ERK1/2 Regulates Epidermal Chemokine Expression and Skin Inflammation
http://www.100md.com
免疫学杂志 2005年第8期
Abstract
Resident cell populations of the skin contribute to the inflammatory response by producing an array of chemokines, which attract leukocytes from the circulation. TNF- is a major inducer of proinflammatory mediators in keratinocytes. We have recently observed that epidermal growth factor receptor (EGFR) signaling affects TNF--driven chemokine expression in epidermal keratinocytes, and its functional impairment increases the levels of crucial chemoattractants such as CCL2/MCP-1, CCL5/RANTES, and CXCL10/IFN--inducible protein-10. In this study, we report evidence that EGFR-dependent ERK1/2 activity is implicated in this mechanism. Abrogation of ERK1/2 activity with specific inhibitors increased chemokine expression in keratinocytes by enhancing mRNA stabilization. In mouse models, inflammatory response to irritants and T cell-mediated contact hypersensitivity were both aggravated when elicited in a skin area previously treated with an EGFR or a MAPK kinase 1/2 inhibitor. In contrast, impairment of p38 MAPK phosphorylation markedly attenuated these responses. Our data indicate that EGFR-dependent ERK1/2 activity in keratinocytes takes part to a homeostatic mechanism regulating inflammatory responses, and emphasize the distinct role of MAPKs as potential targets for manipulating inflammation in the skin.
Introduction
The skin epithelium, which is composed mainly of keratinocytes interspersed with dendritic cells, melanocytes, and rare T lymphocytes and monocytes, is highly committed to host defense. Physical, chemical, or immune-specific insults rapidly evoke an epidermal response characterized by the increased expression of a system of proinflammatory mediators, including chemotactic factors, which initiate the orientated migration of distinct leukocyte subpopulations. In turn, activated monocytes, dendritic cells, and T cells release potent cytokines that act on cells in the local environment to boost the inflammatory response (1, 2). In particular, TNF- induces the expression of numerous chemokines, including CCL2/MCP-1, CCL5/RANTES, CCL20/MIP-3, CCL27/cutaneous T cell-attracting chemokine, CXCL8/IL-8, and CXCL10/IFN--inducible protein-10 in keratinocytes. Epidermal CCL2 is involved in the early response to injury or irritants, and in T cell-mediated skin disorders (3, 4, 5, 6), and controls the recruitment of monocytes/macrophages, dendritic cells, and T cells (7, 8, 9). Active immigration of T cells, monocytes, as well as neutrophils is also supported by the increased expression of CCL5 (5, 6). However, type 1 T cells appear massively attracted into the skin by keratinocyte release of the CXCR3 ligands, CXCL10, CXCL9/monokine induced by IFN-, and CXCL11/IFN-induced T cell -chemoattractant (5, 10). CXCR3 ligands appear deeply involved in type 1 T cell-mediated diseases such as allergic contact dermatitis and psoriasis, but are not relevantly expressed during skin response to irritants (5, 11). Finally, CXCL8 is the best characterized of a group of chemoattractants active in neutrophil recruitment as well as in epithelial and endothelial cell proliferation (3, 12). Enhanced CXCL8 expression is characteristically associated with sustained epidermal growth factor (EGF)3 receptor (EGFR) activation, and is considered a secondary amplification mechanism leading to tissue hyperplasia (13).
EGFR governs the homeostatic maintenance and repair of epithelial tissues. Typical responses to EGFR activation include stimulation of keratinocyte proliferation and migration, and controlled differentiation. We recently found that during their early response to TNF- or IFN-, keratinocytes release EGFR ligands including TGF-, which induce EGFR autophosphorylation and as a consequence activate its signal transduction cascade (14). Enhanced EGFR activation was associated with increased CXCL8 expression, and diminished CCL2, CCL5, and CXCL10 expression. In contrast, impairment of EGFR activation led to an opposite pattern. In the mouse, skin application of a selective EGFR tyrosine kinase inhibitor led to more severe contact hypersensitivity (CH) responses, with increased epidermal levels of CCL2, CCL5, and CXCL10, and a higher number of monocytes/macrophages and T lymphocytes in the skin. These findings suggested that EGFR modulates skin inflammation by affecting chemokine expression in keratinocytes.
EGFR activation leads to the persistent induction of the classical MAPK pathway identified as ERK 1 and 2 (ERK1/2) (15), which plays a fundamental role in the EGFR-driven control of epidermal proliferation (16, 17). In contrast, the other major subgroups of the MAPK family, namely p38 and and the JNK/stress-activated protein kinase 1 and 2 (JNK1/2), are weakly activated by EGFR, whereas they are highly stimulated on exposure to TNF- (15, 18).
In this study, we provide experimental evidence that EGFR is involved in the control of chemokine expression in epidermal keratinocytes via an ERK1/2-dependent mechanism. Our data indicate that ERK1/2 takes part in a homeostatic mechanism that regulates epidermal involvement in the inflammatory response, and point to the MAPKs as promising targets for the modulation of skin inflammation.
Materials and Methods
Chemicals and reagents
Actinomycin D, 5,6-dichloro-1-D-ribofuranosylbenzimidazole (DRB), croton oil (CrO), benzalkonium chloride (BAC), and 2,4-dinitrofluorobenzene (DNFB) were obtained from Sigma-Aldrich. Recombinant human TGF- and TNF- were purchased from R&D Systems. The small molecule, cell-permeant protein kinase inhibitors PD168393 (19), PD98059 (15), U0126 (15), SB203580 (15), and SP600125 (20) were from Calbiochem. Table I summarizes their specific target and the concentration range used in this study.
Table I. Inhibitors used in this study
Cell cultures and treatments
Primary cultures of normal human keratinocytes were obtained as previously described (21), and routinely grown in serum-free keratinocyte growth medium (Cambrex). Primary cultures of dermal macrovasculature endothelial cells were grown in endothelial cell growth medium (Cambrex). Both of these culture media include 10 ng/ml EGF. In the 24 h preceding treatments, 80% confluent cell cultures were switched to EGF-depleted medium. Primary cultures of normal human dermal fibroblasts were grown in DMEM (Invitrogen Life Technologies), supplemented with 10% FCS, 1% penicillin/streptomycin, and 1% L-glutamine.
RNA extraction and RNase protection assay (RPA)
Total RNA was extracted from cell cultures and from mouse ear biopsies using the TRIzol reagent (Invitrogen Life Technologies). The templates of the human and mouse chemokines under investigation, the housekeeping molecule L32, and the kit for RPA were purchased from BD Pharmingen. Total RNA (10 μg) was hybridized overnight with -32UTP-labeled cDNA templates, and reactions were performed as per the manufacturer instructions.
mRNA decay assay
Decay rates of CCL2 and CXCL8 transcripts were measured by the use of RNA polymerase II inhibitors. Briefly, transcription was inhibited by the addition of actinomycin D (5 μg/ml) or DRB (25 μg/ml) to fresh culture medium, and total RNA samples were harvested at selected time points thereafter.
ELISA
Chemokines released in the culture medium were measured with dedicated kits from BD Pharmingen.
Immunoprecipitation and Western blot analysis
Keratinocytes were lysed by adding 1 ml per 10-cm dish of lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM PMSF, 1 mM NaF, 1 mM Na3V04, 0.15 U/ml aprotinin, 1% leupeptin, and 1% pepstatin). EGFR was immunoprecipitated using anti-EGFR Ab from Santa Cruz Biotechnology, and phosphorylated EGFR was immunodetected using anti-phosphotyrosine Ab from BD Transduction Laboratories (BD Pharmingen).
In vitro kinase assay
Cell lysates were prepared by applying the same method as described in the Western blotting protocol. Kinase assays for ERK1/2, JNK1/2, and p38 MAPKs were performed using commercial kits from Cell Signaling Technology.
Transient DNA transfection and luciferase reporter assay
The AP-1-dependent reporter plasmid (AP-1-luc) containing seven tandem copies of the AP-1 site and the NF-B-dependent reporter plasmid (NF-B-luc) containing five tandem copies of the NF-B site were purchased from Stratagene. The inducible reporter vector containing full-length CXCL8 promoter (–1481/+44) (IL-8-luc) was a kind gift of N. Mukaida (Cancer Research Institute, Kanazawa, Japan) (22). The CCL2 promoter luciferase construct containing the proximal promoter (between –107 and +60) and distal enhancer region (between –2742 and –2513) (MCP-1-luc) was kindly provided by M. Goebeler (University of Würzburg, Würzburg, Germany) and has been previously described (23). The reporter plasmid containing full-length CCL5 promoter (–974/+1) was generously provided by P. Nelson (University of Munich, Munich, Germany) (24). Keratinocytes were seeded onto 1.9-cm2 wells (2 x 105 cells/well) and transfected when 60% confluent. Transfection was performed over a 6-h period in 1 ml of medium containing 12 μl of lipofectin (Invitrogen Life Technologies), 0.5 μg of the inducible reporter plasmid, plus 0.5 μg of pCMV · SPORT--galactosidase for the control of transfection efficiency. Transfection medium was then removed, and keratinocytes were allowed to recover in fresh medium for 24 h, starved of EGF for further 24 h, and then stimulated. Cultures were finally washed and harvested with 1 ml of lysis buffer (Roche). Luciferase activity was determined (Promega), and normalized to -galactosidase activity and protein concentration.
Mouse models of skin inflammation
Protocols of skin inflammation in mice were approved by the Institute Ethical Committee. To test for irritant contact dermatitis, BALB/c mice (Charles River Laboratories) received topical application on each side of each ear of 10 μl of the following solutions: 0.5% DNFB, 1% CrO, 10% BAC, or vehicle alone (DMSO/absolute ethanol 1/10 v/v) (25). In selected groups, 10 μl of the vehicle or solutions of PD168393, PD98059, or SB203580 was applied on each side of each ear 30 min before the irritant or vehicle alone. To test for CH, BALB/c mice were sensitized by application of 30 μl of 0.5% DNFB in acetone/olive oil (4/1) on a 2-cm2 area of the shaved abdomen. Five days later, sensitized and unsensitized animals received 10 μl of 0.15% DNFB on each side of each ear. In selected groups, the vehicle or a solution of PD168393, PD98059, or SB203580 was painted on sensitized mice 30 min before challenge (14). Ear thickness was measured before application of the chemical and in subsequent time points. Total RNA was extracted from the ears at the time of maximal inflammation, and chemokine expression was analyzed by RPA. In each experimental group, some mice were sacrificed, and the ears were cut and either paraffin embedded for histological analysis (H&E) or snap frozen for immunohistochemistry.
Immunohistochemistry
Cryostatic sections (4 μm) were processed, as previously described (14). The Abs used for the detection of phosphorylated ERK1/2 and p38 MAPK were from Cell Signaling Technology and BioSource International, respectively. For the detection of CCL2, CCL5, and CXCL10, we used dedicated Abs from Santa Cruz Biotechnology. Anti-mouse CD11b (dendritic cells, monocytes, and neutrophils) and Ly-6G (neutrophils) Abs were from BD Pharmingen. Anti-CD3 (T cells) Ab was from DakoCytomation. Secondary biotinylated Abs and staining kits were from Vector Laboratories. As negative controls, primary Abs were omitted or replaced with isotype-matched Ig. Infiltrating cells positive for CD11b, CD3, or Ly-6G were counted on three different sections from the same biopsy with an eyepiece graticule at a magnification of x200 in 10 adjacent fields.
Statistical analysis
The Wilcoxon signed-rank test or Student’s t test (GraphPad Prism software) was applied to compare differences between groups of data. Significance was assumed at a p value of 0.05 or less.
