Aug 15, 1993 - Victoria Camerini,* Chetan Panwala,+ and Mitchell Kronenberg2+ ... Comprehensive Cancer Center, University of California at Los Angeles, ...
0022-1 767/93/1514-1765$02.00/0 The Journal of Immunology Copyright 0 1993 by The Amerlcan Association of Immunologists
Vol 151, 1765-1 776, No. 4, August 15, 1993 Printed in U.S.A.
Regional Specialization of the Mucosal Immune System lntraepithelial Lymphocytes of the Large Intestine Have a Different Phenotype and Function than Those of the Small Intestine'
Victoria Camerini,* Chetan Panwala,+ and Mitchell Kronenberg2+ *Division of Neonatology, Department of Pediatrics and the +Department of Microbiology and Immunology and Jonsson Comprehensive Cancer Center, University of California at Los Angeles, Los Angeles, CA 90024
ABSTRACT. lntraepithelial lymphocytes (IEL) arefound in boththe small and the large intestine. We demonstrate that there are a number of striking phenotypic and functional differences between the two populations of IEL isolated from mice. Inthe large intestine, the majority of IEL express the (YPTCR, and among these TCR-aP+ lymphocytes, CD4+ cells are as prevalent as CD8+ cells. In contrast, in the small intestine, most of the TCR-aP+ IEL express CD8, and an increased percentage of cells express TCR-yG. In addition, most TCR-y8+ IEL isolated from the large intestine (LI-IEL) are CD4- CD8- cells, as compared to TCR-yG+IEL isolated from the small intestine (SI-IEL), which are predominantly CD8+. Furthermore, CD2 and the lymph node homing receptor, L-selectin, are expressed by most LI-IEL but not by SI-IEL. Furthermore, LI-IEL have much less cytolytic activity than SI-IEL. These data suggest that LI-IEL are a distinct population of lymphocytes that may have a different immunologic role than that of SI-IEL. journal of Immunology, 1993, 151 : 1765.
T
he presence of lymphocytes within the epithelium of the gastrointestinal tract has long been recognized (1). Although the function of these cells is not completely understood, IEL3 represent a unique population of cells that may play a primary role in cellmediated mucosal defense, surveillance of epithelial cell integrity, and in the pathogenesis of a variety of disease states. The phenotype of small intestinal IEL has been extensively characterized in mouse and man ( 2 4 ) . Although mouse SI-IEL are predominantly T cells, they have been found to have a number of unique properties when compared to T lymphocytes found in secondary lymphoid organs. Most notable of these properties are the predomiReceived for publication January 15, 1993. Accepted for publication May 19, 1993.
nance of CD8+ cells ( 3 , the relatively large fraction (2080%) of SI-IEL that express a TCR comprised of y- and 6-chains (6,7), the rapid and high level response these cells display in assays that measure cytotoxicactivity (8), and the thymus-independent development of some SI-IEL (9). Although the existenceof intraepithelial lymphocytes in the largeintestine has been reported in both mouse and man (1, lo), these cells havenot been well characterized in mice. We present phenotypic and functional analyses of mouse LI-IEL. We report that IEL isolated from thelarge intestine share common featureswith those from the small intestine, as might be expected. There are, however, striking differences between these populations both in the expression of a number of cell surface proteins and in assays of cytotoxicity. The localization of LI-IEL and SI-IEL within the epithelium at different sitesmay correlate with the presence of locally distinct and specific immune functions.
by the payment of The costs of publication of this article were defrayed in part page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1765
1766
Materials and Methods Mice
BALB/c, C3H, and C57BL/6 strain mice originally obtained from The Jackson Laboratories (Bar Harbor, ME) or Simonsen Laboratories (Gilroy, CA), were bred and maintained at the UCLA Vivarium(Los Angeles, CA). BALB/c ndnu mice were obtained from the nude mouse facility at UCLA. Female mice were used at 8 to 13 wk of age. Phenotypic analyses were performed withall three mouse strains, but predominantly with BALB/c mice. Functional assays were performed with BALB/c and C3H mice.
