RNA was isolated from mouse ears using an RNeasy kit from Qiagen (Valencia, CA). Quantitative reverse transcriptase–polymerase chain reaction (RT-PCR) for cytokine transcripts was performed using prequalified primers and probes for mouse IL-1β, TNFα, IL-22, IL-17A, IL-21, IL-6, IFNγ, IL-12p35, IL-23p19, myeloperoxidase (MPO), S100A8, S100A9, β-defensin 1, FoxP3, and human S100A7, S100A8, IL-1α, TNFα, IFNα, intercellular adhesion molecule 1 (ICAM-1), and CCL27 (Applied Biosystems, Foster City, CA). The ΔΔCt method was used to normalize transcripts to GAPDH and to calculate fold induction relative to levels detected in control mice.
Serum IL-22, IL-6, IFNγ, and TNFα levels were determined using corresponding enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems). IL-17 MAB721 and IL-17 BAF421 (R&D Systems) were used to detect IL-17 levels.
Ex vivo LN cell activation.
Cervical LN cells were isolated from individual mice and pooled within each treatment group at study termination. LN cells were rested overnight in 10% fetal calf serum RPMI 1640, 5 × 10−5M 2-mercaptoethanol, 0.5 mM sodium pyruvate, 10 mM HEPES buffer, 50 units/ml penicillin, 50 μg/ml streptomycin, and 2 mML-glutamine culture media at 37°C and 5% CO2. Cells (2 × 106/ml) were stimulated in a 24-well plate with anti-mouse CD3/28 Dynabeads (Invitrogen, San Diego, CA) to stimulate T cells according to the recommendations of the manufacturer. Consistent with the in vivo treatment, TNFRII-Fc or isotype control was added at 50 μg/ml to the cultures for 6 hours before RNA was isolated for quantitative RT-PCR analysis of cytokine transcripts. To determine the effect on antigen-presenting cells, LN cells were harvested and rested as described earlier. LN cells (3 × 106/ml) were stimulated in a 24-well plate with IFNγ (100 pg/ml) and lipopolysaccharide (LPS; 1 ng/ml) for 6 hours before RNA was isolated for quantitative RT-PCR analysis of cytokine gene expression.
In vitro T cell activation.
Naive CD4+ T cells were isolated from the splenocytes of BALB/cBy mice as described above. In a 24-well plate, 1 × 106 CD4+ T cells were cultured with an equal number of anti-mouse CD3/28 Dynabeads (Invitrogen) and under the following conditions for Th1 or Th17 cell priming in the presence or absence of various concentrations of TNFα as indicated: 10 ng/ml IL-12 and 10 μg/ml anti–IL-4 for Th1 priming and 1 ng/ml transforming growth factor β (TGFβ), 20 ng/ml IL-6, 10 ng/ml IL-1β, 10 ng/ml IL-23, 10 μg/ml anti-IFNγ, and 10 μg/ml anti–IL-4 for Th17 priming. For restimulation cultures, cells were harvested on day 7 of primary stimulation, washed extensively, and rested overnight. T cells (1 × 106) were restimulated with anti-CD3/28 Dynabeads in the presence or absence of various concentrations of TNFα, as indicated. At the end of 6 hours, T cells were harvested for RNA extraction and analysis of gene expression for various cytokines by quantitative real time RT-PCR.
Primary human keratinocytes (ScienCell, San Diego, CA) were cultured in keratinocyte medium (ScienCell) on human fibronectin–coated plates (BD Biosciences, San Jose, CA). Cells were passaged at 80% confluency, and all experiments were performed between passages 2 and 4. For evaluation of TNFα effects, 15,000 cells were seeded into a 24-well plate and were allowed to adhere for 2 days. Cells were then treated with human TNFα or with TNFα plus human TNFRII-Fc for an additional 2 days. RNA was purified, and quantitative RT-PCR was performed.
Intracellular FoxP3 staining.
FoxP3 staining was performed on pooled cervical LN cells. Cells were first stained for CD4 surface antigen and then treated with Cytofix/Cytoperm according to the recommendations of the manufacturer (PharMingen). Intracellular cytokine staining was performed using APC-conjugated antibodies to FoxP3 and irrelevant IgG isotype controls. All plots were gated on lymphocytes based on forward and side scatter, and positive percentages are shown.
Student's t-test was used to calculate the statistical significance of differences between groups. P values less than or equal to 0.05 were considered significant. Except where indicated otherwise, data are representative of ≥2 independent experiments.
