p38 Mitogen-Activated Protein Kinase Is Activated and Linked to TNF-α Signaling in Inflammatory Bowel Disease1

Inflammatory bowel diseases (IBD)—Crohn’s disease and ulcerative colitis—are relapsing chronic inflammatory disorders which involve genetic, immunological, and environmental factors. The regulation of TNF-α, a key mediator in the inflammatory process in IBD, is interconnected with mitogen-activated protein kinase pathways. The aim of this study was to characterize the activity and expression of the four p38 subtypes (p38α–δ), c-Jun N-terminal kinases (JNKs), and the extracellular signal-regulated kinases (ERK)1/2 in the inflamed intestinal mucosa. Western blot analysis revealed that p38α, JNKs, and ERK1/2 were significantly activated in IBD, with p38α showing the most pronounced increase in kinase activity. Protein expression of p38 and JNK was only moderately altered in IBD patients compared with normal controls, whereas ERK1/2 protein was significantly down-regulated. Immunohistochemical analysis of inflamed mucosal biopsies localized the main expression of p38α to lamina propria macrophages and neutrophils. ELISA screening of the supernatants of Crohn’s disease mucosal biopsy cultures showed that incubation with the p38 inhibitor SB 203580 significantly reduced secretion of TNF-α. In vivo inhibition of TNF-α by a single infusion of anti-TNF-α Ab (infliximab) resulted in a highly significant transient increase of p38α activity during the first 48 h after infusion. A significant infliximab-dependent p38α activation was also observed in THP-1 myelomonocytic cells. In human monocytes, infliximab enhanced TNF-α gene expression, which could be inhibited by SB 203580. In conclusion, p38α signaling is involved in the pathophysiology of IBD.

I nflammatory bowel diseases (IBD) 3 -Crohn's disease (CD) and ulcerative colitis (UC)-are disorders of unknown etiology characterized by chronic relapsing inflammation of the gastrointestinal tract. CD and UC are multifactorial diseases caused by the interplay of genetic, environmental, and immunological factors. It is assumed that a pathological, Ag-driven inflammatory response within a genetically susceptible individual is triggered either by an unrecognized pathogen or by nonpathogenic bowel flora (1). The discrimination between CD and UC is based on clinical, endoscopic, radiological, and histopathological features. Both CD and UC are characterized by an imbalance between pro-and anti-inflammatory cytokines (2). Activated macrophages participate in the mucosal immune response, e.g., by producing proinflammatory cytokines such as TNF-␣, IL-1␤, and the chemokine IL-8 (3)(4)(5).
TNF-␣ plays a central role in mucosal inflammation and is likely to be at the apex of the inflammatory cascade in CD (3, 6 -8). The systemic inhibition of soluble TNF-␣ by a single infu-sion of a chimeric anti-TNF-␣ mAb of IgG1 isotype (infliximab) induced remission in up to 50% of CD patients and significantly improved clinical symptoms in most patients after only a short time (9,10). Clinical responses after a single infusion of infliximab vary in duration (9,10). In some patients, a clinical benefit of a single infusion was seen for as long as 1 year, suggesting that the underlying immunological patterns may be altered beyond the immediate effect of TNF-␣ removal (11). This view is supported by the relationship between mucosal production of inflammatory signaling molecules in remission and clinical relapses (7,12). However, it appears that the increased availability of TNF-␣ and other proinflammatory cytokines is not the primary cause of mucosal inflammation in IBD (13). Two recent studies showed that infliximab induces apoptosis in circulating monocytes (14) as well as in lamina propria T cells of CD patients (15).
Mitogen-activated protein kinases (MAPKs) are conserved among all eukaryotes and participate in multiple cellular processes (16). Four groups of MAPKs have been identified in mammalian cells: the extracellular signal-regulated kinases (ERKs), the c-Jun N-terminal kinases (JNKs) or stress-activated protein kinases (SAPKs), the p38 kinases, and ERK5/big MAPK (16,17). All MAPK cascades cooperate in the orchestration of inflammatory responses, and extensive cross-talk to other inflammatory pathways, such as NF-B and Janus kinase/STAT signaling, has been described (18,19). TNF-␣ is one of the best-characterized agonists of the p38 and JNK pathways and is itself regulated by p38 and JNKs (18,20,21). Other proinflammatory cytokines, like IL-16, which is up-regulated in IBD (22), also activate JNKs and p38 (23). The genes of p38␣ and ERK1 are localized in major IBD susceptibility regions on chromosomes 6 (13) and 16 (24), respectively. In a recent pilot study, the guanylhydrazone JNK/p38 inhibitor CNI-1493 strongly reduced clinical disease activity in CD patients (25). However, no systematic evaluation of the expression, activity, or signal transduction of MAPKs in IBD has been published so far.
The present study focused on the activity and expression of the four p38 subtypes (p38␣-␦) in the inflamed mucosa of CD and UC patients in comparison with healthy normal controls. In addition, JNKs and ERK1/2 (p44/42 MAPK) were investigated. p38␣ showed the most substantial activation in the inflamed mucosa of both UC and CD patients; its activity and localization were further analyzed by in vitro kinase assays and immunohistochemistry, respectively. The role of p38␣ in the TNF-␣ signaling regulation loop was investigated in CD patients by assessing TNF-␣ secretion from inflamed mucosal tissue after in vitro treatment with the p38␣␤ inhibitor SB 203580 and by monitoring p38␣ activity after administration of infliximab in patients, human monocytes, and different cell lines.