Preparation of IEL, LPL, and LAL
IEL from the large and small intestine were prepared according to a modification of a previously published procedure for SI-IEL (11). In several experiments, IEL from the cecum and colon were analyzed separately. We observed no differences between these two populations of IEL. Therefore, the large intestine was analyzed as a single unit and defined as the gastrointestinal tract from the ileocecal junction to the peritoneal reflection at the sigmoid colon. The small intestine was defined as originating from the duodenal bulb and terminating at the ileo-cecal junction. Each intestine was removed, carefully cleaned from its mesentery, and flushed of fecal contents. The lymphoid aggregates of the large intestine and Peyer’s patches of the small intestine were removed before processing. The intestinal tissue was cut longitudinally and then into 2- to 4-cm pieces. It was shaken three times at 200 rpm for 20 to 30 min at 37°C in HBSS without calcium and magnesium (GIBCO, Bethesda, MD) andwith 1 mM DTT (Sigma Chemical Co., St. Louis, MO). The cell suspensions were collected and centrifuged in a discontinuous 40%/70% Percoll (Pharmacia Fine Chemicals, Piscataway, NJ) gradient at 600 X g for 25 min. Cells from the 40%/70% interface were collected, washed several times, and resuspended in complete RPMI media (UCLAMedia Center, Los Angeles, CA) supplemented with 10% FCS. Gentamicin sulfate (Sigma) was added at 5 pgiml in media usedfor functional assays. The purified cell suspensions were stored at 4°C for up to 5 h before use in phenotypic or functional assays. Cells prepared in this way are a mixture of IEL and IEC as judged by light microscopy, light scatter properties in the flow cytometer, and by staining with mAbspecific for these cell types. The average yield per mouse wasapproximately 1.0 X lo6 LI-IEL and 1.0 X lo7 SI-IEL prepared from conventional mice and 4.0 X lo5 for both LI-IEL and SIIEL prepared from athymic ndnu mice. Purified cells were more than 98% viable by exclusion of trypan blue. All phenotypic and functional assays were performed with IEL isolated from large and small intestines in parallel. Pooled samples were prepared from at least twomice
INTRAEPITHELIAL LYMPHOCYTES OF LARGE INTESTINE
matched for sex, age, and strain, and when possible, litter mates were used. After removal of the epithelial cell layer, large intestinal segments were diced into 1- to 2-mm pieces and resuspended in HBSS withoutcalcium or magnesium (GIBCO). Lamina propria lymphocytes (LPL) were prepared using modifications of previously published procedures (11, 12). Briefly, LPL suspensions were prepared after digestion of minced intestine with either collagenase type 11, (C-2139), at 90 U/ml or Dispase at 1.5 mg/ml (both from Sigma) in complete RPMI supplemented with10%FCS. Samples were then incubated at 37°C for 45 to 90 min with shaking at 200 rpm. Cell suspensions were collected, filtered through 60-pm nylonmesh,washedwithHBSS,resuspendedin complete RPMImediawith10%FCS,and placed on discontinuous 40%/70% Percoll gradients as described for the preparation of IEL. After centrifugation, lymphocytes were collected from the 40%/70% interface, washed several times, and prepared for mAb staining as described below. Cells prepared in this way were more than 98% viable as determined by exclusion of trypan blue. Both collagenase and dispase treatments resulted in cell yields of approximately 1.5 X lo6 cells per mouse and similar viability. Lymphoid aggregates, which were removed from large intestine before processing for IEL or LPL isolation, were placed in complete RPMI supplemented with 10% FCS at 4°C for 30 min before either mechanical disruption or digestion with Dispase as described for the preparation of LPL. Supernatants were filtered through nylon mesh and prepared directly for staining with mAb as described. On average, 1 to 3 aggregates were present on the cecum and 6 to 10aggregates were present on the colon. LAL prepared in this way were more than 98% viable by trypan blue exclusion and the average yield was approximately 2.0 X lo6 cells/mouse.
Antibody staining and flow cytometric analysis
IEL prepared as above were resuspended at a concentration of 1 X lo5 cells/ml in PBS staining buffer containing 5% FCS and 0.02% NaN3. Pretitered mAb, either unconjugated or directly conjugated to FITC, PE, or biotin, were added to cell suspensions at 4°C and incubated for 20 to 30 min. After the primary incubation, samples were washed three times in PBS with0.02% NaN3 at 4°C. If the initial reaction was performed with a biotinylated mAb, or an unconjugated primary mAb, the cells were resuspended, as noted above for primary staining reactions, and pretitered FITC or PE-conjugated streptavidin (Becton Dickinson, Emeryville, CA) goat anti-hamster IgG, goat anti-rat IgG (all from CalTag, South San Francisco, CA) were added as appropriate. The secondary reactions were then incubated and washed as described previously. After mAb staining and
Journal of immunology washing, all samples were fixed in PBS with 0.02% NaN, and 1% paraformaldehyde and stored at 4°C before FACScan analysis. Directly conjugated mAb reagents used in the above reactions to characterize LI-IEL and SI-IEL were: CD2-FITC (RM2-5) (13), CD3eFITC (145-2C11) (14), NK cell-FITC (PK136) ( 1 3 , CD8a-PE (53-6.7) (16), CD8P-PE (53-5.8) (17), TCR-aP-FITC(18) TCR-76biotin (GL3) (19), CD45R-FITC (Ly-5), (16A) (20) (all from PharMingen, San Diego, CA), or CD4-FITC (YTS 191.1), CD45R-PE (B220), (RA3-6B2) (21), goat anti mouse IgG-PE, (all from CalTag) or Thy-1.2-biotin (30 H12) (17), CD4-PE (GK1.5) (22), and CD8a-FITC (536.7) (all from Becton Dickinson). Unconjugated primary mAb against the following Ag were incubated with IEL as described above: Vy5 (GLI) (19) (a gift of Dr. Leo Lefrancois, University of Connecticut, Farmington, CT), TCR-ap (H57-597) (18) (a gift of Dr. R. Kubo, University of Colorado, Denver, CO), CD25 or IL-2R a-chain (7D4) (23) (American Type Culture Collection Rockville, MD), L-selectin (MEL- 14) (24) (a gift of Dr. D.Camerini, UCLA, Los Angeles, CA), aIEL chain coexpressed with a p7 chain to form an integrin expressed by IEL (M290) (25) (a gift of Dr. P. J. Kilshaw, AFRC Institute of Animal Physiology and Genetics, Cambridge, UK), and G8.8a (26) (a gift of Dr. A. Fan; University of Washington, Seattle, WA). FITC or PE-conjugated secondary reagents, including streptavidin, goat anti-rat IgG, or goat anti-hamster IgG were used as negative controls where appropriate. The samples were run on a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems, San Jose, CA), equipped with a 15 mW 488 nm air-cooled argon-ion laser in the UCLA Flow Cytometry Core Facility. Daily performance of the instrument was monitored by running Calibrite beads (Becton Dickinson), using the Autocomp software (Becton Dickinson). This software was used to optimize forward angle (FSC) and side angle light scatter (SSC) settings to distinguish IEL from IEC based on their size and granularity. Between 1500 and 5000 gated events, based on the FSC and SSC properties of the cell preparations, were acquired using C30/FACScan Research software (Becton Dickinson). The data were stored and later analyzed on a HP 9000 series model 3 10 computer. Single and multiple parameter analysis using dot plots and histogram overlays with corresponding statistics were used.
