Differential localization of Th1 and Th2 cells in ... - Semantic Scholar

International Immunology, Vol. 10, No. 9, pp. 1325–1334

© 1998 Oxford University Press

Differential localization of Th1 and Th2 cells in autoimmune gastritis Tomoya Katakai1,3, Kazuhiro J. Mori3, Tohru Masuda2 and Akira Shimizu1 1Center

for Molecular Biology and Genetics, Kyoto University, and 2Institute for Immunology, Faculty of Medicine, Kyoto University, Kyoto 606-8501, Japan 3Department of Cell Biology, Faculty of Science, Niigata University, Niigata 950-2181, Japan

Keywords: autoantibody, CD41 T cell subset, neonatal thymectomy, transendothelial migration

Abstract The vast majority of CD4F T cells infiltrating into gastric mucosa (GM) and in the draining (gastric) lymph node (GLN) shows an activated/memory phenotype, CD45RBlowL-selectinlowCD44high, in neonataly thymectomized BALB/c mice bearing autoimmune gastritis (AIG), indicating that these cells are actively involved in this disease. CD4F T cells sort-purified from GLN expressed mRNAs encoding for both IFN-γ and IL-4. However, those infiltrating into GM expressed very low levels of IL-4 mRNA, even though they strongly expressed IFN-γ mRNA. Among CD4F T cells separated from AIG mice expressing detectable levels of either IFN-γ or IL-4 by intracellular staining, less than oneseventh expressed IL-4 and thus most of them expressed IFN-γ in GM, whereas roughly half and one-third expressed IL-4 in GLN and spleen respectively. These findings indicate that the Th1 cells predominantly infiltrate into autoimmune lesions and Th2 cells are mainly resident in the regional LN. We further set up an in vitro model system of transendothelial migration using a murine endothelial cell line, F-2, and found that Th1 cells in CD4F T cells separated from lymphoid tissues of AIG mice preferentially passed through the monolayer of endothelial cells while only a small portion of Th2 cells did so. This differing ability of transendothelial migration and localization might explain the dominance of Th1 cells destroying the tissue in focal lesions without inhibition by the Th2 cells, in spite of both subsets being simultaneously activated in AIG mice, and the functions of each T cell subset seems to be mutually exclusive. Introduction Target- or organ-specific autoimmune diseases can be classified into two categories, autoantibody-induced such as autoimmune hemolytic anemia (1–3) and T cell-mediated such as insulin-dependent diabetes (4). Different from those caused by autoantibodies (3), the onset and effector of the T cellmediated autoimmune diseases are not well characterized, probably due to difficulties in producing simple models that are easy to reproduce. Autoimmune gastritis (AIG) after neonatal thymectomy is one (of few) good experimental model for T cell-mediated autoimmune disorder (5–7) and is easily induced at a high incidence, in ~50–80% of BALB/c mice that have undergone thymectomy at 2–4 days after the birth (d3-Tx). It has been suggested that T cell-mediated autoimmunities are mainly dependent on CD41 T cells (4,8– 11). AIG is also shown to be primarily caused by self-reactive CD41 T cells, because it is transferable into syngeneic nude mice by the transfer of CD41 T cells but not by CD81 T cells nor autoantibodies from suffering mice (7,12), even though

autoantibodies against gastric parietal cell are seen in the sera together with tissue-destructive lymphocytic infiltration into gastric mucosa (GM). Major self-antigens of this disease are suggested to be peptides on MHC class II derived from H1/K1-ATPase (proton pump) (13–15) synthesized in parietal cells. Indeed, a CD41 T cell clone (II-6) which responds to a peptide from the α subunit of H1/K1-ATPase and can induce AIG by transfer into BALB/c nu/nu mice was established (16), and we recently found that T cells carrying TCR similar to the II-6 clone frequently appear in the GM of d3-Tx AIG mice (17). CD41 T cells are further divided into two subpopulations named Th1 and Th2 based on the profiles of their cytokine production (11,18–21). Th1 cells produce IFN-γ, IL-2 and lymphotoxin (so called Th1 cytokines), and promote cellmediated immune responses, whereas Th2 cells produce distinct cytokines (Th2 cytokines) such as IL-4, IL-5, IL-10 and IL-13, which mediate allergic responses or humoral immune responses through supporting B cell activation. Each

Correspondence to: A. Shimizu Transmitting editor: M. Miyasaka

Received 28 April 1998, accepted 25 May 1998

1326 Differential localization of Th1 and Th2 in autoimmune gastritis set of cytokines shows counter-inhibitory effects against the other, and thus the effect of each subset seems to be mutually exclusive in vitro and in vivo (18–21). Actually, in some immune responses, one of them dominates to elicit suitable immune reactions according to the nature of the antigens (18–22). Of the two subsets, Th1 cells are reported to be primarily responsible for promotion of tissue destruction in several autoimmune diseases. The facts that exacerbation of disease can be protected either by administration of antibodies against Th1 cytokines or by Th2-type cytokines themselves (11,23– 27) support this notion. It has been also suggested that Th1 cells are the direct effector which destroy parietal cells in AIG because they express Fas ligand (28), and, on the other hand, parietal cells in AIG express MHC class II, Fas and ICAM-1 which are not expressed normally (29). Indeed, the disease-transferable II-6 clone has the Th1 phenotype, expresses Fas ligand on the surface, produces IFN-γ and can kill the target cells expressing MHC class II (I-Ad) pulsed with antigen peptide (29). In addition to tissue destructive inflammation seen in the mucosal lesion, however, simultaneous production of autoantibody and correlation between disease progression and autoantibody titer (10,30) strongly suggest that not only Th1 but also Th2 cells are activated in AIG mice at the same time. This phenomenon seems to be rather contradictory to the fact that the functions of Th1 and Th2 cells are apparently mutually exclusive. It is very interesting to know how the activation of both Th1 and Th2 subsets is managed not only for a better understanding of T cell-mediated autoimmunity such as AIG but also for that of immune responses in general, since both T cell-mediated killing or tissue destruction and antibody or autoantibody production are often seen in the same individuals in many immune or autoimmune reactions. In this paper we analyze the phenotypic characteristics of CD41 T cells involved in d3-Tx-induced AIG. We further focused on the difference between the activated Th1 and Th2 subsets of AIG mice in localization at the loci of inflammations and in the ability of transendothelial migration (TEM) to elucidate the mechanism of such co-activation.