Patients
Twenty-seven patients with colonic or ileocolonic CD, 16 patients with UC, and 17 age-and sex-matched normal control patients (without signs of pathology; endoscopy mainly for the exclusion of carcinoma) were included in the study (total n ϭ 60; Table I). All IBD patients attended the outpatient clinic of the First Department of Medicine of the Christian-Albrechts-University (Kiel, Germany) because of increased clinical activity. IBD patients included in this study met several requirements: definite diagnosis of either CD or UC along established criteria (26,27), clinical activity (CD activity index Ͼ 150 (28) or clinical activity index for UC Ͼ 4 (29)), moderate to high inflammatory activity confirmed by endoscopy and histology, and exclusion of other diseases (especially irritable bowel syndrome and infectious colitis). None of the patients was treated with cytotoxic drugs or antibiotics. Patients received either no medication, aminosalicylates, or glucocorticoids (Table I). Patients were recruited consecutively along these inclusion and exclusion criteria.
To investigate the influence of in vivo TNF-␣ inhibition on p38␣, five additional CD patients were chosen from an infliximab study population described previously (10). These patients showed a steroid refractory, chronic active rectosigmoidal inflammatory manifestation (CD activity index Ͼ 200) and received a single infusion of infliximab, a humanized anti-TNF-␣ mAb. Responders were defined as patients who did not relapse during 4 wk after a single infusion with infliximab, short-responders relapsed between wk 1 and 4, and nonresponders showed no remission of disease at all. The patients included in the present study were two representative responders (patients 43 and 44), two nonresponders (patients 40 and 41), and one short-responder (patient 42). Written informed consent was obtained from all patients at least 24 h before the procedure, and the project was granted prior approval by the institutional review board.

Samples
From each patient, at least eight colonic biopsies were taken from the same inflamed or noninflamed region. In addition, two biopsies were paraffinembedded and used for histological examination. In eight CD patients (patients 18 -20, 24, 30, 32, 34, and 38) and four UC patients (patients 45, 46, 48, and 58), several sets of biopsy specimens from the same patient including inflamed and noninflamed areas of the colonic mucosa were ex-amined to establish the amount of variation within the samples and the influence of inflammatory activity. A biopsy was attributed to inflamed areas if the macroscopic appearance was confirmed by inflammatory infiltrates in the histological examination. In three additional CD patients (patients 27-29), 24 -30 biopsies from the same inflamed region were taken for biopsy culture experiments (as described below).

Processing of mucosal biopsies for Western blotting and RT-PCR
Biopsy samples were snap-frozen in liquid nitrogen at the time of removal. After mechanical homogenization in liquid nitrogen, specimens were processed for either protein or RNA extraction. Protein extracts were prepared by lysing the tissue homogenates for 5 min in boiling denaturing extraction buffer containing 1% SDS, 10 mM Tris (pH 7.4), and 1% phosphatase inhibitor mixture II (Sigma-Aldrich, St. Louis, MO). After sonication (twice for 5 s), insoluble material was removed by centrifugation for 15 min at 16,000 ϫ g at 4°C. Protein extracts were snap-frozen in liquid nitrogen and stored at Ϫ80°C.
RNA extractions were performed using the RNeasy Mini kit (Qiagen, Hilden, Germany) according to the manufacturer's recommendations. The sample obtained was quantitated by absorbance at 260 nm. RNA integrity was assessed by electrophoresis on a 1% formamide gel, and the absence of genomic DNA contamination was confirmed by PCR for ␤-actin.

Isolation and stimulation of human monocytes
Human monocytes were isolated from 100 ml of blood drawn from three healthy volunteers (two male and one female; age range, 24 -28 years). We used a two-step density centrifugation protocol according to the respective manufacturer's recommendations. After separation with Ficoll-Paque Plus (Amersham Pharmacia Biotech, Piscataway, NJ), mononuclear cells were collected from the interphase and washed in PBS. Monocytes were separated from lymphocytes by resuspension and subsequent centrifugation with isotonic Percoll (density, 1.065 g/cm; Biochrom, Berlin, Germany). After two washing steps in PBS, the monocytes were suspended in monocyte medium (DMEM (Invitrogen, Carlsbad, CA) supplemented with 10% FCS and 1% penicillin/streptomycin (Biochrom)). The cell suspension was adjusted to 1 ϫ 10 6 cells/ml and plated on six-well plates (Falcon; Applied Scientific, San Francisco, CA). Monocytes were further enriched by 90min adherence to the culture plates and washed twice in PBS. Enriched monocytes were allowed to rest for several hours and were subsequently incubated with infliximab (Remicade; 5 g/ml; Centocor, Malvern, PA), a nonspecific human IgG1 mixture from myeloma patients (5 g/ml; Calbiochem, La Jolla, CA), and/or SB 203580 (SB 203580 hydrochloride; Calbiochem) in a concentration of 1 or 10 M. All culture reagents had endotoxin levels of Ͻ 0.01 ng/ml LPS. Viability of the monocytes was Ͼ95% as determined by trypan blue exclusion and purity was at least 85% as assessed by May-Grünwald/Giemsa staining of cytospins (Merck, Darmstadt, Germany). After stimulation, RNA was extracted using the RNeasy Mini kit as described above.

Stimulation of cell lines and preparation of cell lysates
Human THP-1 myelomonocytes (30) and Jurkat T cells (31) were purchased from the American Type Culture Collection (Manassas, VA) and grown according to the supplier's instructions. RPMI 8226 cells (32) were obtained from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). For stimulation experiments, cells below passage number 20 were plated at 1-1.5 ϫ 10 5 /ml in medium, were allowed to rest and grow for 24 h, and were subsequently incubated with infliximab (see above; 5 g/ml), nonspecific IgG1 (see above; 5 or 10 g/ml), or recombinant human TNF-␣ (2.5 or 5 ng/ml; R&D Systems, Minneapolis, MN). After the stimulation period, cell lysates for Western blotting or in vitro p38 MAPK kinase assays were prepared as described for biopsy homogenates.
Following transfer, membranes were blocked and incubated overnight with primary Ab according to the respective manufacturer's recommendations. Primary Abs were as follows: p38 MAPK, phospho-p38 MAPK, p44/42 MAPK, phospho-p44/42 MAPK, SAPK/JNK, phospho-SAPK/JNK (both monoclonal and polyclonal), activating transcription factor-2 (ATF-2), phospho-ATF-2, heat shock protein (Hsp)27, phospho-Hsp27, and poly-(ADP-ribose) polymerase (PARP), all from Cell Signaling Technology; clones E-20 and C-16 for p38␤, clones N-17 and C-19 for p38␦, from Santa Cruz Biotechnology (Santa Cruz, CA); ERK2 and p38␥ from Upstate Biotechnology (Lake Placid, NY); anti-ACTIVE JNK from Promega (Madison, WI ); p38␣ from Zymed Laboratories (San Francisco, CA); and ␤-actin from Sigma-Aldrich. After being washed in TBST (three times for 5 min), membranes were incubated for 30 min with a HRP-conjugated secondary Ab (anti-rabbit (Cell Signaling Technology), anti-goat (Sigma-Aldrich), anti-sheep (Sigma-Aldrich), or anti-mouse (Amersham Pharmacia Biotech), respectively) diluted in blocking buffer. Membranes were subsequently washed, incubated with ECL-Plus Detection Reagent, and exposed to Hyperfilm ECL (both from Amersham Pharmacia Biotech). Between the stainings with phosphospecific Abs, kinase or target Abs, and ␤-actin Ab, blots were stripped in 2% SDS, 62.5 mM Tris, and 100 mM 2-ME for 30 min at 50°C, washed, and blocked again. All measurements of dual-phosphorylated kinase levels and kinase protein expression were normalized by hybridization with Abs against total kinase protein and the housekeeping protein ␤-actin, respectively. Background values from each lane were subtracted to normalize every measurement. The bands were quantified using the densitometry program SigmaGel (Jandel Scientific, San Rafael, CA). All Western blots were exposed to film for varying lengths of time, and only films generating subsaturating levels of intensity were selected for densitometrical and statistical evaluation. Linearity was assured in independent experiments by using different amounts of material and multiple film exposures (data not shown). Each Western blotting experiment was conducted with two separate membranes in parallel to detect potential stripping artifacts.

In vitro p38 MAPK assay
Kinase activity of p38␣ in biopsies and cell lines was determined using a p38 MAPK assay kit (Cell Signaling Technology). The frozen biopsies were homogenized as described previously for Western blotting, but lysed in the provided lysis buffer and processed according to the manufacturer's protocol. The active (i.e., dual-phosphorylated) form of p38␣ was selectively immunoprecipitated, and kinase reactions were conducted with an ATF-2 fusion protein whose Thr 71 -phosphorylated form was detected by Western blotting.

RT-PCR
For the investigation of MAPK mRNA expression in IBD patients, cDNA was synthesized from 500 ng of total RNA from five CD patients (patients 23, 25, 26, 32, and 33), five UC patients (patients 49 -52 and 60), and five normal controls (patients 13-17) using the Advantage RT-for-PCR kit with oligo(dT) primers (Clontech Laboratories, Palo Alto, CA) according to the manufacturer's protocol. Resulting cDNA (5 l of reverse transcriptase reaction) was amplified using GeneAmp PCR buffer and AmpliTaq DNA Polymerase (both from PerkinElmer, Branchburg, NJ) in a 50-l reaction volume containing primer pairs (0.2 M/primer) and 0.2 mM dNTPs (Amersham Pharmacia Biotech). Negative reverse transcriptase and PCR controls included reactions in the absence of RNA sample, reverse transcriptase, and cDNA, respectively. Kinase mRNA expression was normalized by amplifying and analyzing ␤-actin mRNA under the same reaction conditions. All MAPK primer pairs were designed to selectively amplify specific isoforms, and they were optimized to suit the following PCR program: one hold at 94°C for 5 min; 25-35 cycles at 94°C for 30 s, 60°C for 20 s, and 72°C for 30 s; and one hold of 72°C for 7 min. Resulting amplicons were resolved on 2% agarose gels stained with ethidium bromide and visualized through a UV light digital imaging system. Images were evaluated using a densitometrical software (SigmaGel). For each mRNA, the number of cycles directly above detection level (linear phase) was determined and used for evaluation. The efficiency of the primers was confirmed by tests using cDNA from the human monocytic cell line THP-1. The same protocol was used for assessing TNF-␣ mRNA expression in human monocytes and THP-1 cells after stimulation with infliximab.