Results
EGFR activation state controls ERK1/2 activity and AP-1 trans activation
The proinflammatory response of cells to TNF- implies the activation of the MAPK pathways (18, 26). Treatment with TNF- alone caused a transient increase of ERK1/2 and p38 MAPK sp. act., and intense JNK1/2 activation at 15 min stimulation (Fig. 1A). TGF- alone induced a robust and persistent activation of ERK1/2, and a modest induction of JNK1/2 and p38 MAPKs. Costimulation with TNF- and TGF- led to sustained ERK1/2 activity and the synergistic potentiation of JNK1/2 and p38 MAPKs. Conversely, the selective EGFR tyrosine kinase inhibitor PD168393 potently interfered with basal and TNF--associated ERK1/2 activation, confirming the crucial role of EGFR signaling in the control of this MAPK subfamily. Strong suppression of ERK1/2 was also induced by two chemically unrelated MAPK kinase 1/2 (MEK1/2) inhibitors, PD98059 (Fig. 1A) and U0126 (data not shown) (15). SB203580 is a selective inhibitor of p38 MAPK isoforms (15), and induced a strong suppression of p38 MAPK-specific kinase function. The residual p38-associated activity (Fig. 1A, lane 18) could be reasonably due to the p38 isoform, which is abundantly expressed in keratinocytes and coimmunoprecipitates during the kinase assay (27). Quite interestingly, the up-regulated JNK1/2 activity when p38 was inhibited by SB203580 (Fig. 1A, lane 18) and the prolonged p38 activity in the presence of the JNK1/2 inhibitor SP600125 (Fig. 1A, lane 22) indicated p38-JNK1/2 cross-regulation also in keratinocytes (28). Finally, no perturbation of either basal or TNF--induced EGFR phosphorylation could be observed in the presence of the MAPK inhibitors. MAPK activation state is critical for AP-1-dependent gene expression. TGF- promoted AP-1 trans activation with a mean 2.6-fold increase over basal levels (Fig. 1B), whereas TNF- alone enhanced AP-1 transcriptional activity very modestly (Fig. 1B). Noteworthily, MAPK selective inhibitors significantly impaired both basal and TNF--associated AP-1-dependent gene expression, indicating that each MAPK subfamily contributes to AP-1 activation and that the abrogation of distinct MAPKs cannot be compensated by the other pathways. NF-B-dependent gene transcription underwent a strong (mean 4.3-fold) increase following TNF- stimulation, but it was insensitive to EGFR signaling promotion (by TGF-) or blockade (by PD168393), or to MEK1/2 inhibition (by PD98059) (Fig. 1C). Interesting enough, the p38 inhibitor SB203580 and, more potently, the JNK1/2 inhibitor SP600125 impaired NF-B trans activation significantly. No evidence of perturbation of IB cytosolic degradation or p50/p65 nuclear translocation was found (data not shown), supporting the emerging concept that p38 or JNK inhibition could impair NF-B trans activation through IB-independent mechanisms (18, 29, 30, 31, 32).
FIGURE 1. EGFR activation state controls ERK1/2 activity and AP-1 trans activation in human keratinocytes. A, MAPK activity was determined by selective immunoprecipitation (ip) of phosphorylated (P) ERK1/2, p38 MAPKs, or JNK1/2, followed by in vitro kinase assay with the appropriate substrate before SDS-PAGE and immunoblot (ib) analysis. Keratinocytes were stimulated with TNF- (50 ng/ml) and/or TGF- (50 ng/ml) directly, or following 30-min preincubation with 2 μM PD168393 (PD16) or 20 μM PD98059 (PD98), SB203580 (SB20), or SP600125 (SP600). AP-1-regulated (B) or NF-B-regulated (C) luciferase reporter plasmid was transiently transfected into keratinocytes, which were preincubated or not with the indicated inhibitor (or the vehicle, i.e., 0.1% DMSO in culture medium) for 30 min, and then stimulated with TNF- and/or TGF- for 6 h. Values are expressed as the mean ± SD (n = 5 per condition). *, p < 0.05 vs untreated control. **, p < 0.05 vs TNF--treated cultures. Data are representative of three independent experiments.
EGFR activation leads to enhanced CXCL8 and reduced CCL2, CCL5, and CXCL10 expression in skin keratinocytes via ERK1/2 activation
EGFR activation differentially affected both constitutive and TNF--induced expression of a set of chemokines, with increase of CXCL8 and down-regulation of CCL2, CCL5, and CXCL10. These effects were dependent on the dose of TGF- associated with TNF-, and confirmed both at the mRNA and protein levels (Fig. 2, A and B). An opposite outcome was observed when EGFR signaling was impaired (Fig. 2, A and C). Changes in chemokine expression by EGFR signaling appeared restricted to the epithelial cells of the skin. Indeed, TNF--driven CCL2 and CXCL8 expression in dermal fibroblasts (Fig. 2E) and dermal microvasculature endothelial cells was not altered by TGF- or PD168393; although both cell types expressed EGFR and EGFR phosphorylation was induced by TGF- or completely prevented by PD168393 (data not shown), TNF--driven CCL2 and CXCL8 expression was not altered by TGF- or PD168393 in both cell types (Fig. 2, D and E). We then checked the direct impact of MEK1/2 inhibition on chemokine expression. Similar to what we observed in the presence of PD168393, both PD98059 and U0126 up-regulated TNF--induced CCL2, CCL5, and CXCL10, while decreased CXCL8 transcript levels (Fig. 3A). In contrast, the p38 inhibitor SB203580 and the JNK1/2 inhibitor SP600125 invariably down-regulated TNF--induced chemokines. These effects were confirmed at the protein level for CCL2 and CXCL8 (as well as CCL5 and CXCL10; data not shown) in unstimulated keratinocytes (Fig. 3, B and D) and in cells treated with TNF- (Fig. 3, C and E).
FIGURE 2. EGFR activation leads to enhanced CXCL8 and down-regulated CCL2, CCL5, and CXCL10 expression in skin keratinocytes. A, Chemokine gene expression was measured by RPA in total RNA (10 μg) extracted from keratinocytes stimulated for 2 h with TNF- (50 ng/ml) and/or TGF-, directly or following 30-min incubation with PD168393 (PD16). ELISA of TNF- induced chemokine levels after 24-h costimulation with escalating doses of TGF- (B) and PD16 (C). Results are expressed as the mean of three independent experiments ± SD. RPA of CCL2 and CXCL8 expression in dermal fibroblasts (D) and dermal microvasculature endothelial cells (E) stimulated for 2 h with TNF- and/or TGF- (50 ng/ml) directly, or following 30-min preincubation with 2 μM PD16.
FIGURE 3. MEK1/2 inhibitors PD98059 and U0126 reproduce the effects of EGFR inactivation on chemokine expression. A, Chemokine transcript levels were detected in total RNA (10 μg) extracted from keratinocytes stimulated for 2 h with TNF- directly or following 30-min preincubation with 2 μM PD168393 (PD16); 1 μM U0126; 20 μM PD98059 (PD98), SB203580 (SB20), or SP600125 (SP600); or vehicle alone (0.1% DMSO in culture medium). ELISA detection of CCL2 (B and C) and CXCL8 (D and E) levels in the supernatant of unstimulated (open symbols) or TNF--stimulated (filled symbols) cultures at 24 h. Results are expressed as the mean of three independent experiments ± SD.
Inhibition of ERK1/2 signaling affects chemokine expression at the transcriptional and/or posttranscriptional levels
Despite the up-regulated CCL2, CCL5, and CXCL10 expression, we could not find evidence of enhanced gene transcription in association with EGFR or MEK1/2 inhibition. In particular, TNF-, but not TGF-, effectively induced CCL2 promoter (MCP-1-luc) trans activation, which was impaired by SB203580 or SP600125, but not by PD168393 or PD98059 (Fig. 4A), strongly paralleling the profile of NF-B trans activation (Fig. 1C) (33). Similar findings were observed when evaluating CCL5 promoter activity (data not shown). By contrast, CXCL8 promoter (IL-8-luc) was induced by TGF- and/or TNF- (Fig. 4B). Importantly, PD168393 and each selective MAPK inhibitor impaired both basal and TNF--induced CXCL8 promoter activity, confirming that the three MAPKs participate in CXCL8 gene transcription (Fig. 4B), and that both AP-1 and NF-B trans activation are involved in this process (Fig. 1, B and C) (34, 35). Because we did not find evidence of chemokine transcriptional superinduction associated with ERK1/2 inhibition, we then investigated the existence of possible posttranscriptional mechanisms. Actinomycin D was added to prevent further transcription, and the decay of chemokine mRNA was followed by RPA. In cells treated with TNF- alone, CCL2 transcript t1/2 was calculated at 37.5 min, and it underwent a striking increase (>4 h) in cells treated with PD168393 or PD98059 (Fig. 4, C and E), hence providing a mechanism underlying CCL2 up-regulated expression. By contrast, TGF- and SB203580 reduced CCL2 mRNA t1/2 to 30 and 12 min, respectively. Hence, the cumulative effect of SB203580 on TNF--driven CCL2 gene expression was its almost complete abrogation, as observed in Fig. 4C (0-min actinomycin D (Act.D)), due to a reduction of CCL2 gene transcription by more than one-third (Fig. 4A) and a 3 times faster rate of CCL2 transcript decay (Fig. 4E). Both the EGFR and MEK1/2 inhibitors prolonged the t1/2 also of TNF--induced CXCL8 transcript, which passed from 50 min to 215 and 170 min, respectively, whereas TGF- and SB203580 reduced CXCL8 t1/2 to 40 and 18 min, respectively (Fig. 4, D and F). Thus, ERK1/2 blockade was associated with a stabilization of CXCL8 transcript, but also with impaired CXCL8 gene transcription, with the latter mechanism prevailing. By contrast, SB203580 reduced CXCL8 gene expression dramatically, as observed in Fig. 4D (0-min Act.D), being simultaneously responsible for a 2-fold reduction of gene transcription (Fig. 4B) and a 2.5-fold increase in the rate of transcript decay (Fig. 4F), in keeping with previous lines of evidence (34, 35). The same results were obtained with RNA polymerase II inhibition by DRB (data not shown).
FIGURE 4. ERK1/2 inhibition affects chemokine expression at the transcriptional and/or posttranscriptional levels. Luciferase activity was measured in keratinocytes transiently transfected with a plasmid in which the reporter expression was under the transcriptional control of the CCL2 promoter (MCP-1-luc; A) or the CXCL8 promoter (IL-8-luc; B). Transfected keratinocytes were stimulated for 6 h with TNF- (50 ng/ml) alone or in association with TGF- directly, or following 30-min preincubation with 2 μM PD168393 (PD16); 20 μM PD98059 (PD98), SB203580 (SB20), or SP600125 (SP600); or vehicle alone (0.1% DMSO in culture medium). Values are expressed as the mean ± SD (n = 4 per condition). *, p < 0.05 vs untreated control. **, p < 0.05 vs TNF--treated cultures. Data are representative of three independent experiments. RPA of the decay of specific CCL2 (C) and CXCL8 (D) transcript. Keratinocytes were stimulated for 2 h with TNF- alone or in association with TGF- directly, or following 30-min preincubation with the indicated inhibitor. Subsequently, the cultures were washed and the mRNA was extracted (0 min), or they were added to Act.D (5 μg/ml) to evaluate mRNA decay. Semilogarithmic plots of CCL2 (E) and CXCL8 (F) mRNA decay. RPA bands were quantified by laser densitometry, and the chemokine signals were normalized to L32 signals. The ratios were plotted as percentages of values at the time of Act.D addition. Results are expressed as the mean of three independent experiments ± SD.
Impairment of ERK1/2 signaling leads to enhanced inflammatory responses in mouse skin
To verify in vivo our in vitro results, we used mouse models of inflammatory response to irritants and of T cell-mediated immune responses. We found that topical administration of 2–4 mM PD168393, 5–10 mM PD98059, or 5–10 mM SB203580 alone had no effect on ear thickness (Fig. 5A), and did not induce any change in normal skin histology (data not shown). By contrast, the same doses of PD168393 and PD98059 applied 30 min before contact with prototypic skin irritants (25), including DNFB, CrO, and BAC, similarly led to aggravation of the inflammatory response, whereas skin pretreatment with SB203580 resulted in a significant reduction of ear thickness (Fig. 5, B–D). RPA performed on vehicle-pretreated tissue showed the presence of MIP2 (the mouse homologue of human CXCL8) and CCL2, and low level CXCL10 after exposure to DNFB, CrO, or BAC (Fig. 5E). Application of PD168393 or PD98059 led to reduced levels of MIP2 and enhanced expression of CCL2 and CXCL10 transcripts compared with vehicle-treated controls. By contrast, all chemokine transcripts were down-regulated in SB203580-pretreated mice. These results exactly matched those observed in cultured human keratinocytes. Histological analysis at the time point of maximal DNFB-induced reaction visualized a more severe inflammatory response in the skin treated with PD168393 or PD98059, and very limited, if any, inflammation in SB203580-treated skin (Fig. 6). Compared with the faint specific staining for phosphorylated ERK1/2 (P-ERK) and p38 (P-p38) observed in the unperturbed mouse skin (data not shown), up-regulation of these species was observed in the skin affected by irritant contact dermatitis. In PD168393- or PD98059-treated mice, immunohistochemistry confirmed the efficient suppression of P-ERK and an increased number of infiltrating CD11b+ cells in the dermis, whereas no relevant differences could be found in the number of CD3+ or Ly-6G+ cells (data not shown) when compared with vehicle-treated controls. Skin treated with SB203580 showed no change in P-ERK expression compared with controls, and suppression of P-p38, as well as very few CD11b+, CD3+, or Ly-6G+ cells (data not shown). Furthermore, skin treatment with PD168393 or PD98059 led to worsened CH, whereas SB203580 induced a marked attenuation of CH reaction (Figs. 7A and 8). This was associated with a corresponding up- or down-regulation of CXCL10 and CCL2 mRNAs (Fig. 7B) and proteins (Fig. 8). MIP2 mRNA was reduced by all three inhibitors, as expected. More intense CCL2 immunostaining in skin treated with PD168393 or PD98059 was not confined to keratinocytes, but also involved cells with dendritic morphology in the epidermis and dermis. Furthermore, we could observe that the number of CD3+ and CD11b+ cells infiltrating these samples was prominently higher than in controls (Fig. 8). Finally, SB203580-treated skin showed a much lower number of infiltrating CD11b+, CD3+, and Ly-6G+ cells.
