Role of Transcription Factor T-bet Expression by CD4⁺ Cells in Gastritis Due to Helicobacter pylori in Mice

Abstract

Gastritis due to Helicobacter pylori is induced by a Th1-mediated response that is CD4 cell and gamma interferon (IFN-γ) dependent. T-bet is a transcription factor that directs differentiation of and IFN-γ secretion by CD4⁺ Th1 T cells. The goal of this study was to use two mouse models to elucidate the role of T-bet in gastritis due to H. pylori. C57BL/6J mice, congenic T-bet knockout (KO) mutants, or congenic SCID (severe, combined immunodeficient) mutants were given live H. pylori by oral inoculation. SCID mice were given CD4⁺ splenocytes from C57BL/6J or T-bet KO mice by intraperitoneal injection. Twelve or 24 weeks after bacterial inoculation, C57BL/6J mice developed moderate gastritis but T-bet KO mice and SCID mice did not. In contrast, SCID recipients of either C57BL/6J T cells or T-bet KO T cells developed gastritis 4 or 8 weeks after adoptive transfer. In recipients of C57BL/6J CD4⁺ cells but not recipients of T-bet KO cells, gastritis was associated with a delayed-type hypersensitivity response to H. pylori antigen and elevated gastric and serum IFN-γ, interleukin 6, and tumor necrosis factor alpha. In spite of the absence of IFN-γ expression, indicating failure of Th1 differentiation, CD4⁺ T cells from T-bet KO mice induce gastritis in H. pylori-infected recipient SCID mice. This indicates that Th1-independent mechanisms can cause gastric inflammation and disease due to H. pylori.

Gastritis due to Helicobacter pylori is ordinarily induced by a Th1-mediated response that is CD4 cell and gamma interferon (IFN-γ) dependent. The Th1 bias of the host response has been demonstrated for both humans (5, 6, 13, 29) and mice (7, 17, 18, 28) and is consistent across a wide range of experimental and clinical studies. The dependence of gastritis on CD4-dependent cellular immunity is supported by studies showing that CD4⁺ T cells are both necessary and sufficient for induction of gastritis in mice (7, 26) and by studies demonstrating that the absence of CD8-mediated immunity in major histocompatibility class I knockout (KO) mice and humoral immunity μΜΤ KO mice has no effect on either gastritis or the protective immune response to H. pylori (21, 23). Finally, it has been shown in several different experimental models that IFN-γ (classically a Th1 cytokine) is necessary for induction of gastritis due to gastric helicobacter in mice (27, 28, 33) and that interleukin 10 (IL-10) and IL-4, generally associated with Th2 responses, are protective (2, 7, 16, 28). Thus, there is overwhelming evidence from many different types of studies that gastritis due to gastric helicobacter is CD4 and Th1 mediated and that IFN-γ-producing CD4⁺ T cells have a central role in induction of such gastritis.

T-bet (T-box expressed in T cells) is a transcription factor that is required for differentiation of and IFN-γ secretion by CD4⁺ Th1 T cells (31, 32). CD4⁺ T cells from mice lacking T-bet fail to express IFN-γ and are thus limited to non-Th1-type responses (32). In addition, retrovirus gene transduction of T-bet into differentiated Th2 cells resulted in redifferentiation of these cells to a Th1 phenotype by inducing expression of IFN-γ and repressing IL-4 and IL-5 (14, 31).

The close correlation between Th1-biased CD4⁺ T cells and H. pylori-associated gastritis and the central role of T-bet in the development of Th1 T cells predict that T-bet contributes to gastritis due to H. pylori. The goal of this study was to use two different mouse models to determine if T-bet expression by CD4⁺ T cells is essential to induce gastritis due to H. pylori in mice.

MATERIALS AND METHODS

Mice.

