METHODS

Mice

Heterozygous (IL-2^+/−) mice²² from a mixed C57BL/6 H129/Ola background were crossed to obtain IL-2^−/− and IL-2^+/+ mice. Mice were bred under SPF conditions in a barrier sustained facility, or in a completely GF environment at the University of Ulm, Ulm, Germany. The GF state was controlled weekly by microbiological investigations, including culture methods for aerobic and anaerobic bacteria, Gram stain examinations of faeces and intestinal contents, as well as a broad range eubacterial polymerase chain reaction (PCR) of stool samples from mice. The presence of Helicobacter spp could therefore be excluded. Offspring were screened for IL-2 mutation using PCR. In most experiments, littermate controls were used.

Determination of cytokine mRNA expression in intestinal tissue

The intestine was cut into pieces and transferred to 3 ml of Trizol reagent (Gibco BRL Life Technologies, Berlin, Germany).¹⁹ RNA isolation was performed according to the manufacturer's instructions. Extracted RNA was resolved in water containing 0.1% diethyl-pyrocarbonate (DEPC). RNA (20 μg) was used for reverse transcription (RT). RNA solution was mixed with 1 μl of oligo(dT) (Gibco BRL), and DEPC water was added to a final volume of 10 μl. This mixture was incubated at 65°C for 10 minutes. Then, 10 μl of a solution containing 5× reverse transcriptase buffer (100 mM Tris HCl (pH 8.3), 150 mM KCl, 6 mM MgCl₂; Gibco), 40 U RNAsin (Promega Biotec, Madison, Wisconsin, USA), 20 mM dithiothreitol (Gibco), 200 U Superscript II RT RNAse H reverse transcriptase (Gibco BRL), and 2 mM deoxynucleoside triphosphate (dNTP; Roth, Karlsruhe, Germany) were added. The resulting 20 μl were incubated at 37°C for 60 minutes. Finally, the tubes were heated to 90°C for five minutes and 180 μl of DEPC water were added.

PCR assisted amplification of cDNA

cDNA (5 μl) was mixed with a solution containing 1 U Taq DNA polymerase (AmpliTaq; Perkin-Elmer, Branchburg, New Jersey, USA), 200 mM dNTP (Roth), 20 pmol 5′ and 3′ primers (Metabion, Munich, Germany), and Taq DNA polymerase buffer (Perkin-Elmer). This mixture was run on a thermal cycler (Gene Amp 2400; Perkin Elmer) for 20–40 cycles applying the following conditions: denaturation 30 seconds at 94°C, annealing 45 seconds at 60°C, and extension 60 seconds at 72°C. The PCR products were visualised on 2% agarose gels. The primers used were: β-actin, sense TGG AAT CCT GTG GCA TCC ATG AAA C, antisense TAA AAT GCA GCT CAG TAA CAG TCC G; TNF-α sense AGC CCA CGT CGT AGC AAA CCA CCA A, antisense ACA CCC ATT CCC TTC ACA GAG CAA T; IFN-γ sense TGA ACG CTA CAC ACT GCA TCT TGG, antisense TGA CTC CTT TTC CGC TTC CTG AG. The gels were photographed under UV illumination on a Fluoro-S-Multiimager (BioRad, Hercules, California, USA). Quantification of fluorescence signal per band was performed using the manufacturer's software (Multy Analyst). To compare various probes, values were normalised to β-actin.

Preparation of intestinal lymphocytes

Intraepithelial (IEL) and lamina propria lymphocytes (LPL) were prepared from the colon of SPF mice as described by Davies and Parrott²³ with slight modifications. Tissue pieces were washed with Hank's balanced salt solution (BSS), incubated in BSS containing 5 mM dithiothreitol for 10 minutes while stirring to remove mucus, and incubated at 37°C on a shaker in BSS without Ca²⁺ Mg²⁺ containing 1 mM EDTA. The supernatant containing liberated epithelial cells was washed, and lymphocytes were enriched by density gradient centrifugation. The cells were suspended in Percoll 1.055 and layered carefully on top of Percoll 1.085 (Pharmacia, Uppsala, Sweden). The tubes were centrifuged at 560 g at room temperature for 20 minutes. IEL were recovered from the interphase, washed twice, counted, and suspended in Clicks medium. The remaining tissue pieces were digested in the presence of collagenase II and VIII (100 U/ml each; Boehringer Mannheim, Germany) and hyaluronidase (300 U/ml; Boehringer Mannheim). The resulting suspension was enriched for lymphoid cells by density gradient centrifugation as described above. Cells recovered from the interphase (LPL) were washed twice and suspended in Clicks medium. Cell viability was >95%, as determined by trypan blue exclusion dye.