confirm the specificity of autoantibody against GM, control experiments were performed using microsomes from liver or kidney as antigen in the same assay system. Goat anti-mouse Ig (anti-mouse IgG, L 1 H chain specific), anti-mouse IgG1 or anti-mouse IgG2a (Caltag, Burlingame, CA) conjugated with horseradish peroxidase was used for developing the antibody. For histological examination, stomachs were fixed in 10% formalin, embedded in paraffin, sectioned and stained with hematoxylin & eosin.

Preparation of infiltrating cells To collect mononuclear cells infiltrating into GM, inflamed glandular stomachs were isolated from AIG mice and their contents washed with Hanks’ balanced solution. Then 10% FCS/RPMI medium (RPMI 1640 supplemented with 10% FCS, 50 µM 2-mercaptoethanol, 100 U/ml penicillin and 100 µg/ml streptomycin) was injected into submucosa of stomachs followed by dissection into small pieces to release infiltrating cells. Cells were used for further experiments after removing large tissue debris by passing through cotton in a funnel.

Flow cytometric analysis and cell sorting Approximately 106 cells were stained by combinations of antibodies described below. For direct staining; phycoerythrin (PE)–anti-CD4 (GK1.5; Becton Dickinson, Mountain View, CA). For indirect staining; biotin–anti-CD45RB (16A; PharMingen, San Diego, CA), anti-L-selectin (MEL-14; Caltag) or culture supernatants of hybridomas; KM201 (anti-CD44, rat IgG1), FD441.8 (anti-LFA-1, rat IgG2b) and PC-61 (anti-CD25, rat IgG1) using FITC–streptavidin or Cy5–anti-rat IgG (Amersham, Tokyo, Japan) as the secondary reagents. Stained cells were analyzed by a FACScalibur flow cytometer and CellQuest software (Becton Dickinson). For separation of CD41 T cells into CD45RBhigh and CD45RBlow populations, infiltrating cells in GM or cells from regional lymph nodes (LN) were stained with anti-CD45RB followed by Cy5–anti-rat IgG and PE–antiCD4, and CD41 cells were fractionated into two populations based on the high or low expression of CD45RB by a FACS Vantage cell sorter (Becton Dickinson).

RT-PCR and Southern hybridization Methods

Mice and neonatal thymectomy BALB/c or BALB/c nu/nu mice of appropriate age, purchased from Japan SLC (Shizuoka, Japan), were maintained and mated at the Experimental Animal Facility, Faculty of Medicine or the facility in the Center for Molecular Biology and Genetics, Kyoto University. Three days after birth, normal BALB/c mice underwent thymectomy (d3-Tx), as described previously (31). After 2–4 months, AIG-developed thymectomized mice (AIG mice) were used for experiments.

ELISA and histology Autoimmune gastritis was diagnosed by the detection of autoantibodies in sera of thymectomized mice and confirmed by histological examination. Autoantibodies in sera or culture supernatants were detected by ELISA using total microsomal fractions extracted from glandular stomachs of BALB/c mice as antigen and o-phenylenediamine as the substrate (32). To

RT-PCR and Southern hybridization were performed basically as described (33,34). Total RNA was extracted from sorted cells using TRIzol Reagent (Gibco/BRL, Gaithersburg, MD) according to the manufacturer’s instructions. To synthesize first-strand cDNA, 100 ng–2 µg of total RNA was reverse transcribed by avian myeloblastosis virus RT (Seikagaku, Tokyo, Japan) and oligo(dT)12–18 primer (Pharmacia Biotech, Tokyo, Japan). Specific cDNAs were amplified with Taq DNA polymerase (Takara Shuzo, Otsu, Japan), using specific primers. Complementary DNAs corresponding to the reverse transcript of 0.25–1 ng of total RNA were amplified by PCR of a total 25 µl. All PCR were carried out with 35 cycles of denaturation at 95°C for 20 s, annealing at 55°C for 2 min, extension at 72°C for 1 min with 2 s increase of each cycle followed by polymerization for 10 min at 72°C using a DNA thermal cycler (Perkin-Elmer Cetus, Norwalk, CT). We chose these amounts of template cDNAs and amplification cycles so that the amount of the products are still exponentially increasing and reflecting the content of the specific mRNA.