C57BL/6J mice and congenic mutant mice were used. T-bet KO mice and C57BL/6J controls were obtained from our breeding colony established at the University of Michigan from breeders kindly supplied by Laurie Glimcher from the Harvard School of Public Health. The colony was helicobacter-free during the first part of this study but was contaminated with lower-bowel helicobacter (determined by helicobacter-specific PCR of the cecal tip) upon moving to a different facility (the later experiments in this study). No other known mouse pathogens are present in the mouse colony as determined by routine periodic screening of sentinel mice. Mice were maintained in static microisolator cages and offered nonsupplemented commercial mouse chow and water ad libitum. Construction of the T-bet KO mutant has been described (32). These mice have a deletion of exon 1 and surrounding sequences of the T-bet gene and fail to express IFN-γ in CD4⁺ T cells, NK cells, or NK T cells. Helicobacter-free specific-pathogen-free C57BL/6J-Prkdc^scid (severe combined immunodeficient [SCID]) mice were obtained from Jackson laboratories. All animal experiments were approved by the University of Michigan Animal Care and Use Committee. The number of mice used in each experimental group is indicated in Table 1.

TABLE 1.

Experimental mouse groups

Mouse strain	Infection	Donor cells transferred	Sacrifice interval (wk)	No. of mice
C57BL/6J	No	None	12-24	21
C57BL/6J	Yes	None	1	6
C57BL/6J	Yes	None	12	21
C57BL/6J	Yes	None	24	11
T-bet KO	No	None	12-24	11
T-bet KO	Yes	None	12	26
T-bet KO	Yes	None	24	8
SCID	No	None	8	15
SCID	Yes	None	8	14
SCID	No	C57BL/6J	4	5
SCID	No	C57BL/6J	8	15
SCID	Yes	C57BL/6J	4	4
SCID	Yes	C57BL/6J	8	15
SCID	No	T-bet KO	4	5
SCID	No	T-bet KO	8	15
SCID	Yes	T-bet KO	4	4
SCID	Yes	T-bet KO	8	15

Experimental design, simple inoculation model.

Six- to 8-week-old female C57BL/6J and T-bet KO mice were orally inoculated once with 1 × 10⁸ CFU of broth-cultured H. pylori, strain SS1, in 0.1 ml of broth. Mice were killed 12 or 24 weeks after inoculation, and gastric tissue was collected for histologic examination and quantitative bacterial culture as described below. Control mice were not given bacteria.

Experimental design, adoptive transfer.

Six- to eight-week-old SCID mice were inoculated as described above with H. pylori SS1. Two weeks after bacterial inoculation, mice were given congenic CD4⁺ splenocytes by intraperitoneal injection. For splenocyte isolation, spleens were removed from uninfected C57BL/6 or T-bet KO mice immediately after death and disaggregated in RPMI medium at 4°C. Cells were sedimented and resuspended briefly in ACK buffer (0.15 M NH₄Cl, 10 mM KHCO₃, 0.1 mM Na₂EDTA, pH 7.2 to 7.4, filter sterilized) to lyse red blood cells. Single-cell splenocyte suspensions were washed and resuspended in 80 μl of degassed MACS buffer (1× phosphate-buffered saline [Gibco], 0.5% bovine serum albumin, 100 mM EDTA, pH 7.2), 20 μl of CD4 MACS beads per 10⁷ cells were added, and samples were incubated for 15 min at 4C. Cells were washed and loaded onto a magnetized MACS magnetic separation column, and CD4-negative cells were removed by flushing with MACS buffer. The column was removed from the magnet, and CD4⁺ cells were washed out of the column, spun down, and resuspended in RPMI medium, and viable cells were counted. We have previously shown that this procedure results in greater than 95% pure CD4⁺ T cells and that these cells, but not the negatively selected population, induce gastritis in H. pylori-infected SCID mice (7; also unpublished data). Each mouse received 5 × 10⁵ viable CD4⁺ cells. Control groups were not inoculated with bacteria, not given splenocytes, or were neither inoculated nor transferred. Mice were killed 4 or 8 weeks after transfer. Mice were randomly assigned to experimental groups, and the numbers of mice per group varied depending on the number of times the individual experiment was repeated (Table 1).

Tissue collection.

At necropsy, gastric tissue was collected for histopathologic examination, PCR, and bacterial culture. For histologic examination, two adjacent longitudinal strips from the greater curvature of the glandular stomach were fixed in neutral buffered formalin and embedded in paraffin. Five-micron sections were stained with hematoxylin and eosin and examined blind. In a subset of mice, the glandular gastric mucosa from approximately one-half of the gastric mucosa was removed, weighed, and stored at −70C for quantitative reverse transcription-PCR (qRT-PCR) to detect cytokines as described below. In another subset, one-half of the gastric mucosa was removed and weighed, and bacterial colonization was quantified by plate dilution.

Delayed-type hypersensitivity (DTH) assay.

Twenty-four hours before necropsy, mice were challenged with 10 μg of H. pylori sonicate in 30 μl of phosphate-buffered saline injected into the right hind footpad as previously described (9). The left hind footpad was challenged with sterile phosphate-buffered saline. Foot thickness was measured with a dial thickness gauge before and 24 h after injection and expressed as the difference between the two measurements for each foot.

Histopathologic scoring.

The extent of gastritis was scored as previously described (9). Briefly, both strips of gastric mucosa from each mouse (see above) were examined in their entirety. Each ×20 field was examined and scored as positive or negative for neutrophilic inflammation or mononuclear inflammation. Neutrophilic inflammation was defined as the presence of at least three clusters of neutrophils within the lamina propria. Mononuclear inflammation was defined as mucosal infiltrates of lymphocytes, plasma cells, histiocytes, or a mixture of cells sufficient to replace or displace gastric glands. Extent of inflammation was expressed as percent affected mucosa.

qRT-PCR.

Total RNA was isolated from gastric mucosal scrapings using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. mRNA was further purified from the total RNA by processing the samples through QIAGEN′s RNeasy MiniKit and stored at −80°C. cDNAs were generated using Invitrogen's SuperScript First Strand kit and random hexamer primers. Primers for murine glycerol-3-phosphate dehydrogenase (GAPDH), IFN-γ, IL-6, IL-12 p35 (IL-12), and tumor necrosis factor alpha (TNF-α) were designed using Stratagene's on-line Lab Tools suite for use in a SYBR Green assay. All primers were designed to either directly span exon junctions or include at least one exon junction in the amplified sequence. Primers used were as follows: GAPDH, 5′-CGTCCCGTAGACAAAATG and 3′-TGGCAACAATCTCCACTT; IFN-γ, 5′-CCTTTGGACCCTCTGACT and 3′-AGCCAAGATGCAGTGTGT; IL-6, 5′-CACTTCACAAGTCGGAGG and 3′-AATTGCCATTGCACAACT; IL-12, 5′-ATGAAGACATCACACGGG and 3′-CAGCTCCCTCTTGTTGTG; and TNF-α, 5′-CCCAAAGGGATGAGAAGT and 3′-ACAGGCTTGTCACTCGAA.

qRT-PCR mixes were made in triplicate with Stratagene's Brilliant SYBR Green QPCR Master Mix, and the reactions were carried out on a thermocycler/fluorimeter (Stratagene Mx3000P QPCR System) and analyzed with the associated software. In all instances, cytokine assays were performed in parallel to GAPDH assays, and the cytokine cycle threshold (C_T:Cyt) numbers were normalized to the cycle threshold value for GAPDH (C_T:GAPDH) using the equation .

Serum cytokine determination.

Cytokine levels in serum were determined with a commercially available cytometric bead array mouse inflammation kit (catalog no. 551287; BD Biosciences), performed according to the manufacturer's instructions.

Dendritic cell isolation and stimulation.

Bone marrow dendritic cells were isolated from uninfected C57BL/6J or T-bet KO mice as follows. Mice were killed by a pentobarbital overdose, and both femurs and both tibias were aseptically removed and placed in RPMI at 4°C. Bone marrow was removed by flushing, the cells were disaggregated and washed in cold RPMI, and red blood cells were lysed in 5 ml ACK buffer for 3 min. Bone marrow cells from individual mice were washed and resuspended in 30 ml of RPMI-10% fetal calf serum with antibiotics and 10 ng/ml each of IL-4 and granulocyte-macrophage colony-stimulating factor. Cells were incubated for 6 days in 5% CO₂ at 37°C. Additional IL-4 and granulocyte-macrophage colony-stimulating factor (10 ng/ml each) were added on day 3. On day 6, loosely adherent and nonadherent cells were collected, washed, and resuspended in 6 ml of Hanks buffered salt solution (Gibco). A 2.3-ml volume of Opti-prep concentrate (Sigma) was added to 9.7 ml of buffer C (0.88% [wt/vol] NaCl, 1 mM EDTA, 0.5% [wt/vol] bovine serum albumin, 10 mM HEPES, pH 7.4, filter sterilized). The cell suspension was gently layered on top of 6 ml of the Opti-prep mixture and then spun at 600 × g for 5 min. The interface was collected, and viable cells were counted by Trypan blue exclusion. Dendritic cell purity was 85% as determined by flow cytometric detection of CD11c. For H. pylori stimulation, dendritic cells were plated in six-well plates at 10⁶ cells/ml, 10⁸/ml of live H. pylori strain SS1 bacteria were added, and cells were incubated overnight at 37°C. Control wells were incubated without H. pylori. For splenocyte coculture, adherent and nonadherent cells were removed and irradiated at 2,000 rad for 5 min and 23 s and then kept on ice.

Splenocyte proliferation assay.

Splenocytes were harvested from uninfected or SS1-infected C57BL/6J or T-bet KO mice as described above, and 2 × 10⁵ splenocytes/well were plated in 96-well plates. Irradiated H. pylori-pulsed or nonpulsed C57BL/6J dendritic cells were added (2 × 10⁵, 2 × 10⁴, or 2 × 10³ cells/well), and the plates were incubated for 72 h at 37°C. After 72 h, aliquots of cell culture supernatant were removed and stored at −80°C for cytokine determination by using a cytometric bead array mouse Th1/Th2 cytokine kit (catalog no. 551287; BD Biosciences), performed according to the manufacturer's instructions. The proliferation rate was determined with the ViaLight Plus Cell Proliferation and Cytotoxicity BioAssay kit (catalog no. LT07-221; Cambrex) according to the manufacturer's instructions. The labeling index was calculated as light production by (splenocytes + dendritic cells)/splenocytes alone.

Statistics.

Groups were compared by using the Mann-Whitney U test (for two group samples) or analysis of variance (for samples with more than two groups). Fisher's protected least significant difference was used post hoc to determine significant differences between multiple groups. Significance level was set at <0.05. Error bars on graphs indicate standard errors of the means.

RESULTS

Simple inoculation model.

As shown in Fig. 1, gastritis in infected C57BL/6J mice was significantly more extensive than in infected T-bet KO mice or in uninfected mice at both 12 and 24 weeks after inoculation. Gastritis was minimal in uninfected mice in both groups. Figure 2 illustrates the typical appearance of gastritis in infected C57BL/6J mice compared to infected T-bet-KO mice and uninfected mice. In H. pylori-infected C57BL/6J mice, gastritis consisted of mixed inflammatory infiltrates distributed in a multifocal pattern throughout the gastric fundus. Inflammatory infiltrates were rare or nonexistent in infected T-bet KO or uninfected mice.

FIG. 1.

By 12 weeks after inoculation, both neutrophilic and mononuclear inflammatory infiltrates were significantly more extensive in H. pylori-infected C57BL/6J mice than in uninfected mice and infected T-bet KO mice. Inflammation in C57BL/6J mice continued to increase over the course of the experiment (24 weeks). In contrast, inflammatory infiltrates did not differ between infected and uninfected T-bet KO mice either 12 or 24 weeks after inoculation.

FIG. 2.

Moderate lymphocytic, plasmacytic, and neutrophilic gastritis and adentitis (arrow) and mild lymphoplasmacytic infiltrates deep in the lamina propria (arrowhead) are present in C57BL/6J mice infected with H. pylori (A). Inflammation was minimal in infected T-bet KO mice (arrowhead, B) and not identifiable in uninfected mice (C). Original magnification, ×400.

Bacterial colonization did not differ markedly between the groups of mice. At 3 months after inoculation, colonization in C57BL/6J mice was about 10-fold less than in T-bet KO mice (P < 0.05), but by 6 months after inoculation there was no significant difference. (Fig. 3).

FIG. 3.

At 12 weeks after bacterial inoculation, bacterial colonization in C57BL/6J mice was significantly less than in T-bet KO mice, but by 24 weeks the difference was no longer significant.

Adoptive transfer model.

As shown in Fig. 4, nonrecipient SCID mice infected with H. pylori did not develop gastritis. In contrast, SCID recipients of CD4⁺ splenocytes from naive C57BL/6J mice developed extensive gastritis that in some animals involved 100% of the gastric mucosa. These results are consistent with those of previous studies showing dense colonization with no inflammatory response in SCID mice and severe, rapidly progressive gastritis in SCID recipients of CD4⁺ T cells (7, 9). Surprisingly, SCID recipients of T-bet KO CD4⁺ splenocytes also developed gastritis. By 8 weeks after transfer, gastritis in both groups of mice was greater than in uninfected control mice, and approached 100% of the gastric mucosa in some animals in both groups. Gastric lesions in C57BL/6J recipients and T-bet KO recipients could not be distinguished by subjective examination alone (Fig. 5), although quantitation of the lesions revealed a statistically significant difference in extent of mononuclear gastritis between the groups 8 weeks after adoptive transfer (see Fig. 4). Consistent with previous studies (7-9), uninfected recipient mice had variable neutrophilic infiltrates, but mononuclear gastritis remained below 20% of the gastric mucosa (Fig. 4).

FIG. 4.

A: Nontransferred H. pylori-infected SCID mice did not develop inflammatory infiltrates, but adoptive transfer with CD4⁺ splenocytes from either C57BL/6J mice or T-bet KO mice resulted in both neutrophilic and mononuclear inflammatory cell infiltration that was detectable by 4 weeks after transfer and was significantly greater than that for uninfected mice by 8 weeks after transfer. Inflammation was present in infected recipients of either C57BL/6J cells or T-bet KO cells, although by 8 weeks after transfer, mononuclear gastritis was significantly more extensive in recipients of C57BL/6J cells. The histologic appearance of gastritis was indistinguishable in the two groups of mice (see Fig. 5). B: Uninfected SCID mice and recipient SCID mice developed only mild gastritis.

FIG. 5.

In H. pylori-infected SCID recipients of both C57BL/6J (A) and T-bet KO (B) CD4⁺ splenocytes, gastritis was severe, characterized by marked, widespread lymphocytic, plasmacytic, and neutrophilic infiltrates and complete loss of the normal fundic morphology. Note the widespread inflammatory infiltrate and loss of parietal cells characteristic of gastritis due to H. pylori in mice. Original magnification, ×400.

Bacterial colonization in both groups of recipient mice was markedly suppressed compared to that for nonrecipient SCID mice (Fig. 6). This is consistent with several previous studies in which marked gastric inflammation is associated with suppression of H. pylori colonization (7-9). There was no difference in the degree of colonization suppression between the two adoptive transfer groups (see Fig. 6).

FIG. 6.

Marked inflammation in SCID recipients of CD4⁺ splenocytes from either C57BL/6J mice or T-bet KO mice was associated with marked, significant suppression of bacterial colonization that reached approximately 4 logs by 8 weeks after transfer. Bacterial colonization at the 4-week time point was not determined in nontransferred SCID mice (ND).

DTH and gastric cytokine expression.

DTH was evaluated only for SCID and recipient SCID mice, because previous studies showed that in C57BL/6J mice DTH responses to H. pylori are not detectable by footpad challenge (9). As expected, the DTH response as measured by footpad swelling in response to H. pylori antigen indicated that recipients of C57BL/6J T cells but not recipients of T-bet KO T cells developed a response (Fig. 7). Twenty-four hours after challenge, footpad thickness in recipients of C57BL/6J CD4⁺ cells was significantly greater than in all other groups. In recipients of T-bet KO cells, footpads did not swell in response to H. pylori antigen, indicating a lack of systemic DTH response.

FIG. 7.

DTH response as measured by footpad swelling in response to injection of H. pylori antigen is a reproducible finding with H. pylori-infected SCID recipients of CD4⁺ splenocytes and likely reflects a Th1-type immune response (7). In the current study, H. pylori-infected recipients of CD4⁺ splenocytes from C57BL/6J mice developed footpad swelling that was significantly greater than in H. pylori-infected recipients of T-bet KO CD4 splenocytes, uninfected mice, and nontransferred mice.

Consistent with the absence of systemic evidence of DTH, qRT-PCR detection of IFN-γ mRNA in gastric scrapings revealed that elevated IFN-γ expression was present only in recipients of C57BL/6J T cells and in chronically infected C57BL/6J mice (Fig. 8). Recipients of T-bet KO T cells had minimal gastric IFN-γ expression that was not significantly different from that for uninfected mice or nonrecipient SCID mice (P < 0.05). To determine if other gastric proinflammatory cytokines differed between mouse groups, IL-6, IL-12, and TNF-α mRNA expression were measured in gastric epithelium from infected and uninfected C57BL/6J mice and from SCID and recipient SCID mice killed 8 weeks after inoculation (Fig. 9). In gastric tissue from H. pylori-infected SCID recipients of C57BL/6J cells, IL-6 and TNF-α mRNA levels were significantly higher than in gastric mucosa from nontransferred SCID mice and recipients of T-bet KO cells. In contrast, cytokine mRNA in gastric tissue from infected T-bet KO recipients did not differ from that from nontransferred controls. Gastric IL-12 mRNA levels were highly variable, and there were no statistically significant differences between groups.

FIG. 8.

In conjunction with marked gastritis, H. pylori-infected SCID recipients of C57BL/6J splenocytes developed evidence of upregulation of IFN-γ in situ, as reflected by elevated IFN-γ mRNA levels normalized to GAPDH mRNA. Infected SCID mice that either did not receive splenocytes or received CD4⁺ splenocytes from T-bet KO mice did not develop elevated IFN-γ mRNA in gastric tissue. IFN-γ was upregulated in chronically infected C57BL/6J mice (12 to 24 weeks after inoculation) but not in acutely infected (1 week after inoculation) mice or in uninfected mice.

FIG. 9.

qRT-PCR evaluation demonstrated that IL-6 mRNA (A) was significantly elevated in gastric mucosa from infected SCID recipients of C57BL/6J cells compared to levels from uninfected SCID mice, C57BL/6J mice, and SCID recipients of C57BL/6J cells as well as infected SCID mice, infected C57BL/6J mice, and infected recipients of T-bet KO cells. TNF-α (B) was significantly greater in recipients of C57BL/6J cells than in all other groups of mice. There were no significant differences in IL-12 levels between the groups (C).

Serum cytokines.

As measured by the cytokine bead assay, serum IFN-γ, IL-6, IL-12, and TNF-α largely paralleled gastric cytokine levels (Fig. 10). Serum IFN-γ, IL-6, and TNF-α as well as serum MCP-1 were elevated in SCID recipients of C57BL/6J CD4⁺ cells and were significantly greater than in most other groups. In a few cases the differences did not reach statistical significance, most likely because of large individual variations for individual mice. Serum IL-12 levels did not differ between groups. IL-10 was not detectable in serum from any group (not shown).

FIG.10.

Serum proinflammatory cytokines levels in all groups largely recapitulated gastric cytokine expression. Serum IFN-γ (A), IL-6 (B), and TNF-α (C), as well as serum MCP-1 (E), were elevated in SCID recipients of C57BL/6J CD4⁺ cells. Serum IL-12 (D) levels did not differ between groups. IL-10 was not detectable in serum from any group (not shown).

In vitro stimulation of immune cells.

In spite of the absence of Th1 differentiated cells characteristic of lymphocytes from T-bet KO mice, there were no differences in the H. pylori-mediated proliferative responses of splenocytes from T-bet and C57BL/6 mice in vitro. For this determination, splenocytes were cocultured with C57BL/6J bone marrow dendritic cells that had been pulsed with live H. pylori and then irradiated to prevent proliferation while retaining antigen-presenting ability. Figure 11 shows that splenocytes from both H. pylori-infected and uninfected C57BL/6J and T-bet KO mice proliferated when cocultured with dendritic cells that had been pulsed with H. pylori (labeling index > 1) compared to splenocytes cultured with dendritic cells that had not been pulsed. However, there were no differences between labeling indices of splenocytes from C57BL/6J and T-bet KO mice, indicating that T-bet does not mediate proliferative response of splenocytes to H. pylori antigen. Labeling indices did not differ markedly between splenocytes from infected and uninfected mice, although splenocytes from H. pylori-infected T-bet KO mice had significantly higher labeling indices than did splenocytes from uninfected mice.

FIG. 11.

Cultured bone marrow dendritic cells from C57BL/6J mice were pulsed overnight with live H. pylori, irradiated, and cocultured with splenocytes from H. pylori-infected or uninfected C57BL/6J or T-bet KO mice. In all groups of mice, coculture with pulsed dendritic cells resulted in increased proliferation compared to coculture with nonpulsed dendritic cells. There were no differences between C57BL/6J and T-bet splenocytes.

To determine if differences in dendritic cell cytokine responses could account for differences in responses of T-bet KO and C57BL/6J mice, cultured bone marrow dendritic cells were pulsed overnight with live H. pylori, and cytokine release into the medium was determined. Figure 12 demonstrates that pulsing with H. pylori resulted in elevated cytokine production in dendritic cells from both mouse strains. Production of IL-10, MCP-1, IL-6, and TNF-α were all elevated compared to levels for unstimulated dendritic cells. IL-12p70 was minimal in all groups. There were no differences between dendritic cells from C57BL/6J and T-bet KO mice.

FIG. 12.

Cytokine secretion by cultured bone marrow dendritic cells after 24-h coculture with H. pylori. Both C57BL/6 and T-bet-KO dendritic cells expressed elevated levels of IL-6, TNF-α, MCP-1, and IL-10 and low to nonexistent levels of IL-12p70 compared to unstimulated cells, but there was no difference between the two mouse strains. Dendritic cells were isolated from uninfected mice (see Methods).

DISCUSSION

The results reported here confirm in part the role of Th1-mediated immunopathology in the development of gastritis due to H. pylori. We showed that gastritis developed in C57BL/6J mice infected with H. pylori but not in mice lacking T-bet, a transcription factor that directs Th1 differentiation of CD4⁺ T cells. These results are compatible with reports that IFN-γ KO mice also fail to develop gastritis due to gastric helicobacter (27, 28, 33) and support the hypothesis that H. pylori gastritis is at least in part Th1 mediated.

In contrast to the above, results of the adoptive transfer experiments indicate that non-Th1-mediated mechanisms also contribute to gastritis due to H. pylori. These results were initially a surprise. Because of the strong association of IFN-γ with gastritis in many different models of H. pylori-related disease and because of reports of IFN-γ and T-bet dependence of other Th1-mediated adoptive transfer models of disease, such as inflammatory bowel disease (22), we expected markedly less gastritis in H. pylori-infected SCID recipients of T-bet KO cells than in recipients of C57BL/6J cells. Surprisingly, however, gastritis was almost indistinguishable in the two groups. The morphological appearance of the gastritis in the two groups was identical. Consistent with Th1-mediated disease, it was characterized by a severe, widespread, mixed inflammatory infiltrate consisting of lymphocytes, plasma cells, histiocytes, and neutrophils. Inflammation was numerically slightly less extensive in the T-bet KO recipients (see Fig. 3), but the differences were small, and only the difference in mononuclear cell infiltration reached statistical significance. Thus, there were minimal differences in H. pylori-associated gastritis induced by adoptive transfer of CD4⁺ T cells that were capable of expressing IFN-γ and CD4⁺ T cells that were not.

These results have some overlap with those of previous studies in our laboratory in which moderate gastritis due to H. pylori developed in SCID recipients of splenocytes from IFN-γ-KO mice (7). In that study, however, gastritis was significantly less extensive in recipients of IFN-γ KO splenocytes than in recipients of C57BL/6J splenocytes and did not progress beyond 4 weeks after transfer. In the current study, in contrast, recipients of T-bet KO CD4⁺ cells developed gastritis that was only slightly less extensive than that of recipients of C57BL/6J cells and that continued to progress for the duration of the experiment. The mild gastritis in recipients of IFN-γ KO splenocytes was attributed to the presence of non-T cells and CD8⁺ cells and was not investigated further. However, in light of current results, it is likely that IFN-γ-deficient CD4 T cells contributed to gastritis in the previous study. In the current study, gastritis in recipients of T-bet KO CD4⁺ cells was extensive and could not be attributed to elevated IFN-γ from non-CD4 cells, because CD4⁺ cells only, rather than unfractionated splenocytes, were transferred, and IFN-γ was not elevated in recipient mice. Thus, the appearance of marked gastritis in recipients of Th1-deficient CD4⁺ cells in T-bet KO recipients combined with mild gastritis in recipients of IFN-γ-deficient splenocytes suggests that a Th1-independent mechanism for H. pylori-associated gastritis was either unmasked or deregulated in infected recipient mice.

To explore mechanisms whereby T-bet-deficient CD4⁺ cells could induce gastritis, we examined gastric mucosa of recipient SCID mice to determine if we could detect proinflammatory cytokines that could account for IFN-γ-independent gastritis in T-bet KO recipients. Surprisingly, not only IFN-γ but also the proinflammatory cytokines IL-6 and TNF-α were low or absent in recipients of T-bet KO cells. Furthermore, serum cytokine levels paralleled gastric cytokine levels, further indicating that recipients of T-bet KO cells failed to mount a conventional proinflammatory response. The presence of severe gastritis in the absence of elevated IL-6 and TNF-α, in addition to IFN-γ, was unexpected. TNF-α and IL-6 are nonspecific indicators of inflammation that are produced by several different cell types and would not be expected to be dependent on Th1-biased CD4⁺ T cells. Yet in this case they appeared to be dependent on the transfer of Th1 cells for their induction in response to H. pylori infection.

In order to address the dichotomy between induction of gastritis in recipients of T-bet KO CD4⁺ cells and the absence of gastritis in the T-bet KO mice themselves, we determined in vitro responses of splenocytes and dendritic cells to H. pylori, comparing these responses to those of cells from C57BL/6 mice. For this, we determined if splenocytes from the two strains differed in their ability to respond to H. pylori in vitro, and we compared the in vitro cytokine responses of dendritic cells to H. pylori. As shown above, we were unable to detect differences between cells from C57BL/6J and T-bet KO mice in any of these assays. These results indicate that the failure of T-bet KO mice to develop gastritis could not be attributed to abnormal splenocyte proliferative responses to H. pylori or to dendritic cell dysfunction and that T-bet-deficient T cells retain their ability to proliferate in response to H. pylori, consistent with their ability to induce gastritis in vivo.

There is a precedent for the suggestion that gastritis due to H. pylori has a Th1-independent component. First, although there is good evidence that B cells and antibody are not essential for gastritis due to H. pylori (21, 23), infected individuals do mount a strong local and systemic humoral immune response, indicating that many cells of the immune system, including those generally associated with Th2-biased responses, are activated by H. pylori antigens (3, 20). It is possible that in the absence of a Th1 response, other arms of the immune system could be derepressed and lead to gastritis. Second, adoptive transfer studies have suggested that Th2-biased T-cell clones as well as Th1 clones are capable of inducing gastritis in H. pylori-infected mice (18). In those studies, it was suggested that Th2 responses may be less inflammatory but more protective than Th1 responses. More-recent studies have also shown that gastritis due to gastric helicobacters in mice may have a Th2-like component (10, 12). Finally, although most chronic enteric inflammatory diseases are considered largely Th1 mediated, some are not. For example, Th2-type responses have been associated with histologic evidence of enteric inflammation in such diseases as ulcerative colitis in humans (4) and oxazolone-induced inflammatory bowel disease among other models in mice (30). Thus, although enterogastric inflammation is generally Th1 mediated, Th1-independent pathways exist.

In addition to the above, descriptions of Th1-like disease in the absence of either Th1 or Th2 cells and cytokines are not unprecedented (1, 15, 19). For example, some mouse models of autoimmune disease, such as experimental allergic encephalitis in mice and autoimmune arthritis, until recently were generally thought to be Th1 mediated. It has recently been demonstrated, however, that these diseases are associated with Th1-independent inflammation induced by IL-17-producing T cells that appear to be functionally and developmentally distinct from either Th1 or Th2 cells (1, 11, 15, 19). Of relevance to the current report, T-bet KO mice were shown to have elevated numbers of IL-17-producing T helper cells (Th17 cells) (11), leading to the possibility that dysregulation of Th17 cells in recipients of T-bet KO cells could induce gastritis, as does dysregulation of Th1 cells in recipients of C57BL/6J T cells either in our model (7) or in other adoptive transfer models of inflammatory disease (24, 25). Whether Th2 cells, Th17 cells, or some other as yet undescribed T-cell subset is responsible for gastritis in H. pylori-infected recipients of T-bet KO cells awaits further study.