Histologic and immunocytochemical studies
The large intestine was removed as described in the previous section, rinsed free of fecal contents, and stored in PBS at 4°C until use. Sections (3-5 cm) of large intestine were either fixed in 10% formalin or immersed in Optimal Cutting Temperature (OCT) Compound (Miles Inc, Elkhart, IN) and snap frozen in liquid nitrogen. Formalinfixed tissue was embedded in paraffin and 5-pm sections
1767
were prepared and stained with hematoxylin-eosin. Frozen sections were also cut at 5 pm on a cryostat microtome and placed on glass slides. These sections were air-dried and fixed in acetone. Before mAb staining, frozen tissue sections were rehydrated with PBS and then incubated with PBS staining buffer with 20% FCS and 0.02% NaN, at 4°C before incubation with pretitered mAb CD4-PE (22) and CD8-PE (16) in PBS staining buffer with 5% FCSand 0.02% NaN, for 60 min. The sections were rinsed withPBS before analysis. Tissue sections stained with hematoxylineosin were photographed through a Nikon Labophot microscope with attached camera. Frozen sections were analyzed by epifluorescence microscopy with an IMT2-DMB dichroic mirror unit on an Olympus IMT-2 microscope. TCR stimulation of IEL
Microtiter plates were precoated with 50 p1 of a solution containing 10 to 25 pg/ml of pretitered CD3 mAb (1452 C l l ) in PBS. The plates were incubated either at room temperature for 2 h or overnight at 4°C. The wells were subsequently washed three times withcomplete RPMI supplemented with 10% FCS. To measure IEL IL-2R expression, freshly isolated LI-IEL and SI-IEL were plated at a density of 1.0 X lo5 cells/well in flat-bottom microtiter plates (Coming Glass Works, Corning, NY), with or without CD3 mAb precoating, in 200p1 of complete RPMI with 10% FCS and gentamicin sulfate. The cells were incubated at 37°C with 5% (v/v) C 0 2 for 48 h, harvested, and prepared for CD25(7D4) mAb staining as described above. To measure IL-2 production from TCR-stimulated IEL, LI-IEL, and SI-IEL were cultured as described above and their supernatants harvested at 48 h. The supernatants were frozen and thawed, and 50 p1 were added to cultures of the indicator cell line CTLL-20 (a gift from Dr. Eli Sercarz, UCLA, Los Angeles, CA). The CTLL-20 cells, which are responsive to IL-2 but not to IL-4 (27), were plated at a density of 1.O X lo4cells/well in 150 p1 of complete RPMI. After 20 to 24 h, the cultures were pulsed with 1 pCi/well of [,H]TdR for 24 h. The wells were harvested with an automated cell harvester and [3H]TdRincorporation measured by liquid scintillation counting. Redirected lysis assays
FcR+ P388Dl target cells (28) were prepared for use by incubating 5 X lo6 cells with 300 pCi of NaS'Cr2O3 (NENDu Pont, Wilmington, DE) for 4 h at 37°C with 5% (v/v) COz. Labeled target cells were washed four times and plated at 5 X lo3 cells/well in U-bottom 96-well microtiter plates (Costar, Cambridge, MA). LI-IEL and SI-IEL were titrated in separate wells to give an E:T ratio ranging from 1.25:1 to 40: 1 in a total volumeof 200 pl. Where indicated, CD3 mAb was added at the same time as IEL to give a final concentration of 2 pg/ml. The assays were incubated at
1768
LYMPHOCYTES INTRAEPITHELIAL
OF LARGE INTESTINE
(Table I). Furthermore, CD4- CD8- (double-negative) cells consistently represented a greater percentage of the total LI-IEL population as compared with SI-IEL (Fig. 1 and Cell Types.' Table I). Ag LI-IEL SI-IEL Further analyses of IEL were carried outusingtriple mAb stainingwith CD3-FITC plus CD4 and CD8 both Mean Yo f '>SD ( n ) Mean Yo +'> SD ( n ) conjugated to PE. The results in Figure IB, show that most CD2 16 f 10 (9) 85 f 10 (9) IEL in both sites are T lymphocytes, but at least 30% of 75 f 13 (2) 78 f 11 (2) CD3 24 f 14 (9) 11 f 4 (9) CD4 LI-IEL and approximately 10%of SI-IEL areCD4-, C D K , 31 f 13 (10) CD8 70 f 10 (10) and CD3- (triple-negative). A small proportion of these 1 f 1 (6) 8 f 8 (6) CD4+CD8+' 45 f 10 (6) 21 5 7 (6) CD4-CD8" triple-negative cells, in both small and large intestine, are 63 f 8 (4) 57 f 3 (4) n p TCR B lymphocytes (see below). Theremainder are probably not 31 f 11 (4) 12 f 8 (4) y6 TCR NK cells, as LI-IEL from C3H mice didnot display epitopes 77 f 15 ( 3 ) 64 * 11 (3) Thy 1.2 27 f 23 (3) MEL-14 3 f 4 (3) recognized by the NK cell mAb PK136 (data not shown) LI-IEL and SI-IEL were prepared i n parallel as described In Materials and (1 5 ) . Although the identity of the majority of these tripleMethods from either BALB/c, C57B/6. or C3H mice. negative cells has not been determined, we conclude that ')Results are expressed as the mean percent positive fluorescence. n, Numher of experlments. they belong to the hematopoietic cell lineage as more than Results from simultaneous analysis for CD4 and CD8 Ag. 95% of the cells in a preparation of LI-IEL react with the CD45R mAb (clone 16A). The triple-negative cells may 37°C with 5 % (v/v) COz. After12 h, 100 pl of supernatant represent an undefined population of NK cells, or lymphowerecollectedand counted in a y-counter. The percent cyte precursors that have not yet acquired markers charspecifickillingwascalculated as the sample release(in acteristic of mature cells. In addition, both populations of cprn), minus the spontaneous release (in cpm), divided by LI-IEL and SI-IEL contain cells that areCD4-CD8the total release (incpm), minusthe spontaneous release (in double-negative but CD3+ (Fig. 1B). Further studies, discpm). Spontaneous release was measured in supernatants cussed later, show that many of these double-negative T collected from labeled target cells incubated with media lymphocytes in the large intestine are TCR-y8+. alone, whereas total release was measured in supernatants The differences between the phenotypes of LI-IEL and collected from labeled target cells after treatment with 1% SI-IEL that we observed were unexpected. To ensure the (v/v) Triton-X 100 (Sigma). integrity of the method of isolation of LI-IEL, and to rule out contamination of the epithelial compartment by either Results LPLorLAL, we performed histologicand immunocyLI-IEL are predominantly T lymphocytes tochemical analyses of thin sections of the large intestine. By staining LI-IELwith mAb specific for CD4, CD8, CD3, Figure 2 shows photomicrographsof the large intestine before and after removal of the epithelial layer (A and B, and TCR isotypes, we determined that most LI-IEL, like respectively). In A, a villus with its associated columnar SI-IEL, areT lymphocytes. A summary of the staining data epithelium (solid arrow), underlying lamina propria (LP) from multiple experiments, in which LI-IEL and SI-IEL and numerous goblet cells oriented down the vertical axis were analyzed in parallel from the same mice, is presented of the crypt is shown. In B , a similar section of the large in Table I, and representative histograms from these analintestine,stainedafterstandardpreparation of IEL,is yses are presented in the figures. Themajority of the lymphocytes in both IEL populations expresses either CD4 or shown. This section is devoid of the columnar epithelium although the underlying lamina propria remains intact. AlCD8. There are, however, differences in the percentage of though in the treated intestine theremay be some widening CD4 and CD8 expressed. CD8 single positive cellsare of the lamina propria, and partial disruption of the linear less prevalent among LI-IEL,representing 31% of the lymarray of goblet cells down the vertical axis of the crypt, the phocyte population onaverage, whereas CD8+SI-IEL, lamina propria remains highly cellular and appears unafobtained from the same subset of mice, represented, on fected. Based on this, and analysis of numerous other secaverage, 70% of the population (Table I). Representative tions, we conclude that the method of preparation does not two-colorFACScan analyses, comparing CD4 and CD8 lead to significant disruption of the lamina propria tissue expression on LI-IEL and SI-IEL from the same of mice, set and release of LPL. Data obtained from staining frozen are shown in Figure 1A. LI-IEL could be further distintissuesectionswith CD4 (Fig. 2C) and CD8 (Fig. 2 0 ) guished from SI-IEL by the low or near complete absence mAb, confirm the presence of IEL expressing these molof CD4+ CD8+ (double-positive) cells. In Figure lA, less ecules. The arrows in C and D show both CD4+ and CD8+ than 1% of LI-IEL are double-positive, although 4% of IEL, respectively, located within the epithelial layer borSI-IEL display this phenotype; a similar difference in the dering the lumen at the top and right sides. Additional percent of double positive cells was consistently observed Table I
Phenotypic profile of /EL
1
1769
Journal of Immunology
SI IEL
LI IEL
A
30.1 I
r
0.5
I
FIGURE 1. CD4 and CD8 expression by IEL. A, Dualcolor immunofluorescence of LI-IEL and SI-IEL prepared from the same pool of BALB/c were reacted with rnAb CD4-FITC and CD8-PE. Data areexpressed in dot plot format of log fluorescence; the numbers in the quadrants are the percentage ofpositively staining cells. B, Dual color immunofluorescence of LI-IEL and SI-IEL reacted with rnAb CD4 andCD8, bothconjugated to PE and CD3-FITC. LI-IEL and SI-IEL prepared fromthe same pool of BALB/c mice were compared.
I-,""""""
-4
I
CD8
CD8
B
.. "
3.3 I
47.0
I
77.7
" "
10.0
..