Differential localization of Th1 and Th2 in autoimmune gastritis 1327 We confirmed this titration of mRNAs by experiments changing the cycles of amplification in which the increment of template cDNAs and additional cycles of amplification necessary to give the same amount of products corresponded very well. Primers used were as follows; IFN-γ sense 59-ATCCTGCAGAGCAGATTATC-39, antisense 59-TACTCGAGTCAGCAGCGACTCCTTT-39, IL-4 sense 59-ATGGGTCTCAACCCCCAGCTAGT-39, antisense 59-GCTCTTTAGGCTTTCCAGGAAGTC39, β-actin sense 59-TCAGAAGGACTCCTATGTGG-39, antisense 59-TCTCTTTGATGTCACGCACG-39. Then 5 or 1 µl each of PCR product was electrophoresed, transferred onto a nylon membrane and hybridized with 32P-labeled cDNA probes. Probes were purified inserts of cDNA clones made by cloning PCR amplified fragments into pGEM-T vector (Promega, Madison, WI). Autoradiographs were analyzed using a BAS 5000 Bio-image analyzer (Fuji Film, Tokyo, Japan).

Intracellular cytokine staining Expression of IFN-γ and IL-4 in CD41 T cells was examined by the intracellular staining method, using a Cytostain kit (PharMingen). Cells (106) were incubated with phorbol myristate acetate (50 ng/ml) and ionomycin (500 ng/ml) in 1 ml of 10% FCS/RPMI medium for 3 h followed by an additional 3 h incubation with 0.6 µl of GolgiStop (containing monensin; PharMingen). Then, cells were stained with anti-CD4 antibody (GK1.5), followed by Cy5–anti-rat IgG as the secondary reagent after harvesting and blocking by normal mouse serum with 0.1 U/µl of DNase I (Takara Shuzo) to prevent nonspecific binding of antibodies. After fixing and permeabilizing the cells with Cytofix/Cytoperm solution (PharMingen), cells were stained with FITC–anti-IFN-γ and PE–anti-IL-4 (Caltag), and analyzed on a FACScalibur flow cytometer. TEM assay CD41 T cells used in the TEM assay were prepared from AIG mice by a panning method to adsorb out other cell types. Briefly, spleen or lymph node cells were stained by the mixture of anti-HSA (J11D), MHC class II (TIB120), FcγRII/III (2.4G2) and CD8 (HO-2.2) culture supernatant of hybridomas followed by plating onto culture dishes coated by goat anti-rat IgG antibody (polyclonal; Caltag) and unbound cells were collected. The purity of CD41 was monitored by flow cytometry and was .90%. Then 2.53105 cells of the murine endothelial cell line, F-2 (35,36), were cultured on Transwell plates (24 mm diameter, 8 µm pore size; Corning Coaster, Tokyo, Japan) with 10% FCS/DMEM for 2 days to make a confluent monolayer. For the TEM assay, AIG-derived CD41 T cells were applied onto the F-2 cell monolayer with 10% FCS/RPMI, and, after 4 h of incubation, cells were recovered from the upper and lower compartments. The recovered and input cells were used for RNA extraction to semi-quantify mRNAs by RT-PCR followed by Southern hybridization or for intracellular cytokine staining as described above.

autoimmune diseases such as gastritis, thyroiditis, oophoritis or orchitis (5–7,12,37). Among such symptoms, gastritis appears most frequently; ~60% of animal treated with d3-Tx. Previous studies have shown that CD41 T cells in lesions of some other autoimmune disorders have an activated/memory phenotype (38–40), i.e. low level or no expression of naive/ memory marker CD45RB and putative LN homing receptor L-selectin, and high level expression of various cell adhesion molecules, such as CD44, LFA-1. To clarify the effector cells in AIG, the phenotypic characteristics of CD41 T cells from GM and GLN (draining lymph node of GM) of AIG mice were analyzed by staining with antibodies against CD45RB and other surface markers. In AIG mice, .90 and 70% of GM infiltrating and GLN cells respectively expressed a low level of or no CD45RB molecules, while .70% of CD41 T cells in spleen, LN and GLN (almost identical profile to LN) of normal mice were CD45RBhigh (Fig. 1a). Although CD41 cells isolated from GM of normal mice were insufficient to give a conclusive result, these cells seem to express low levels or no CD45RB (data not shown). Increased numbers of CD45RBlowCD41 T cells were also seen in spleen (62%) and LN (58%) of AIG mice,, although the percentages of activated/memory cells were obviously lower than that of GLN (Fig. 1a). The majority of CD41 T cells in AIG mice expressed little L-selectin, and a high level of CD44 and LFA-1 (Fig. 1b), in contrast to normal LN in which most CD41 T cells showed a L-selectinhighCD44lowLFA-1low (naive) phenotype. Other than adhesion molecules, the expression of CD25 was up-regulated, especially in GM CD41 T cells (Fig. 1b). These results indicated that a number of CD41 T cells in AIG mice systemically show an activated/memory phenotype due to the effect of d3Tx and correspond well to the systemic appearance of autoimmune symptoms other than AIG. In particular, we demonstrated that GLN of AIG mice have a higher content of CD41 T cells with an activated/memory phenotype compared to other lymphoid organs. In addition, the fact that almost all the infiltrating CD41 T cells into GM have this phenotype is noteworthy. We then measured autoantibodies produced by equal numbers of mononuclear cells from GM, GLN, spleen and other lymph nodes of AIG mice which were cultured in vitro without exogenous antigen. Anti-GM microsome antibody was produced most effectively in the culture supernatants of GLN or GM-infiltrating cells (Fig. 2), whereas autoantibody against microsomes from liver or kidney were undetectable (data not shown). Therefore, autoantibody specific to GM components is produced in a relatively restricted area, GLN and GM, in which higher contents of CD41 T cells with an activated/ memory phenotype were observed (Fig. 1a). These findings suggest that further reactions against GM self-antigens occur at the GM and its draining LN, and thus these organs are the major sites of autoimmune reaction in AIG.

Results and discussion

Th1 cells predominantly infiltrate into GM, while Th2 cells mainly reside in GLN

CD41 T cells in AIG mice show a marked activated/memory phenotype in gastric lymph node (GLN) and GM After neonatal thymectomy, between 2 and 4 days after birth (d3-Tx), BALB/c mice develop various organ-specific

Tissue destructive inflammation at sites of GM and anti-GM autoantibody production co-exist in our autoimmune model, and it seems that disease progression correlates to autoantibody titer (our unpublished data, 10,30). In addition, as

1328 Differential localization of Th1 and Th2 in autoimmune gastritis

Fig. 1. (a) Expression of CD45RB on CD41 T cells in lymphoid organs and inflamed GM of AIG mice. Histograms of fluorescence intensities of cells from spleen and LN of normal and AIG mice, and GLN and GM of AIG mice stained with anti-CD45RB antibody after gating on CD41 cells are shown. Numbers in the panels indicate percentages of cells in CD45RBlow and CD45RBhigh fractions. (b) Phenotypic characterization of CD41 T cells in AIG mice. Three-color flow cytometric analyses of cells from LN of normal mice, and GLN and GM of AIG mice were performed using antibodies against CD4, CD45RB and one of CD44, LFA-1, L-selectin or CD25. Contour plots of fluorescence intensities between CD45RB (horizontal axes) and another marker (vertical axes) of cells in the CD41 fraction are shown.

Fig. 2. Production of anti-GM autoantibody by the cells infiltrating into GM and those in lymphoid organs of AIG mice. Equal numbers (53105) of mononuclear cells from GM, GLN and spleen were cultured in 96-well plates for 2 days. Amounts of autoantibody secreted in the culture media were measured by ELISA as described in Methods and are shown as absorbance at 405 nm. Aliquots of culture media subjected to ELISA were chosen so that the resultant absorbance values were within the range linearly correlated to the amount of autoantibody (0.02–1.0).

shown in Fig. 3, autoantibodies not only of IgG2a subclass, production of which is augmented by IFN-γ (41), but also of IgG1 subclass, class switching to which IL-4 enhances (41), were detected in the sera of AIG mice. These findings strongly suggested that not only Th1 but also Th2 specific to GM selfantigens are sufficiently activated in the same individuals carrying AIG, but in relatively a restricted area, i.e. GLN and GM, as described above. These observations, however, seem to be rather contradictory with the fact that the functions of the two subsets are mutually exclusive (18–22), and raised the question whether the Th1 and Th2 subsets of CD41 cells are activated at the same place in AIG mice, especially at GM where suborganic compartmentation is difficult to imagine. Therefore, we analyzed the location of these subsets in AIG mice using expression of IFN-γ and IL-4 as markers for Th1 and Th2 cells respectively. Both CD41CD45RBhigh (naive) and CD41CD45RBlow (activated/memory) T cells were purified from GM or GLN cells using a cell sorter (Fig. 4a), and the mRNA expression of IFN-γ and IL-4 in both populations was assessed by RT-PCR followed by Southern hybridization. Since almost all the GM CD41 T cells belonged to the CD45RBlow, activated/memory population, we could not collect enough of the naive population from infiltrating CD41 T cells. As shown in Fig. 4(b), little cytokine mRNA was detected in the CD41CD45RBhigh populations from every source, in

Differential localization of Th1 and Th2 in autoimmune gastritis 1329

Fig. 3. IgG subclasses (IgG1 and IgG2a) of autoantibodies against GM in the sera of AIG mice. Autoantibodies in each serum were detected by ELISA as described in Methods. Titers of autoantibodies were defined as the highest dilution point at which the absorbance at 405 nm is .0.1 when test sera are serially diluted by 3-fold from 1/40 dilution.

agreement with their ‘naive’ phenotype. In the case of the CD41CD45RBlow population, patterns of mRNA expression were different from each source. The collected cells from GLN of AIG mice expressed both of IFN-γ, and evidently a high level of IL-4 mRNAs. In sharp contrast, those from infiltrating CD41 T cells into the inflamed GM expressed much less IL-4 mRNA but an increased level of IFN-γ mRNA as compared to GLN. CD41CD45RBlow cells sort-purified from LN of normal mice, which are ~20% of the total CD41 cells in LN, also expressed mRNAs of both cytokines at a lower amount than those produced by GLN CD41CD45RBlow cells in AIG. We further confirmed the expression of IFN-γ or IL-4 in CD41 T cells as protein at the single cell level by an intracellular staining technique. Lymphocytes derived from normal or AIG mice were first stimulated by phorbol myristate acetate and ionomycin in vitro, followed by the treatment with monensin to stop the secretion of cytokines from the Golgi apparatus, and such cytokines trapped into Golgi were stained together with CD4 antigen on the cell surface. Roughly 20–30% of CD41 T cells from spleen, GLN and GM of AIG expressed detectable levels of either IFN-γ or IL-4 (Fig. 5), showing that the majority of CD41 T cells in AIG mice have an activated/memory phenotype. On the other hand, ,6% and ~3% of CD41 T cells from normal spleen and LN respectively expressed such cytokines (Fig. 5), also correlating the finding that CD41CD45RBlow activated/memory cells were ,30% of normal spleen or LN cells (Fig. 1a). Among cells in GM of AIG mice producing either cytokine, only less than one-seventh (3.18% of total CD41 cells) expressed IL-4 and most of them (20.65% of total CD41 cells) expressed IFN-γ. In contrast, significant populations, roughly half (14.23% of total CD41 cells) and one-third (6.63% of total CD41 cells) of such cells in GLN and spleen of AIG mice respectively, expressed IL-4 (Fig. 5). These results both on the expression pattern of the landmark cytokine mRNAs in the sort-purified CD41 T cells (Fig. 4) and

on the distribution of such cytokine-expressing cells observed by cytoplasmic staining (Fig. 5) are quite consistent with each other, and clearly indicated that IFN-γ producers, i.e. Th1 cells are dominant at GM and IL-4 producers, and Th2 cells are mainly resident in GLN. Our observations are not from the auto-reactive T cells alone but from the activated T cells at the inflammation sites in total. However, auto-reactive T cells might not be insignificant among them since mRNA encoding TCR very similar to that of a T cell clone specific to the proton pump (autoantigen) in frequently detectable in GM of mice bearing AIG after d3-Tx (17). Therefore, it is possible to assume that auto-reactive Th1 and Th2 cells in this AIG model also show the similar, differential localization, although formal evidence is lacking. This differential localization of Th1/Th2 subsets even at closely located, but two anatomically distinct sites for autoimmune gastritis, i.e. inflamed lesions of GM and its GLN, might explain why tissue destructive inflammation can occur at GM in spite of the fact that the two subsets co-exist in the same animal, i.e. Th1 cells predominantly localized in GM can further be activated without inhibitory effects of Th2 cells by either autoantigen specific or non-specific manner. On the other hand, both Th1 and Th2 cells were observed in GLN— this apparent co-activation of both subsets might be managed by sub-organic structures such as germinal centers or follicles.

Th1 but not Th2 cells from AIG mice could easily transmigrate through a monolayer of endothelial cells in vitro Since the primary cause of infiltration of effector cells into the inflamed tissue is migration of such cells across the blood endothelial wall into the region of inflammation (42), the difference in the transmigration ability between two subsets of CD41 T cells into the inflamed lesion could be one of the reasons for the selective accumulation of Th1 cells but not Th2 cells in the GM of AIG mice described above. In order to test the possibility that the two subsets have different TEM ability, we set up an assay system using a murine endothelial cell line, F-2, as separator layer cells (Fig. 6). Similar systems were already shown to be a useful tool for studying TEM (43,44). In our system, CD41 T cells prepared from AIG mice were applied onto the confluent monolayer of F-2 cells growing on a pored polycarbonate membrane which is topologically equivalent to inner blood vessels, and after 4 h of incubation ~5–10% of input T cells reproducibly migrated across the endothelial monolayer and could be collected from the lower compartment equivalent to the tissue fluid of outer blood vessels (Fig. 6a). We analyzed the expression levels of IFN-γ and IL-4 mRNAs of these transmigrated cells and compared them to those of input cells or cells remaining in the upper compartment. The transmigrated cells expressed substantial amounts of IFN-γ but little or no detectable IL-4 mRNA, in contrast to input cells or cells in the upper compartment they expressed plenty of both types of cytokines (Fig. 7a). Almost the same results were obtained using cells from both of spleen and GLN of AIG mice (Fig. 7a, GLN and spleen). This ability to support the selective transmigration seems to be specific to some kinds of endothelial cells since very few cells were recovered from the lower compartment when NIH 3T3 cells were used as the separator layer instead of F-2 cells. Furthermore, in such cells (,1% of input cells) recovered


Fig. 4. (a) Purification of CD45RBlow and CD45RBhigh fractions of CD41 T cells by cell sorting. Cells from tissues indicated were stained with antibodies against CD4 and CD45RB, and fractionated into CD41CD45RBlow and CD41CD45RBhigh populations. Gates for sorting are shown by bold squares in the upper panels. Purity of the sorted cells in each population was confirmed by re-analysis on the same sorter as shown in the lower panel. (b) Expression of IFN-γ and IL-4 mRNAs in the sorted populations. Total RNA extracted from each population was subjected to RT-PCR analysis followed by Southern hybridization as described in Methods. cDNAs serially diluted 2 times were used for the PCR reaction and starting amounts of cDNAs for the dilution were normalized by the amounts of β-actin mRNA as shown in the bottom panels. RT-PCR/ hybridization results for IFN-γ mRNA (upper panels) and IL-4 mRNA (middle panels) of three dilution points of cDNAs from the sort-purified cells of CD41CD45RBlow and CD41CD45RBhigh populations derived from the tissues indicated on top are shown.

from the lower compartment using NIH 3T3, neither of IFN-γ nor IL-4 mRNA were detected by RT-PCR even after normalization of the RNA amount (data not shown). On the other hand, no bias to the IFN-γ dominancy was seen in the cells which have simply passed through the Transwell pored membrane without a cell monolayer (data not shown). Intracellular staining of IFN-γ was also performed to analyze expression at the

single cell level, as the marker for activated Th1 cells. As shown in Fig. 7(b), IFN-γ-producing Th1 cells were highly concentrated in the transmigrated compartment of the TEM assay using spleen CD41 cells from AIG mice. The fine molecular mechanism(s) to differentially localize Th1 and Th2 subsets in AIG mice is yet to be elucidated, but it is likely that cells of each subset have a distinct ability to

Differential localization of Th1 and Th2 in autoimmune gastritis 1331

Fig. 5. IFN-γ- and IL-4-producing CD41 T cells in lymphoid organs and inflamed GM of AIG mice detected by intracellular staining. Fluorescence intensities after anti-IFN-γ (horizontal axes) and anti-IL4 (vertical axes) staining of the cells in the CD41 fraction from the organs indicated are shown as dot plots. Percentages of positively stained cells (in the enclosed areas) are indicated in the panels.

infiltrate from the blood vessels into and/or to stay at the autoimmune lesion. In fact, our findings demonstrate that this is the case. We do not know, at this time, if this difference in TEM activity can be generalized to all the activated Th1 and Th2 cells, including those stimulated with given exogenous antigens, or not. However, the Th1 predominance observed in the inflamed GM of AIG mice could be explained by the difference in TEM activity in vitro. It is now well established that activated T cells differentiate into two distinct types of effector subpopulations, i.e. Th1 and Th2, and they follow distinct developmental pathways in response to quite different stimuli. Thus, it is likely that the differences in expression and/or functions of adhesion molecules could be responsible for the differential TEM ability of Th1 and Th2, possessing distinct function. At this time, differential expression of known adhesion molecules on these subsets has not been recognized; however, Austrup et al. have reported undetermined ligand(s) for P- or E-selectin exclusively expressed on the Th1 cells at inflamed skin and arthritic joints (45). In fact, P- and E-selectins are expressed on activated endothelial cells, and might act as the initial adhesive molecules for effector cells to migrate to inflamed

Fig. 6. Schematic representation (a) of the TEM assay and the cells (b–d) used for the assay. Experimental details are given in Methods. Morphology of murine endothelial cell line, F-2, on the conventional culture plate (3400), on the Transwell plate (3400) and that of NIH 3T3 on the Transwell plate (3400) are shown in (b), (c) and (d) respectively. Cells on the Transwell plates were fixed by ethanol:acetic acid (85:15) solution and stained with Hemacolor (Merck, Darmstadt, Germany).


Fig. 7. (a) Expression of IFN-γ and IL-4 mRNAs in CD41 T cells purified from GLN or spleen of AIG mice after the TEM assay. Total RNAs were extracted from CD41 T cells recovered from the upper or lower compartment of Transwell plates after the TEM assay or input cells kept in culture medium, and subjected to RT-PCR to quantify mRNAs. cDNAs serially diluted 4 times were used for PCR and starting amounts of cDNAs for the dilution were normalized by the amounts of β-actin mRNA. (b) Expression of IFN-γ in the individual cells recovered from the TEM assay. Spleen CD41 T cells purified from AIG mice were subjected to the TEM assay, and expression of IFN-γ was detected by intracellular staining and flow cytometry. Fluorescence intensities after anti-IFN-γ staining in the CD41 fraction of the cells recovered from the upper and lower compartments or input cells were plotted as histograms. Percentages of negatively and positively stained cells in the region M1 and M2 respectively are also shown.

sites and across the blood endothelium (46,47). Since multistep binding of lymphocytes to endothelium is required for many situations of TEM (46), it is also possible that the difference is dependent on further transmigration steps other than the initial adhesion. We have no data on the differences in adhesive properties to F-2 cells between Th1 and Th2 cells from AIG mice. We are now investigating which steps and/or molecules are different between the Th1 and Th2 of AIG mice, and also whether such differences are intrinsic to Th1 and Th2 cells or reflection of difference of their activation in AIG. It is very likely that selective transmigration of Th1 cells is independent of the antigen specificity of them. Naturally, many of the GM infiltrating cells are not self-reactive but the cells activated as bystanders. However, preferential TEM ability of whole Th1 cells including such bystanders could produce the environment of Th1 dominance in the inflamed site and might facilitate to elicit self-antigen-specific Th1 responses. Our findings described here lead to the idea that, during the onset of T cell-mediated autoimmune disorders, selfantigen-reactive Th1 cells, partly activated, that have escaped from negative selection or peripheral suppression (triggered Th1 cells, due to d3-Tx in this experimental model) might selectively migrate into the locus of autoimmune reaction (GM in our model) and be further activated by contact with the

self-antigens presented by MHC on the target cells (parietal cells in our model) and thus destroy them to induce inflammation. Selective migration of Th1 cells might also facilitate subsequent inflammatory processes. Th2 cells (also triggered) might stay at the draining LN (GLN in our case) and be further activated by the self-antigens released from the lesion by tissue destruction. They support autoantibody production and might become memory cells. Triggering of Th2 cells may also occur at the draining LN by encountering self-antigen, but a small fraction of circulating Th2 cells also could pass through the focal lesion and be further activated. Such a scheme might be applicable to many T cell-mediated autoimmune disorders, but the actual mechanisms of the disease onset and subsequent processes of inflammation remain to be elucidated.

Acknowledgements We are grateful to Dr Y. Agata and H. Kitao for useful suggestions and discussions, and Professor S. Nishikawa for cell sorting and valuable discussions. We also thank Dr S. Hirano for excellent technical assistance. This work was supported in part by Grants-inAid for Science Research from the Ministry of Education, Science Sports and Culture of Japan. T. K. is a Research Fellow of Japan Society for the Promotion of Science.

Differential localization of Th1 and Th2 in autoimmune gastritis 1333 Abbreviations AIG d3-Tx GLN GM LN PE TEM

autoimmune gastritis day 3 thymectomy gastric lymph node gastric mucosa lymph node phycoerythrin transendothelial migration

18

19 20

21

References 1 DeHeer, D. H., Linder, E. J. and Edgington, T. S. 1978. Delineation of spontaneous erythrocyte autoantibody responses of NZB and other strains of mice. J. Immunol. 120:825. 2 Shibata, T., Berney, T., Reininger, L., Chicheportiche, Y., Ozaki, S., Shirai, T. and Izui, S. 1990. Monoclonal anti-erythrocyte autoantibodies derived from NZB mice cause autoimmune hemolytic anemia by two distinct pathogenic mechanisms. Int. Immunol. 2:1133. 3 Murakami, M., Tubata, T., Okamoto, M., Shimizu, A., Kumagai, S., Imura, H. and Honjo, T. 1992. Antigen-induced apoptotic death of Ly-1 B cells responsible for autoimmune disease in transgenic mice. Nature 357:77. 4 Tisch, R. and McDevitt, H. 1996. Insulin-dependent diabetes mellitus. Cell 85:291. 5 Kojima, A., Taguchi, O. and Nishizuka, Y. 1980. Experimental production of possible autoimmune gastritis followed by macrocytic anemia in athymic nude mice. Lab. Invest. 42:387. 6 Kojima, A. and Prehn, R. T. 1981. Genetic susceptibility of postthymectomy autoimmune diseases in mice. Immunogenetics 14:15. 7 Sakaguchi, S., Fukuma, K., Kuribayashi, K. and Masuda, T. 1985. Organ-specific autoimmune diseases induced in mice by elimination of T cell subset. I. Evidence for the active participation of T cells in natural self-tolerance; deficit of a T cell subset as a possible cause of autoimmune disease. J. Exp. Med. 161:72. 8 Zamvil, S. S. and Steinman, L. 1990. The T lymphocyte in experimental allergic encephalomyelitis. Annu. Rev. Immunol. 8:579. 9 Katz, J. D., Wang, B., Haskins, K., Benoist, C. and Mathis, D. 1993. Following a diabetogenic T cell from genesis through pathogenesis. Cell 74:1089. 10 Asano, M., Toda, M., Sakaguchi, N. and Sakaguchi, S. 1996. Auto-immune disease as a consequence of developmental abnormality of a T cell subpopulation. J. Exp. Med. 184:387. 11 Liblau, R. S., Singer, S. M. and McDevitt, H. O. 1995. Th1 and Th2 CD41 T cells in the pathogenesis of organ-specific autoimmune diseases. Immunol. Today. 16:34. 12 Smith, H., Lou, Y.-H., Lacy, P. and Tung K. S. K. 1992. Tolerance mechanism in experimental ovarian and gastric autoimmune diseases. J. Immunol. 149:2212. 13 Suri-Payer, E., Kehn, P. J., Cheever, A. W. and Shevach, E. M. 1996. Pathogenesis of post-thymectomy autoimmune gastritis. Identification of anti-H/K adenosine triphosphatase-reactive T cells. J. Immunol. 157:1799. 14 Mori, Y., Fukuma, K., Adachi, Y., Shigeta, K., Kannagi, R., Tanaka, H., Sakai, M., Kuribayashi, K., Uchino, H. and Masuda, T. 1989. Parietal cell autoantigens involved in neonatal thymectomyinduced murine autoimmune gastritis. Studies using monoclonal autoantibodies. Gastroenterology 97:364. 15 Jones, C. M., Callaghan, J. M., Gleeson, P. A., Mori, Y., Masuda, T. and Toh, B. H. 1991. The parietal cell autoantibodies recognized in neonatal thymectomy-induced murine gastritis are the α and β subunit of gastric proton pump. Gastroenterology 101:287. 16 Nishio, A., Hosono, M., Watanabe, Y., Sakai, M., Okuma, M. and Masuda, T. 1994. A conserved epitope on H1,K1-adenosine triphosphatase of parietal cells discerned by a murine gastritogenic T cell clone. Gaseroenterology 107:1408. 17 Katakai, T., Agata, Y., Shimizu, A., Ohshima, C., Nishio, A., Inaba, M., Kasakura, S., Mori, K. J. and Masuda, T. 1997. Structure of T cell receptor expressed on a gastritogenic T cell clone, II-6, and