FIGURE 5. Topical application of PD168393 or PD98059 aggravates, whereas SB203580 reduces irritant contact dermatitis. Change in ear thickness following topical application of vehicle alone (A), 0.5% DNFB (B), 1% CrO (C), or 10% BAC (D). In selected groups, the irritants were painted (10 μl each side of each ear) 30 min after local application of 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), 10 mM SB203580 (SB20), or vehicle alone. Data represent the mean of the mean changes in ear thickness at each time point from three different experiments (n = 5 per treatment group). In B, significance at p < 0.05 at 2- to 48-h time points in mice groups treated with the three inhibitors vs vehicle-treated controls. In C, p < 0.05 vs vehicle-treated controls at 4- to 48-h time points. In D, p < 0.05 vs vehicle-treated controls at 6- to 48-h time points. E, RPA of chemokine expression in total RNA (10 μg) extracted from the whole ears 6 h after DNFB (a), 8 h after 1% CrO (b), or 24 h after 10% BAC (c). Results are representative of three independent experiments.
FIGURE 6. Augmented inflammatory response to DNFB in the skin of nonimmune mice pretreated with PD168393 or PD9059. Histology (H&E) and immunohistochemical detection of P-ERK and P-p38 as well as CD11b+ and CD3+ cells in ear samples collected 6 h after DNFB painting onto the skin pretreated with vehicle alone (10% DMSO in absolute ethanol), 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), or 10 mM SB203580 (SB20). Magnification: H&E, x40; P-ERK and P-p38, x200; CD11b and CD3, x100.
FIGURE 7. Topical application of PD168393 or PD98059 enhances, whereas SB203580 attenuates CH responses. A, Change in ear thickness due to application of DNFB on the ear skin of immunized BALB/c mice 30 min after local painting of vehicle alone (10% DMSO in absolute ethanol), 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), or 10 mM SB203580 (SB20). Data represent the mean of the mean changes in ear thickness at each data point from three different experiments (n = 5 mice per group). In A, significance at p < 0.05 at 24- to 72-h time points in mice groups treated with the three inhibitors vs vehicle-treated controls. B, RPA of chemokine expression in total RNA (10 μg) extracted from the whole ear 24 h after CH elicitation. RPA representative of three independent experiments.
FIGURE 8. Up-regulated chemokine expression and increased numbers of infiltrating CD11b+ and CD3+ cells in the skin treated with PD168393 or PD98059 before CH elicitation. In contrast, SB203580 suppresses CH reaction. Histology (H&E) and immunohistochemical detection of CCL2, CXCL10, and CD11b+ and CD3+ cells in ear samples collected 24 h after DNFB elicitation onto the skin pretreated with vehicle alone (10% DMSO in absolute ethanol), 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), or 10 mM SB203580 (SB20). Magnification: H&E, x40; CCL2 and CXCL10, x200; CD11b and CD3, x100.
Discussion
In the present study, we have investigated the role of the distinct MAPKs in the modulation of chemokine expression in epidermal keratinocytes. The results indicate that ERK1/2 suppresses proinflammatory chemokines and eventually skin inflammatory responses, whereas p38 MAPK promotes chemokine expression and amplificates skin inflammation. In epithelial cells, autocrine EGFR ligands provide the main extracellular signals for high steady state ERK1/2 activity, which in turn drives intracellular proproliferation and prosurvival programs. We found that ERK1/2 activity strictly depended on EGFR signaling in skin keratinocytes, with selective EGFR inhibition suppressing constitutive ERK1/2 and preventing its activation by TGF- or the proinflammatory cytokines TNF- and IFN- (14). Of note, we uncovered the existence of an EGFR-mediated differential regulation of cytokine-driven chemokine expression in these cells, with higher levels of EGFR activation associated with enhanced CXCL8, but suppressed CCL2, CCL5, and CXCL10 expression, whereas opposite events were registered when EGFR function was blocked. This mechanism appeared to be specific of the epithelium. Apart from epidermal keratinocytes, EGFR-mediated regulation of chemokine expression was reproduced in normal human bronchial, prostate, and mammary epithelial cells (data not shown), but not in nonepithelial components of the skin such as dermal fibroblasts and dermal microvascular endothelial cells. In this study, we provided pharmacological evidence that ERK1/2 is implicated in this mechanism. Impairment of ERK1/2 obtained either by EGFR or MEK1/2 inhibition was similarly associated with down-regulated CXCL8 and enhanced TNF--driven CCL2, CCL5, and CXCL10 in skin keratinocytes.
The specific contribution of each MAPK subfamily to gene expression could be coordinated by their reciprocal interplay. In particular, a mutually inhibitory cross talk between ERK1/2 and p38 has been described in numerous cell types, including HeLa cells (36, 37, 38). Hence, we initially checked whether ERK1/2 blockade could lead to enhanced JNK1/2 or p38 activity, possibly supporting a stronger gene transcription in TNF--stimulated keratinocytes. We could not collect any evidence of ERK1/2 involvement in regulatory cross talk mechanisms. By contrast, we found that ERK1/2, JNK1/2, or p38 selective inhibition each impaired AP-1 trans activation, whereas JNK1/2 or p38 inhibition also reduced NF-B trans activation through IB-independent mechanisms. This last finding could help to explain current and previous evidence that JNK1/2 or p38 inhibition decreases the promoter activity of proinflammatory genes known to strictly depend on NF-B trans activation, such as CCL2 (33) and CCL5 (39, 40). Only CXCL8 promoter activity could be impaired by ERK1/2 inhibition, confirming that ERK1/2-driven AP-1 trans activation plays a relevant role in CXCL8 transcription (35). In their whole, these data suggested that the enhanced expression of CCL2, CCL5, and CXCL10 was rather due to posttranscriptional events.
Increased gene expression during inflammatory responses involves not only the transcription of unstable mRNAs, but also the dynamic regulation of their turnover (18). We found that ERK1/2 inhibition was associated with an impressive increase in the stability of CCL2 transcript. ERK1/2 inhibition was also associated with a considerable stabilization of CXCL8 transcript, which, however, could not counteract its suppressive effect on CXCL8 gene transcription, eventually leading to a reduced CXCL8 expression. Most of the studies that analyzed the role of MAPKs on gene transcript turnover have implicated p38 MAPKs or JNKs as relevant contributors to mRNA stability (41, 42). Accordingly, we observed that p38 MAPK inhibition was associated with a reduced stability of both CXCL8 and CCL2 transcripts, hence opposing chemokine expression both at the transcriptional and posttranscriptional level. CCL2 and CXCL8 (but also CXCL10) transcripts possess adenylate/uridylate-rich elements (ARE) in their 3'-untranslated regions. Activation of p38 MAPK pathway was found to specifically stabilize transcripts containing these elements (42). However, CCL5 mRNA does not contain these motifs (41). In the reasonable hypothesis that ERK1/2 inhibition leads to increased stability of ARE-containing (e.g., CCL2, CXCL8, and CXCL10) and also non-ARE-containing transcripts (e.g., CCL5) in keratinocytes, a general p38 MAPK-independent mechanism should be involved. Although the molecular pathway(s) linking ERK1/2 activity to mRNA turnover still remains to be identified, ERK1/2 has been previously associated with mRNA stability. In human neutrophils, enhanced ERK1/2 activity increases GM-CSF mRNA stability, thus contributing to guarantee the presence of this important growth factor (43). By contrast, ERK1/2 activity was found to correlate with reduced type I collagen mRNA stability in skin fibroblasts (44). This latter mechanism may provide the inhibitory signal on extracellular matrix deposition during wound repair and tumor growth. Importantly, these effects could be specifically reversed by the MEK1/2 inhibitor PD98059.
We looked for the possible pathogenic relevance of MAPK-mediated control on epidermal chemokines using mouse models of skin inflammation. Compared with the faint specific staining found in the unperturbed mouse skin, we confirmed that the phosphorylated forms of both ERK1/2 and p38 were up-regulated during skin response to irritant chemicals (45). Noteworthily, the pharmacological suppression of ERK1/2 activation produced by an EGFR or a MEK1/2 inhibitor led to an aggravation of the local inflammatory response, with enhanced expression of CCL2 and more numerous CD11b+ inflammatory cells, which comprise monocytes, dendritic cells, and neutrophils. The observation that the count of Ly-6G+ cells was not increased in the skin treated with the EGFR inhibitor PD168393 or the MEK1/2 inhibitor PD98059 suggests that the higher number of CD11b+ cells was essentially due to monocytes and dendritic cells, the principal targets of CCL2 (7, 8). In keeping with previous reports in humans (11), we could not detect significant CXCL10 up-regulation during the acute response to irritants. This chemokine was, however, abundantly expressed during CH, and underwent a striking up-regulation in the epidermis of skin pretreated with PD168393 or PD98059, thus providing a strong molecular background to the increased recruitment of CD3+ cells observed in these biopsies. These data confirmed the ERK1/2 is involved in the regulation of a number of potent proinflammatory chemokines in epidermal keratinocytes in vivo, and suggest that it is part of a homeostatic mechanism that tends to oppose skin inflammation (Fig. 9). By contrast, the local application of a p38 MAPK inhibitor led to a reduced inflammatory response in terms of resulting tissue perturbation and number of infiltrating inflammatory cells. SB203580 is known to block the kinase activity of p38 MAPK, but not the phosphorylation of p38 MAPK by upstream kinases (15). Hence, we believe that our evidence of a suppressed p38 MAPK phosphorylation is an indirect consequence of SB203580 skin prepainting, deriving from SB203580-induced (15, 18, 46) local abrogation of TNF- and IL-1 expression in keratinocytes. Indeed, these cytokines are the main activators of the signaling pathways leading to p38 MAPK phosphorylation and eventually to increased chemokine expression in the inflamed skin (1, 2). An impairment of their local expression may thus provide the mechanism underlying SB203580-associated decrease of p38 MAPK phosphorylation. In keeping with our in vitro data, the epidermal (and dermal) expression of the chemokines tested was invariably down-regulated during both irritant dermatitis and CH, further indicating that p38 MAPKs may well represent a relevant therapeutic target in some inflammatory skin diseases. Recently, systemic administration of SB202190 to mice was shown to protect the epidermis against the damaging effects of acute UVB radiation (47). Furthermore, local painting of the same inhibitor was reported to inhibit the expression of CH (48). In this study, we confirmed this finding using the homologue SB203580, and extended its anti-inflammatory effect toward a variety of primary skin irritants. Numerous p38 inhibitors are now being developed for human use for the treatment of common inflammatory diseases such as Crohn’s disease, chronic obstructive pulmonary disease, rheumatoid arthritis, and psoriasis, and some of them are now in clinical trials (49, 50). Serious concerns in their systemic use still derive from p38 MAPK involvement in vital processes such as cardiac and platelet function, hemopoiesis, and bone resorption (15, 51). Our data indicate that a topical use may effectively suppress skin inflammatory disorders, possibly with a better systemic safety profile.
FIGURE 9. Schematic representation of cytokine-driven EGFR activation and its effect on chemokine expression in keratinocytes. Through activation of potent transcription factors, the cytokines TNF- and IFN- boost the expression of crucial chemokines such as CCL2, CCL5, and CXCL10 (A). Activation of p38 MAPK has a primary repercussion on this process by enhancing both proinflammatory gene transcription and transcript stability. However, cytokine receptor (R) stimulation also promotes a rapid metalloproteinase (MMP)-mediated TGF- shedding from its membrane precursor (proTGF-) (14 ), leading to activation of the EGFR/ERK1/2 pathway. This pathway drives a down-regulatory effect on the expression of chemokines by a reduction of their transcript stability. When the EGFR/ERK1/2 down-regulatory loop is abrogated either by the EGFR tyrosine kinase inhibitor PD168393 or the MEK1/2 inhibitor PD98059 (B), chemokine expression is deranged. Topical application of these inhibitors predisposes the skin to more severe inflammatory responses.
Our data also suggest that pharmacological abrogation of EGFR/ERK1/2 signaling pathway worsens skin inflammation by increasing chemokine expression in keratinocytes. However, when EGFR superactivation was forced in the suprabasal epidermal layers by transgenic overexpression of 1 integrins, a patchy skin hyperproliferation and inflammation were observed, which were correlated with enhanced ERK1/2 activity and IL-1 overproduction in keratinocytes (52, 53). Also, the transgenic expression of activated MEK1 in suprabasal keratinocytes led to hyperproliferative and inflammatory skin lesions (54). These results apparently contradict our understanding of ERK1/2 as a regulator of skin inflammation, but they derive from complex conditions of extraordinary ERK1/2 up-regulation, and could be interpreted as a further clue of the fine, homeostatic role of this kinase in the skin. Finally, sustained EGFR/ERK1/2 activation has been shown to provide the major pathway fueling abnormal cellular growth in a number of epithelial tumors, and EGFR inhibition represents an effective cancer treatment strategy (55). Our results possibly suggest that the ultimate anticancer activity due to EGFR inhibition may also rely on a stronger recruitment of inflammatory cells and hence a more pronounced antitumor immune response.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We thank Domenico Rosi for his technical help in the processing of mouse histological samples, and Maurizio Inzillo for help with the artwork.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants from Ministero della Salute and Ministero dell’Istruzione, dell’Università e della Ricerca (Programmi di Ricerca Scientifica di Rilevante Interesse Nazionale).
2 Address correspondence and reprint requests to Dr. Saveria Pastore, Laboratory of Immunology, Istituto Dermopatico dell’Immacolata, Instituto di Ricovero e Cura a Carattere Scientifico, via Monti di Creta 104, 00167 Roma, Italy. E-mail address: pastore{at}idi.it
3 Abbreviations used in this paper: EGF, epidermal growth factor; Act.D, actinomycin D; ARE, adenylate/uridylate-rich element; BAC, benzalkonium chloride; CH, contact hypersensitivity; CrO, croton oil; DNFB, 4-dinitrofluorobenzene; DRB, 5,6-dichloro-1--D-ribofuranosylbenzimidazole; EGFR, EGF receptor; MEK, MAPK kinase; P-ERK, phosphorylated ERK; P-p38, phosphorylated p38; RPA, RNase protection assay.
Received for publication December 2, 2004. Accepted for publication February 3, 2005.
References
Pasparakis, M., G. Courtois, M. Hafner, M. Schmidt-Supprian, A. Nenci, A. Toksoy, M. Krampert, M. Goebeler, R. Gillitzer, A. Israel, et al 2002. TNF-mediated inflammatory skin disease in mice with epidermis-specific deletion of IKK2. Nature 417:861.
Pastore, S., A. Cavani, C. Albanesi, G. Girolomoni. 2005. Chemokines of the skin. J. D. Bos, ed. Skin Immune System 3rd Ed.373. CRC Press, Boca Raton.
Gillitzer, R., M. Goebeler. 2001. Chemokines in cutaneous wound healing. J. Leukocyte Biol. 69:513
Kim, B. H., Y. S. Lee, K. S. Kang. 2003. The mechanism of retinol-induced irritation and its application to anti-irritant development. Toxicol. Lett. 146:65.
Goebeler, M., A. Trautmann, A. Voss, E. V. Br?cker, A. Toksoy, R. Gillitzer. 2001. Differential and sequential expression of multiple chemokines during elicitation of allergic contact hypersensitivity. Am. J. Pathol. 158:431.
Giustizieri, M. L., F. Mascia, A. Frezzolini, O. De Pità, L. M. Chinni, A. Giannetti, G. Girolomoni, S. Pastore. 2001. Keratinocytes from patients with atopic dermatitis and psoriasis show a distinct chemokine production profile in response to T cell-derived cytokines. J. Allergy Clin. Immunol. 107:871.
Nakamura, K., I. W. Williams, T. S. Kupper. 1995. Keratinocyte-derived chemoattractant protein-1 (MCP-1): analysis in a transgenic model demonstrates that MCP-1 can recruit dendritic and Langerhans cells to the skin. J. Invest. Dermatol. 105:635.
Lu, B., B. J. Rutledge, L. Gu, J. Fiorillo, N. W. Lukacs, S. L. Kunkel, R. North, C. Gerard, B. J. Rollins. 1998. Abnormalities in monocyte recruitment and cytokine expression in monocyte chemoattractant protein 1-deficient mice. J. Exp. Med. 187:601.
Carr, M. W., S. J. Roth, E. Luther, S. S. Rose, T. A. Springer. 1994. Monocyte chemoattractant protein 1 acts as a T-lymphocyte chemoattractant. Proc. Natl. Acad. Sci. USA 91:3652.
Rottman, J. B., T. L. Smith, K. G. Ganley, T. Kikuchi, J. G. Krueger. 2001. Potential role of the chemokine receptors CXCR3, CCR4, and the integrin E7 in the pathogenesis of psoriasis vulgaris. Lab. Invest. 8:335.
Flier, J., D. M. Boorsma, D. P. Bruynzeel, P. J. van Beek, T. J. Stoof, R. J. Scheper, R. Willemze, T. P. Tensen. 1999. The CXCR3 activating chemokines IP-10, Mig, and IP-9 are expressed in allergic but not in irritant patch test reactions. J. Invest. Dermatol. 113:574.
Addison, C. L., T. O. Daniel, M. D. Burdick, H. Liu, J. E. Ehlert, Y. Y. Xue, L. Buechi, A. Walz, A. Richmond, R. Strieter. 2000. The CXC chemokine receptor 2, CXCR2, is the putative receptor for ELR+ CXC chemokine-induced angiogenic activity. J. Immunol. 165:5269.
Reich, K., C. Garbe, V. Blaschke, C. Maurer, P. Middel, G. Westphal, U. Lippert, C. Neumann. 2001. Response of psoriasis to interleukin-10 is associated with suppression of cutaneous type 1 inflammation, down-regulation of the epidermal interleukin-8/CXCR2 pathway and normalization of keratinocyte maturation. J. Invest. Dermatol. 116:319.
Mascia, F., V. Mariani, G. Girolomoni, S. Pastore. 2003. Blockade of the EGF receptor induces a deranged chemokine expression in keratinocytes leading to enhanced skin inflammation. Am. J. Pathol. 163:303.
Pearson, G., F. Robinson, T. Beers Gibson, B. E. Xu, M. Karandikar, K. Berman, M. H. Cobb. 2001. Mitogen-activated protein (MAP) kinase pathways: regulation and physiological functions. Endocr. Rev. 22:153.
Iordanov, M. S., R. J. Choi, O. P. Ryabinina, T. H. Dinh, R. K. Bright, B. E. Magun. 2002. The UV (ribotoxic) stress response of human keratinocytes involves the unexpected uncoupling of the Ras-extracellular signal-regulated kinase signaling cascade from the activated epidermal growth factor receptor. Mol. Cell. Biol. 22:5380.
Cai, T., K. Nishida, T. Hirano, P. A. Khavari. 2002. Gab1 and SHP-2 promote Ras/MAPK regulation in epidermal growth and differentiation. J. Cell Biol. 159:103.
Wajant, H., K. Pfizenmaier, P. Scheurich. 2003. Tumor necrosis factor signaling. Cell Death Differ. 10:45.
Fry, D. W., A. J. Bridges, W. A. Denny, A. Doherty, K. D. Greis, J. L. Hicks, K. E. Hook, P. R. Keller, W. Leopold, J. A. Loo, et al 1998. Specific, irreversible inactivation of the epidermal growth factor receptor and erbB2, by a new class of tyrosine kinase inhibitors. Proc. Natl. Acad. Sci. USA 95:12022.
Bennett, B. L., D. T. Sasaki, B. W. Murray, E. C. O’Leary, S. T. Sakata, W. Xu, J. C. Leisten, A. Motiwala, S. Pierce, Y. Satoh, et al 2001. SP600125, an anthrapyrazolone inhibitor of Jun N-terminal kinase. Proc. Natl. Acad. Sci. USA 98:13681
Pastore, S., E. Fanales-Belasio, C. Albanesi, L. M. Chinni, A. Giannetti, G. Girolomoni. 1997. Granulocyte/Macrophage colony-stimulating factor is overproduced by keratinocytes in atopic dermatitis. J. Clin. Invest. 99:3009.
Smith, R. S., E. R. Fedyk, T. A. Springer, N. Mukaida, B. H. Iglewski, R. P. Phipps. 2001. IL-8 production in human lung fibroblasts and epithelial cells activated by the Pseudomonas autoinducer N-3-oxododecanoyl homoserine lactone is transcriptionally regulated by NF-B and activator protein-1. J. Immunol. 167:366.
Ueda, A., Y. Ishigatsubo, T. Okubo, T. Yoshimura. 1997. Transcriptional regulation of the human monocyte chemoattractant protein-1 gene. J. Biol. Chem. 272:31092.
Nelson, P. J., H. T. Kim, W. C. Manning, T. J. Goralski, A. M. Krensky. 1993. Genomic organization and transcriptional regulation of the RANTES chemokine gene. J. Immunol. 151:2601.
Mizumoto, N., T. Kumamoto, S. C. Robson, J. Sevigny, H. Matsue, K. Enjyoji, A. Takashima. 2002. CD39 is the dominant Langerhans cell-associated ecto-NTPase: modulatory roles in inflammation and immune response. Nat. Med. 8:358.
Yang, S. H., A. D. Sharrocks, A. J. Whitmarsh. 2003. Transcriptional regulation by the MAP kinase signaling cascades. Gene 320:3.
Xiao, Y. Q., K. Malcolm, G. S. Worthen, S. Gardai, W. P. Schiemann, V. A. Fadok, D. L. Bratton, P. M. Henson. 2002. Cross-talk between ERK and p38 MAPK mediates selective suppression of pro-inflammatory cytokines by transforming growth factor-. J. Biol. Chem. 277:14884
Lee, A. H., J. H. Hong, Y. S. Seo. 2000. Tumor necrosis factor- and interferon- synergistically activate the RANTES promoter through nuclear factor B and interferon regulatory factor (IRF-1) transcription factors. Biochem. J. 350:131.
Kujime, K., S. Hashimoto, Y. Gon, K. Shimizu, T. Horie. 2000. p38 mitogen-activated kinase and c-Jun-N-terminal kinase regulate RANTES production by influenza virus-infected human bronchial epithelial cells. J. Immunol. 164:3222.
Clark, A. R., J. L. E. Dean, J. Saklatvala. 2003. Post-transcriptional regulation of gene expression by mitogen-activated protein kinase p38. FEBS Lett. 546:37
Tchen, C. R., M. Brook, J. Saklatvala, A. R. Clark. 2004. The stability of tristetraproline mRNA is regulated by mitogen-activated protein kinase p38 and by tristetraproline itself. J. Biol. Chem. 279:32393.
Esnault, S., J. S. Malter. 2002. Extracellular signal-regulated kinase mediates granulocyte-macrophage colony-stimulating factor messenger RNA stabilization in tumor necrosis factor- plus fibronectin-activated peripheral blood eosinophils. Blood 99:4048.
Hobbs, R. M., V. Silva-Vargas, R. Groves, F. M. Watt. 2004. Expression of activated MEK1 in differentiating epidermal cells is sufficient to generate hyperproliferative and inflammatory skin lesions. J. Invest. Dermatol. 123:503
Huang, S., E. A. Armstrong, S. Benavente, P. Chinnaiyan, P. M. Harari. 2004. Dual-agent molecular targeting of the epidermal growth factor receptor (EGFR): combining anti-EGFR antibody with tyrosine kinase inhibitor. Cancer Res. 64:5355.(Saveria Pastore, Francesc)
Resident cell populations of the skin contribute to the inflammatory response by producing an array of chemokines, which attract leukocytes from the circulation. TNF- is a major inducer of proinflammatory mediators in keratinocytes. We have recently observed that epidermal growth factor receptor (EGFR) signaling affects TNF--driven chemokine expression in epidermal keratinocytes, and its functional impairment increases the levels of crucial chemoattractants such as CCL2/MCP-1, CCL5/RANTES, and CXCL10/IFN--inducible protein-10. In this study, we report evidence that EGFR-dependent ERK1/2 activity is implicated in this mechanism. Abrogation of ERK1/2 activity with specific inhibitors increased chemokine expression in keratinocytes by enhancing mRNA stabilization. In mouse models, inflammatory response to irritants and T cell-mediated contact hypersensitivity were both aggravated when elicited in a skin area previously treated with an EGFR or a MAPK kinase 1/2 inhibitor. In contrast, impairment of p38 MAPK phosphorylation markedly attenuated these responses. Our data indicate that EGFR-dependent ERK1/2 activity in keratinocytes takes part to a homeostatic mechanism regulating inflammatory responses, and emphasize the distinct role of MAPKs as potential targets for manipulating inflammation in the skin.
Introduction
The skin epithelium, which is composed mainly of keratinocytes interspersed with dendritic cells, melanocytes, and rare T lymphocytes and monocytes, is highly committed to host defense. Physical, chemical, or immune-specific insults rapidly evoke an epidermal response characterized by the increased expression of a system of proinflammatory mediators, including chemotactic factors, which initiate the orientated migration of distinct leukocyte subpopulations. In turn, activated monocytes, dendritic cells, and T cells release potent cytokines that act on cells in the local environment to boost the inflammatory response (1, 2). In particular, TNF- induces the expression of numerous chemokines, including CCL2/MCP-1, CCL5/RANTES, CCL20/MIP-3, CCL27/cutaneous T cell-attracting chemokine, CXCL8/IL-8, and CXCL10/IFN--inducible protein-10 in keratinocytes. Epidermal CCL2 is involved in the early response to injury or irritants, and in T cell-mediated skin disorders (3, 4, 5, 6), and controls the recruitment of monocytes/macrophages, dendritic cells, and T cells (7, 8, 9). Active immigration of T cells, monocytes, as well as neutrophils is also supported by the increased expression of CCL5 (5, 6). However, type 1 T cells appear massively attracted into the skin by keratinocyte release of the CXCR3 ligands, CXCL10, CXCL9/monokine induced by IFN-, and CXCL11/IFN-induced T cell -chemoattractant (5, 10). CXCR3 ligands appear deeply involved in type 1 T cell-mediated diseases such as allergic contact dermatitis and psoriasis, but are not relevantly expressed during skin response to irritants (5, 11). Finally, CXCL8 is the best characterized of a group of chemoattractants active in neutrophil recruitment as well as in epithelial and endothelial cell proliferation (3, 12). Enhanced CXCL8 expression is characteristically associated with sustained epidermal growth factor (EGF)3 receptor (EGFR) activation, and is considered a secondary amplification mechanism leading to tissue hyperplasia (13).
EGFR governs the homeostatic maintenance and repair of epithelial tissues. Typical responses to EGFR activation include stimulation of keratinocyte proliferation and migration, and controlled differentiation. We recently found that during their early response to TNF- or IFN-, keratinocytes release EGFR ligands including TGF-, which induce EGFR autophosphorylation and as a consequence activate its signal transduction cascade (14). Enhanced EGFR activation was associated with increased CXCL8 expression, and diminished CCL2, CCL5, and CXCL10 expression. In contrast, impairment of EGFR activation led to an opposite pattern. In the mouse, skin application of a selective EGFR tyrosine kinase inhibitor led to more severe contact hypersensitivity (CH) responses, with increased epidermal levels of CCL2, CCL5, and CXCL10, and a higher number of monocytes/macrophages and T lymphocytes in the skin. These findings suggested that EGFR modulates skin inflammation by affecting chemokine expression in keratinocytes.
EGFR activation leads to the persistent induction of the classical MAPK pathway identified as ERK 1 and 2 (ERK1/2) (15), which plays a fundamental role in the EGFR-driven control of epidermal proliferation (16, 17). In contrast, the other major subgroups of the MAPK family, namely p38 and and the JNK/stress-activated protein kinase 1 and 2 (JNK1/2), are weakly activated by EGFR, whereas they are highly stimulated on exposure to TNF- (15, 18).
In this study, we provide experimental evidence that EGFR is involved in the control of chemokine expression in epidermal keratinocytes via an ERK1/2-dependent mechanism. Our data indicate that ERK1/2 takes part in a homeostatic mechanism that regulates epidermal involvement in the inflammatory response, and point to the MAPKs as promising targets for the modulation of skin inflammation.
Materials and Methods
Chemicals and reagents
Actinomycin D, 5,6-dichloro-1-D-ribofuranosylbenzimidazole (DRB), croton oil (CrO), benzalkonium chloride (BAC), and 2,4-dinitrofluorobenzene (DNFB) were obtained from Sigma-Aldrich. Recombinant human TGF- and TNF- were purchased from R&D Systems. The small molecule, cell-permeant protein kinase inhibitors PD168393 (19), PD98059 (15), U0126 (15), SB203580 (15), and SP600125 (20) were from Calbiochem. Table I summarizes their specific target and the concentration range used in this study.
Table I. Inhibitors used in this study
Cell cultures and treatments
Primary cultures of normal human keratinocytes were obtained as previously described (21), and routinely grown in serum-free keratinocyte growth medium (Cambrex). Primary cultures of dermal macrovasculature endothelial cells were grown in endothelial cell growth medium (Cambrex). Both of these culture media include 10 ng/ml EGF. In the 24 h preceding treatments, 80% confluent cell cultures were switched to EGF-depleted medium. Primary cultures of normal human dermal fibroblasts were grown in DMEM (Invitrogen Life Technologies), supplemented with 10% FCS, 1% penicillin/streptomycin, and 1% L-glutamine.
RNA extraction and RNase protection assay (RPA)
Total RNA was extracted from cell cultures and from mouse ear biopsies using the TRIzol reagent (Invitrogen Life Technologies). The templates of the human and mouse chemokines under investigation, the housekeeping molecule L32, and the kit for RPA were purchased from BD Pharmingen. Total RNA (10 μg) was hybridized overnight with -32UTP-labeled cDNA templates, and reactions were performed as per the manufacturer instructions.
mRNA decay assay
Decay rates of CCL2 and CXCL8 transcripts were measured by the use of RNA polymerase II inhibitors. Briefly, transcription was inhibited by the addition of actinomycin D (5 μg/ml) or DRB (25 μg/ml) to fresh culture medium, and total RNA samples were harvested at selected time points thereafter.
ELISA
Chemokines released in the culture medium were measured with dedicated kits from BD Pharmingen.
Immunoprecipitation and Western blot analysis
Keratinocytes were lysed by adding 1 ml per 10-cm dish of lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM PMSF, 1 mM NaF, 1 mM Na3V04, 0.15 U/ml aprotinin, 1% leupeptin, and 1% pepstatin). EGFR was immunoprecipitated using anti-EGFR Ab from Santa Cruz Biotechnology, and phosphorylated EGFR was immunodetected using anti-phosphotyrosine Ab from BD Transduction Laboratories (BD Pharmingen).
In vitro kinase assay
Cell lysates were prepared by applying the same method as described in the Western blotting protocol. Kinase assays for ERK1/2, JNK1/2, and p38 MAPKs were performed using commercial kits from Cell Signaling Technology.
Transient DNA transfection and luciferase reporter assay
The AP-1-dependent reporter plasmid (AP-1-luc) containing seven tandem copies of the AP-1 site and the NF-B-dependent reporter plasmid (NF-B-luc) containing five tandem copies of the NF-B site were purchased from Stratagene. The inducible reporter vector containing full-length CXCL8 promoter (–1481/+44) (IL-8-luc) was a kind gift of N. Mukaida (Cancer Research Institute, Kanazawa, Japan) (22). The CCL2 promoter luciferase construct containing the proximal promoter (between –107 and +60) and distal enhancer region (between –2742 and –2513) (MCP-1-luc) was kindly provided by M. Goebeler (University of Würzburg, Würzburg, Germany) and has been previously described (23). The reporter plasmid containing full-length CCL5 promoter (–974/+1) was generously provided by P. Nelson (University of Munich, Munich, Germany) (24). Keratinocytes were seeded onto 1.9-cm2 wells (2 x 105 cells/well) and transfected when 60% confluent. Transfection was performed over a 6-h period in 1 ml of medium containing 12 μl of lipofectin (Invitrogen Life Technologies), 0.5 μg of the inducible reporter plasmid, plus 0.5 μg of pCMV · SPORT--galactosidase for the control of transfection efficiency. Transfection medium was then removed, and keratinocytes were allowed to recover in fresh medium for 24 h, starved of EGF for further 24 h, and then stimulated. Cultures were finally washed and harvested with 1 ml of lysis buffer (Roche). Luciferase activity was determined (Promega), and normalized to -galactosidase activity and protein concentration.
Mouse models of skin inflammation
Protocols of skin inflammation in mice were approved by the Institute Ethical Committee. To test for irritant contact dermatitis, BALB/c mice (Charles River Laboratories) received topical application on each side of each ear of 10 μl of the following solutions: 0.5% DNFB, 1% CrO, 10% BAC, or vehicle alone (DMSO/absolute ethanol 1/10 v/v) (25). In selected groups, 10 μl of the vehicle or solutions of PD168393, PD98059, or SB203580 was applied on each side of each ear 30 min before the irritant or vehicle alone. To test for CH, BALB/c mice were sensitized by application of 30 μl of 0.5% DNFB in acetone/olive oil (4/1) on a 2-cm2 area of the shaved abdomen. Five days later, sensitized and unsensitized animals received 10 μl of 0.15% DNFB on each side of each ear. In selected groups, the vehicle or a solution of PD168393, PD98059, or SB203580 was painted on sensitized mice 30 min before challenge (14). Ear thickness was measured before application of the chemical and in subsequent time points. Total RNA was extracted from the ears at the time of maximal inflammation, and chemokine expression was analyzed by RPA. In each experimental group, some mice were sacrificed, and the ears were cut and either paraffin embedded for histological analysis (H&E) or snap frozen for immunohistochemistry.
Immunohistochemistry
Cryostatic sections (4 μm) were processed, as previously described (14). The Abs used for the detection of phosphorylated ERK1/2 and p38 MAPK were from Cell Signaling Technology and BioSource International, respectively. For the detection of CCL2, CCL5, and CXCL10, we used dedicated Abs from Santa Cruz Biotechnology. Anti-mouse CD11b (dendritic cells, monocytes, and neutrophils) and Ly-6G (neutrophils) Abs were from BD Pharmingen. Anti-CD3 (T cells) Ab was from DakoCytomation. Secondary biotinylated Abs and staining kits were from Vector Laboratories. As negative controls, primary Abs were omitted or replaced with isotype-matched Ig. Infiltrating cells positive for CD11b, CD3, or Ly-6G were counted on three different sections from the same biopsy with an eyepiece graticule at a magnification of x200 in 10 adjacent fields.
Statistical analysis
The Wilcoxon signed-rank test or Student’s t test (GraphPad Prism software) was applied to compare differences between groups of data. Significance was assumed at a p value of 0.05 or less.
Results
EGFR activation state controls ERK1/2 activity and AP-1 trans activation
The proinflammatory response of cells to TNF- implies the activation of the MAPK pathways (18, 26). Treatment with TNF- alone caused a transient increase of ERK1/2 and p38 MAPK sp. act., and intense JNK1/2 activation at 15 min stimulation (Fig. 1A). TGF- alone induced a robust and persistent activation of ERK1/2, and a modest induction of JNK1/2 and p38 MAPKs. Costimulation with TNF- and TGF- led to sustained ERK1/2 activity and the synergistic potentiation of JNK1/2 and p38 MAPKs. Conversely, the selective EGFR tyrosine kinase inhibitor PD168393 potently interfered with basal and TNF--associated ERK1/2 activation, confirming the crucial role of EGFR signaling in the control of this MAPK subfamily. Strong suppression of ERK1/2 was also induced by two chemically unrelated MAPK kinase 1/2 (MEK1/2) inhibitors, PD98059 (Fig. 1A) and U0126 (data not shown) (15). SB203580 is a selective inhibitor of p38 MAPK isoforms (15), and induced a strong suppression of p38 MAPK-specific kinase function. The residual p38-associated activity (Fig. 1A, lane 18) could be reasonably due to the p38 isoform, which is abundantly expressed in keratinocytes and coimmunoprecipitates during the kinase assay (27). Quite interestingly, the up-regulated JNK1/2 activity when p38 was inhibited by SB203580 (Fig. 1A, lane 18) and the prolonged p38 activity in the presence of the JNK1/2 inhibitor SP600125 (Fig. 1A, lane 22) indicated p38-JNK1/2 cross-regulation also in keratinocytes (28). Finally, no perturbation of either basal or TNF--induced EGFR phosphorylation could be observed in the presence of the MAPK inhibitors. MAPK activation state is critical for AP-1-dependent gene expression. TGF- promoted AP-1 trans activation with a mean 2.6-fold increase over basal levels (Fig. 1B), whereas TNF- alone enhanced AP-1 transcriptional activity very modestly (Fig. 1B). Noteworthily, MAPK selective inhibitors significantly impaired both basal and TNF--associated AP-1-dependent gene expression, indicating that each MAPK subfamily contributes to AP-1 activation and that the abrogation of distinct MAPKs cannot be compensated by the other pathways. NF-B-dependent gene transcription underwent a strong (mean 4.3-fold) increase following TNF- stimulation, but it was insensitive to EGFR signaling promotion (by TGF-) or blockade (by PD168393), or to MEK1/2 inhibition (by PD98059) (Fig. 1C). Interesting enough, the p38 inhibitor SB203580 and, more potently, the JNK1/2 inhibitor SP600125 impaired NF-B trans activation significantly. No evidence of perturbation of IB cytosolic degradation or p50/p65 nuclear translocation was found (data not shown), supporting the emerging concept that p38 or JNK inhibition could impair NF-B trans activation through IB-independent mechanisms (18, 29, 30, 31, 32).
FIGURE 1. EGFR activation state controls ERK1/2 activity and AP-1 trans activation in human keratinocytes. A, MAPK activity was determined by selective immunoprecipitation (ip) of phosphorylated (P) ERK1/2, p38 MAPKs, or JNK1/2, followed by in vitro kinase assay with the appropriate substrate before SDS-PAGE and immunoblot (ib) analysis. Keratinocytes were stimulated with TNF- (50 ng/ml) and/or TGF- (50 ng/ml) directly, or following 30-min preincubation with 2 μM PD168393 (PD16) or 20 μM PD98059 (PD98), SB203580 (SB20), or SP600125 (SP600). AP-1-regulated (B) or NF-B-regulated (C) luciferase reporter plasmid was transiently transfected into keratinocytes, which were preincubated or not with the indicated inhibitor (or the vehicle, i.e., 0.1% DMSO in culture medium) for 30 min, and then stimulated with TNF- and/or TGF- for 6 h. Values are expressed as the mean ± SD (n = 5 per condition). *, p < 0.05 vs untreated control. **, p < 0.05 vs TNF--treated cultures. Data are representative of three independent experiments.
EGFR activation leads to enhanced CXCL8 and reduced CCL2, CCL5, and CXCL10 expression in skin keratinocytes via ERK1/2 activation
EGFR activation differentially affected both constitutive and TNF--induced expression of a set of chemokines, with increase of CXCL8 and down-regulation of CCL2, CCL5, and CXCL10. These effects were dependent on the dose of TGF- associated with TNF-, and confirmed both at the mRNA and protein levels (Fig. 2, A and B). An opposite outcome was observed when EGFR signaling was impaired (Fig. 2, A and C). Changes in chemokine expression by EGFR signaling appeared restricted to the epithelial cells of the skin. Indeed, TNF--driven CCL2 and CXCL8 expression in dermal fibroblasts (Fig. 2E) and dermal microvasculature endothelial cells was not altered by TGF- or PD168393; although both cell types expressed EGFR and EGFR phosphorylation was induced by TGF- or completely prevented by PD168393 (data not shown), TNF--driven CCL2 and CXCL8 expression was not altered by TGF- or PD168393 in both cell types (Fig. 2, D and E). We then checked the direct impact of MEK1/2 inhibition on chemokine expression. Similar to what we observed in the presence of PD168393, both PD98059 and U0126 up-regulated TNF--induced CCL2, CCL5, and CXCL10, while decreased CXCL8 transcript levels (Fig. 3A). In contrast, the p38 inhibitor SB203580 and the JNK1/2 inhibitor SP600125 invariably down-regulated TNF--induced chemokines. These effects were confirmed at the protein level for CCL2 and CXCL8 (as well as CCL5 and CXCL10; data not shown) in unstimulated keratinocytes (Fig. 3, B and D) and in cells treated with TNF- (Fig. 3, C and E).
FIGURE 2. EGFR activation leads to enhanced CXCL8 and down-regulated CCL2, CCL5, and CXCL10 expression in skin keratinocytes. A, Chemokine gene expression was measured by RPA in total RNA (10 μg) extracted from keratinocytes stimulated for 2 h with TNF- (50 ng/ml) and/or TGF-, directly or following 30-min incubation with PD168393 (PD16). ELISA of TNF- induced chemokine levels after 24-h costimulation with escalating doses of TGF- (B) and PD16 (C). Results are expressed as the mean of three independent experiments ± SD. RPA of CCL2 and CXCL8 expression in dermal fibroblasts (D) and dermal microvasculature endothelial cells (E) stimulated for 2 h with TNF- and/or TGF- (50 ng/ml) directly, or following 30-min preincubation with 2 μM PD16.
FIGURE 3. MEK1/2 inhibitors PD98059 and U0126 reproduce the effects of EGFR inactivation on chemokine expression. A, Chemokine transcript levels were detected in total RNA (10 μg) extracted from keratinocytes stimulated for 2 h with TNF- directly or following 30-min preincubation with 2 μM PD168393 (PD16); 1 μM U0126; 20 μM PD98059 (PD98), SB203580 (SB20), or SP600125 (SP600); or vehicle alone (0.1% DMSO in culture medium). ELISA detection of CCL2 (B and C) and CXCL8 (D and E) levels in the supernatant of unstimulated (open symbols) or TNF--stimulated (filled symbols) cultures at 24 h. Results are expressed as the mean of three independent experiments ± SD.
Inhibition of ERK1/2 signaling affects chemokine expression at the transcriptional and/or posttranscriptional levels
Despite the up-regulated CCL2, CCL5, and CXCL10 expression, we could not find evidence of enhanced gene transcription in association with EGFR or MEK1/2 inhibition. In particular, TNF-, but not TGF-, effectively induced CCL2 promoter (MCP-1-luc) trans activation, which was impaired by SB203580 or SP600125, but not by PD168393 or PD98059 (Fig. 4A), strongly paralleling the profile of NF-B trans activation (Fig. 1C) (33). Similar findings were observed when evaluating CCL5 promoter activity (data not shown). By contrast, CXCL8 promoter (IL-8-luc) was induced by TGF- and/or TNF- (Fig. 4B). Importantly, PD168393 and each selective MAPK inhibitor impaired both basal and TNF--induced CXCL8 promoter activity, confirming that the three MAPKs participate in CXCL8 gene transcription (Fig. 4B), and that both AP-1 and NF-B trans activation are involved in this process (Fig. 1, B and C) (34, 35). Because we did not find evidence of chemokine transcriptional superinduction associated with ERK1/2 inhibition, we then investigated the existence of possible posttranscriptional mechanisms. Actinomycin D was added to prevent further transcription, and the decay of chemokine mRNA was followed by RPA. In cells treated with TNF- alone, CCL2 transcript t1/2 was calculated at 37.5 min, and it underwent a striking increase (>4 h) in cells treated with PD168393 or PD98059 (Fig. 4, C and E), hence providing a mechanism underlying CCL2 up-regulated expression. By contrast, TGF- and SB203580 reduced CCL2 mRNA t1/2 to 30 and 12 min, respectively. Hence, the cumulative effect of SB203580 on TNF--driven CCL2 gene expression was its almost complete abrogation, as observed in Fig. 4C (0-min actinomycin D (Act.D)), due to a reduction of CCL2 gene transcription by more than one-third (Fig. 4A) and a 3 times faster rate of CCL2 transcript decay (Fig. 4E). Both the EGFR and MEK1/2 inhibitors prolonged the t1/2 also of TNF--induced CXCL8 transcript, which passed from 50 min to 215 and 170 min, respectively, whereas TGF- and SB203580 reduced CXCL8 t1/2 to 40 and 18 min, respectively (Fig. 4, D and F). Thus, ERK1/2 blockade was associated with a stabilization of CXCL8 transcript, but also with impaired CXCL8 gene transcription, with the latter mechanism prevailing. By contrast, SB203580 reduced CXCL8 gene expression dramatically, as observed in Fig. 4D (0-min Act.D), being simultaneously responsible for a 2-fold reduction of gene transcription (Fig. 4B) and a 2.5-fold increase in the rate of transcript decay (Fig. 4F), in keeping with previous lines of evidence (34, 35). The same results were obtained with RNA polymerase II inhibition by DRB (data not shown).
FIGURE 4. ERK1/2 inhibition affects chemokine expression at the transcriptional and/or posttranscriptional levels. Luciferase activity was measured in keratinocytes transiently transfected with a plasmid in which the reporter expression was under the transcriptional control of the CCL2 promoter (MCP-1-luc; A) or the CXCL8 promoter (IL-8-luc; B). Transfected keratinocytes were stimulated for 6 h with TNF- (50 ng/ml) alone or in association with TGF- directly, or following 30-min preincubation with 2 μM PD168393 (PD16); 20 μM PD98059 (PD98), SB203580 (SB20), or SP600125 (SP600); or vehicle alone (0.1% DMSO in culture medium). Values are expressed as the mean ± SD (n = 4 per condition). *, p < 0.05 vs untreated control. **, p < 0.05 vs TNF--treated cultures. Data are representative of three independent experiments. RPA of the decay of specific CCL2 (C) and CXCL8 (D) transcript. Keratinocytes were stimulated for 2 h with TNF- alone or in association with TGF- directly, or following 30-min preincubation with the indicated inhibitor. Subsequently, the cultures were washed and the mRNA was extracted (0 min), or they were added to Act.D (5 μg/ml) to evaluate mRNA decay. Semilogarithmic plots of CCL2 (E) and CXCL8 (F) mRNA decay. RPA bands were quantified by laser densitometry, and the chemokine signals were normalized to L32 signals. The ratios were plotted as percentages of values at the time of Act.D addition. Results are expressed as the mean of three independent experiments ± SD.
Impairment of ERK1/2 signaling leads to enhanced inflammatory responses in mouse skin
To verify in vivo our in vitro results, we used mouse models of inflammatory response to irritants and of T cell-mediated immune responses. We found that topical administration of 2–4 mM PD168393, 5–10 mM PD98059, or 5–10 mM SB203580 alone had no effect on ear thickness (Fig. 5A), and did not induce any change in normal skin histology (data not shown). By contrast, the same doses of PD168393 and PD98059 applied 30 min before contact with prototypic skin irritants (25), including DNFB, CrO, and BAC, similarly led to aggravation of the inflammatory response, whereas skin pretreatment with SB203580 resulted in a significant reduction of ear thickness (Fig. 5, B–D). RPA performed on vehicle-pretreated tissue showed the presence of MIP2 (the mouse homologue of human CXCL8) and CCL2, and low level CXCL10 after exposure to DNFB, CrO, or BAC (Fig. 5E). Application of PD168393 or PD98059 led to reduced levels of MIP2 and enhanced expression of CCL2 and CXCL10 transcripts compared with vehicle-treated controls. By contrast, all chemokine transcripts were down-regulated in SB203580-pretreated mice. These results exactly matched those observed in cultured human keratinocytes. Histological analysis at the time point of maximal DNFB-induced reaction visualized a more severe inflammatory response in the skin treated with PD168393 or PD98059, and very limited, if any, inflammation in SB203580-treated skin (Fig. 6). Compared with the faint specific staining for phosphorylated ERK1/2 (P-ERK) and p38 (P-p38) observed in the unperturbed mouse skin (data not shown), up-regulation of these species was observed in the skin affected by irritant contact dermatitis. In PD168393- or PD98059-treated mice, immunohistochemistry confirmed the efficient suppression of P-ERK and an increased number of infiltrating CD11b+ cells in the dermis, whereas no relevant differences could be found in the number of CD3+ or Ly-6G+ cells (data not shown) when compared with vehicle-treated controls. Skin treated with SB203580 showed no change in P-ERK expression compared with controls, and suppression of P-p38, as well as very few CD11b+, CD3+, or Ly-6G+ cells (data not shown). Furthermore, skin treatment with PD168393 or PD98059 led to worsened CH, whereas SB203580 induced a marked attenuation of CH reaction (Figs. 7A and 8). This was associated with a corresponding up- or down-regulation of CXCL10 and CCL2 mRNAs (Fig. 7B) and proteins (Fig. 8). MIP2 mRNA was reduced by all three inhibitors, as expected. More intense CCL2 immunostaining in skin treated with PD168393 or PD98059 was not confined to keratinocytes, but also involved cells with dendritic morphology in the epidermis and dermis. Furthermore, we could observe that the number of CD3+ and CD11b+ cells infiltrating these samples was prominently higher than in controls (Fig. 8). Finally, SB203580-treated skin showed a much lower number of infiltrating CD11b+, CD3+, and Ly-6G+ cells.