-
m
CD3
CD4+ and CD8+ cells are located within the lamina propria. To further examinethe possibility that T cells fromlarge intestine LPL or LAL could have contaminated the IEL preparations, cell suspensions of LPL and LAL were analyzed by flow cytometry after staining with mAb. As summarized in Table 11, the yield per mouse of lymphocytes from epithelium, lamina propria, and lymphoid aggregates is similar. Therefore, if the majority of LI-IEL were derived from another compartment, agross cross-contamination of the IEL preparation would have been required. Furthermore, as shown in Table 11, the lymphocytes in allthree sites have distinct phenotypes. The majority of LAL, similar to Peyer's patch lymphocytes, is B cells. In addition, LI-IEL are greatly enriched for TCR-yG+ cells as compared to LPL and LAL. Nearly 100% of the B220+ cells in the LI-IEL preparation are T cells (CD3+), although only a minority of the B220+ LPL or LAL are CD3+. Collectively, these data demonstrate that contamination by adjacent compartments does not contribute significantly to the distinct differences between LI-IEL and SI-IEL. Expression of y6 and aP TCR by LI-IEL
LI-IEL contain a smaller percentage of TCR-yG cells than SI-IEL, although the percentage in LI-IEL is still greater than that found in nonepithelial lymphoid compartments. An average of 12% of LI-IEL expresses a yG TCR isotype,
CD3
n
whereas 3 1% of SI-IEL from the same mice, express this TCR isotype (Table I). Further comparison of IEL bearing either TCR receptor isotype, demonstrated differences in CD4 and CD8coreceptor expression. Most LI-IEL bearing the TCR-aP receptor, expressed either CD4 or CD8. Representative data from a two-color flow cytometric analysis of LI-IEL are shown in Figure 3A. Approximately 55% of the cells in this LI-IEL preparation express TCR-(UP,and 58% of these TCR-aP+ LI-IEL are CD4+ (Fig. 3A, l e f ) whereas 35% are CD8+ (Fig. 3A, right panel). As this preparation (data not shown), and other LI-IEL preparations, had very few double-positive cells (Table I), we conclude that less than 10% of the TCR-aP+ lymphocytes are double-negative cells. However, consistent with previous results (3), we found that the majority of TCR aP+ SI-IEL was CD4CD8+ (data not shown). Figure 3B depicts representative data from staining of LI-IEL and SI-IEL for TCR-yG and coreceptors, CD4 and CD8. IEL that are CD4+ and TCRyGt are practically undetectable (Fig. 3B, top). The majority of TCR yG LI-IEL is double negative, although a smaller percentage is CD8+ (Fig. 3B, bottom left). In contrast, consistent with previous reports (3-3, TCR-yG SIIEL are predominantly CD8+ (Fig. 3B, bottom right). The majority of TCR-yG IEL, from both the large and small intestine, expresses the Vy5 gene segment as determined by immunofluorescence staining with the mAb GL1 that is
LYMPHOCYTES INTRAEPITHELIAL
1770
FIGURE 2. Light photomicrographs of sagittalsections of thelarge intestine prepared before ( A ) or after ( B )removal of the epithelium and stained with hemotoxylin and eosin. Magnification, 1 0 0 ~ Photomicrographs . of frozen sagittal sectionstakenunderfluorescence observation after staining with either mAb CD4 (C) or CD8 ( D )both directly conjugated to PE. Positively staining cells are denoted by the arrows. Magnification, SOX.
INTESTINE OF LARGE
e
@hi? A,
Table II Mucosal lymphocyte compartments of large intestine Mean Yo Cellsc Cell Type'
LI-IEL
LPL LAL
Cell Yield"
9.0 x 10% 1.5 x 10' 2.7 x 1Oh
TCR-up
TCR-yti
71 48 15
14 2 0
13
6220' CD3*
6220CD3-
47 6 17
5 17 65
518
2 81
.' LI-IEL and LPL were prepared sequentially from the large intestinesof BALBlc mice as described in Materials andMethods. In a separate preparation, LAL were prepared and analyzed as described. Representative data from one of two experiments are shown. "Cell yield is the number of viable cells per mouse and wasdetermined by light microscopy after purification oflymphocytes on Percoll gradients and staining with trypan blue. 1 The data are expressed as the percent positive mean fluorescence after staining with directly conjugated mAb and FACScan analysis. specific for all or most cells that express the Vy5 TCR (30). Although most of these Vy5+ IEL are CD8+ in the small intestine, the majorityof GLl LI-IEL are double-negative (data not shown). +
Expression by IEL of molecules related to adhesion and activation
We found that most LI-IEL express CD2, which in mice has been reported to be present on thymocytes, peripheral T
lymphocytes, and a subset of B lymphocytes (1 3). In contrast, only a minority of SI-IEL prepared in parallel from the same mice, expresses this molecule (Table I). Representative data, derived from the analysis of LI-IEL and SIIEL prepared from BALB/c mice, are shown in Figure 4. Additional two-color FACScan analysis of LI-IEL for coexpression of CD2 with both TCR isotypes, revealed that the majority of TCR-aP cells coexpress CD2, although only a minority of TCR-$3 cells does (data not shown).
1771
Journal of Immunology
A
i
i
19.7
30.1
gi
I
.
----SIIEL
-LI IEL
;.*
m
TCR @
TCR ap
LI IEL
SI IEL
CD2 FIGURE 4. CD2 expression on IEL. Histogram analysis depicting comparative expression of CD2-FITC on LI-IEL and SI-IEL prepared in parallel from a single BALB/c mouse. Approximately 85% of the analyzed LI-IEL are CD2+, whereas 15% of the SI-IEL express CD2, compared to the stainingwith an isotype control, FITC-conjugated mAb.
-control ""
LI IEL
................SI
TCR @
TCR
fl
FIGURE 3. TCR isotype and coreceptor expression on IEL. Dual colorimmunofluorescence of LI-IEL and SI-IEL prepared in parallel from a pool of BALB/c mice. A, Coreceptor expression on TCR-aP+ LI-IEL. Purified LI-IEL were reacted first with either CD4-FITC or CD8-FITC mAb followed by TCR-aP mAb andthen by anti-hamster IgG-PE. 6,Coreceptor expression on TCR-yS+ IEL. Dualcolor immunofluorescence of LI-IEL (left column) and SI-IEL (right column) reacted separately with either CD4-FITC (top row) or CD8-FITC rnAb (bottom row) and then with TCR yS mAb followed bystreptavidin-PE.
IEL
MEL-14
-control
- - - - LI
................ SI
IEL IEL
M290 L-Selectin, an adhesion molecule previously characterized as a homing receptor for high endothelial venules in peripheral lymph nodes (identified by the mAb MEL-14) is also differentially expressed on IEL (Table I). Our data indicate that some LI-IEL express L-selectin, although most SI-IEL do not. Representative histograms obtained by staining LI-IEL and SI-IEL, prepared from the same mice, with the mAb MEL-14, are presented in Figure 5A. In contrast, however, both LI-IEL and SI-IEL were found to express the unique integrin a-chain, identified by the mAb M290 (Fig. 5s).This a-chain, coexpressed with a p7 chain, was previously shown to beexpressed on almost all SI-IEL (31). Although it is expressed by the majority of both LIIEL and SI-IEL in the samples tested, the expression level is more heterogenous in IEL found in the large intestine.
FIGURE 5 . Expression of adhesion markers by IEL. A, Histogram analysis depicting comparative expression of t-selectin. Presence of the lymph node homing receptor was identified using the mAb MEL-14 on LI-IEL (widely spaced dotted line, 46% positive) and SI-IEL (closeiy spaced dots, 8% positive). 6,Expression of aIEL P7 integrin, identified bythe rnAb M290, on LI- and SI-IEL (symbols as in A ) . LI-IEL and SI-IEL in A and 6 were prepared from a pool of C3Hmice. Incubations with MEL-14 mAb ( A ) and M290 rnAb (6)were followed by incubation with goat anti-rat IgG-PE, staining with this reagent alone is indicated by a solid line.
Expression of markers that may depend on thymus maturation
It has been proposed that the form of CD8 expressed by SI-IEL or the presence of Thy- 1.2 expression, is indicative
1772
LYMPHOCYTES INTRAEPITHELIAL
55.2
INTESTINE OF LARGE
on TCR ligation with CD3 mAb. However, the levels of incorporation are less than those produced by stimulated spleen cells (data not shown). In addition, LI-IEL express IL-2R a-chain after TCR stimulation. Freshly isolated LIIEL, or those kept in culture in the absence of CD3 mAb, do not express IL-2R a-chain (Fig. 8). In contrast to LI-IEL, ithasbeen reported that SI-IEL do not express IL-2R a-chain, even after stimulation with anti-CD3 mAb (33, 34).
CDBa
FIGURE 6. Expression of CD8 a- and P-chains by IEL. Dual color immunofluorescence of LI-IEL and SI-IEL prepared in parallel from a pool of BALB/c mice. IEL were reacted with CD8a-FITC and CD8P-PE mAb.
of the site of maturation. According to this view, cells that express the CD8 alp heterodimer differentiate in the thymus, whereas those that express the CD8 a/ahomodimer, mature extrathymically (32). It also has been reported that Thy-1.2 expression is correlated with thymus maturation (32). We found that preparations of LI-IEL, like SI-IEL, from euthymic mice contained cells bearing the CD8 d a homodimer. Figure 6 demonstrates that approximately 50%of CD8+ LI-IEL expressed the homodimeric form of CD8, whereas approximately 60% of SI-IEL did. Further analyses of IEL isolated from athymic ndnu mice showed a reduction in the numbers of IEL that were present in both the large and small intestine (see Materials and Methods). The phenotype of the mature T cells that remained in these mice, however, were predominantly TCR-yG+ and CD8+, with CD8 comprised of the d a homodimer. The reduction of mature T cells among IEL was moredramatic in the large intestine (Fig. 7A, top leftpanel) than in the small intestine (Fig. 7A, top right panel). We also found that LI-IEL and SI-IEL from euthymic mice expressed Thy- 1.2 at levels that were notsignificantly different when the two populations were analyzed in parallel (Table I). As reported for SI-IEL (8). two-color analysis demonstrated that there were CD3+ LI-IEL that were not Thy-1.2+ (data not shown). When athymic ndnu mice were examined, we found that the Thy-1.2+ cells were still present, although in reduced numbers in both LI-IEL and SI-IEL (data not shown), supporting the thymicindependent differentiation of at least some of the Thy- 1.2+ cells. In summary, these results suggest that some LI-IEL, like SI-IEL, expressing TCR-yG, Thy-1.2, andor CD8 may mature extrathymically. IL-2 synthesis and IL-2R expression by LI-IEL
Both LI-IEL and SI-IEL produced measurable IL-2 after TCR stimulation. The results of assays for IL-2 production, using the indicator cell line CTLL-20, are presented in Table 111. The data show that cytokhe synthesis is dependent
LI-IEL are not cytolytic in redirected lysis assay
Freshly isolated LI-IEL from BALBk mice did not have activity in a redirected lysis assay (Fig. 9). In contrast, in agreement with previous reports (8, 19) SI-IEL prepared from the same mice showed significant cytotoxic activity. Killing of CD3 mAb-coated targets by SI-IEL was obtained at E:T ratios as low as 1.25: 1. Flow cytometric analysis demonstrated that the SI-IEL in this sample contained 3.5 times more CD8+ IEL thandid the LI-IEL. The preparation from small intestine, therefore, may have had a higher concentration of cytotoxic cells. However, usingdata corrected for CD8+ cell number by dividing the LI-IEL effectors by 3.5, it is apparent from Figure 9 that freshly isolated LI-IEL did not have significant cytotoxic activity for the P388Dl targets even at the highest concentration tested, a corrected E:T ratio of more than 1 1: 1.
Discussion The resident lymphocytes of the smallintestinal epithelium of the mouse have been wellcharacterized ( 2 4 ) , although intraepithelial lymphocytes of the large intestine have not. Through comparative studies of LI-IEL and SI-IEL isolated from several different inbred strains of mice, we have demonstrated striking phenotypic andfunctional differences between these cells. Compared to SI-IEL, a greater fraction of LI-IEL have a phenotype similar to peripheral T cells, as characterized by their expression of a number of markers, including TCR-ap, CD2, and L-selectin. The results from several experiments indicate, however, that this increased representation of T lymphocytes with a phenotype typical of the periphery was not due to contamination by lymphocytes that reside outside of the epithelium. First, some of the phenotypic differences between SI-IEL and LI-IEL are so great, such as the expression of CD2, that if LI-IEL were not different from SI-IEL, the differences could only occur if nearly the entire preparation were comprised of cells from another site. This is probably not the case as flow cytometric analysis of LPL and LAL, adjacent compartments that could contaminate IEL preparations, demonstrated that these cells have a phenotype that is distinct from IEL. If we assume that all the surface Ig+ B cells in the LI-IEL preparation are contaminants, and make the further assumption
1773
Journal of Immunology
SI IEL
LI IEL A
1
I
I
FIGURE 7. Expression of CD8 a- and p-chains and TCR isotypes by athymic nu/nu mice. Dual color immunofluorescence of LI-IEL and SI-IEL prepared from a poolof four mice. IEL were reacted with CD8a-FITC and CD8p-PEmAb ( A ) , or with TCR-ap-FITC and TCR $-biotin mAb followed by streptavidin-PE ( B ) .
CDBa
CDBa
7I
" .
"
4.0 I I
I
=I
I -"____I . .:.;,.. 94.0,
"
""""_
2.0
-.,.
.
I
1
ITm
TCR a0
TCR ap Table 111 IL-2 production by /EL Treatment Cell Typed
CD3 cpm i SD
None cpmb
* SD
Expt. 1 LI-IEL SI-IEL
7995 f 5060 3409 It 1 1 04
228 f 147 67 f 4
Expt. 2 LI-IEL SI-IEL
3246 i 1801 3386 f 472
327 i 9 180 f 48
a Supernatants were collected from cultures of IEL preparedfrom C3H mice (Expt. 1 ) and BALB/c mice (Expt. 2) incubated in the presence or absence of plate-bound mAb CD3. Data represent mean of triplicate samples, expressed as the cpm of 3H TdR incorporated. Data from two of three representative experimentsis shown. SD = standard deviation.
that B cells from LPL or LAL are as likely to be contaminants as T cells in these sites; then for the LI-IEL preparation in TableI, no more than15% could have come from LPL or 3% from LAL. These levels of contamination would not greatly affect any of the results, except perhaps the assay for IL-2 secretion. This level of contamination would, however, be greatly overestimated if even 1 to 2% of LI-IEL were in fact B cells. Second, analyses of thin sections of the large intestine, before and after preparation of IEL,revealed preservation of the laminapropria cytoarchitecture. In addition, in situimmunofluorescentantibody
IL-2R FIGURE 8. IL2-RexpressionbyLI-IEL.LI-IEL were stained with mAb7D4 andwerethenreacted with goat anti-rat (Heavy and Light chain) IgG-PE. The broken line depicts the profile offreshly isolatedLI-IEL, which is similar to the profile of thenegative control, stained with goat anti-rat IgG-PE alone. The dotted line represents the results from LI-IEL cultured for 72 h without stimulation, and the solid line is LI-IEL stimulated with plate-bound CD3 mAb for 72 h.
staining demonstrated the presence of CD8+ and CD4+ lymphocytes within theepithelial layer. Finally, the number of cells within the separate compartments of the large intestine is nearly equivalent. Therefore, a significant crosscontamination would require that most of the cells in the lamina propria or lymphoid aggregates would have ended up in the IEL preparation. This is unlikely due to the fact that thelymphoid aggregates were removed and the lamina
1774
LYMPHOCYTES INTRAEPITHELIAL
-.
I M
"--t
U-IELJCll
"-e-.U-IEL-X1 --t
1
SI-IEL+Xll SI-IEL-2C11
Y
W
-20 J I 2 .15. :2l5 : l
I 5:l
1O:l 4 0 : 2 10 : l
E:T RATIO
FIGURE 9. Cytotoxic activity of IEL. Freshly isolated LI-IEL and SI-IEL from a pool of BALB/c mice were mixed with 5'Cr-labeled P388D1 target cells at the E:T ratios noted on the abscissa. Effectors are total cells in the preparation (not corrected for CD8+ number). The data are expressed as percent specific killing.
propria wasnot grossly affected by the preparation of IEL (Fig. 2). Despite some overlap, the data suggest that T lymphocytes in the two IEL populations may differ regarding function, specificity and homing. The experimental evidence is strongest with regard to a functional difference. The consequences of in vitro TCR cross-linking in the two populations are quite distinct; LI-IEL exhibited no cytotoxic potential in a redirected lysis assay, whereas SI-IEL proliferated to a lesser extent but rapidly killed target cells. The LI-IEL proliferative response was greater than that of SIIEL (data not shown) and was similar to the responses of splenic or lymph node T cells, which proliferate in response to TCR-mediated signals in vitro. The expression of CD2 on a large percentage of LI-IEL, as compared with its near absence on SI-IEL, may correlate with the functional difference seen between these cells. This difference in CD2 expression may be significant for the functional difference as signals derived through CD2 can activate T cells and augment TCR-mediated signals (35). Alternatively, LI-IEL may proliferate better than SI-IEL because of the greater number of CD4+ cells. Analysis of separated cell populations will be required to answer these questions. The in vivo specificities of IEL are poorly characterized. Potential differences in the specificity of LI-IEL and SI-IEL populations are implied, however, by the differences in coreceptor expression. In LI-IEL, CD4 single-positive cells were nearly as common as CD8 single-positive cells. The average CD4:CDS ratio for the LI-IEL preparations was 0.77: 1. The fraction of CD4+ cells was even higher among those cells that express an aP TCR. These data suggest that MHC class 11-restricted cells are more common in the large
OF LARGE INTESTINE
intestine. In contrast, for SI-IEL in the preparations analyzed, the ratio of CD4:CD8 lymphocytes was 0.13: l , suggesting thatthe great majority of these lymphocytes is MHC class I restricted. The finding that LI-IEL have a larger percentage of double-negative cells in the TCR-yC?+population, raises the possibility that fewer of these cells may recognize a CD8 binding class I, or class I like-molecule. It has been proposed that nonclassical class I molecules expressed by intestinal epithelial cells, including the TLa and CD1, may present Agto TCR-yC?+IEL (36,37). Interestingly, the pattern of expression of TLa, which binds CD8 (38), is consistent with the pattern of TCR yC?and coreceptor expression by IELin the large andsmall intestine. TLa is expressed by small intestinal epithelial cells and is therefore localized where the predominant population of TCR"y8 are CD8+. In contrast, CD1, which conceivably could have the capability to interact with double-negative cells, is present in both the small and largeintestine. In the latter site, more of the TCR-yC?+ IEL are double-negative. Results from recent experiments using parabiotic mice suggest that SI-IEL do not recirculate rapidly (39). Although there are no data on homing and recirculation of IEL in thelarge intestine, the expression of L-selectin, the lymph node homing receptor, by LI-IEL but not by SI-IEL, suggests that IEL in the two compartments may traffic differently in vivo. Although the FACScan staining pattern is somewhat different, the expression of the integrin containing aIEL, by both LI-IEL and SI-IEL, may imply that a common integrin is important in epithelial cell interactions shared by both compartments. Further studies are required, however, to determine the functional significance of differential expression of L-selectin or thepresence of a common integrin a chain on LI-IEL and SI-IEL. LI-IEL have characteristics similar to other peripheral T cells that can participate in bothafferent and efferent limbs of the immune response. Among these characteristics are the presence of CD4+ cells and the expression of L-selectin that may permit recirculation and homing of LI-IEL to other sites. However, SI-IEL have several characteristics of differentiated cytotoxic cells that may function primarily in the efferent limb of the immune response. These properties include, a brisk cytotoxic response in redirected lysis assays, the presence of cytotoxic granules in most SI-IEL (40), a relatively poor proliferative response to TCRmediated signals (33, 34) and the absence of at least one peripheral homing receptor, L-selectin. The phenotypic and functional differences between LIIEL and SI-IEL may reflect differences in the microenvironments in which these cells reside. The characteristics of the resident microbial flora, the composition of luminal Ag, and the characteristics of epithelial cells at these sites, may account for the phenotypic and functional differences that
Journal of Immunology
1775
pria of murine small intestine. Gut 22:481. 12. Taguchi, T., J. R. McGhee, R. L. Coffman, K. W. Beagley, J. H. Eldridge, K. Takatsu, and H. Kiyono. 1990. Analysis of Thl and Th2 cells in murine gut-associated tissues, frequencies of CD4+ and CD8+ cells that secrete IFN-y and IL-5. J. Immunol. 145:68. 13. Yagita, H., T. Nakamura, H., Karasuyama, and K. Okumura. 1989. Monoclonal antibodies specific for murine CD2 reveal its presence on B as well as Tcells. Proc. Natl. Acad.Sci. USA 86:645. 14. Leo, O., M. Foo, D. H.Sachs,L. E. Samelson, and J. A. Bluestone. 1987. Identification of a monoclonal antibody specific for a murine T3 polypeptide. Proc. Natl.Acad. Sci. USA 84:1374. 15. Koo, G., and J. Peppard. 1984. Establishment of monoclonal anti-NK 1.1 antibody. Hybridoma 3:301. