Transmembrane tumor necrosis factor is a potent inducer of colitis even in the absence of its secreted form

GASTROENTEROLOGY 2004;127:816 – 825

Transmembrane Tumor Necrosis Factor Is a Potent Inducer of Colitis Even in the Absence of its Secreted Form NADIA CORAZZA,* THOMAS BRUNNER,* CAROLINE BURI,* SILVIA RIHS,* MARTIN A. IMBODEN,*,‡ INGE SEIBOLD,* and CHRISTOPH MUELLER* *Department of Pathology, University of Bern, Bern; and ‡Berna Biotech, Bern, Switzerland

Background & Aims: Tumor necrosis factor (TNF) is cleaved proteolytically from a 26-kilodalton transmembrane precursor protein into secreted 17-kilodalton monomers. Transmembrane (tm) and secreted trimeric TNF are biologically active and may mediate distinct activities. We assessed the consequences of a complete inhibition of TNF processing on the course of colitis in recombination activating gene (RAG)2ⴚ/ⴚ mice on transfer of CD4 CD45RBhi T cells. Methods: TNFⴚ/ⴚ mice, transgenic for a noncleavable mutant TNF gene, were used as donors of CD4 T cells, and, on a RAG2ⴚ/ⴚ background, also as recipients. Kinetics of disease development were compared in the absence of TNF, in the absence of secreted TNF, and in the presence of secreted and tmTNF. The analysis at the end of the observation period included the histopathologic assessment of the intestine and the localization of TNF and interferon ␥ (IFN␥)-expressing cells. Results: The complete prevention of TNF secretion in tmTNF transgenic RAG2ⴚ/ⴚ mice neither prevented nor delayed disease induction by transferred transgenic for a noncleavable transmembrane mutant of mouse TNF (tmTNF tg) CD4 CD45RBhi T cells. tmTNF expression by transferred CD4 T cells, however, was not required for disease induction because severe colitis and weight loss also were observed in tmTNF RAG2ⴚ/ⴚ recipients of TNFⴚ/ⴚ CD4 CD45RBhi T cells. In the presence of tmTNF, the absence of secreted TNF did not affect frequency and distribution of TNF and interferon-␥ messenger RNA (mRNA)-expressing cells. Conclusions: These results indicate that specific inhibitors of TNF processing are not appropriate for modulating the pro-inflammatory and disease-inducing effects of TNF in chronic inflammatory disorders of the intestine.

umor necrosis factor (TNF) family members are expressed on the surface as type II transmembrane (tm) proteins. Although the majority of the TNF-related proteins exert their biological function preferentially via cell-contact– dependent interactions with their receptors, its archetypal member, TNF, is cleaved preferentially from the tm 26-kilodalton precursor into 17-kilodalton secreted monomers to form biologically active trimers.1 Soluble trimeric TNF binds with high affinity to TNF

T

receptor 1 (TNF-R1),2 which contains a death-domain– mediating apoptosis induction, whereas tmTNF appears to activate TNF-R2 preferentially. Engagement of TNF-R2 leads to the recruitment of TNF receptorassociated factor (TRAF)1 and TRAF2, which subsequently may activate nuclear factor ␬ B and protect from caspase-8 –mediated apoptosis induction.3 Although TNF-R2 contains no death domain, it may participate indirectly in the induction of cell death.4 The proteolytic cleavage of 26-kilodalton tmTNF into secreted TNF is mediated by a cell membrane– bound metalloprotease, the TNF␣ converting enzyme (TACE), a member of the ADAM (a disintegrin and metalloprotease) family (ADAM-17).5–7 The TNF processing activity of TACE/ADAM-17 can be blocked efficiently by specific inhibitors, and mice treated with TACE inhibitors are fully protected from lipopolysaccharide- and galactosamine-induced septic shock.7–9 Hence, administration of specific TACE inhibitors has been proposed as a novel therapeutic regimen to modulate the deleterious effects of excessive TNF secretion in inflammatory disorders such as rheumatoid arthritis or inflammatory bowel disease. In the CD4 CD45RBhi T-cell transfer model of colitis, TNF plays a crucial nonredundant role in the induction of disease.10 –12 Hence, we used this experimental mouse model of colitis to assess the consequences of a complete inhibition of TNF cleavage on the course of disease. To this end, TNF⫺/⫺ mice, transgenic for a noncleavable mutant TNF gene, were used as donor and, after backcrossing to the recombination activating gene (RAG)2⫺/⫺ Abbreviations used in this paper: ADAM, a disintegrin and metalloprotease; ICAM-1, intercellular adhesion molecule 1; IFN, interferon; mAb, monoclonal antibody; RAG, recombination activating gene; tm, transmembrane; tmTNF tg, transgenic for a noncleavable transmembrane mutant of mouse tumor necrosis factor; TNF, tumor necrosis factor; TNF-R, tumor necrosis factor receptor; TACE, tumor necrosis factor ␣ converting enzyme; VCAM-1, vascular cell adhesion molecule 1; wt, wild type. © 2004 by the American Gastroenterological Association 0016-5085/04/$30.00 doi:10.1053/j.gastro.2004.06.036

September 2004

background, also as recipient animals. In these tmTNF transgenic mice the 3 potential TNF cleavage sites at positions ⫺10, ⫹1, and ⫹11 are inactivated by amino acid deletion or substitution. In the tmTNF mutant used in the present study, the proline residue at position ⫹12 (pro⫹12) is conserved because the mouse transmembrane 1–12 deletion mutant of TNF has been reported to exert altered bioactivity, possibly owing to the absence of pro⫹12.13 These transgenic for a noncleavable transmembrane mutant of mouse TNF (tmTNF tg) mice are protected from lipopolysaccharide- and D-galactosamine–induced mortality and no TNF bioactivity is found in the serum of lipopolysaccharide-challenged mice. On activation, ex vivo isolated tmTNF tg CD4 T cells still exert cell-contact– dependent cytotoxic activity against TNF-sensitive indicator cells.14 Here we show that in the TNF-dependent CD4 CD45RBhi T-cell transfer model of colitis induction the complete inhibition of TNF processing cannot prevent the onset of an exacerbating inflammatory reaction in the colonic mucosa.

Materials and Methods Animals RAG2⫺/⫺ C57BL/6J ⫻ 129/SvEv-Gpi1c (wild-type [wt]TNF RAG2⫺/⫺) mice, derived from the embryonic stem cell line EK.CCE (129/SvEv-Gpi1c); were kindly provided by Dr. H. Bluethmann and E. Wagner (Basel Institute for Immunology). TNF⫺/⫺ lymphotoxin␣⫺/⫺ C57BL/6J ⫻ 129/SvEv mice (TNF⫺/⫺ mice) were generated using the embryonic stem cell line GS1 as described earlier.15 C57BL/6J ⫻ 129SvJ mice were obtained from Dr. H. Bluethmann and the Biotechnology and Animal Breeding Division (Füllinsdorf, Switzerland). TNF⫺/⫺ mice and mice transgenic for a noncleavable transmembrane mutant of mouse TNF (tmTNF tg),14 were backcrossed with RAG2⫺/⫺ mice in our laboratory. Polymerase chain reaction procedures were used to determine the RAG2, TNF, and tmTNF genotype. Three different primers were used to characterize the RAG2-genotype (RAG1 5=-gggAggACACTCACTTgCCAgTA-3=, RAG2 5=-AgTCAggAgTCTCCATCT CACTgA-3=, and RAGNeo 5=-CggCCggAgAACCTgCgTgCAA-3=) to yield a 350-bp fragment for the mutated, and a 263-bp fragment for the wild-type allele. TNF/lymphotoxin␣ genotype was characterized using the TNF10 (5=CCTCAgCAAACCACCAAgTggA-3=) and TNF12 (5=-TTgggCAgATTgACCTCAgCg-3=) primers in which a 350-bp fragment corresponded to wild-type allele and no band was detected for the mutant allele. In the tmTNF tg mice the transgene is under the control of the TNF promoter. The regulatory elements at the 3= end of the TNF gene, including the AU-rich regulatory elements, have been maintained to ensure appropriate gene regulation. To generate a noncleavable mutant of murine TNF, 2 deletions ( [Leu⫺12–Leu⫺10] and 

TRANSMEMBRANE TNF INDUCES COLITIS

817

[Leu⫺2–Leu⫹1]) and an amino acid substitution Lys⫹11 3 Glu⫹11 (positions according to secreted TNF) were introduced into a genomic TNF clone.14 Mice were kept in laminar flows in specific pathogen-free conditions in the central animal facility of the Medical School, University of Bern, Switzerland.

Monoclonal Antibodies Used for Cell Sorting and Immunostainings Phycoerythrin (PE)-conjugated anti-CD4 (clone GK1.5) and fluorescein isothiocyanate– conjugated antiCD45RB (clone 16A) were purchased from Pharmingen (San Diego, CA). Protein G–purified antibodies from the supernatants of clone RA3-6B2 (anti-B220), clone F4/80 (macrophage-specific mAb), clone 53– 6.7 (anti-CD8␣), and clone M1/70 (anti-Mac-1) were biotinylated or fluorescein isothiocyanate– conjugated according to standard protocols. For the immunohistochemical detection of adhesion molecules in the colonic mucosa the following primary antibodies were used for staining of frozen sections: biotinylated rat anti-mouse mucosal addressin cell adhesion molecule-1 MAdCAM-1; clone R3-3; kindly provided by Dr. B. Holzmann, Munich, Germany), biotinylated hamster anti–intercellular adhesion molecule 1 (ICAM-1; clone 3E2), and biotinylated anti–vascular cell adhesion molecule 1 (VCAM-1; clone 429) (both BD-PharMingen, San Diego, CA), followed by an incubation with avidin, conjugated to horseradish peroxidase (Dako, Glostrup, Denmark).

Isolation and Purification of CD4 CD45RBhi and CD4 CD45RBlo Splenocytes After osmotic lysis of erythrocytes, splenocytes of donor mice were resuspended at 2 ⫻ 106 cells/100 ␮L phosphatebuffered saline (PBS)/5% horse serum and incubated with 0.5–1 ␮g biotinylated anti-CD8␣ and anti-B220 monoclonal antibodies (mAbs) per 106 cells for 15 minutes on ice. Second-step stainings were performed using avidin-magnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany) and the CD8␣⫹ and B220⫹ splenocytes attached to the beads were removed by magnetic separation. The negative fraction, enriched for CD4 T cells, was stained using fluorescein isothiocyanate– conjugated CD45RB mAb and PE-conjugated CD4 mAb. Subsequently, CD4 T cells were sorted according to the expression of CD45RB on a FACSVantage (Becton Dickinson, San Jose, CA). CD4 T cells were divided into 3 different subpopulations according to the expression level of CD45RB (low, intermediate, or high). The CD4 cell subset with the lowest expression of CD45RB was collected as CD45RBlo, the subset with the highest expression of CD45RB was termed CD45RBhi. CD4 T cells with intermediate cell surface expression of CD45RB were discarded.

Reconstitution of Recipient Mice With CD4 T-Cell Subpopulations Sorted CD4 CD45RBhi and CD4 CD45RBlo T cells were washed and resuspended at 1 ⫻ 106 cells/mL in sterile PBS. Two ⫻ 105 cells each were injected intraperitoneally into

818

CORAZZA ET AL.

GASTROENTEROLOGY Vol. 127, No. 3

8 –12-week-old recipient mice. On transfer, the body weight of recipient mice was determined every other day. The mice were killed and analyzed between 21 and 24 days after cell transfer, that is, when experimental mice either showed a weight loss of ⱖ20% of their initial weight or on day 24 after cell transfer (end of observation period). To exclude the possibility of graft-versus-host reactions caused by differences in minor histocompatibility antigens between donor T cells and recipients, several controls were included. In particular, in all experiments control transfers of CD4 CD45RBlo T cells into the corresponding recipients were performed. In all of these instances, at the end of the observation period (day 24 after cell transfer), no clinical signs of inflammation in the skin, liver, or small bowel were observed. In all control recipients, the body weight always remained ⬎98% of the initial body weight at any time point during the experiment. Furthermore, mixed lymphocyte reactions revealed no indications for a possible reactivity of donor T cells against irradiated spleen cells from wtTNF, and mutant TNF RAG2⫺/⫺ recipients (data not shown).

Histology Intestinal tissue sections from the large and small bowel were fixed in 4% paraformaldehyde (in 1⫻ PBS) for subsequent paraffin embedding. After preparation of tissue sections, deparaffinized sections were stained with H&E for histopathologic analysis. To compare the histopathologic alterations in the colon induced by the adoptive transfer of CD4 T-cell subpopulations from different donors, a scoring system was established using the following parameters12: (1) infiltration of the lamina propria of large bowel (score from 0 –3), (2) mucin depletion (score from 0 –2), (3) crypt abscesses (score from 0 –2), (4) epithelial erosion (score from 0 –2), (5) hyperemia (score from 0 –3), and (6) thickness of the colonic mucosa (score from 1–3). Hence, the range of histopathologic scores was from 1 (no alteration) to 15 (most severe signs of colitis).

Preparation of

35S-Labeled

RNA Probes

A 1108-bp complementary DNA fragment of the murine TNF gene (position 1–1,108; kindly provided by Genentech Inc., San Francisco, CA) and a 900-bp EcoRI/HindIII complementary DNA fragment of interferon ␥ (IFN␥) gene (generously provided by Dr. K. Arai, DNAX, Palo Alto, CA) were subcloned into the expression vector pGEM-2 (Promega Catalys AG, Wallisellen, Switzerland). After linearization of the plasmid, sense and antisense RNA probes were prepared using the appropriate RNA polymerase as previously described.16

In Situ Hybridization Serial frozen sections of intestinal tissue were hybridized in situ with an antisense RNA probe, and as a negative control with a sense RNA probe, specific for the TNF gene. In situ hybridizations were performed as previously described.16 Briefly, cryostat sections were fixed in 4% paraformaldehyde in PBS for 20 minutes, washed in PBS, and incubated in the

presence of 1 ␮g/mL proteinase K (Roche Diagnostics, Mannheim, Germany) at 37°C for 30 minutes. After postfixation and acetylation, the hybridization was performed with 2 ⫻ 105 cpm of 35S-labeled RNA probe per ␮L of hybridization solution for 18 hours at 48°C. After digestion of nonhybridized single-stranded RNA and washing, the slides were dipped in NTB2 nuclear track emulsion (Eastman Kodak Co., Rochester, NY). The slides were exposed for 3 weeks in the dark at 4°C, developed, and counterstained with nuclear fast red by standard techniques.

TNF Bioassay Biologically active TNF in the serum of recipient mice was detected by using the TNF-sensitive L929 fibroblast bioassay as described.17 As a specificity control, a neutralizing polyclonal rabbit anti mouse TNF antibody (IP-400; Genzyme, Cambridge, MA) was added to the L929-containing wells. Recombinant mouse TNF was used as a positive control.

Detection of Apoptotic Cells on Tissue Sections For the detection of apoptotic cells on tissue sections, the single-strand DNA–specific mAb F7-26 (Alexis Corporation, Lausen, Switzerland) was used as a primary reagent followed by incubation with a rabbit-anti mouse immunoglobulin (Ig)M-peroxidase conjugate (Zymed, South San Francisco, CA) according to the suggestions of the manufacturer of the single-strand DNA–specific mAb. In brief, sections of formalin-fixed, paraffin-embedded tissues were heated in formamide, stained with mAb F7-26, and the secondary anti-mouse IgMperoxidase conjugate, developed with 3,3,-diaminobenzidine, and counterstained with Gill’s hematoxylin.

Results The transfer of 2 ⫻ 105 wtTNF CD4 CD45RBhi T cells into syngeneic wtTNF RAG2⫺/⫺ mice leads to the onset of clinical signs of colitis with persistent diarrhea and progressive weight loss after more than 12 days after cell transfer whereas transfer of TNF⫺/⫺ CD4 CD45RBhi T cells into TNF⫺/⫺ RAG2⫺/⫺ recipients does not induce clinical signs of disease (Figure 1). When, however, TNF⫺/⫺ mice transgenic for tmTNF are used as donors of colitogenic CD4 CD45RBhi T cells and as recipients (in a RAG2⫺/⫺ background), progressive wasting disease is seen consistently from day 10 after adoptive cell transfer (Figure 1). To confirm the absence of secreted TNF in tmTNF tg RAG2⫺/⫺ recipients with colitis, bioactive TNF was measured in the serum of tmTNF tg RAG2⫺/⫺, and of wtTNF RAG2⫺/⫺ recipients with active disease. In the serum of tmTNF tg RAG2⫺/⫺ recipient mice no TNF activity was detected, however, wtTNF RAG2⫺/⫺ mice contained up to 250 pg/mL of TNF in the serum. Hence, complete inhibition of TNF processing in TNF-producing cells of the

September 2004

Figure 1. TmTNF induces persistent colitis and weight loss in a TNF-dependent model of colitis. Reconstitution of wtTNF RAG2⫺/⫺ mice with 2 ⫻ 105 wtTNF CD4 CD45RBhi donor T cells (
) and of tmTNF tg RAG2⫺/⫺ mice with tmTNF tg CD4 CD45RBhi T cells (⽧) leads to progressive weight loss whereas TNF⫺/⫺ RAG2⫺/⫺ recipients, reconstituted with 2 ⫻ 105 TNF⫺/⫺ CD4 CD45RBhi T cells () show no signs of weight loss. The change in body weight is expressed as percentage of the weight at the time of cell transfer. Data represent mean values ⫾ SD of 6 –12 mice per group.

recipient and the donor neither prevents nor delays the onset of persistent diarrhea or progressive weight loss. The histopathologic scores obtained from colonic tissue sections of tmTNF tg RAG2⫺/⫺ recipients of tmTNF CD4 CD45RBhi T cells at the end of the observation period are in the same range as those obtained from wtTNF RAG2⫺/⫺ recipients of wtTNF CD4 CD45RBhi T cells, thus confirming that the absence of secreted TNF does not prevent induction of severe inflammation of the colon (Figure 2A). For each single parameter assessed in the histopathologic analysis (crypt hyperplasia, cellularity, goblet cell depletion, crypt abscesses, hyperemia, and epithelial erosion), no statistically significant difference was observed between tmTNF and wtTNF recipients of colitogenic CD4 T cells (data not shown). Representative histologic pictures from the 3 different groups of recipients of CD4 CD45RBhi T cells are shown in Figure 2B–D. At the end of the observation period (days 21–24 after cell transfer), a comparable extent of crypt abscesses, epithelial erosion, and cellular infiltration of the colonic mucosa are observed in wtTNF RAG2⫺/⫺ and tmTNF tg RAG2⫺/⫺ recipients of CD4 CD45RBhi T cells from wtTNF and tmTNF tg donors, respectively. To assess how the absence, or presence, of secreted TNF may affect the frequency and distribution of TNF-expressing cells, in situ hybridizations were performed for the localization of TNF messenger RNA (mRNA)expressing cells in the colonic mucosa. As shown in Figure 3, in both groups with active colitis TNF mRNA-expressing cells are found preferentially in the lamina propria adjacent to the gut lumen, often at sites with erosion of the colonic epithelium. With the exception of the colon, no differences in the extent of CD4 T-cell infiltration of the

TRANSMEMBRANE TNF INDUCES COLITIS

819

small intestine or of other organs, including liver and lung, were observed between wtTNF, tmTNF tg, and TNF⫺/⫺ RAG2⫺/⫺ recipients of CD4 CD45RBhi T cells (data not shown). It has been shown previously that TNF production by non–T cells of the RAG2⫺/⫺ recipients of TNF⫺/⫺ CD4 CD45RBhi donor T cells is sufficient for the onset of colitis.12 Therefore, we determined whether in the absence of TNF expression by transferred T cells, tmTNF tg expression in non–T cells of the recipients is capable of mediating the onset of colitis. As shown in Figure 4, the transfer of TNF⫺/⫺ CD4 CD45RBhi T cells induces weight loss and histopathologic alterations in both tmTNF tg and wtTNF RAG2⫺/⫺ recipients. The TNF⫺/⫺ CD4 CD45RBhi T-cell–induced onset of weight loss is even more pronounced in tmTNF tg RAG2⫺/⫺ mice than in wtTNF RAG2⫺/⫺ recipients. At the end of

Figure 2. Histopathologic assessment of (A) individual wtTNF RAG2⫺/⫺ recipients of 2 ⫻ 105 wtTNF CD4 CD45RBhi donor T cells (
) (N ⫽ 11), tmTNF tg RAG2⫺/⫺ recipients of tmTNF CD4 CD45RBhi T cells (⽧) (N ⫽ 6), and of TNF⫺/⫺ RAG2⫺/⫺ recipients of TNF⫺/⫺ CD4 CD45RBhi T cells () (N ⫽ 12). The colitis scores represent the total of individual scores for cellularity, mucin depletion, crypt abscesses, epithelial erosion, hyperemia, and thickness of the colon, as described in the Materials and Methods section. Annotated numbers represent the mean colitis score of N mice per group. Representative tissue sections of the colon from a (B) wtTNF CD4 CD45RBhi 3 wtTNF RAG2⫺/⫺ mouse, (C) tmTNF tg CD4 CD45RBhi 3 tmTNF tg RAG2⫺/⫺ mouse, and (D) TNF⫺/⫺ CD4 CD45RBhi 3 TNF⫺/⫺ RAG2⫺/⫺ mouse are shown (H&E staining, magnification: 12⫻).

820

CORAZZA ET AL.

Figure 3. Distribution of TNF mRNA-expressing cells in the colon of a (A) wtTNF CD4 CD45RBhi T-cell–reconstituted wtTNF RAG2⫺/⫺ mouse, (B) tmTNF CD4 CD45RBhi T-cell–reconstituted tmTNF tg RAG2⫺/⫺ mouse, and (C) TNF⫺/⫺ CD4 CD45RBhi T-cell–reconstituted TNF⫺/⫺ RAG2⫺/⫺ mouse on day 24 after adoptive cell transfer as assessed by in situ hybridization with a 35S labeled TNF antisense probe. (D) As a control, a section from the same tissue as in (A) was hybridized with a sense probe of the TNF gene. Nuclear fast red counterstaining, magnification: 50⫻.

Figure 4. tmTNF produced by recipient non–T cells is sufficient to induce clinical signs of colitis on transfer of TNF⫺/⫺ CD4 CD45RBhi T cells. Reconstitution of tmTNF tg RAG2⫺/⫺ mice with 2 ⫻ 105 TNF⫺/⫺ CD4 CD45RBhi T cells (⽧) induces colitis and weight loss comparable with wtTNF RAG2⫺/⫺ recipients of TNF⫺/⫺ colitogenic T cells (
). As negative controls, both groups of recipient mice were injected with 2 ⫻ 105 CD4 CD45RBlo T cells (〫, □). Data are the mean ⫾ SD of 4 –14 mice per group.

GASTROENTEROLOGY Vol. 127, No. 3

Figure 5. Histopathologic assessment of (A) individual wtTNF RAG2⫺/⫺ (
, □) and tmTNF tg RAG2⫺/⫺ recipients (⽧, 〫) of TNF⫺/ ⫺CD4 CD45RBhi T cells (
, 〫) and CD4 CD45RBlo T cells (□, 〫) at the end of the observation period. Annotated numbers represent mean colitis scores of all individual mice in each group. Representative tissue sections of the colon from a (B) TNF⫺/⫺ CD4 CD45RBhi 3 wtTNF RAG2⫺/⫺ mouse, (C) a TNF⫺/⫺ CD4 CD45RBhi 3 tmTNF tg RAG2⫺/⫺ mouse, and (D) a TNF⫺/⫺ CD4 CD45RBlo 3 tmTNF tg RAG2⫺/⫺ mouse are shown (H&E staining, magnification: 12⫻).

the observation period the histopathologic alterations of the colon of tmTNF tg and wtTNF RAG2⫺/⫺ recipients are comparable whereas TNF⫺/⫺ CD4 CD45RBlo memory-type control T cells failed to induce weight loss or severe signs of histopathologic alterations in both wtTNF RAG2⫺/⫺ and tmTNF tg RAG2⫺/⫺ recipients (Figures 4 and 5). As seen in Figure 1, weight loss is accelerated slightly in tmTNF tg RAG2⫺/⫺ recipients of tmTNF tg CD4 CD45RBhi T cells when compared with wtTNF RAG2⫺/⫺ recipients wtTNF CD4 CD45RBhi T cells. Hence, we subsequently assessed whether tmTNF tg CD4 CD45RBhi T cells may be more potent inducers of colitis and even may induce signs of colonic inflammation in TNF⫺/⫺ RAG2⫺/⫺ recipients. As shown in Figure 6, some TNF⫺/⫺ RAG2⫺/⫺ recipients of tmTNF tg CD4 CD45RBhi T cells indeed developed moderate to severe signs of colitis. Severe histopathologic alterations (score, ⬎9) were always associated with weight loss and persistent diarrhea whereas TNF⫺/⫺ RAG2⫺/⫺ recipients of tmTNF tg CD4 CD45RBhi T cells with a colitis score of less than 9 developed no clinical signs of colitis. To determine how presence or absence of secreted TNF may affect the frequency and the localization of

September 2004

Figure 6. tmTNF tg CD4 CD45RBhi T cells, but not wtTNF CD4 CD45RBhi T cells, occasionally induce inflammatory alterations in the colons of TNF⫺/⫺ RAG2⫺/⫺ mice. Colitis scores were determined as described in the Materials and Methods section on paraffin sections from TNF⫺/⫺ RAG2⫺/⫺ recipients of CD4 CD45RBhi (), and CD4 CD45RBlo control cells (Œ) from wtTNF, or tmTNF tg donor mice on day 24 after cell transfer. Annotated numbers represent mean colitis scores of all individual mice in each group.

TNF- and IFN␥-expressing cells in the colonic mucosa, we screened colonic tissue sections from all recipients for the presence of TNF and IFN␥ mRNA-expressing cells by in situ hybridization. As shown in Figure 7, the absence of proteolytic TNF shedding in tmTNF RAG2⫺/⫺ recipients did not alter the distribution and the frequency of TNF mRNA-expressing cells in the affected mucosa substantially. These analyses also revealed no difference in the distribution pattern of IFN␥ mRNA-expressing cells in the colonic mucosa, thus indicating that secreted TNF is not required for the functional differentiation or the recruitment of IFN␥-expressing cells. Since recently compelling evidence was provided for a TNF-mediated regulation of apoptosis in distinct cell subsets during intestinal inflammation in patients with inflammatory bowel disease, but also in experimental models of colitis and ileitis,18 –20 we directly assessed how the absence of secreted TNF might affect the frequency and distribution of apoptotic cells in situ. To this end we identified apoptotic cells on formamide-treated tissue sections by immunostainings with a primary antibody against single-strand DNA. As shown in Figure 8, numerous apoptotic cells are detected in the affected colonic mucosa of wtTNF and tmTNF recipients of colitogenic CD4 CD45RBhi T cells, but not in TNF⫺/⫺ RAG2⫺/⫺ recipients.

Discussion Compelling evidence for a crucial role of TNF in the pathogenesis of chronic inflammatory diseases of the intestine has been obtained in various experimental an-

TRANSMEMBRANE TNF INDUCES COLITIS

821

imal models of colitis and has been validated by the often striking results of the treatment with the anti-TNF mAb infliximab of patients with fistulizing Crohn’s disease.21,22 With the increasing number of patients treated with TNF-targeting strategies, it becomes evident that complete blocking of TNF also may have deleterious effects, particularly in patients with latent mycobacterial infections that may become activated on initiation of an anti-TNF therapy.23 Hence, more selective TNF-targeting strategies that allow the control of latent infections of intracellular bacteria, but attenuate the deleterious effects of an excessive release of TNF, would represent a clear advantage. The finding that distinct functions may be ascribed to the secreted 17-kilodalton and the 26-kilodalton transmembrane form of TNF24,25 prompted several groups to construct mouse mutant tmTNF transgenic mice,14,26 or tmTNF knock-in mice.27 This allowed direct assessment of the differential TNF-mediated effects in the absence, or presence, of secreted TNF. The functional analysis of these various tmTNF constructs revealed a split phenotype because some tmTNF-mediated effects, such as

Figure 7. TNF and IFN␥ mRNA-expressing cells show similar distribution patterns in the colons of tmTNF tg and wtTNF RAG2⫺/⫺ recipients. Distribution of (A–D) TNF mRNA- and (E–H) IFN␥-expressing cells in the colons of a (A, E) wtTNF CD4 CD45RBhi T-cell–reconstituted wtTNF RAG2⫺/⫺ mouse, (B, F) tmTNF tg CD4 CD45RBhi T-cell–reconstituted tmTNF tg RAG2⫺/⫺ mouse, (C, G) TNF⫺/⫺ CD4 CD45RBhi T-cell– reconstituted tmTNF tg RAG2⫺/⫺ mouse, and (D, H) a TNF⫺/⫺ CD4 CD45RBhi T-cell–reconstituted TNF⫺/⫺ RAG2⫺/⫺ mouse on day 24 after adoptive cell transfer as assessed by in situ hybridization with a 35S-labeled TNF and IFN␥ antisense probe, respectively. L, intestinal lumen. Nuclear fast red counterstaining, magnification: 15⫻.

822

CORAZZA ET AL.

GASTROENTEROLOGY Vol. 127, No. 3

Figure 8. Apoptotic cells are more frequent in the affected colon of (A) tmTNF and (B) wtTNF RAG2⫺/⫺ recipients of CD4 CD45RBhi T cells from tmTNF tg and wtTNF donors, respectively, than in (C) TNF⫺/⫺ RAG2⫺/⫺ recipients of TNF⫺/⫺ CD4 CD45RBhi T cells. Apoptotic cells (indicated by arrows) were identified on formamide-treated paraffin colonic tissue sections obtained from recipients on day 24 after CD4 T-cell transfer by immunostainings with a primary antibody specific for single-strand DNA. Immunoperoxidase staining, magnification: 200⫻.

induction of VCAM-1 and ICAM-1 on endothelial cells14 or reconstituting an intact splenic microarchitecture, which may be seen equally in tmTNF and wtTNF mice, whereas other properties, such as protection from lipopolysaccharide- and D-galactosamine–induced septic shock, resemble more the phenotype of TNF-deficient mice. Occasionally, tmTNF mice showed functional phenotypes that clearly were distinct from both wtTNF and TNF⫺/⫺ mice, for example, increased serum levels of interleukin 12 on stimulation with lipopolysaccharide,12 or an enhanced expression of the chemokine genes TCA3 and RANTES in normal spleens.27 The concept that preventing proteolytic TNF processing may indeed lead to the attenuation of immunopathologic reactions while sparing some of the vital functions of TNF (e.g., in controlling mycobacterial infection by inducing granuloma formation) has been supported by the finding that under suboptimal conditions the induction of experimental allergic encephalomyelitis is impaired in tmTNF knock-in mice27 whereas tmTNF transgenic mice still are competent to eliminate mycobacterial infection.28 These results thus indicated that prevention of TNF secretion may lead to an attenuation of inflammatory disorders while maintaining sufficient protection from infection with intracellular pathogens, including mycobacteria. The recent finding of enhanced TACE activity in patients with ulcerative colitis29 further spurred interest in a strategy that specifically targets the proteolytic processing of TNF in this severe inflammatory disorder. However, despite all these positive indications, results obtained in the present study clearly show that colitis is induced in immunodeficient recipients of colitogenic CD4 CD45RBhi T cells even if secretion of TNF is abrogated completely. This shows that the systemic effects of TNF are not required for induction and perpetuation of a chronic inflammation of the large intestine. This finding is somewhat unexpected because in experimental allergic encephalomyelitis most of the disease-

promoting effects of TNF have been linked to TNF-R1 engagement,30 whereas tmTNF has been described as the main TNF-R2–activating ligand.31 The preferential binding of tmTNF to TNF-R2 is intriguing because in addition to its induced expression on hematopoietic cells in the affected lamina propria of patients with Crohn’s disease and in mice with experimental colitis,19 TNF-R2 also is induced in colonic epithelial cells during onset of experimental colitis in mice and in patients with active inflammatory bowel disease.32 Hence, tmTNF expressed on recipient lamina propria cells may contribute to the progression and perpetuation of the disease by cognate interactions with hematopoietic cells such as dendritic cells, macrophages, and also donor-derived CD4 T cells in the lamina propria, but also via direct interactions with colonic epithelial cells. tmTNF-induced effects in the epithelium may include accelerated induction of epithelial cell apoptosis,18,33 disrupted tight junctions and barrier function of the epithelium,34 and induction of pro-inflammatory gene expression.35 Although this preferential binding of mutant tmTNF to TNF-R2 may depend in part on the mutations introduced to generate a noncleavable mutant TNF,27,36 it is intriguing that TNF-R2 is up-regulated in colonic epithelial cells from patients with inflammatory bowel disease and mouse models of colitis and the absence of TNF-R2 attenuates the spontaneous development of colitis in the T-cell receptor ␣⫺/⫺ mice.37 Hence, both TNF-R1- and TNF-R2-mediated effects may be required for the development of colitis. Similar to our previous findings in wtTNF RAG2⫺/⫺ recipients of TNF⫺/⫺ CD4 T cells, expression of tmTNF by the colitogenic CD4 T cells is not required for disease induction and tmTNF by host-derived non–T cells alone is sufficient for mediating colitis. Remarkably, the distribution of IFN␥ and TNF mRNA-expressing cells is comparable in tmTNF RAG2⫺/⫺ and wtTNF RAG2⫺/⫺ recipients of colitogenic cells, hence secreted TNF does not seem to greatly influence frequency and localization

September 2004

TRANSMEMBRANE TNF INDUCES COLITIS

823

Figure 9. tmTNF tg and wtTNF RAG2⫺/⫺ recipients of CD4 CD45RBhi T cells show increased expression of ICAM-1 and VCAM-1 in the colonic mucosa. Colonic frozen sections of wtTNF, tmTNF tg, and TNF⫺/⫺ RAG2⫺/⫺ recipients, obtained on day 24 after transfer of CD4 CD45RBhi T cells were stained for MAdCAM-1, ICAM-1, and VCAM-1. Immunoperoxidase staining, magnification: 75⫻.

of IFN␥-producing cells in the affected mucosa (Figure 6). In the absence of secreted TNF, the tmTNF-expressing cells in the affected colon often were found in close vicinity to the colonic lumen of tmTNF RAG2⫺/⫺ recipients (Figures 3 and 6). This was particularly true early after adoptive transfer of colitogenic T cells where TNF mRNA-expressing cells preferentially were found in the (sub)epithelial area in direct vicinity to the colonic lumen (data not shown). Hence, the distribution of tmTNF-expressing cells during development of transfer colitis is comparable with the previously reported distribution of TNF-expressing cells after the transfer of wtTNF CD4 CD45RBhi T cells into wtTNF RAG2⫺/⫺ mice.12 Unfortunately, we were unable to define clearly the phenotype of the TNF-producing cells by immunostainings of tissue sections or fluorescence-activated cell sorter analysis of isolated cells using a variety of mouse TNF-specific reagents. Reverse-transcription polymerase chain reaction analysis of sorted colonic epithelial cells, N418-positive dendritic cells, and F4/80-positive macrophages of the colonic lamina propria of mice with active colitis, however, revealed the presence of TNF mRNA in all these cell subsets (data not shown). To date the cellular target(s) and the essential TNFmediated effects involved in the pathogenesis of colitis in experimental animals and in patients with inflammatory bowel disease have not been defined exactly. Results obtained with bone marrow chimeras in the TNF-overexpressing, AU-rich regulatory elements model of colitis indicate that TNF-responding hematopoietic and

nonhematopoietic stromal cells may be equally important in mediating the disease-promoting effects of TNF.38 A variety of TNF-mediated effects have been shown to be operative in the inflamed intestinal mucosa.39 – 41 These effects include the direct apoptosis induction in intestinal epithelial cells,33 up-regulation of adhesion molecules on intestinal microvascular endothelial cells,42 activation of matrix metalloproteases, and triggering of the maturation of dendritic cells and macrophages. Immunostainings of frozen colonic tissue sections revealed that tmTNF RAG2⫺/⫺ recipients of colitogenic CD4 T cells of TNF⫺/⫺ donor mice indeed expressed VCAM-1 and ICAM-1 on endothelial cells of colonic venules similar to wtTNF RAG2⫺/⫺ recipients (Figure 9). However, up-regulation of these adhesion molecules is likely to be a result of the progressing inflammatory reaction that further sustains the recruitment of inflammatory cells rather than a disease-inducing event. Cell-surface expression of tmTNF in intestinal epithelial cells may affect the barrier function not only by accelerating apoptosis induction, but together with IFN␥ also by disrupting intestinal barrier functions.34 Although some of the detected apoptotic nuclei in the epithelium may be derived from intraepithelial lymphocytes, the increased appearance of apoptotic cells in the affected epithelium in tmTNF recipients compared with TNF⫺/⫺ recipients (Figure 8) may indicate that tmTNF also can induce apoptosis in the epithelium. Such a TNF-mediated apoptosis of intestinal epithelial cells has

824

CORAZZA ET AL.

been suggested in the SAMP1/YitFc mouse, in which neutralization of TNF in vivo abrogates intestinal epithelial cell apoptosis that occurs during the spontaneous onset of ileitis in this mouse model of Crohn’s disease.18 The frequency of apoptotic cells in the lamina propria of tmTNF RAG2⫺/⫺ mice was comparable with wtTNF RAG2⫺/⫺ recipients, although owing to the high frequency of macrophages and the efficient phagocytic removal of apoptotic cells in this compartment, minor differences in the apoptosis rate may not be detected on tissue sections. In conclusion, the complete inhibition of the proteolytic processing of the 26-kilodalton tmTNF cannot abrogate induction of colitis in a strictly TNF-dependent mouse model of disease and kinetics of disease induction and histopathologic alterations in the affected colonic mucosa are comparable in the absence or presence of secreted TNF. These results thus indicate that the specific inhibition of TNF cleavage alone will most likely not result in an attenuation of inflammatory bowel disease. In the absence of secreted TNF, tmTNF-expressing cells may become even more colitogenic as has been previously reported for TNFR2– overexpressing CD4 T cells, which induce an accelerated onset of colitis on transfer into immunodeficient mice.19 It remains to be seen whether the blocking of additional TACE-mediated effects by TACE inhibitors, such as shedding of the chemokine fractalkine (CX3CL1),43 may attenuate excessive inflammatory reactions. Furthermore, TACE inhibitors with additional specificities and anti-inflammatory properties still may prove valid although some of the current TACE inhibitors that prevent shedding of TNF receptors also may have diseaseenhancing effects.44,45

References 1. Vassalli P. The pathophysiology of tumor necrosis factors. Annu Rev Immunol 1992;10:411– 452. 2. Grell M, Wajant H, Zimmermann G, Scheurich P. The type 1 receptor (CD120a) is the high-affinity receptor for soluble tumor necrosis factor. Proc Natl Acad Sci U S A 1998;95:570 –575. 3. Rothe M, Sarma V, Dixit VM, Goeddel DV. TRAF2-mediated activation of NF-kappa B by TNF receptor 2 and CD40. Science 1995;269:1424 –1427. 4. Grell M, Zimmermann G, Gottfried E, Chen CM, Grunwald U, Huang DC, Wu Lee YH, Durkop H, Engelmann H, Scheurich P, Wajant H, Strasser A. Induction of cell death by tumour necrosis factor (TNF) receptor 2, CD40 and CD30: a role for TNF-R1 activation by endogenous membrane-anchored TNF. EMBO J 1999;18:3034 –3043. 5. Black RA, Rauch CT, Kozlosky CJ, Peschon JJ, Slack JL, Wolfson MF, Castner BJ, Stocking KL, Reddy P, Srinivasan S, Nelson N, Boiani N, Schooley KA, Gerhart M, Davis R, Fitzner JN, Johnson RS, Paxton RJ, March CJ, Cerretti DP. A metalloproteinase disintegrin that releases tumour-necrosis factor-alpha from cells. Nature 1997;385:729 –733.

GASTROENTEROLOGY Vol. 127, No. 3

6. Moss ML, Jin SL, Milla ME, Bickett DM, Burkhart W, Carter HL, Chen WJ, Clay WC, Didsbury JR, Hassler D, Hoffman CR, Kost TA, Lambert MH, Leesnitzer MA, McCauley P, McGeehan G, Mitchell J, Moyer M, Pahel G, Rocque W, Overton LK, Schoenen F, Seaton T, Su JL, Warner J, Willard D, Becher. Cloning of a disintegrin metalloproteinase that processes precursor tumour-necrosis factor-alpha. Nature 1997;385:733–736. 7. Gearing AJ, Beckett P, Christodoulou M, Churchill M, Clements J, Davidson AH, Drummond AH, Galloway WA, Gilbert R, Gordon JL, Leber TM, Mangan M, Miller K, Nayee P, Owen K, Patel S, Thomas W, Wells G, Wood LM, Woolley K. Processing of tumour necrosis factor-alpha precursor by metalloproteinases. Nature 1994;370: 555–557. 8. McGeehan GM, Becherer JD, Bast RC, Boyer CM, Champion B, Connolly KM, Conway JG, Furdon P, Karp S, Kidao S, McElroy AB, Nichols J, Pryzwansky KM, Schoenen F, Sekut L, Truesdale A, Verghese M, Warner J, Ways JP. Regulation of tumour necrosis factor-alpha processing by a metalloproteinase inhibitor. Nature 1994;370:558 –561. 9. Mohler KM, Sleath PR, Fitzner JN, Cerretti DP, Alderson M, Kerwar SS, Torrance DS, Otten-Evans C, Greenstreet T, Weerawarna K, Kronheim SR, Petersen M, Gerhart M, Kozlosky CJ, March CJ, Black RA. Protection against a lethal dose of endotoxin by an inhibitor of tumour necrosis factor processing. Nature 1994;370:218–220. 10. Powrie F, Leach MW, Mauze S, Caddle LB, Coffman RL. Phenotypically distinct subsets of CD4⫹ T cells induce or protect from chronic intestinal inflammation in C. B-17 scid mice. Int Immunol 1993;5:1461–1471. 11. Morrissey PJ, Charrier K, Braddy S, Liggitt DD, Watson JD. CD4⫹ T cells that express high levels of CD45RB induce wasting disease transferred into congenic severe combined immunodeficient mice. Disease development is prevented by cotransfer of purified CD4⫹ T cells. J Exp Med 1993;178:237–244. 12. Corazza N, Eichenberger S, Eugster HP, Mueller C. Nonlymphocyte-derived tumor necrosis factor is required for induction of colitis in recombination activating gene (RAG)2(⫺/⫺) mice upon transfer of CD4(⫹)CD45RB(hi) T cells. J Exp Med 1999;190: 1479 –1492. 13. Decoster E, Vanhaesebroecks B, Vandenabeele P, Grooten J, Fiers W. Generation and biological characterization of membranebound, uncleavable murine tumor necrosis factor. J Biol Chem 1995;270:18473–18478. 14. Mueller C, Corazza N, Trachsel-Loseth S, Eugster HP, BuhlerJungo M, Brunner T, Imboden MA. Noncleavable transmembrane mouse tumor necrosis factor-alpha (TNFalpha) mediates effects distinct from those of wild-type TNFalpha in vitro and in vivo. J Biol Chem 1999;274:38112–38118. 15. Eugster H-P, Müller M, Karrer U, Car BD, Schnyder B, Eng VM, Woerly G, Le Hir M, di Padova F, Aguet M, Zinkernagel R, Bluethmann H, Ryffel B. Multiple immune abnormalities in tumor necrosis factor and lymphotoxin-alpha double-deficient mice. Int Immunol 1996;8:23–36. 16. Mueller C, Gershenfeld HK, Lobe CG, Okada CY, Bleackley RC, Weissman IL. A high proportion of T lymphocytes that infiltrate H-2-incompatible heart allografts in vivo express genes encoding cytotoxic cell-specific serine proteases, but do not express the MEL-14 defined lymph node homing receptor. J Exp Med 1988; 167:1124 –1136. 17. Wahl SM, Dougherty SF. Measurement of tumor necrosis factor alpha and beta. In: Coligan JE, Kruisbeek AM, Margulies DH, Shevach EM, Strober W, eds. Current protocols in immunology. New York: Wiley & Sons, Inc., 1993:6.10.1– 6.11.6. 18. Marini M, Bamias G, Rivera-Nieves J, Moskaluk CA, Hoang SB, Ross WG, Pizarro TT, Cominelli F. TNF-␣ neutralization ameliorates the severity of murine Crohn’s-like ileitis by abrogation of intestinal epithelial cell apoptosis. Proc Natl Acad Sci U S A 2003;100:8366 – 8371.

September 2004

19. Holtmann MH, Douni E, Schutz M, Zeller G, Mudter J, Lehr HA, Gerspach J, Scheurich P, Galle PR, Kollias G, Neurath MF. Tumor necrosis factor-receptor 2 is up-regulated on lamina propria T cells in Crohn’s disease and promotes experimental colitis in vivo. Eur J Immunol 2002;32:3142–3151. 20. Van den Brande JM, Braat H, van den Brink GR, Versteeg HH, Bauer CA, Hoedemaeker I, van Montfrans C, Hommes DW, Peppelenbosch MP, van Deventer SJ. Infliximab but not etanercept induces apoptosis in lamina propria T-lymphocytes from patients with Crohn’s disease. Gastroenterology 2003;124:1774 –1785. 21. Targan SR, Hanauer SB, van Deventer SJ, Mayer L, Present DH, Braakman T, De Woody KL, Schaible TF, Rutgeerts PJ. A shortterm study of chimeric monoclonal antibody cA2 to tumor necrosis factor alpha for Crohn’s disease. Crohn’s Disease cA2 Study Group. N Engl J Med 1997;337:1029 –1035. 22. van Dullemen HM, van Deventer SJ, Hommes DW, Bijl HA, Jansen J, Tytgat GN, Woody J. Treatment of Crohn’s disease with antitumor necrosis factor chimeric monoclonal antibody (cA2). Gastroenterology 1995;109:129 –135. 23. Keane J, Gershon S, Wise RP, Mirabile-Levens E, Kasznica J, Schwieterman WD, Siegel JN, Braun MM. Tuberculosis associated with infliximab, a tumor necrosis factor alpha-neutralizing agent. N Engl J Med 2001;345:1098 –1104. 24. Kriegler M, Perez C, DeFay K, Albert I, Lu SD. A novel form of TNF/ cachectin is a cell surface cytotoxic transmembrane protein: ramifications for the complex physiology of Tnf. Cell 1988;53:45–53. 25. Perez C, Albert I, DeFay K, Zachariades N, Gooding L, Kriegler M. A nonsecretable cell surface mutant of tumor necrosis factor (TNF) kills by cell-to-cell contact. Cell 1990;63:251–258. 26. Alexopoulou L, Pasparakis M, Kollias G. A murine transmembrane tumor necrosis factor (TNF) transgene induces arthritis by cooperative p55/p75 TNF receptor signaling. Eur J Immunol 1997;27:2588 –2592. 27. Ruuls SR, Hoek RM, Ngo VN, McNeil T, Lucian LA, Janatpour MJ, Korner H, Scheerens H, Hessel EM, Cyster JG, McEvoy LM, Sedgwick JD. Membrane-bound TNF supports secondary lymphoid organ structure but is subservient to secreted TNF in driving autoimmune inflammation. Immunity 2001;15:533–543. 28. Olleros ML, Guler R, Corazza N, Vesin D, Eugster HP, Marchal G, Chavarot P, Mueller C, Garcia I. Transmembrane TNF induces an efficient cell-mediated immunity and resistance to Mycobacterium bovis bacillus Calmette-Guerin infection in the absence of secreted TNF and lymphotoxin-alpha. J Immunol 2002;168:3394–3401. 29. Brynskov J, Foegh P, Pedersen G, Ellervik C, Kirkegaard T, Bingham A, Saermark T. Tumour necrosis factor alpha converting enzyme (TACE) activity in the colonic mucosa of patients with inflammatory bowel disease. Gut 2002;51:37– 43. 30. Kassiotis G, Kollias G. Uncoupling the proinflammatory from the immunosuppressive properties of tumor necrosis factor (TNF) at the p55 TNF receptor level: implications for pathogenesis and therapy of autoimmune demyelination. J Exp Med 2001;193:427– 434. 31. Grell M, Douni E, Wajant H, Lohden M, Clauss M, Maxeiner B, Georgopoulos S, Lesslauer W, Kollias G, Pfizenmaier K, Scheurich P. The transmembrane form of tumor necrosis factor is the prime activating ligand of the 80 kDa tumor necrosis factor receptor. Cell 1995;83:793– 802. 32. Mizoguchi E, Mizoguchi A, Takedatsu H, Cario E, de Jong YP, Ooi CJ, Xavier RJ, Terhorst C, Podolsky DK, Bhan AK. Role of tumor necrosis factor receptor 2 (TNFR2) in colonic epithelial hyperplasia and chronic intestinal inflammation in mice. Gastroenterology 2002;122:134 –144. 33. Piguet PF, Vesin C, Donati Y, Barazzone C. TNF-induced enterocyte apoptosis and detachment in mice: induction of caspases and prevention by a caspase inhibitor, ZVAD-fmk. Lab Invest 1999;79:495–500. 34. Bruewer M, Luegering A, Kucharzik T, Parkos CA, Madara JL, Hopkins AM, Nusrat A. Proinflammatory cytokines disrupt epithe-

TRANSMEMBRANE TNF INDUCES COLITIS

35.

36.

37.

38.

39.

40. 41. 42.

43.

44.

45.

825

lial barrier function by apoptosis-independent mechanisms. J Immunol 2003;171:6164 – 6172. Panja A, Goldberg S, Eckmann L, Krishen P, Mayer L. The regulation and functional consequence of proinflammatory cytokine binding on human intestinal epithelial cells. J Immunol 1998; 161:3675–3684. Decoster E, Vanhaesebroeck B, Vandenabeele P, Grooten J, Fiers W. Generation and biological characterization of membranebound, uncleavable murine tumor necrosis factor. J Biol Chem 1995;270:18473–18478. Mizoguchi E, Mizoguchi A, Takedatsu H, Cario E, De Jong YP, Ooi CJ, Xavier RJ, Terhorst C, Podolsky DK, Bhan AK. Role of tumor necrosis factor receptor 2 (TNFR2) in colonic epithelial hyperplasia and chronic intestinal inflammation in mice. Gastroenterology 2002;122:134 –144. Kontoyiannis D, Boulougouris G, Manoloukos M, Armaka M, Apostolaki M, Pizarro T, Kotlyarov A, Forster I, Flavell R, Gaestel M, Tsichlis P, Cominelli F, Kollias G. Genetic dissection of the cellular pathways and signaling mechanisms in modeled tumor necrosis factor-induced Crohn’s-like inflammatory bowel disease. J Exp Med 2002;196:1563–1574. Holtmann MH, Schutz M, Galle PR, Neurath MF. Functional relevance of soluble TNF-alpha, transmembrane TNF-alpha and TNFsignal transduction in gastrointestinal diseases with special reference to inflammatory bowel diseases. Z Gastroenterol 2002; 40:587– 600. Mueller C. Tumour necrosis factor in mouse models of chronic intestinal inflammation. Immunology 2002;105:1– 8. Van Deventer SJ. Tumour necrosis factor and Crohn’s disease (see comments). Gut 1997;40:443– 448. Haraldsen G, Kvale D, Lien B, Farstad IN, Brandtzaeg P. Cytokineregulated expression of E-selectin, intercellular adhesion molecule-1 (ICAM-1), and vascular cell adhesion molecule-1 (VCAM-1) in human microvascular endothelial cells. J Immunol 1996;156: 2558 –2565. Garton KJ, Gough PJ, Blobel CP, Murphy G, Greaves DR, Dempsey PJ, Raines EW. Tumor necrosis factor-alpha-converting enzyme (ADAM17) mediates the cleavage and shedding of fractalkine (CX3CL1). J Biol Chem 2001;276:37993–38001. Dekkers PE, Lauw FN, ten Hove T, te Velde AA, Lumley P, Becherer D, van Deventer SJ, van der Poll T. The effect of a metalloproteinase inhibitor (GI5402) on tumor necrosis factoralpha (TNF-alpha) and TNF-alpha receptors during human endotoxemia. Blood 1999;94:2252–2258. Williams LM, Gibbons DL, Gearing A, Maini RN, Feldmann M, Brennan FM. Paradoxical effects of a synthetic metalloproteinase inhibitor that blocks both p55 and p75 TNF receptor shedding and TNF alpha processing in RA synovial membrane cell cultures. J Clin Invest 1996;97:2833–2841.

Received August 28, 2003. Accepted May 27, 2004. Address requests for reprints to: Christoph Mueller, Ph.D., Division of Immunopathology, Institute of Pathology, Murtenstrasse 31, CH-3010 Bern, Switzerland. e-mail: [email protected]; fax: (41) 31-381-8764. The authors thank Claudio Vallan for excellent assistance in cell sorting; Susanne Eichenberger for expert histopathologic analysis; Stefan Müller, Chris Wasem, and Myriam Bühler for discussions and assistance; Marianne Bärtschi for the preparation of tissue sections; Dr. Bernhard Holzmann, Munich, and Dr. Rudolf Lucas, Konstanz, Germany, for providing primary antibodies; and Ernst Wagner, HansPietro Eugster, and Horst Bluethmann for providing mouse strains. Supported by grants 31-53961.98 and 31-65307.01 from the Swiss National Science Foundation and a Senior Research Award from the Colitis and Crohn’s Foundation of America (to C.M.).