Immunohistochemical studies
Biopsies were embedded in cryomatrix and snap-frozen in liquid nitrogen. Cryostat sections (7 m) were thaw-mounted onto Superfrost slides (Erie, Portsmouth, NH), post-fixed for 5 min in acetone, air-dried, and stored at Ϫ20°C. Two slides of each biopsy were stained with H&E for routine histological evaluation. The other slides were permeabilized by incubation with 0.1% Triton X-100 in 0.1 M PBS, washed three times in PBS, and blocked with 0.75% BSA in PBS for 20 min. Sections were subsequently incubated with the primary Abs (p38 MAPK (1/200), see Western blot analysis; CD68 (Ki-M6; 1/500) from Dianova (Hamburg, Germany); CD4 (1/500) from BD PharMingen (San Diego, CA); eosinophil peroxidase (1/ 1000) from Biotrend (Köln, Germany); pan-human neutrophilic peptide (HNP; 1/1000) from Novocastra (Newcastle-upon-Tyne, U.K.)) diluted in 0.75% BSA for 1 h at room temperature. After washing in PBS, tissuebound Ab was detected using biotinylated goat-anti rabbit IgG Abs (Vector Laboratories, Burlingame, CA) followed by FITC-conjugated avidin, both diluted at 1/100 in PBS, and/or Cy3-conjugated rabbit-anti mouse (Jackson ImmunoResearch Laboratories, West Grove, PA) at 1/350 in PBS. Controls were included using normal rabbit or mouse sera (Jackson ImmunoResearch Laboratories) as irrelevant primary Abs as well as omitting the primary Abs using only secondary Abs and/or FITC-conjugated avidin. Fluorescence was detected by an Axiophot microscope (Zeiss, Jena, Germany) with appropriate filter systems, and pictures were taken using a digital camera system (Axiocam; Zeiss).

Biopsy cultures and ELISA
To assess the release of TNF-␣ after inhibition of p38␣ by SB 203580, the supernatants of short-term (4-h) cultures of biopsy specimens from three representative CD patients (two untreated, one aminosalicylate-treated; patients [27][28][29] were screened for soluble TNF-␣ by standard ELISA. Immediately after removal, the biopsies were placed in ice-cold wash medium (RPMI 1640 supplemented with 10% FCS; both from Biochrom) and immediately transferred to the laboratory. After being washed 10 times in wash medium, single specimens were placed in 500 l of culture medium (wash medium supplemented with 1% penicillin/streptomycin; Biochrom) in the wells of a sterile 48-well cell culture plate. Half the biopsies were incubated in culture medium with 10 M SB 203580 (SB 203580 hydrochloride; Calbiochem). After 4 h at 37°C and 5% CO 2 , supernatants and biopsies were snap-frozen in liquid nitrogen and stored at Ϫ80°C until analysis. TNF-␣ levels in the supernatants were determined by standard TNF-␣-ELISA (capture Ab MAB610 and detection Ab BAF210; R&D Systems). The biotinylated detection Ab was coupled to extravidin-peroxidase (Sigma-Aldrich), and immune complexes were detected by incubation with o-phenylendiamine-dihydrochloride and H 2 O 2 (both from Sigma-Aldrich) according to the manufacturer's instructions. After stopping the enzyme reaction by addition of 1 M HCl, the OD 490 was measured in an ELISA reader (Milenia/DPC, Los Angeles, CA).

Statistical analysis and replication rate
The normality of the data was checked by calculating Lilliefors probabilities based on the Kolmogorov-Smirnov test. Statistical significance of the non-normally distributed patient data was tested using the Mann-Whitney U test, the obtained p values were corrected for ties, and results were expressed as medians (quartiles). Multiple testing corrections were performed using the Bonferroni method. Measurements for each kinase were conducted three to seven times per patient, resulting in an overall average replication rate of four independent experiments per patient. Both the ELISA and cell culture data followed a normal distribution; their significance was determined by the t test for dependent (ELISA) or independent (Western blots from cell extracts) samples, and the respective results were displayed as means Ϯ SD.

Detection of active MAPKs
For activation, both a threonine and a tyrosine residue in a characteristic Thr-X-Tyr motif must be phosphorylated in MAPKs (16). Therefore, Abs specific for the dual-phosphorylated (i.e., active) forms of the MAPKs were used in Western blot analyses to determine kinase activities in the colonic mucosal biopsies.
While the activities of p38␣ and JNK1/2 were similar in IBD patients regardless of disease and treatment (data not shown), significant differences in ERK1/2 activity were observed between un-treated and glucocorticoid-treated patients in CD (Fig. 2). This effect was not due to a difference in glucocorticoid dosage, as the median glucocorticoid dose in the CD patients was 15 mg/day compared with 20 mg/day in the UC patients (Table I).  p38 kinase activity p38␣ activation was not dependent on disease category or treatment (see above), but only on the presence and severity of inflammation, as indicated by the results of both in vitro kinase assays (Fig. 3) and Western blots for dual-phosphorylated p38 (Fig. 4A). Virtually all p38 activity could be attributed to the p38␣ isoform, because p38␤ and p38␥ proteins were not found in significant amounts (see below) and phosphorylated p38␦ was generally not detectable (Fig. 4A). The variance of p38␣ activity within both the normal controls and the patients was considerable, ranging from moderate to very high activities in IBD patients and from almost none to moderate activities in normal controls, which may reflect the heterogeneity of individuals (Fig. 4A).

Expression of MAPKs
The protein expression of p38␣ displayed no significant differences between patients with IBD and normal controls (Fig. 4B), whereas the amount of p38␦ protein showed a tendency toward a lower level in the inflamed mucosa of both CD and UC patients (NS; Fig. 4C). In general, differences in protein expression were (similar to MAPK activation) more pronounced in inflamed than in noninflamed mucosal specimens of the same patient (Fig. 4C). p38␤ and p38␥ proteins were below detection level, even when 30 g of total protein (instead of 10 g) were separated on each lane of the polyacrylamide gel.
Similar to p38␦, JNK protein was tendentially diminished in IBD without significant differences to normal controls (data not shown). In contrast to the nonsignificant differences in p38 and JNK protein contents, ERK1/2 showed a significantly lower level of protein expression in all IBD patient groups in comparison to normal controls (CD: 39% reduction, p Ͻ 0.001; UC: 48% reduction, p Ͻ 0.0001; Fig. 5A). Similar to the phosphorylation pattern, no differences between ERK1 and ERK2 were observed in the regulation of protein expression.
Linear phase RT-PCR assessment of MAPK transcripts in IBD patients and controls showed no differences in mRNA expression (Fig. 5B). The amount of p38␣ and p38␦ mRNA was similar: in both cases, 30 cycles of the described PCR program were sufficient to produce evaluable amounts of PCR products. The low protein expression of p38␤ and p38␥ was mirrored on the mRNA level: while 35 cycles were barely sufficient to produce detectable quantities of p38␤ amplicons, p38␥ could not be detected even with 40 cycles. In control experiments, transcripts of all p38 isoforms were detectable in the human monocytic cell line THP-1 (data not shown). JNK1, JNK2, ERK1, and ERK2 were transcribed similarly (30 -32 cycles).

Localization of p38␣ expression
As described above, the protein expression of p38␣ showed no significant differences between IBD patients and normal controls. p38␣ protein in the inflamed lamina propria mainly colocalized with CD68 (Ki-M6) specific for monocytes/macrophages and with HNP specific for neutrophils. Double-stained lamina propria macrophages were frequently observed near the epithelial lining of eroded crypts. Fig. 6 shows representative results obtained from FIGURE 3. Kinase activity of p38␣ in vitro. After selective immunoprecipitation of the dual-phosphorylated form of p38␣, kinase reactions were conducted with an ATF-2 fusion protein whose Thr 71 -phosphorylated form (P-ATF-2) was detected by Western blotting. p38␣ kinase activity was significantly higher in IBD patients than in normal controls and depended on the severity of inflammation (H, high; M, moderate; L, low). The experiment shown was repeated twice with similar results.

FIGURE 4.
Western blot analysis of dual-phosphorylated p38, p38␣ protein, and p38␦ protein in denatured extracts from colonic mucosal biopsy specimens. The membrane was successively probed with phosphospecific p38 (Thr 180 /Tyr 182 ) Ab (A), p38␣ Ab (B), ␤-actin Ab, and p38␦ Ab (C). Every experiment was conducted three times with two separate membranes in parallel to detect potential stripping artifacts. Only phospho-p38␣ (P-p38␣) was detected (A), as p38␤ and p38␥ protein expression was generally below detection level, and no phospho-p38␦ (P-p38␦) signal was found at the height of the p38␦ band (indicated by square brackets and a dashed arrow in A). Comparison between inflamed (i) and noninflamed (n) biopsy samples of the same patient (marked by a horizontal line above the lanes) showed that the significant increase in p38␣ activity and the tendential decrease in p38␦ protein expression were related only to the inflammatory activity in the mucosa (A and C) and not to the treatment with glucocorticoids or aminosalicylates. p38␣ expression displayed no significant differences (B). ASA, Aminosalicylate treated; GC, glucocorticoid treated. one of four identical experiments conducted with biopsies from four normal controls, three CD patients, and three UC patients. In further tests, stainings with Abs against eosinophil peroxidase and CD4 ϩ lymphocytes were performed, but none of them showed significant colocalization with the p38␣ signal (data not shown).

Inhibition of p38␣ in cultured biopsies from CD patients
From three representative CD patients (patients [27][28][29], whole colonic mucosal biopsies were cultured for 4 h and incubated with the p38␣␤ inhibitor SB 203580 at a concentration of 10 M to ensure both sufficient inhibitor concentrations within the tissue and specificity of inhibition (34,35). Concentrations of 5-20 M showed a linear relationship between the dose of SB 203580 and TNF-␣ secretion (data not shown). Samples from one patient showed a low degree of inflammation (patient 27, aminosalicylatetreated), while specimens from the other two patients (patients 28 and 29, both untreated) displayed moderate to severe inflammation. From patient 29, two separate biopsy culture sets were obtained from two different anatomical locations, moderately and highly inflamed, respectively. After 4 h of incubation, TNF-␣ release into the supernatant was assessed by standard ELISA. All TNF-␣ concentrations measured were within the sensitivity range specified by the manufacturer. Biopsies from the mildly inflamed mucosa of patient 27 secreted ϳ5 pg TNF-␣ per 1-mg specimen, while the moderately and highly inflamed tissue explants of patients 28 and 29 released 47-51 pg TNF-␣ per 1-mg specimen. The inhibitory activity of SB 203580 was inversely correlated to the severity of inflammation (Fig. 7A). In highly inflamed tissue of patient 29, TNF-␣ secretion was reduced by only 8% (NS). However, a highly significant decrease of TNF-␣ release ( p Ͻ 0.01) was observed in mildly inflamed (Ϫ88%) and moderately inflamed (Ϫ37 and Ϫ38%, respectively) mucosa of all patients.
The specificity of p38␣ inhibition was controlled by examining the phosphorylation levels of Hsp27, which is a specific target for p38␣␤ signaling (19). As p38␤ was barely expressed in the inflamed mucosa (see above), Hsp27 phosphorylation was a precise indicator for p38␣ activity. Fig. 7B shows that incubation with SB 203580 significantly reduced Hsp27 phosphorylation in both weakly and highly inflamed CD biopsies. As Hsp27 production is up-regulated by stress (36), the differences between the patients in baseline Hsp27 expression reflected the inflammatory activity.

Activation of p38␣ by infliximab in vivo and in vitro
Patients with CD were treated with infliximab, a mAb directed against TNF-␣. Two responders to infliximab treatment, two nonresponders, and one short-responder were chosen. The phosphorylation and protein expression of p38␣ and JNK1/2 in the affected sigmoid mucosa were determined by Western blotting experiments using denatured extracts of mucosal biopsy specimens taken immediately before, 24 h after, and 48 h (n ϭ 3) after a single infusion of infliximab. All patients showed a highly significant increase of p38␣, but not JNK1/2, dual phosphorylation (between 2-and 4-fold) 24 h after infusion ( p Ͻ 0.000001; Fig. 8A). After 48 h, p38␣ activity dropped to a level still significantly higher than before infusion ( p Ͻ 0.01), but also significantly lower than 24 h after infusion ( p Ͻ 0.01).
To find out more about the mechanisms and consequences of infliximab-induced p38␣ activation, we stimulated freshly isolated peripheral monocytes from three healthy volunteers as well as human THP-1 (myelomonocyte), Jurkat (T lymphocyte), and RPMI FIGURE 6. Immunohistochemical analysis of the localization of p38␣ in cryosections of biopsies from a normal control (A) and IBD patients (B-F). In the inflamed mucosa of CD (C) and UC (D) specimens, immune cells within the lamina propria showed a strong staining for p38␣. The specificity of the signal was confirmed by using a normal serum as irrelevant primary Ab under identical conditions (B). In the inflammatory infiltrates of CD specimens (E-F), the bulk of p38␣ staining colocalized with CD68 (Ki-M6) staining specific for monocytes/macrophages (E) and HNP specific for neutrophils (F). Similar results were obtained with UC specimens (data not shown). All results displayed are representative of four separate experiments with biopsy sections obtained from four normal controls, three CD patients, and three UC patients. 8226 (multiple myeloma) cells with infliximab (5 g/ml, corresponding to the therapeutic dose of 5 mg/kg body weight). The three cell lines were selected because THP-1 cells express TNF-␣ constitutively on their cell membrane, while resting Jurkat and RPMI-8226 cells do not.
Infliximab induced a highly significant ( p Ͻ 0.01) increase in p38␣, but not JNK1/2, dual phosphorylation in THP-1 cells (Fig.  8, B-E). This activation was visible already after 0.5 h, gradually increased toward 24 h (Fig. 8B), and was sustained until 48 h after stimulation (data not shown). In vitro kinase assays confirmed the Western blotting results (Fig. 8C). The significant activation of p38␣ after 24 h could not be induced by nonspecific IgG1 and did not occur in Jurkat or RPMI 8226 cells (Fig. 8D). This finding was mirrored on the transcription factor level by a strong phosphorylation of ATF-2 on Thr 71 in THP-1 cells, but not in Jurkat cells (Fig. 8E). The protein expression of the MAPKs and ATF-2 (Fig.  8, A, B, and E) was not significantly altered by infliximab when compared with ␤-actin expression. The MAPK activity and expression patterns observed in the infliximab-treated patients (Fig.  8A) corresponded to the findings in THP-1 cells.
Treatment with infliximab (5 g/ml) induced a significant increase in TNF-␣ mRNA in human monocytes (Fig. 8F) and THP-1 cells (data not shown), as determined by linear phase RT-PCR. TNF-␣ gene induction was observed after 6 h until 24 h after stimulation (maximum, 6-fold) and could be abrogated completely by treatment with a low dose (1 M) of SB 203580 (Fig. 8F). In contrast, inhibition of p38␣ by 1 M SB 203580 did not influence the infliximab-induced apoptosis of monocytes mediated by caspase-3 (14), as determined by cleavage of nuclear PARP, a main cleavage substrate of caspase-3 (Fig. 8G). This result was confirmed by FACS analysis of annexin V-stained monocytes (data not shown), which indicated a 15% increase in apoptosis by treatment with 5 g/ml infliximab, regardless of coincubation with SB 203580.

Discussion
In this study, we investigated the activity and expression of MAPKs in IBD patients and the influence of the therapeutic anti-TNF-␣ mAb infliximab on MAPK pathways. Our results show that p38␣, JNK1/2, and ERK1/2 were significantly activated in the inflamed colonic mucosa of IBD patients, with p38␣ exhibiting the strongest activation in both CD and UC. In the meantime, this result (37) has been confirmed by others (25). Moreover, infliximab induced TNF-␣ gene expression in human monocytes via a transient p38␣ activation.
The activation of p38␣, JNK1/2, and ERK1/2 in IBD is consistent with previous reports implicating these enzymes in several cascades of inflammatory signal transduction (18,21,38). Except for ERK1/2 in glucocorticoid-treated CD patients, the activation of MAPKs in IBD was dependent only on the severity of inflammation, not on aminosalicylate or glucocorticoid therapy. All p38 activity observed could be attributed to p38␣. As the relative activities of JNK1/2 and ERK1/2 were lower than the activity of p38␣, and as p38␣ was the only activated enzyme not showing a tendential (JNK1/2) or significant (ERK1/2) down-regulation on the protein level, p38␣ exhibited by far the highest increase of active enzyme in the inflamed mucosa, suggesting an exceptional role for this kinase in IBD.
While p38␣ and p38␦ protein was expressed in similar amounts in all samples, p38␤ and p38␥ protein contents were below detection level, which was mirrored by a low mRNA expression. This is consistent with several studies demonstrating that p38␤ expression is very low in the intestine and in peripheral leukocytes, and that p38␥ is almost exclusively expressed in skeletal muscle (39 -42). p38␣ and p38␦ have been demonstrated to be the major isoforms in peripheral leukocytes, with p38␣ clearly emerging as the most important isoform in inflammatory cells and especially in macrophages (41)(42)(43). Therefore, p38␣ is a first-rate candidate enzyme for targeted inhibition. Our immunohistochemical analysis revealed that the main p38␣ expression observed in IBD mucosal biopsies colocalized with lamina propria macrophages and neutrophils, thus affirming the key role of these cells in IBD.
TNF-␣ secretion is regulated by p38␣ and JNK activation. In CD, TNF-␣ blockade by infliximab is used for therapy. We chose the model systems of SB 203580-treated CD biopsy cultures and colonic tissue from CD patients before and after infliximab treatment to investigate the interconnection of p38␣ and TNF-␣ signaling in vivo. The anti-inflammatory effects of p38␣␤-inhibiting pyridinyl imidazole derivatives, such as SB 203580, have been demonstrated in several in vivo models (44 -46). These effects can be attributed in part to the ability of the inhibitors to suppress monocyte/macrophage production of TNF-␣, IL-1␤, and other cytokines (47,48). Several studies have demonstrated that SB 203580 inhibits TNF-␣ production and/or release in human monocytes (49), THP-1 cells (50), and T cells (21). To specifically inhibit p38␣␤, SB 203580 had to be applied at a concentration of 10 M or below, as influences on other kinases have been observed with higher concentrations (34,35). As p38␤ is barely expressed in FIGURE 7. Inhibition of p38␣ in cultured biopsies from CD patients. A, Reduction of TNF-␣ secretion from short-term (4-h) CD biopsy cultures. Fresh colonic biopsy specimens were obtained from three CD patients (patient 27 received aminosalicylates; patients 28 and 29 were untreated). Half of the biopsies were incubated with the p38␣ inhibitor SB 203580 (SB) at a concentration of 10 M. Results are expressed as means Ϯ SD. Each column represents ELISA data from eight (controls) or four (SB 203580) biopsies obtained in two separate ELISA series. The reduction of TNF-␣ release was inversely correlated to the severity of inflammation (low in patient 27, moderate (mod.) in patient 28, and moderate to high in patient 29). B, Representative Western blots of denatured extracts from control (-) and SB 203580-treated (SB) biopsies from patient 27 and patient 29. The membrane was successively probed with phosphospecific Hsp27 Ab (P-Hsp27) and Hsp27 Ab. SB treatment significantly reduced Hsp27 phosphorylation in all samples. leukocytes or the intestine (see above), practically all effects observed could be attributed to p38␣ inhibition.
The disease status-related reduction of TNF-␣ secretion by specific inhibition of p38␣ in mucosal biopsies from CD patients demonstrates that p38␣ regulates TNF-␣ production in CD and that p38␣ repression can significantly diminish inflammatory activity in this system. The significant reduction of Hsp27 phosphorylation confirmed the specificity of p38␣ inhibition in all patients. The fact that the highly inflamed tissue of patient 29 showed a strong reduction in Hsp27 phosphorylation, but only a tendential decrease in TNF-␣ secretion, suggests that p38␣ inhibition may prove especially rewarding to avoid TNF-␣ production in inactive patients (i.e., remission maintenance). For induction of remission in highly active patients, the additional inhibition of JNKs could be necessary to FIGURE 8. Infliximab activates p38␣ both in vivo and in vitro. A, p38␣ is activated by infliximab in the sigmoid mucosa of CD patients (P-p38␣), while JNK1/2 activity is not altered (P-JNK1/2). Biopsies were taken immediately before (b), 24 h after, and 48 h after a single infusion of infliximab (5 mg/kg). Shown are representative Western blots of the same denatured extracts from biopsy specimens of a responder (R; patient 44) and a nonresponder (NR; patient 40) to infliximab treatment. All patients featured a highly significant transient increase in dual-phosphorylated (i.e., active) p38␣ (P-p38␣). B, Representative Western blot of denatured extracts from THP-1 cells. Infliximab (5 g/ml) strongly enhanced p38␣ dual phosphorylation (P-p38␣), increasing from 0.5 to 24 h. The short-term activation caused by 5 ng/ml recombinant human TNF-␣ was investigated for comparison. This experimental setting was used in three independent stimulation series and produced consistent results. C, Infliximab treatment (5 g/ml, 24 h) increased p38␣ kinase activity in THP-1 cells, as determined by in vitro kinase assay. After selective immunoprecipitation of the dual-phosphorylated form of p38␣, kinase reactions were conducted with an ATF-2 fusion protein whose Thr 71 -phosphorylated form (P-ATF-2) was detected by Western blotting. The data shown are representative of four kinase assay experiments. D, Stimulation with infliximab (Ifx; 5 g/ml, 24 h), but not with a nonspecific IgG1 mixture, induced a highly significant (p Ͻ 0.01) activation of p38␣ in resting THP-1 cells. In contrast, no significant activation of p38␣ was observed in resting Jurkat or RPMI 8226 cells. Data represent a total of seven independent stimulation series with THP-1 cells, six series with Jurkat cells, and four series with RPMI 8226 cells. E, Phosphorylation of ATF-2 (P-ATF-2) Thr 71 in response to infliximab (ϩ) was restricted to THP-1 cells. Infliximab treatment did not influence JNK1/2 activity (P-JNK1/2), which was confirmed by the use of three different primary Abs and purchased positive control cell extracts (data not shown). The Western blots displayed represent the same denatured extracts from control (-) and infliximab-treated (ϩ; 5 g/ml, 24 h) THP-1 and Jurkat cells, respectively. MAPK and ATF-2 protein expression were not significantly altered by infliximab (see A and B). F, Induction kinetic of TNF-␣ mRNA expression by infliximab (Ifx; 5 g/ml) in monocytes obtained from a healthy individual. Coincubation with 1 M SB 203580 completely abrogated TNF-␣ induction. Control IgG1 did not induce TNF-␣ expression after 6 or 24 h (data not shown). Data shown are representative for stimulation series with monocytes obtained from three healthy volunteers. G, PARP cleavage in response to infliximab treatment (Ifx; 5 g/ml) was not influenced by coincubation with SB 203580 (SB; 1 M). Nonspecific human IgG1 did not induce PARP cleavage. SB 203580 alone had no influence on apoptosis (data not shown). The Western blot shown is representative for experiments with monocytes obtained from three healthy volunteers. reduce TNF-␣ secretion to normal levels (25). However, ongoing clinical studies using specific p38 inhibitors, such as BIRB 796 BS (Boehringer-Ingelheim, Ridgefield, CT), in active CD will clarify this issue.
Interestingly, induction of TNF-␣ by p38␣ was also seen after treatment with infliximab. Infliximab enhanced TNF-␣ gene expression in human peripheral monocytes from healthy individuals and in THP-1 myelomonocytic cells. This effect could be completely abrogated by coincubation with the p38␣␤ inhibitor SB 203580 (1 M). In parallel to these findings, we demonstrated a highly significant, transient increase of p38␣ activity in sigmoidal biopsies of five representative CD patients during the first 48 h after a single infusion of infliximab (5 mg/kg body weight), while JNK1/2 activity was not altered. A strong increase in circulating TNF-␣-most likely bound to infliximab-during the first days after treatment with a single infusion of infliximab (5 mg/kg) has been reported in rheumatoid arthritis patients (51).
To investigate the mechanisms underlying this novel signaling effect, we performed extensive in vitro studies with THP-1, Jurkat, and RPMI 8226 cells. While a constitutive secretion of TNF-␣ has been shown in naive, resting THP-1 myelomonocytes (52,53), resting Jurkat T cells and RPMI 8226 plasma cells do not express TNF-␣ (54 -56). Activation of p38␣ by infliximab was detected only in THP-1 cells, corresponding to the constitutive production of TNF-␣ in this cell line. The specific p38␣ activation also accounts for the increase in phosphorylated ATF-2, as the AP-1 component ATF-2 is regulated by p38 and JNKs, and JNK1/2 were not activated by infliximab.
During the last years, outside-to-inside (reverse) signaling through transmembrane TNF-␣ by ligation of soluble TNF-␣ receptor (57) or anti-TNF-␣ Abs (54,58) has emerged as a new pathway in inflammatory signal transduction. Recently, the specific induction of apoptosis by infliximab via transmembrane TNF-␣ has been shown in peripheral blood monocytes from CD patients and healthy individuals (14) as well as in lamina propria T cells (15). Prior findings demonstrating that the ex vivo TNF-␣ production of LPS-stimulated whole blood from CD patients decreased after infliximab treatment (10) could be explained by an infliximab-induced apoptosis of TNF-␣-producing leukocytes. In addition, the ELISA used would not detect infliximab-bound TNF-␣, in contrast to the assay used to study TNF-␣ serum levels in rheumatoid arthritis patients (10,51).
In view of the previous findings linking transmembrane TNF-␣ to infliximab-induced apoptosis (14), we investigated the influence of p38␣ inhibition on the apoptosis of infliximab-stimulated human monocytes. Treatment with 1 M SB 203580, which was able to abrogate infliximab-induced, p38␣-mediated TNF-␣ induction, had no influence on apoptosis. Therefore, we conclude that p38␣ activation, although involved in TNF-␣ regulation and most likely in the maintenance of an inflammatory environment, is not interfering with immune cell apoptosis.