Suppression of skin inflammation by blocking the IL-12/23 pathway and exacerbation of skin inflammation by TNFα neutralization.
Previously, we and other investigators have shown that anti–IL-12/23p40 antibody effectively prevents the development of skin lesions in the CD45RBhigh adoptive transfer model of skin inflammation (22, 28). To compare the efficacy of neutralizing IL-12/23 and TNFα pathways in this model, mice were administered a neutralizing antibody that binds to the p40 subunit of IL-12 and IL-23 or a soluble murine TNFRII-Fc fusion protein or the corresponding isotype controls rat IgG1 or mouse IgG2a. Consistent with the results of prior studies, we found that mice given IL-12/23p40 antibody did not develop disease, whereas mice given the same amount of isotype control antibody developed severe skin lesions (Figure 1A). The mean ± SEM disease severity scores for mice treated with IL-12/23p40 antibody and the corresponding isotype control were 0.2 ± 0.12 and 2.4 ± 0.41, respectively. Histopathologic results were consistent with the macroscopic observations showing that the IL-12/23p40 antibody almost completely abrogated acanthosis, parakeratosis, and inflammatory cell infiltrates in the ear skin biopsy specimens as compared with mice treated with isotype control antibody (Figures 1B and C).
The efficacy of TNFα neutralization was determined by administering soluble TNFRII-Fc or isotype control antibody to recipient mice. Notably, compared with the mice treated with isotype control, mice that received soluble TNFRII-Fc fusion protein developed significantly more severe disease starting on day 42. Mice treated with TNFRII-Fc had a mean ± SEM disease severity score of 3.4 ± 0.3, as compared with 2.2 ± 0.23 in the control mice (Figures 1A and B). Microscopic analysis demonstrated that the affected skin from the TNFRII-Fc–treated mice showed slightly more acanthosis and inflammatory infiltrates than did skin from the mice receiving isotype control antibodies (Figure 1C). Overall, the above results show that IL-12/23 neutralization prevented the development of skin disease, while TNFα blockade triggered more severe skin lesions.
The effect of TNFα neutralization on skin inflammation is not strain dependent. When similar studies were performed in B6 scid/scid mice, soluble TNFRII-Fc once again exacerbated skin lesions. This result was confirmed by histopathologic analysis, in that there was a slight increase in epidermal hyperplasia and inflammation (data not shown).
Different effects of IL-12/23 and TNFα neutralization on gene expression in affected skin.
To further understand the factors that contribute to skin inflammation, we first examined the effect of IL-12/23p40 and TNFα neutralization on gene expression in the skin. Quantitative RT-PCR analyses of messenger RNA from mouse ears demonstrated that IL-12/23 pathway neutralization led to a reduction in several antimicrobial peptide genes, including S100A8, S100A9, and β-defensin (calgranulin B). However, these antimicrobial peptide genes were strongly up-regulated in the TNFRII-Fc–treated mice (Figure 2A). Additionally, IL-12/23p40 antibody treatment significantly decreased IL-1β and IL-6 transcripts, accompanied by a trend toward reduced IFNγ and TNFα expression (Figure 2B).
TNFRII-Fc treatment led to a significant increase in IL-1β but a decrease in TNFα transcripts, whereas IL-6 and IFNγ transcripts were relatively unchanged (Figure 2B). Importantly, mice treated with anti-p40 antibody experienced notable reductions in IL-17A and IL-22 (Th17) transcripts, while expression of these 2 genes was highly elevated in mice receiving TNFRII-Fc (Figure 2B). In addition, IL-21 transcripts, which are usually below the detection limit in isotype control–treated mice, were increased in mice that received TNFRII-Fc (Figure 2B). These results indicate that blockade of the IL-12/23 pathway dramatically down-modulates Th17 cytokine genes, whereas TNFα neutralization significantly enhances Th17 cytokine transcripts. These findings also suggest that Th17 cell–associated cytokine transcripts correlate strongly with the development of skin disease in this model.
Different effects of IL-12/23 and TNFα neutralization on serum cytokine profile.
We also determined the effect of blocking IL-12/23 and TNFα pathways on the serum levels of cytokines at the end of the study. Mice treated with anti-p40 antibody produced significantly less serum IL-22 and showed a trend toward reduced production of IL-17A when compared with the isotype-treated control mice (Figure 3A). In contrast, serum concentrations of IL-17A and IL-22 were increased in mice treated with TNFRII-Fc (Figure 3A). Of note, serum IL-21 levels were below the detection limit (4 pg/ml) under all treatment conditions. Low levels of IL-6 and IFNγ were detected in the isotype control–treated mice, and the levels of these 2 cytokines were not significantly changed by neutralizing either pathway (Figure 3B). TNFα levels in mice treated with isotype control and mice treated with anti-p40 antibody were below the detection limit (6 pg/ml), whereas TNFRII-Fc–treated mice experienced a dramatic increase in TNFα levels. Since the ELISA used for the detection of TNFα can detect TNFα complexed with the soluble TNFRII-Fc administered during treatment, detection of this elevated level of TNFα suggests that the TNFα ELISA captured and stabilized biologically inactive TNFα. A similar observation of serum TNFα has been described in patients with RA who were treated with a TNFα inhibitor (infliximab) (29).
Overall, these findings demonstrate that anti-p40 treatment blocks, whereas TNFRII-Fc enhances, Th17 cytokine production, correlating well with gene expression profiles in the ear. Inhibition of TNFα alone is not sufficient to prevent skin inflammation, since TNFα neutralization enhances Th17 cytokine production and leads to more severe skin lesions.
TNFα neutralization enhances T cell activity but suppresses the functions of antigen-presenting cells and keratinocytes.
Previous results indicate that skin infiltrates, including T cells, macrophages, dendritic cells (DCs), neutrophils, and keratinocytes, as well as the proinflammatory mediators released by these cells, play critical roles in driving skin inflammation (30). We further dissected the effect of long-term TNFα blockade on CD4 cells, antigen-presenting cells, and keratinocytes in this model. First, we stimulated the LN cells from TNFRII-Fc–treated or isotype-treated mice with anti-CD3/28 Dynabeads. Consistent with the in vivo treatment, 50 μg/ml of TNFRII-Fc or isotype control was added to the cultures. Cells were then harvested for RNA isolation and quantitative RT-PCR analysis of the cytokine genes. Cells from mice treated with TNFRII-Fc expressed increased transcripts for IFNγ, IL-17A, IL-22, and IL-21, but not IL-2, as compared with those from the isotype control–treated group (Figure 4A). These data indicate that TNF blockade facilitates Th17 and Th1 cell activity.
We next examined the effect of neutralizing the TNFα pathway on antigen-presenting cells. This was achieved by adding IFNγ and LPS to the LN cells from mice treated with TNFRII-Fc or isotype control antibody. Quantitative RT-PCR analysis revealed that compared with isotype control–treated mice, LN cells from TNFRII-Fc–treated mice expressed fewer transcripts for IL-12p35, IL-23p19, and MPO but similar levels of IFNα (Figure 4B), suggesting that TNFα neutralization in vivo correlates with negative regulation of the major functions of antigen-presenting cells (DCs and macrophages).
We then evaluated the direct effect of TNF on keratinocytes. Since murine keratinocyte cell lines are not available, we used a human primary keratinocyte cell line in our study. Human keratinocytes were cultured in the presence or absence of TNFα, TNFα plus TNFRII-Fc, or TNFα plus isotype control. Keratinocytes were then harvested, and RNA was isolated for quantitative RT-PCR for various genes. As indicated in Figure 4C, addition of TNFα led to activation of keratinocytes and increased antimicrobial peptide (S100A7 and S100A8), ICAM-1, chemokine (CCL27), and proinflammatory cytokine (IL-1α and TNFα) gene expression compared with cultures without TNFα. The presence of TNFRII-Fc reduced these transcripts. Collectively, these findings suggest that TNF blockade promotes T cell activity but inhibits antigen-presenting cell and keratinocyte functions.
TNFα has divergent functions during priming and reactivation of naive T cells.