FIGURE 5. Topical application of PD168393 or PD98059 aggravates, whereas SB203580 reduces irritant contact dermatitis. Change in ear thickness following topical application of vehicle alone (A), 0.5% DNFB (B), 1% CrO (C), or 10% BAC (D). In selected groups, the irritants were painted (10 μl each side of each ear) 30 min after local application of 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), 10 mM SB203580 (SB20), or vehicle alone. Data represent the mean of the mean changes in ear thickness at each time point from three different experiments (n = 5 per treatment group). In B, significance at p < 0.05 at 2- to 48-h time points in mice groups treated with the three inhibitors vs vehicle-treated controls. In C, p < 0.05 vs vehicle-treated controls at 4- to 48-h time points. In D, p < 0.05 vs vehicle-treated controls at 6- to 48-h time points. E, RPA of chemokine expression in total RNA (10 μg) extracted from the whole ears 6 h after DNFB (a), 8 h after 1% CrO (b), or 24 h after 10% BAC (c). Results are representative of three independent experiments.
FIGURE 6. Augmented inflammatory response to DNFB in the skin of nonimmune mice pretreated with PD168393 or PD9059. Histology (H&E) and immunohistochemical detection of P-ERK and P-p38 as well as CD11b+ and CD3+ cells in ear samples collected 6 h after DNFB painting onto the skin pretreated with vehicle alone (10% DMSO in absolute ethanol), 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), or 10 mM SB203580 (SB20). Magnification: H&E, x40; P-ERK and P-p38, x200; CD11b and CD3, x100.
FIGURE 7. Topical application of PD168393 or PD98059 enhances, whereas SB203580 attenuates CH responses. A, Change in ear thickness due to application of DNFB on the ear skin of immunized BALB/c mice 30 min after local painting of vehicle alone (10% DMSO in absolute ethanol), 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), or 10 mM SB203580 (SB20). Data represent the mean of the mean changes in ear thickness at each data point from three different experiments (n = 5 mice per group). In A, significance at p < 0.05 at 24- to 72-h time points in mice groups treated with the three inhibitors vs vehicle-treated controls. B, RPA of chemokine expression in total RNA (10 μg) extracted from the whole ear 24 h after CH elicitation. RPA representative of three independent experiments.
FIGURE 8. Up-regulated chemokine expression and increased numbers of infiltrating CD11b+ and CD3+ cells in the skin treated with PD168393 or PD98059 before CH elicitation. In contrast, SB203580 suppresses CH reaction. Histology (H&E) and immunohistochemical detection of CCL2, CXCL10, and CD11b+ and CD3+ cells in ear samples collected 24 h after DNFB elicitation onto the skin pretreated with vehicle alone (10% DMSO in absolute ethanol), 4 mM PD168393 (PD16), 10 mM PD98059 (PD98), or 10 mM SB203580 (SB20). Magnification: H&E, x40; CCL2 and CXCL10, x200; CD11b and CD3, x100.
Discussion
In the present study, we have investigated the role of the distinct MAPKs in the modulation of chemokine expression in epidermal keratinocytes. The results indicate that ERK1/2 suppresses proinflammatory chemokines and eventually skin inflammatory responses, whereas p38 MAPK promotes chemokine expression and amplificates skin inflammation. In epithelial cells, autocrine EGFR ligands provide the main extracellular signals for high steady state ERK1/2 activity, which in turn drives intracellular proproliferation and prosurvival programs. We found that ERK1/2 activity strictly depended on EGFR signaling in skin keratinocytes, with selective EGFR inhibition suppressing constitutive ERK1/2 and preventing its activation by TGF- or the proinflammatory cytokines TNF- and IFN- (14). Of note, we uncovered the existence of an EGFR-mediated differential regulation of cytokine-driven chemokine expression in these cells, with higher levels of EGFR activation associated with enhanced CXCL8, but suppressed CCL2, CCL5, and CXCL10 expression, whereas opposite events were registered when EGFR function was blocked. This mechanism appeared to be specific of the epithelium. Apart from epidermal keratinocytes, EGFR-mediated regulation of chemokine expression was reproduced in normal human bronchial, prostate, and mammary epithelial cells (data not shown), but not in nonepithelial components of the skin such as dermal fibroblasts and dermal microvascular endothelial cells. In this study, we provided pharmacological evidence that ERK1/2 is implicated in this mechanism. Impairment of ERK1/2 obtained either by EGFR or MEK1/2 inhibition was similarly associated with down-regulated CXCL8 and enhanced TNF--driven CCL2, CCL5, and CXCL10 in skin keratinocytes.
The specific contribution of each MAPK subfamily to gene expression could be coordinated by their reciprocal interplay. In particular, a mutually inhibitory cross talk between ERK1/2 and p38 has been described in numerous cell types, including HeLa cells (36, 37, 38). Hence, we initially checked whether ERK1/2 blockade could lead to enhanced JNK1/2 or p38 activity, possibly supporting a stronger gene transcription in TNF--stimulated keratinocytes. We could not collect any evidence of ERK1/2 involvement in regulatory cross talk mechanisms. By contrast, we found that ERK1/2, JNK1/2, or p38 selective inhibition each impaired AP-1 trans activation, whereas JNK1/2 or p38 inhibition also reduced NF-B trans activation through IB-independent mechanisms. This last finding could help to explain current and previous evidence that JNK1/2 or p38 inhibition decreases the promoter activity of proinflammatory genes known to strictly depend on NF-B trans activation, such as CCL2 (33) and CCL5 (39, 40). Only CXCL8 promoter activity could be impaired by ERK1/2 inhibition, confirming that ERK1/2-driven AP-1 trans activation plays a relevant role in CXCL8 transcription (35). In their whole, these data suggested that the enhanced expression of CCL2, CCL5, and CXCL10 was rather due to posttranscriptional events.
Increased gene expression during inflammatory responses involves not only the transcription of unstable mRNAs, but also the dynamic regulation of their turnover (18). We found that ERK1/2 inhibition was associated with an impressive increase in the stability of CCL2 transcript. ERK1/2 inhibition was also associated with a considerable stabilization of CXCL8 transcript, which, however, could not counteract its suppressive effect on CXCL8 gene transcription, eventually leading to a reduced CXCL8 expression. Most of the studies that analyzed the role of MAPKs on gene transcript turnover have implicated p38 MAPKs or JNKs as relevant contributors to mRNA stability (41, 42). Accordingly, we observed that p38 MAPK inhibition was associated with a reduced stability of both CXCL8 and CCL2 transcripts, hence opposing chemokine expression both at the transcriptional and posttranscriptional level. CCL2 and CXCL8 (but also CXCL10) transcripts possess adenylate/uridylate-rich elements (ARE) in their 3'-untranslated regions. Activation of p38 MAPK pathway was found to specifically stabilize transcripts containing these elements (42). However, CCL5 mRNA does not contain these motifs (41). In the reasonable hypothesis that ERK1/2 inhibition leads to increased stability of ARE-containing (e.g., CCL2, CXCL8, and CXCL10) and also non-ARE-containing transcripts (e.g., CCL5) in keratinocytes, a general p38 MAPK-independent mechanism should be involved. Although the molecular pathway(s) linking ERK1/2 activity to mRNA turnover still remains to be identified, ERK1/2 has been previously associated with mRNA stability. In human neutrophils, enhanced ERK1/2 activity increases GM-CSF mRNA stability, thus contributing to guarantee the presence of this important growth factor (43). By contrast, ERK1/2 activity was found to correlate with reduced type I collagen mRNA stability in skin fibroblasts (44). This latter mechanism may provide the inhibitory signal on extracellular matrix deposition during wound repair and tumor growth. Importantly, these effects could be specifically reversed by the MEK1/2 inhibitor PD98059.
We looked for the possible pathogenic relevance of MAPK-mediated control on epidermal chemokines using mouse models of skin inflammation. Compared with the faint specific staining found in the unperturbed mouse skin, we confirmed that the phosphorylated forms of both ERK1/2 and p38 were up-regulated during skin response to irritant chemicals (45). Noteworthily, the pharmacological suppression of ERK1/2 activation produced by an EGFR or a MEK1/2 inhibitor led to an aggravation of the local inflammatory response, with enhanced expression of CCL2 and more numerous CD11b+ inflammatory cells, which comprise monocytes, dendritic cells, and neutrophils. The observation that the count of Ly-6G+ cells was not increased in the skin treated with the EGFR inhibitor PD168393 or the MEK1/2 inhibitor PD98059 suggests that the higher number of CD11b+ cells was essentially due to monocytes and dendritic cells, the principal targets of CCL2 (7, 8). In keeping with previous reports in humans (11), we could not detect significant CXCL10 up-regulation during the acute response to irritants. This chemokine was, however, abundantly expressed during CH, and underwent a striking up-regulation in the epidermis of skin pretreated with PD168393 or PD98059, thus providing a strong molecular background to the increased recruitment of CD3+ cells observed in these biopsies. These data confirmed the ERK1/2 is involved in the regulation of a number of potent proinflammatory chemokines in epidermal keratinocytes in vivo, and suggest that it is part of a homeostatic mechanism that tends to oppose skin inflammation (Fig. 9). By contrast, the local application of a p38 MAPK inhibitor led to a reduced inflammatory response in terms of resulting tissue perturbation and number of infiltrating inflammatory cells. SB203580 is known to block the kinase activity of p38 MAPK, but not the phosphorylation of p38 MAPK by upstream kinases (15). Hence, we believe that our evidence of a suppressed p38 MAPK phosphorylation is an indirect consequence of SB203580 skin prepainting, deriving from SB203580-induced (15, 18, 46) local abrogation of TNF- and IL-1 expression in keratinocytes. Indeed, these cytokines are the main activators of the signaling pathways leading to p38 MAPK phosphorylation and eventually to increased chemokine expression in the inflamed skin (1, 2). An impairment of their local expression may thus provide the mechanism underlying SB203580-associated decrease of p38 MAPK phosphorylation. In keeping with our in vitro data, the epidermal (and dermal) expression of the chemokines tested was invariably down-regulated during both irritant dermatitis and CH, further indicating that p38 MAPKs may well represent a relevant therapeutic target in some inflammatory skin diseases. Recently, systemic administration of SB202190 to mice was shown to protect the epidermis against the damaging effects of acute UVB radiation (47). Furthermore, local painting of the same inhibitor was reported to inhibit the expression of CH (48). In this study, we confirmed this finding using the homologue SB203580, and extended its anti-inflammatory effect toward a variety of primary skin irritants. Numerous p38 inhibitors are now being developed for human use for the treatment of common inflammatory diseases such as Crohn’s disease, chronic obstructive pulmonary disease, rheumatoid arthritis, and psoriasis, and some of them are now in clinical trials (49, 50). Serious concerns in their systemic use still derive from p38 MAPK involvement in vital processes such as cardiac and platelet function, hemopoiesis, and bone resorption (15, 51). Our data indicate that a topical use may effectively suppress skin inflammatory disorders, possibly with a better systemic safety profile.