22 23 24

25

26

27

28

29

30

31

32

33

34 35

36

frequent appearance of the similar clonotypes in mice bearing autoimmune gastritis. Int. Immunol. 9:1849. Seder, R. A. and Paul, W. E. 1994. Acquisition of lymphokineproducing phenotype by CD41 T cells. Annu. Rev. Immunol. 12:635. Abbas, A. K., Murphy, K. M. and Sher, A. 1996. Functional diversity of helper T lymphocytes. Nature 383:787. Powrie, F. and Coffman, R. L. 1993. Cytokine regulation of T cell function: potential for therapeutic intervention. Immunol. Today. 14:270. Fitch, F. W., McKisic, M. D., Lancki, D. W. and Gajewski, T. F. 1993. Differential regulation of murine T lymphocyte subsets. Annu. Rev. Immunol. 11:29. Reiner, S. L. and Locksley, R. M., 1995. The regulation of immunity to Leishmania major. Annu. Rev. Immunol. 13:151. Katz, J. D., Benoist, C. and Mathis, D. 1995. T helper cell subsets in insulin-dependent diabetes. Nature 268:1185. Trembleau, S., Penna, G., Bosi, E., Mortara, A., Gately, M. K. and Adorini, L. 1995. Interleukin 12 administration induces T helper type 1 cells and accelerates autoimmune diabetes in NOD mice. J. Exp. Med. 181:817. Rapoport, M. J., Jaramillo, A., Zipris, D., Lazarus, A. H., Serreze, D. V., Leiter, E. H., Cyopick, P., Danska, J. S. and Delovitch, T. L. 1993. Interleukin 4 reverses T cell proliferative unresponsiveness and prevents the onset of diabetes in nonobese diabetic mice. J. Exp. Med. 178:87. Khoury, S. J., Hancock, W. W. and Weiner, H. L. 1992. Oral tolerance to myelin basic protein and natural recovery from experimental autoimmune encephalomyelitis are associated with downregulation of inflammatory cytokines and differential upregulation of transforming growth factor β, interleukin 4, and prostaglandin E expression in the brain. J. Exp. Med. 176:1355. Mueller, R., Krahl, T. and Sarvetnick, N. 1996. Pancreatic expression of interleukin-4 abrogates insulitis and autoimmune diabetes in nonobese diabetic (NOD) mice. J. Exp. Med. 184:1093. Ramsdell, F., Seaman, M. S., Miller, R. E., Picha, K. S., Kennedy, M. K. and Lynch, D. H. 1994. Differential ability of Th1 and Th2 cells to express Fas ligand and to undergo activation-induced cell death. Int. Immunol. 6:1545 Nishio, A., Katakai, T., Oshima, C., Kasakaura, S., Sakai, M., Yonehara, S., Suda, T., Nagata, S. and Masuda, T. 1996. A possible involvement of Fas–Fas ligand signaling in the pathogenesis of murine autoimmune gastritis. Gastroenterology 111:959. Martinelli, T. M., van Driel, I. R., Alderuccio, F., Gleeson, P. A. and Toh, B. H. 1996. Analysis of mononuclear cell infiltrate and cytokine production in murine autoimmune gastritis. Gastroenterology 110:1791. Fukuma, K., Sakaguchi, S., Kuribayashi, K., Chen, W. L., Morishita, R. Sekita, K., Uchino, H. and Masuda, T. 1988. Immunologic and clinical studies on murine experimental autoimmune gastritis induced by neonatal thymectomy. Gastroenterology 94:274. Mori, Y., Hosono, M., Murakami, K., Katoh, H., Yoshikawa, Y., Kuribayashi, K., Kannagi, R., Sakai, M., Okuma, M. and Masuda, T. 1991. Genetic studies on experimental autoimmune gastritis induced by neonatal thymectomy using recombinant inbred strains between a high-incidence strain, BALB/c, and a lowincidence strain, DBA/2. Clin. Exp. Immunol. 84:145. Han, H., Okamoto, M., Honjo, T. and Shimizu, A. 1991. Regulated expression of immunoglobulin trans-mRNA consisting of the variable region of a transgenic µ chain and constant regions of endogenous isotypes. Int. Immunol. 3:1197. Shimizu, A. and Honjo, T. 1993. Synthesis and regulation of transmRNA encoding immunoglobulin ε heavy chain. FASEB J. 7:149. Toda, Y., Tsujioka, K., Maruguchi, Y., Ishii, K., Miyauchi, Y., Kuribayashi, K. and Imamura, S. 1990. Establishment and characterization of a tumorigenic vascular endothelial cell line (F2). Cancer Res. 50:5526. Maruyama, H., Toda, K., Uno, K., Miyake, K., Matsushima, K., Yamamoto, K., Mori, K. J. and Masuda, T. 1993. Murine endothelial cell line cells, F-2: interaction with leukocytes and cytokines production. Microbiol. Immunol. 37:895.

1334 Differential localization of Th1 and Th2 in autoimmune gastritis 37 Taguchi, O. and Nishizuka, Y. 1981. Experimental autoimmune orchitis after neonatal thymectomy in the mouse. Clin. Exp. Immunol. 46:425. 38 Lafaille, J. J., Nagashima, K., Katsuki, M. and Tonegawa, S. 1994. High incidence of spontaneous autoimmune encephalomyelitis in immunodeficient anti-myelin basic protein T cell receptor transgenic mice. Cell 78:399. 39 Renno, T., Zeine, R., Girard, J. M., Gillani, S., Dodelet, V. and Owens, T. 1994. Selective enrichment of Th1 CD45RBlowCD41 T cells in auto-immune infiltrates in experimental allergic encephalomyelitis. Int. Immunol. 6:347. 40 Scott, B., Liblau, R., Degermann, S., Marconi, L. A., Ogata, L., Caton, A. J., McDevitt, H. O. and Lo, D. 1994. A role for nonMHC genetic polymorphism in susceptibility to spontaneous autoimmunity. Immunity 1:73. 41 Finkelman, F. D., Holmes, J., Katona, I. M., Urban, Jr., J. F., Beckmann, M. P., Park, L. S., Schooley, K. A., Coffman, R. L., Mosmann, T. R. and Paul, W. E. 1990. Lymphokine control of in vivo immunoglobulin isotype selection. Annu. Rev. Immunol. 8:303.

42 Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte emigration: the multistep paradigm. Cell 76:301. 43 Pietschmann, P., Cush, J. J., Lipsky, P. E. and OppenheimerMarks, N. 1992. Identification of subsets of human T cells capable of enhanced transendothelial migration. J. Immunol. 149:1170. 44 Ro¨hnelt, R. K., Hoch, G., Reiß, Y. and Engelhardt, B. 1997. Immunosurveillance modelled in vitro: naive and memory T cells spontaneously migrate across unstimulated microvascular endothelium. Int. Immunol. 9:435. 45 Austrup, F., Vestweber, D., Borges, E., Lo¨hning, M., Bra¨uer, R., Herz, U., Renz, H., Hallmann, R., Scheffold, A., Radbruch, A. and Hamann, A. 1997. P- and E-selectin mediate recruitment of Thelper-1 but not T-helper-2 cells into inflamed tissues. Nature 385:81. 46 Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte emigration: the multistep paradigm. Cell 76:301. 47 Lasky, L. A. 1992. Selectins: interpreters of cell-specific carbohydrate information during inflammation. Science 258:964.