To further examine the role that TNF plays during naive T cell differentiation and reactivation, we differentiated naive T cells (CD4+CD45RBhighCD25−) purified from BALB/cBy mice into the Th1 (IL-12 and anti-IL-4) and Th17 (TGFβ, IL-6, IL-1β, IL-23, anti-IFNγ, and anti–IL-4) lineages cultured with or without TNFα for 7 days (31). We then examined the expression of cytokine transcripts using quantitative RT-PCR after restimulation with anti-CD3/28 beads and various concentrations of TNFα. Th1 cells expressed the highest amounts of IFNγ transcripts. Th17 cells produced the greatest abundance of IL-17A and IL-22 (data not shown), demonstrating that these cells were successfully differentiated. TNFα further enhanced IFNγ transcripts from Th1 cells when present during primary stimulation but suppressed IFNγ expression during restimulation (Figure 5A). Similarly, TNFα enhanced the expression of Th17 cytokine genes (IL-17, IL-22, and IL-21) during priming but suppressed Th17 cytokine transcripts in a dose-dependent manner when present during restimulation (Figure 5B). Our results indicate that TNFα has divergent effects on cytokine production from T cells during priming and reactivation. In particular, long-term exposure to TNFα can result in the suppression of T cell function.
Negative regulation of FoxP3+ Treg cells by TNFα neutralization.
In addition to effector T cells, Treg cells have been implicated in the maintenance of immune homeostasis in the skin (32). We have previously shown that Treg cells suppress disease progression in a dose-dependent manner (22). Treg cells not only arise as thymus-derived natural Treg cells but can also be induced in the periphery upon encountering antigen in a specific cytokine milieu (33). We investigated whether the FoxP3 gene transcript, which defines Treg cells, is expressed in this adoptive transfer model. To do this, we performed a time course study in which mouse ears were collected from CB17 scid/scid recipient mice on weeks 1, 2, 4, and 13 after adoptive transfer. Figure 6A indicates disease progression with time after adoptive transfer where FoxP3 gene expression in mouse ear RNA was determined by quantitative RT-PCR. FoxP3 transcripts were not detected in CB17 scid/scid mice 1 week after cell transfer but became apparent at 2 weeks. FoxP3 transcripts were further increased during week 4 and remained elevated until week 13 after adoptive transfer (Figure 6B). This finding implies that Treg cells are induced in the periphery and gradually increase in number as disease progresses.
To determine whether there is a correlation between Treg cells and the severity of disease, we compared FoxP3 gene expression in ear RNA from mice treated with isotype control with mice treated with TNFRII-Fc; we found that ear RNA from mice treated with TNFRII-Fc had significantly fewer FoxP3 transcripts than did that from isotype control–treated mice (Figure 6C). This result was consistent with the intracellular FoxP3 staining data, where mice receiving TNFRII-Fc had an ∼2-fold decrease in the percentage of FoxP3+ Treg cells as compared with that of isotype control–treated mice (from 5.56 to 2.38) in their draining LNs (Figure 6D). In addition, TNFRII-Fc treatment did not have an obvious effect on the overall percentage of CD4+ T cells in the recipient mice. Collectively, these results demonstrate that TNFα regulates the expansion of Treg cells.
TNF antagonists have demonstrated efficacy in treating a number of rheumatic diseases, including RA, psoriatic arthritis, and psoriasis. Nevertheless, in a small subset of patients, TNF neutralization can also lead to worsening or induction of psoriasis (16). The immune mechanisms involved in soluble TNFRII-Fc–induced psoriasis remission are well studied (34). Patients with psoriasis who responded to TNF blockade had reduced inflammatory infiltrates, including DCs and T cells, as well as proinflammatory cytokine gene expression in the lesional skin (34). However, the mechanisms by which TNF blockade enhances skin inflammation are not well understood. It is possible that certain patients exhibit new or worsened psoriasis after anti-TNF treatment as a consequence of their genetic makeup. Using a psoriasis-like skin inflammation model, we demonstrated that administration of murine soluble TNFRII-Fc reproducibly exacerbated skin lesions by activating Th17 and Th1 cells or/and preventing the expansion of the FoxP3+ Treg population.
It has been suggested that the paradoxical occurrence of psoriasis as an adverse event of anti-TNF therapy may be due to the relationship between TNFα and type I IFNα (16, 18). Dermal plasmacytoid DCs, which produce IFNα, have been shown to play a pivotal role in the early phase of induction of psoriasis (35). TNFα normally inhibits plasmacytoid DC maturation from hematopoietic progenitors and thus IFNα production (18). The inhibition of TNFα may induce sustained IFNα production that, in certain patients, may lead to an outbreak of psoriasis; some patients have been reported to have locally increased type I IFNα activity while receiving anti-TNFα treatment (18). In our model, however, IFNα transcripts were not significantly up-regulated in the lesional skin from mice receiving TNF blockade (data not shown). Also, addition of TNFα to LPS-stimulated draining LN cells from TNFRII-Fc–treated mice did not further enhance IFNα expression. These findings point toward additional mechanism(s) that may contribute to anti-TNF–induced skin inflammation.
The immunomodulatory effect of TNFα could be due to a direct suppressive activity on T cell function. Consistent with the results of previous studies (8, 36), we found that acute exposure to TNFα in vitro enhances T cell activity while chronic TNFα exposure suppresses T cell responses (Figure 5). Also, we demonstrated that long-term in vivo TNFα blockade enhanced proinflammatory cytokine (IL-1β and IL-6) and Th17 cytokine (IL-17A and IL-22) transcripts in the mouse ear and Th17 cytokines in the serum. Restimulation of draining LN cells from anti-TNF–treated mice revealed that TNFα neutralization enhanced the expression of IL-17, IL-22, and IL-21 transcripts from T cells but suppresses IL-12p35 and IL-23p19 gene expression from antigen-presenting cells (Figure 4). The divergent effects of TNFα on T cells and antigen-presenting cells may imply that IL-21 produced by Th17 cells, but not the IL-23 pathway, is critical in maintaining Th17 cells in vivo when TNFα is limiting. Whether similar conclusions apply to patients with exacerbated skin inflammation undergoing anti-TNF treatment will require further investigation.
In addition to the direct immunosuppressive effect of TNFα on T cells, a recent study by Chen et al (10) has shown that TNFα may indirectly suppress T cell function by promoting the expansion and activity of Treg cells. Although in our model CD4+CD25+ Treg cells are depleted before adoptive transfer, Treg cells can still differentiate from mature CD4+ T cells in the periphery, as indicated in our time course study and by the results of previous studies (33). Strikingly, TNFα neutralization down-modulates FoxP3 transcripts in the affected ears and reduces the percentage of the FoxP3+ Treg population in the draining LNs. The enhanced expression of IL-21 transcripts by TNF blockade in our model suggests that IL-21 may further suppress the activity and function of Treg cells, since IL-21 regulates the reciprocal developmental pathways of Th17 and Treg cells (37, 38).
Consistent with our observations, TNF blockade has been shown to expand Th1 and Th17 cells in a CIA model (39). Patients receiving TNFα antagonist treatment have increased peripheral T cell activity (40). Interestingly, in both studies, TNF inhibition correlated with decreased T cell trafficking to the sites of joint destruction, which resulted in an accumulation and sequestration of these cells in the circulation or draining LNs (39, 41). In support of an antiinflammatory role of TNFα, repeated administration of TNFα was shown to protect (NZB × NZW)F1 and Lewis rats from developing lupus-like nephritis (42, 43) and autoimmune arthritis, respectively (44, 45). The fact that more severe experimental autoimmune encephalomyelitis and CIA are induced in TNFα-deficient mice (46) and that TNFRII−/− mice are subjected to more severe inflammation in the cecal ligation and puncture model (10) strongly highlights a negative regulatory role of TNFα in some immune models.
Taken together, our findings demonstrate that TNF blockade facilitates disease progression in a CD4+ T cell–mediated psoriasis-like skin inflammation model by directly enhancing Th17 cell function and/or indirectly down-modulating the expansion of Treg cells. The lack of T cell suppression by Treg cells may further enhance expansion of Th17 cells. We have shown previously that cytokines produced by Th17 cells, including IL-22, can act directly on keratinocytes and are required for the development of Th17 cell–dependent disease in this model (30). The enhanced production of Th17 cytokines, in particular IL-22, may further activate keratinocytes to release antimicrobial peptides that act in a positive feedback loop to recruit more immune cells to the site of skin inflammation and cause tissue damage. The pathologic consequences induced by TNF blockade in this model need to be further explored in patients with RA who develop psoriasis or psoriatic arthritis after treatment with TNF inhibitors. Nevertheless, our study highlights a novel negative feedback mechanism for TNFα to limit the intensity and/or the duration of chronic inflammation in some disease settings.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Young had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Ma, Nickerson-Nutter, Young.
Acquisition of data. Ma, Napierata, Stedman, Benoit.
Analysis and interpretation of data. Ma, Stedman, Benoit, Collins, Young.
We thank Adam Root for preparing the antibodies and the cytokines. We also thank Marina Shen for providing the human reagents, Sarah Eichenberger, Richard Maylor, and John-Paul Jimenez for excellent technical support in tissue collection, and Jameel Syed for histologic slide preparation and staining.