FIGURE 9. Schematic representation of cytokine-driven EGFR activation and its effect on chemokine expression in keratinocytes. Through activation of potent transcription factors, the cytokines TNF- and IFN- boost the expression of crucial chemokines such as CCL2, CCL5, and CXCL10 (A). Activation of p38 MAPK has a primary repercussion on this process by enhancing both proinflammatory gene transcription and transcript stability. However, cytokine receptor (R) stimulation also promotes a rapid metalloproteinase (MMP)-mediated TGF- shedding from its membrane precursor (proTGF-) (14 ), leading to activation of the EGFR/ERK1/2 pathway. This pathway drives a down-regulatory effect on the expression of chemokines by a reduction of their transcript stability. When the EGFR/ERK1/2 down-regulatory loop is abrogated either by the EGFR tyrosine kinase inhibitor PD168393 or the MEK1/2 inhibitor PD98059 (B), chemokine expression is deranged. Topical application of these inhibitors predisposes the skin to more severe inflammatory responses.
Our data also suggest that pharmacological abrogation of EGFR/ERK1/2 signaling pathway worsens skin inflammation by increasing chemokine expression in keratinocytes. However, when EGFR superactivation was forced in the suprabasal epidermal layers by transgenic overexpression of 1 integrins, a patchy skin hyperproliferation and inflammation were observed, which were correlated with enhanced ERK1/2 activity and IL-1 overproduction in keratinocytes (52, 53). Also, the transgenic expression of activated MEK1 in suprabasal keratinocytes led to hyperproliferative and inflammatory skin lesions (54). These results apparently contradict our understanding of ERK1/2 as a regulator of skin inflammation, but they derive from complex conditions of extraordinary ERK1/2 up-regulation, and could be interpreted as a further clue of the fine, homeostatic role of this kinase in the skin. Finally, sustained EGFR/ERK1/2 activation has been shown to provide the major pathway fueling abnormal cellular growth in a number of epithelial tumors, and EGFR inhibition represents an effective cancer treatment strategy (55). Our results possibly suggest that the ultimate anticancer activity due to EGFR inhibition may also rely on a stronger recruitment of inflammatory cells and hence a more pronounced antitumor immune response.
Disclosures
The authors have no financial conflict of interest.
Acknowledgments
We thank Domenico Rosi for his technical help in the processing of mouse histological samples, and Maurizio Inzillo for help with the artwork.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by grants from Ministero della Salute and Ministero dell’Istruzione, dell’Università e della Ricerca (Programmi di Ricerca Scientifica di Rilevante Interesse Nazionale).
2 Address correspondence and reprint requests to Dr. Saveria Pastore, Laboratory of Immunology, Istituto Dermopatico dell’Immacolata, Instituto di Ricovero e Cura a Carattere Scientifico, via Monti di Creta 104, 00167 Roma, Italy. E-mail address: pastore{at}idi.it
3 Abbreviations used in this paper: EGF, epidermal growth factor; Act.D, actinomycin D; ARE, adenylate/uridylate-rich element; BAC, benzalkonium chloride; CH, contact hypersensitivity; CrO, croton oil; DNFB, 4-dinitrofluorobenzene; DRB, 5,6-dichloro-1--D-ribofuranosylbenzimidazole; EGFR, EGF receptor; MEK, MAPK kinase; P-ERK, phosphorylated ERK; P-p38, phosphorylated p38; RPA, RNase protection assay.
Received for publication December 2, 2004. Accepted for publication February 3, 2005.
References
Pasparakis, M., G. Courtois, M. Hafner, M. Schmidt-Supprian, A. Nenci, A. Toksoy, M. Krampert, M. Goebeler, R. Gillitzer, A. Israel, et al 2002. TNF-mediated inflammatory skin disease in mice with epidermis-specific deletion of IKK2. Nature 417:861.
Pastore, S., A. Cavani, C. Albanesi, G. Girolomoni. 2005. Chemokines of the skin. J. D. Bos, ed. Skin Immune System 3rd Ed.373. CRC Press, Boca Raton.
Gillitzer, R., M. Goebeler. 2001. Chemokines in cutaneous wound healing. J. Leukocyte Biol. 69:513
Kim, B. H., Y. S. Lee, K. S. Kang. 2003. The mechanism of retinol-induced irritation and its application to anti-irritant development. Toxicol. Lett. 146:65.
Goebeler, M., A. Trautmann, A. Voss, E. V. Br?cker, A. Toksoy, R. Gillitzer. 2001. Differential and sequential expression of multiple chemokines during elicitation of allergic contact hypersensitivity. Am. J. Pathol. 158:431.
Giustizieri, M. L., F. Mascia, A. Frezzolini, O. De Pità, L. M. Chinni, A. Giannetti, G. Girolomoni, S. Pastore. 2001. Keratinocytes from patients with atopic dermatitis and psoriasis show a distinct chemokine production profile in response to T cell-derived cytokines. J. Allergy Clin. Immunol. 107:871.
Nakamura, K., I. W. Williams, T. S. Kupper. 1995. Keratinocyte-derived chemoattractant protein-1 (MCP-1): analysis in a transgenic model demonstrates that MCP-1 can recruit dendritic and Langerhans cells to the skin. J. Invest. Dermatol. 105:635.
Lu, B., B. J. Rutledge, L. Gu, J. Fiorillo, N. W. Lukacs, S. L. Kunkel, R. North, C. Gerard, B. J. Rollins. 1998. Abnormalities in monocyte recruitment and cytokine expression in monocyte chemoattractant protein 1-deficient mice. J. Exp. Med. 187:601.
Carr, M. W., S. J. Roth, E. Luther, S. S. Rose, T. A. Springer. 1994. Monocyte chemoattractant protein 1 acts as a T-lymphocyte chemoattractant. Proc. Natl. Acad. Sci. USA 91:3652.
Rottman, J. B., T. L. Smith, K. G. Ganley, T. Kikuchi, J. G. Krueger. 2001. Potential role of the chemokine receptors CXCR3, CCR4, and the integrin E7 in the pathogenesis of psoriasis vulgaris. Lab. Invest. 8:335.
Flier, J., D. M. Boorsma, D. P. Bruynzeel, P. J. van Beek, T. J. Stoof, R. J. Scheper, R. Willemze, T. P. Tensen. 1999. The CXCR3 activating chemokines IP-10, Mig, and IP-9 are expressed in allergic but not in irritant patch test reactions. J. Invest. Dermatol. 113:574.
Addison, C. L., T. O. Daniel, M. D. Burdick, H. Liu, J. E. Ehlert, Y. Y. Xue, L. Buechi, A. Walz, A. Richmond, R. Strieter. 2000. The CXC chemokine receptor 2, CXCR2, is the putative receptor for ELR+ CXC chemokine-induced angiogenic activity. J. Immunol. 165:5269.
Reich, K., C. Garbe, V. Blaschke, C. Maurer, P. Middel, G. Westphal, U. Lippert, C. Neumann. 2001. Response of psoriasis to interleukin-10 is associated with suppression of cutaneous type 1 inflammation, down-regulation of the epidermal interleukin-8/CXCR2 pathway and normalization of keratinocyte maturation. J. Invest. Dermatol. 116:319.
Mascia, F., V. Mariani, G. Girolomoni, S. Pastore. 2003. Blockade of the EGF receptor induces a deranged chemokine expression in keratinocytes leading to enhanced skin inflammation. Am. J. Pathol. 163:303.
Pearson, G., F. Robinson, T. Beers Gibson, B. E. Xu, M. Karandikar, K. Berman, M. H. Cobb. 2001. Mitogen-activated protein (MAP) kinase pathways: regulation and physiological functions. Endocr. Rev. 22:153.
Iordanov, M. S., R. J. Choi, O. P. Ryabinina, T. H. Dinh, R. K. Bright, B. E. Magun. 2002. The UV (ribotoxic) stress response of human keratinocytes involves the unexpected uncoupling of the Ras-extracellular signal-regulated kinase signaling cascade from the activated epidermal growth factor receptor. Mol. Cell. Biol. 22:5380.
Cai, T., K. Nishida, T. Hirano, P. A. Khavari. 2002. Gab1 and SHP-2 promote Ras/MAPK regulation in epidermal growth and differentiation. J. Cell Biol. 159:103.
Wajant, H., K. Pfizenmaier, P. Scheurich. 2003. Tumor necrosis factor signaling. Cell Death Differ. 10:45.
Fry, D. W., A. J. Bridges, W. A. Denny, A. Doherty, K. D. Greis, J. L. Hicks, K. E. Hook, P. R. Keller, W. Leopold, J. A. Loo, et al 1998. Specific, irreversible inactivation of the epidermal growth factor receptor and erbB2, by a new class of tyrosine kinase inhibitors. Proc. Natl. Acad. Sci. USA 95:12022.
Bennett, B. L., D. T. Sasaki, B. W. Murray, E. C. O’Leary, S. T. Sakata, W. Xu, J. C. Leisten, A. Motiwala, S. Pierce, Y. Satoh, et al 2001. SP600125, an anthrapyrazolone inhibitor of Jun N-terminal kinase. Proc. Natl. Acad. Sci. USA 98:13681
Pastore, S., E. Fanales-Belasio, C. Albanesi, L. M. Chinni, A. Giannetti, G. Girolomoni. 1997. Granulocyte/Macrophage colony-stimulating factor is overproduced by keratinocytes in atopic dermatitis. J. Clin. Invest. 99:3009.
Smith, R. S., E. R. Fedyk, T. A. Springer, N. Mukaida, B. H. Iglewski, R. P. Phipps. 2001. IL-8 production in human lung fibroblasts and epithelial cells activated by the Pseudomonas autoinducer N-3-oxododecanoyl homoserine lactone is transcriptionally regulated by NF-B and activator protein-1. J. Immunol. 167:366.
Ueda, A., Y. Ishigatsubo, T. Okubo, T. Yoshimura. 1997. Transcriptional regulation of the human monocyte chemoattractant protein-1 gene. J. Biol. Chem. 272:31092.
Nelson, P. J., H. T. Kim, W. C. Manning, T. J. Goralski, A. M. Krensky. 1993. Genomic organization and transcriptional regulation of the RANTES chemokine gene. J. Immunol. 151:2601.
Mizumoto, N., T. Kumamoto, S. C. Robson, J. Sevigny, H. Matsue, K. Enjyoji, A. Takashima. 2002. CD39 is the dominant Langerhans cell-associated ecto-NTPase: modulatory roles in inflammation and immune response. Nat. Med. 8:358.
Yang, S. H., A. D. Sharrocks, A. J. Whitmarsh. 2003. Transcriptional regulation by the MAP kinase signaling cascades. Gene 320:3.
Xiao, Y. Q., K. Malcolm, G. S. Worthen, S. Gardai, W. P. Schiemann, V. A. Fadok, D. L. Bratton, P. M. Henson. 2002. Cross-talk between ERK and p38 MAPK mediates selective suppression of pro-inflammatory cytokines by transforming growth factor-. J. Biol. Chem. 277:14884
Lee, A. H., J. H. Hong, Y. S. Seo. 2000. Tumor necrosis factor- and interferon- synergistically activate the RANTES promoter through nuclear factor B and interferon regulatory factor (IRF-1) transcription factors. Biochem. J. 350:131.
Kujime, K., S. Hashimoto, Y. Gon, K. Shimizu, T. Horie. 2000. p38 mitogen-activated kinase and c-Jun-N-terminal kinase regulate RANTES production by influenza virus-infected human bronchial epithelial cells. J. Immunol. 164:3222.
Clark, A. R., J. L. E. Dean, J. Saklatvala. 2003. Post-transcriptional regulation of gene expression by mitogen-activated protein kinase p38. FEBS Lett. 546:37
Tchen, C. R., M. Brook, J. Saklatvala, A. R. Clark. 2004. The stability of tristetraproline mRNA is regulated by mitogen-activated protein kinase p38 and by tristetraproline itself. J. Biol. Chem. 279:32393.
Esnault, S., J. S. Malter. 2002. Extracellular signal-regulated kinase mediates granulocyte-macrophage colony-stimulating factor messenger RNA stabilization in tumor necrosis factor- plus fibronectin-activated peripheral blood eosinophils. Blood 99:4048.
Hobbs, R. M., V. Silva-Vargas, R. Groves, F. M. Watt. 2004. Expression of activated MEK1 in differentiating epidermal cells is sufficient to generate hyperproliferative and inflammatory skin lesions. J. Invest. Dermatol. 123:503
Huang, S., E. A. Armstrong, S. Benavente, P. Chinnaiyan, P. M. Harari. 2004. Dual-agent molecular targeting of the epidermal growth factor receptor (EGFR): combining anti-EGFR antibody with tyrosine kinase inhibitor. Cancer Res. 64:5355.(Saveria Pastore, Francesc)