Preparation of single cell suspensions from spleen

The spleen was removed, disrupted by passage through a sterile stainless steel mesh, and erythrocytes were lysed using a standard procedure. The cells were washed three times, counted, and suspended in Clicks medium.

Splenocytes obtained from SPF IL-2^+/− mice were irradiated with 3500 rad and used as antigen presenting cells.

Flow cytometry

Cells were washed three times with phosphate buffered saline (PBS) containing 1% bovine serum albumin and 0.1% NaN₃. Cells were resuspended in 10 μl of antibody containing solution and incubated at 4°C for 30 minutes. The following monoclonal antibodies (mAbs) were used: anti-CD3 mAb-PE (Pharmingen, Hamburg, Germany), anti-CD4 mAb-biotin (clone YTL 191), anti-CD8 mAb-FITC (clone YTL 169), anti-αβTCR mAb-FITC (clone 5759721), and anti-γδTCR mAb-biotin (clone GL3). Then, PBS was added, and cells were pelleted and washed twice with PBS. Cells were resuspended in 10 μl PBS containing 2.75% streptavidin Cy-Chrome (Pharmingen) and incubated for 30 minutes at 4°C. After several washes the cells were resuspended in 1 ml PBS and analysed on a Coulter Epics XL (CoulterImmunotech, Krefeld, Germany). Propidium iodide was added to controls and viable cells were gated. At least two independent experiments were performed per group of mice. A minimum of 2000 CD3⁺ cells were analysed per measurement.

Caecal antigen

Caecal antigens were prepared as described elsewhere.¹⁶ Briefly, the contents of the caecum from 20 SPF IL-2^+/− mice were pooled in 5 ml of PBS. DNAse I (100 μg/ml; Boehringer Mannheim) was added and the mixture was vortexed for 30 minutes. Subsequently, bacterial cells were disrupted by sonification for five minutes on ice (Cell disruptor B15; Branson, Schwäbisch Gmünd, Germany). The suspension was then centrifuged to remove particles and passed through a 0.22 μm filter. Sterility was confirmed by incubation on agar plates under aerobic and anaerobic conditions.

Bacteroides vulgatus antigen

Bacteroides vulgatus was grown from the colon of SPF IL-2^−/− mice kept in our facility. The strain was identified and characterised by standard laboratory procedures, including API Rapid ID (BioMerieux) as well as sequencing of 16S rDNA, as described elsewhere.^24,25 Bacteria were grown on blood agar plates at 37°C under anaerobic conditions for two days. Bacteria were washed from the plates with sterile PBS, passed through a cell strainer (70 μm; Falcon) to retain agar pieces, and incubated at 60°C for two hours. This suspension was stored at 4°C and used as the antigen preparation. The sterility of the antigen preparation was established by growing the bacterial suspension on blood agar plates. Concentration of protein was determined by colorimetric reaction with Bradford reagent (BioRad, München, Germany).

Cytokine production by mononuclear cells

Splenocytes, mesenteric lymph nodes cells, LPL, and IEL (10⁶/ml) were cultured in the presence of 2×10⁶/ml irradiated splenic cells in Clicks/RPMI 1640 medium (Biochrom, Berlin, Germany) routinely supplemented with 2 mM l-glutamine (Biochrom), 10 mM HEPES (Biochrom), 5×10⁻⁵ 2-ME (Biochrom), 10 μg/ml streptomycin (Biochrom), 100 U/ml penicillin (Biochrom), 10% heat inactivated fetal calf serum (Biochrom) and 30 μg/ml gentamicin, with or without addition of 30 μg/ml of bacterial antigen. Human recombinant IL-2 (20 U/ml) was also added. As positive controls, cells were stimulated with 3 μg/ml Con A or 10 μg/ml anti-CD3 mAb (14.5 2c11). After 44 and 70 hours, supernatants were harvested. In some of the experiments anti-IL-10 mAbs (R&D Systems, Wiesbaden, Germany) were added to the cultures at a final concentration of 0.0125 μg/ml. The ED₅₀ for in vitro inhibition of the biological effects of 2.5 ng/ml recombinant IL-10 is 0.005–0.015 μg/ml. Where indicated, 10 μg/ml polymyxin B was added to the cultures to neutralise lipopolysaccharide. At least two independent experiments were performed per group.

The supernatants collected after 48 hours were analysed for IFN-γ concentration by ELISA, as reported previously,²⁶ using anti-IFN-γ mAb clones AN18 and R46A2. Finally, 50 μl/well of a solution containing streptavidin and biotinylated alkaline phosphatase were incubated for 45 minutes at 37°C. Then, a solution of 1 mg/ml p-nitrophenyl-phosphate (Sigma, St Louis, Missouri, USA) was added and optical density was measured. Concentrations of IFN-γ protein were calculated on the basis of a IFN-γ standard curve and linear regression. The lower limit of detection was 1.0 ng/ml. IL-10 was determined by ELISA purchased from R&D Systems (Wiesbaden, Germany). The ELISA was performed according to the manufacturer's instructions. The limit of detection of IL-10 was 30 pg/ml. IL-4 was measured by bioassay using the IL-4 sensitive cell line CT4S, as described previously.²⁷ The limit of detection was 1 U/ml.

Adoptive transfer and tracking of T cells in vivo

Cells from mesenteric, axial, cerebral, and inguinal lymph nodes from IL-2^+/− mice were incubated for 15 minutes at 4°C with the following antibodies: YTL 169 (anti-CD8), M 1.70 (anti-MAC-1), and PC61 (anti-CD25). Lymph node cells were washed and added to petri dishes coated overnight with 10 μg/ml of rabbit antimouse Ig antibody (Dako, Hamburg, Germany). Non-adherent cells were resuspended after 30 minutes by gently shaking and harvesting the suspension. This process was repeated once. The purity of this preparation was examined by FACS analysis. At least 95% of cells were T cells of which at least 97% were CD4⁺CD25⁻. The cells were labelled with 20 μCi/ml ⁵¹Cr (ICN-Biochemicals, Eschwege, Germany). Then, cells were purified by density gradient centrifugation (Nykodenz 17%; Nykomed, Oslo, Norway), washed three times, and injected intravenously into recipient mice (10⁶ cells/mouse). After one hour, animals were sacrificed, the respective organs removed, and radioactivity per organ was measured on a gamma counter (1480 Wizard; Wallac, Turku, Finland). The results were corrected for background counts and expressed as percentage of the total amount of radioactivity applied per mouse. Mean (SD) values were determined from 3–7 mice. In some experiments mice were injected intraperitoneally with 250 μg Fab fragment of MECA₃₆₇ mAb^28,29 or 400 μg anti-VCAM-1 mAb (IgG)³⁰ 2–3 hours before transfer of the cells. Control animals were injected with PBS.

Cell lines and adoptive transfer

Splenocytes and LPL from IL-2^+/+ and IL-2^−/− mice were stimulated with caecal bacterial antigen (CBA) at 10 μg/ml every 14 days. Recombinant IL-2 was added at 20 U/ml. For determination of cytokine production, cells were cultured without stimulus, with ConA (3 μg/ml), or in anti-CD3 mAb coated wells (10 μg/ml). For these experiments, cells were used 14 days after antigen stimulation. After 48 and 72 hours the supernatants were collected and assayed for cytokine content by ELISA, as described above. For adoptive transfer experiments, cells were used nine days after antigen stimulation. Before transfer, cells were purified by Ficoll density gradient centrifugation. Viable cells were above 98%, as determined by trypan blue exclusion. Then, 8×10⁶ cells were injected intravenously into recipient mice without clinical signs of colitis. The recipient mice were examined at regular intervals for clinical signs of colitis. The following score was used to assess the clinical status of the mice: hunched posture: 0=absent, 1=present; movement: 0=normal, 1=reduced; fur: 0=normal, 1=bristled; rectal prolapse: 0=absent, 4=present; faeces: 0=normal, 2=liquid, 4=blood; body weight loss: 0=<10%; 2=10–20%; 4=>20%. An index of <3 was considered as slight or no colitis, >3 to 5 as moderate colitis, and >5 as severe colitis.

Statistics

Differences between mean values were analysed using the Student's t test. Survival times after the adoptive transfer of cell lines were compared applying the rank log test and the Cox test. Values of p<0.05 were considered significant.

RESULTS

Increased cytokine mRNA expression in the colon of SPF IL-2^−/− but not GF IL-2^−/− or SPF IL-2^+/+ mice

To analyse the impact of the indigenous microflora on intestinal cytokine mRNA expression, levels of mRNA of various cytokines in the colon of 20 week old GF and SPF IL-2^+/+ and IL-2^−/− mice were compared. The results are expressed as a ratio of β-actin. As shown in fig 1A, significantly increased mRNA expression of IFN-γ and TNF-α was found in the colon of SPF IL-2^−/− mice compared with SPF IL-2^+/+ mice (21 and 4.8-fold increase, respectively; p<0.05). In contrast, there was no significant difference in cytokine mRNA levels between GF IL-2^+/+ and IL-2^−/− mice.

Based on these results, we hypothesised that colitis in GF IL-2^−/− mice might develop later than in SPF mice or may not occur at all. To test this hypothesis, we examined colonic mRNA expression levels of 33 week old mice. As shown in fig 1B, no significant differences in colonic IFN-γ and TNF-α mRNA expression were found between 33 week old GF IL-2^+/+ and IL-2^−/− mice. Nevertheless, the slight increase in cytokine mRNA expression of aged GF IL-2^−/− mice is in keeping with the mild focal infiltrations³¹ of the colon detected by histological examinations (data not shown). In summary, these results confirm that the GF state prevents the development of colitis in IL-2^−/− mice.

αβT cells are increased in intestinal IEL and LPL of SPF IL-2^−/− mice but less pronounced in GF IL-2^−/−

CD4 T cells have been demonstrated to be involved in colitis in IL-2^−/− mice.³² As the GF state may modulate the development of the GALT,^33–35 T cell subpopulations of IEL and LPL were analysed by flow cytometry in 20 week old GF and SPF IL-2^+/+ and IL-2^−/− mice.

The average yield of IEL isolated from the colon of SPF IL-2^+/+ and IL-2^−/− mice was 3.8 (2.2)×10⁶ and 9.8 (5.5)×10⁶ cells per mouse, respectively. Among IEL isolated from the colon of SPF IL-2^−/− mice, we found a pronounced increase in CD4⁺ αβTCR⁺ T cells compared with SPF IL-2^+/+ mice (39.5 v 5.6% and 93.4 v 74.4%, respectively; p< 0.05) (fig 2). In parallel, CD8⁺ and γδTCR⁺ T cells were decreased (56.9 v 69.8% (p<0.05) and 7.8 v 23.2%, respectively). In contrast, IEL from GF IL-2^−/− mice showed only non-significantly increased numbers of CD4⁺ T cells (18.3 v 13.6%), and less abundant γδ T cells (12.9 v 20.6%) compared with GF IL-2^+/+ mice. In contrast with the observations in SPF mice, GF IL-2^−/− mice showed a significantly higher percentage of CD8⁺ αβTCR⁺ T cells (67.0 v 43.8%, and 84.3 v 69.6 %, respectively; p<0.05) compared with GF IL-2^+/+ mice (fig 2). The average yield of IEL isolated from the colon of GF IL-2^+/+ and IL-2^−/− mice was 5.7 (1.3)×10⁶ and 5.7 (2.4)×10⁶ cells per mouse, respectively.

Colon LPL from SPF IL-2^−/− mice contained significantly less γδTCR⁺ (1.6 v 7.2%; p<0.05) and more αβTCR⁺ cells (97.1 v 90.4%; p<0.05) compared with SPF IL-2^+/+ mice. Moreover, CD8⁺ LPL were significantly increased in IL-2^−/− mice (31.4 v 20.1%; p<0.01) while CD4⁺ cells remained largely unchanged (fig 2). The average yield of LPL isolated from the colon of SPF IL-2^+/+ and IL-2^−/− mice was 3.6 (1.3)×10⁶ and 22.8 (17.3)×10⁶ cells per mouse, respectively.

Although the average yield of LPL isolated from the colon of GF mice was comparable for IL-2^+/+ and IL-2^−/− mice (6.1 (1.4)×10⁶v 7.0 (3.7)×10⁶ cells per mouse), relative proportions of the T cell subpopulations showed slight but significant differences (fig 2). CD4⁺ T cells were found significantly less frequently in GF IL-2^−/− mice while CD8⁺ T cells were significantly increased (61.6 v 71.1% (p<0.05) and 35.8 v 16.7% (p<0.01)). Together, these data implicate an increased influx of peripheral T cells in the colon of IL-2^−/− mice with colitis.

Increased migration of adoptively transferred CD4⁺CD25⁻ T cells into the colon of SPF IL-2^−/− mice is mediated by MAdCAM-1

The abovementioned observations argue for an increased influx of T cells into the colon of SPF IL-2^−/− mice in the course of intestinal inflammation. Therefore, migration of T cells in SPF and GF IL-2^−/− and IL-2^+/+ mice was directly investigated. For this purpose, CD4⁺CD25⁻ T cells were ⁵¹Cr labelled and adoptively transferred into SPF and GF IL-2^−/− and IL-2^+/+ mice, respectively. After one hour, recipients were sacrificed and radioactivity of various organs was determined. The results depicted in fig 3 demonstrate that the influx of T cells into the colon of 10–13 week old SPF IL-2^−/− mice was significantly increased compared with SPF IL-2^+/+ or GF IL-2^−/− mice (1.57 v 0.73% and 0.53%, respectively; p<0.05 and p<0.01). The increased T cell migration into the colon was even more pronounced in older mice at 15–20 weeks of age (3.28 v 0.5%; p<0.01) (fig 3C). At 5–8 weeks of age, when the first histological manifestations of inflammation appeared, there was a slight statistically insignificant increase in migration of T cells into the colon of SPF IL-2^−/− mice compared with SPF IL-2^+/+ and GF IL-2^−/− mice (0.66 v 0.41% and 0.39%) (fig 3A). The lymph nodes of SPF IL-2^−/− mice also displayed an age dependent increased influx of radiolabelled cells. Examination of the mesenteric lymph nodes revealed that retention of radiolabelled cells in SPF IL-2^−/− mice increased with age which is in agreement with the lymphoproliferative disease in these animals.⁷ At the age of 5–8 and 10–13 weeks, retention of radiolabelled cells in the small intestine of SPF IL-2^−/− mice was significantly less than that in SPF IL-2^+/+ mice (1.2 v 1.9% and 1.8 v 3.6%, respectively), possibly due to competition of the large and small intestine for cells with similar “homing” properties.

In another series of experiments, blocking anti-MAdCAM-1 mAb^28,29 or blocking anti-VCAM-1 mAb³⁰ were injected intraperitoneally into SPF mice prior to the adoptive transfer of T cells. As shown in fig 4A, blocking of MAdCAM-1 significantly reduced migration of cells to the intestine as well as to the mesenteric lymph nodes. Inhibition by anti-MAdCAM-1 mAbs was comparable in healthy IL-2^+/+ and diseased IL-2^−/− mice (69 and 79%, respectively). In contrast, administration of anti-VCAM-1 mAbs did not significantly modulate T cell migration (fig 4B). Taken together, these data show that colitis in SPF IL-2^−/− mice is associated with an increased influx of CD4⁺CD25⁻ αβT cells, which is largely mediated by the addressin MAdCAM-1 but not VCAM-1.

Increased spontaneous IFN-γ production by IEL and LPL of SPF IL-2^−/− mice is not suppressed by microflora induced IL-10

To determine whether T cells from mice with colitis may be reactive with antigens from the indigenous bacterial microflora, we determined IFN-γ and IL-10 production in mononuclear cells from spleen, mesenteric lymph nodes, and colonic IEL and LPL of SPF mice on exposure to microflora antigens. Heat killed Bacteroides vulgatus was selected as an antigen because B vulgatus may trigger colitis.^14,15 In addition, a sonicate of the caecal content of SPF IL-2^+/+ mice termed CBA was included, as previously described by Cong and colleagues.¹⁶

As depicted in fig 5A, IEL and LPL from IL-2^−/− but not those from IL-2^+/+ mice produced significant amounts of IFN-γ without antigen stimulation in vitro (7.1 v <1.0 ng/ml and 30.2 v 4.1 ng/ml, respectively; p<0.05 for LPL). Likewise, anti-CD3 stimulated IEL and LPL from the colon of SPF IL-2^−/− mice secreted much higher amounts of IFN-γ than IEL or LPL from IL-2^+/+ mice (40.6 v <1.3 ng/ml and 211.7 v 15.3 ng/ml, respectively). These results are in keeping with cytokine mRNA analysis and suggest a role for Th1 cells in colitis of IL-2^−/− mice. Exposure of IEL or LPL to microflora antigens in the presence of antigen presenting cells did not induce additional IFN-γ production in IEL and LPL from IL-2^+/+ or IL-2^−/− mice. However, addition of microflora antigens induced a slight decrease in IFN-γ production in some of our experiments although IFN-γ levels produced by IEL and LPL from IL-2^−/− mice were still much higher than those produced by IEL and LPL from IL-2^+/+ mice. Conversely, IEL, LPL, and spleen cells from both IL-2^+/+ and IL-2^−/− mice produced high levels of IL-10 in the presence of microflora antigens (fig 5B). IEL and MLN from IL-2^−/− mice produced even higher IL-10 levels than those from IL-2^+/+ mice.

To address whether IL-10 modulates IFN-γ production in response to flora related antigens, neutralising anti-IL-10 mAb were added to the cultures. As depicted in fig 6, anti-IL-10 antibodies did not modulate IFN-γ production in vitro.

As splenocytes produced higher quantities of IL-10 in response to heat killed B vulgatus than to anti-CD3 mAb, it may be that cells other than T cells contributed to IL-10 production. To neutralise lipopolysaccharide in our system, cells were stimulated in the presence or absence of polymyxin B. Addition of polymyxin B did not decrease the amount of IL-10 secreted in response to heat killed B vulgatus, suggesting that the production of IL-10 is not due to lipopolysaccharide stimulation of B cells (data not shown). Depletion of LPL from CD4⁺ T cells resulted in a variable decrease in IL-10 production in response to heat killed B vulgatus, suggesting that CD4⁺ T cells are only partially involved in IL-10 production (data not shown).

Adoptive transfer of microflora specific T cell lines

To investigate whether microflora specific T cells may induce colitis in IL-2^−/− mice, microflora specific LPL T cell lines from the colon of SPF IL-2^−/− mice with colitis were established by stimulation of these cells with CBA. The resulting cell line (LCK98) was CD3⁺CD4⁺αβTCR⁺, and produced high quantities of IL-10, some transforming growth factor β1, and very low quantities of IFN-γ and IL-4 (table 1). These cells were adoptively transferred into SPF IL-2^+/+ and IL-2^−/− mice, the latter of which were used at 6–10 weeks of age prior to the appearance of clinical signs of colitis. The mice were observed for up to five months for clinical manifestations of colitis. As shown in fig 7, transfer of T cell line LCK98 did not significantly change the survival of IL-2^−/− recipient mice, although slightly prolonged survival was found. Consistently, the development of colitis was not changed in these mice, as determined by a clinical colitis score (fig 8). Transfer of these cells into healthy IL-2^+/+ mice did not exhibit any clinical effect (data not shown).

Further T cell lines were established from spleen cells of IL-2^−/− and IL-2^+/+ mice (SW98.1 and SK98.2, respectively). These CD3⁺ αβTCR⁺ cell lines produced substantial levels of both IFN-γ and IL-10 (table 1). IL-2^−/− mice that received SW98.1 cells showed a trend (p<0.1) towards shortened survival. However, the colitis score of these mice was not significantly different compared with control mice (fig 8).

DISCUSSION

Several animal models of IBD have shown that colitis is triggered by the indigenous intestinal microflora.^36–38 Thus while for example GF IL-2^−/− mice do not develop disease, SPF IL-2^−/− mice develop colitis by an average age of 12–15 weeks.^7,39 CD4⁺ T cells are thought to play a crucial role in the development of colitis in IL-2 deficient mice as mice deficient in both IL-2 and T cells fail to develop colitis whereas IL-2^−/−×JH1^−/− B cell deficient mice⁴⁰ as well as IL-2^−/−×β2 microglobulin^−/− mice develop colitis.³² Here, we have analysed functions of intestinal T cells from IL-2^−/− mice with a special emphasis on their reactivity with microflora antigens. The most important findings of this study were: (i) in contrast with SPF IL-2^−/− mice, GF IL-2^−/− did not exhibit significantly increased proinflammatory mRNA expression; (ii) CD4⁺ CD25⁻ αβTCR⁺ cells migrate more frequently into the colon of SPF IL-2^−/− mice compared with GF IL-2^−/− mice, which was mediated mainly by MAdCAM-1, but not VCAM-1; (iii) IEL and LPL from IL-2^−/− mice spontaneously produce high levels of IFN-γ in vitro; and (iv) IEL and LPL from IL-2^−/− produce high quantities of IL-10 but no IFN-γ on stimulation with microflora antigens which however did not reduce IFN-γ production.

Studies in other animal models of IBD^37,41 as well as data obtained from patients with IBD^17,18 argue for loss of T cell tolerance against antigens from the intestinal microflora. Thus it is believed that CD4⁺ T cells of the GALT become reactive towards microflora antigens. To address this issue, we focused our interest on T cells within the IEL and LPL populations of the colon of SPF and GF IL-2^−/− mice. To elucidate T cell reactivity towards microflora antigens, IEL and LPL were stimulated with CBA, as described by Cong and colleagues.¹⁶ Furthermore, B vulgatus was included as an antigen due to its ability to induce colitis in other models of IBD.^14,15 These experiments revealed that IEL and LPL of IL-2^−/− mice spontaneously produced high quantities of IFN-γ which were not increased on in vitro stimulation with microflora antigens.

In the search for mediators that may modulate IFN-γ, we determined IL-10 production in cultures of intestinal mononuclear cells, as IL-10 is known to inhibit IFN-γ production by T cells and is thought to play an important role in the immunoregulation of the GALT.^42–44 IL-10 was produced in high levels by IEL and LPL in response to CBA and heat killed B vulgatus. Therefore, it seemed conceivable that IL-10 is produced in an antigen specific manner. To test whether IL-10 modulates IFN-γ production in our system, experiments were performed in the presence and absence of a monoclonal antibody which blocks IL-10 bioactivity. The presence of anti-IL-10 mAb did not increase IFN-γ levels spontaneously secreted by colon LPL from IL-2^−/− mice. Likewise, IFN-γ secretion was not increased on antigen stimulation. Regulatory T (Tr1) cells have recently been described as playing an important role in immunoregulation of the GALT. Tr1 cells emerge in vitro after repeated restimulation in the presence of IL-10 and produce large amounts of IL-10 as well as residual levels of IFN-γ and no IL-4.^43,45 Therefore, we were interested in the cellular source of IL-10 produced by colon LPL. To address this issue, we depleted colon LPL of CD4⁺ T cells and neutralised lipopolysaccharide by polymyxin B to exclude non-specific IL-10 induction by this bacterial component.⁴⁶ As both modifications did not consistently modulate IL-10 production, the cellular source of IL-10 is not yet clear and requires further investigation. However, we cannot exclude the fact that the in vitro experimental conditions may have been inappropriate for intestinal T cell stimulation although comparable conditions were used previously by others.^16,41

Our results suggest that in IL-2^−/− mice the microflora triggers production of proinflammatory cytokines, including TNF-α and IFN-γ. Although IL-10 producing Tr1 cells are present in the intestinal tissue of IL-2^−/− mice, IL-10 does not efficaciously exert a significant negative regulatory effect on IFN-γ production. Whether this is due to an imbalance of Th1 and Th2 cytokines present in the colon tissue or because of disturbed IL-10 receptor function in colonic T cells of IL-2^−/− mice is not known.

Flow cytometric analysis of IEL and LPL populations in the colon of SPF and GF IL-2^−/− and IL-2^+/+ mice revealed increased numbers of αβTCR⁺CD4⁺ T cells in SPF IL-2^−/− mice. In keeping with these observations, we found increased migration of adoptively transferred CD4⁺CD25⁻ T cells into the colon and mesenteric lymph nodes of SPF IL-2^−/− compared with SPF IL-2^+/+ and GF IL-2^−/− mice. Whether accumulation of cells in the mesenteric lymph nodes is secondary to colonic inflammation associated with increased recruitment and local proliferation or due to the systemic lymphoproliferative disease observed in this animal model⁷ cannot be answered from the data presented in this study. Furthermore, increased numbers of radiolabelled cells in the mesenteric lymph nodes may reflect increased efflux of cells from the inflamed colon. Migration of T cells into the intestine is largely dependent on MAdCAM-1 in healthy mice.⁴⁷ Similar observations have also been made in our experiments as well as in IL-10 deficient mice.⁴⁸ In contrast, administration of anti-VCAM-1 antibodies did not modulate T cell migration in the IL-2^−/− mouse whereas it reduced the influx of T cells into the inflamed colon of the IL-10^−/− mouse (A Hamann, personal communication). Therefore, the role of VCAM-1 remains a matter of controversial discussion in various models of IBD. The increased influx of CD4⁺ T cell into the colon of IL-2^−/− mice also suggests that MAdCAM-1 was upregulated. This should be confirmed by further histological studies. Nevertheless, these findings may be of therapeutic importance as they may provide a site specific target for anti-inflammatory intervention. Future studies will have to show if blocking MAdCAM-1 is a promising therapeutic approach in modulating the course of colitis. In other models of IBD, a similar approach to inhibiting ICAM-1 was beneficial.⁴⁹ Therefore, other adhesion molecules are also likely to participate in the mediation of increased intestinal T cell migration.

To study the function of T cells in our model in vivo, microflora reactive T cell lines were established from colon LPL of IL-2^−/− mice with colitis.¹⁶ These CD4⁺ αβTCR⁺ T cell lines produced very high levels of IL-10, some transforming growth factor β1 but only marginal IFN-γ and IL-2 whereas T cell lines established from spleen cells of IL-2^−/− and IL-2^+/+ mice produced less IL-10 but much higher IFN-γ levels. To test whether these T cells would modulate the course of colitis, adoptive transfer experiments were performed. IL-2^+/+ recipient mice did not develop signs of colitis on transfer of colon LPL from IL-2^−/− mice. Likewise, the adoptive transfer of LPL from IL-2^−/− mice into IL-2^−/− recipient mice did not significantly accelerate the development of the disease but rather increased mean survival time. As these cells produced high levels of IL-10, some transforming growth factor β1, but only marginal levels of IFN-γ, these cells can be considered as Tr1 cells. Consistently, these cells did not exacerbate the disease but rather exerted a beneficial effect. However, this is in contrast with observations reported by Cong and colleagues.¹⁶ Thus in C3H/Bir mice, microflora reactive CD4 T cells from spleen and mesenteric lymph nodes induced colitis on adoptive transfer into SCID mice. These cells produced IL-2, IL-3, IFN-γ, but not IL-4, and thus can be considered as Th1 cells. However, it is not known whether these cells also produced IL-10.

Our results as well as data of others^32,40 suggest that in IL-2^−/− mice there is at least one T cell population that produces proinflammatory cytokines such as IFN-γ which in concert with other proinflammatory cytokines such as TNF-α possibly produced by macrophages, promote bowel inflammation. We did not obtain evidence that such a proinflammatory colitogenic T cell population is reactive with a particular bacterial antigen. On the other hand however we have presented evidence that even in the inflamed colon of IL-2^−/− mice there is a second regulatory T cell population (possibly Tr1 cells present which produce substantial quantities of IL-10 on stimulation with microflora antigens). However, IL-10 does not significantly reduce IFN-γ levels produced in vitro, suggesting that the proinflammatory T cell population may be unresponsive to IL-10. It is not yet clear whether these results can be extrapolated to the in vivo situation. In fact, cytokine mRNA analysis revealed that in IL-2^−/− mice there is increased mRNA expression of both IFN-γ and IL-10 in the colon.¹⁹ At present, the role of IL-10 is not clear and should be further investigated to elucidate the pathomechanisms operating in colon inflammation in IL-2^−/− mice. Moreover, macrophages can be stimulated by bacterial products such as lipopolysaccharide⁵⁰ and might play an important role in the early phase of colitis in IL-2^−/− mice as cytokines such as IL-12,⁵¹ IL-1, and TNF-α are involved and may precede Th1 cytokine expression in situ.^19,21 Taken together, these data and the data presented in this study suggest that T cell activation in IL-2^−/− mice is not directly triggered by flora antigens but rather is secondary to initial activation of the innate immune system.

Finally, data on mRNA levels in colon tissue of SPF and GF IL-2^−/− mice confirmed the absence of significant inflammation in GF IL-2^−/− mice.^7,31,39 Nevertheless, GF mice developed a mild focal inflammation at 33 weeks of age which was accompanied by only marginally increased Th1 cytokines. Therefore, in addition to the microflora, food antigens should be taken into account in the pathogenesis of IBD. This issue however remains to be elucidated in mouse models of IBD.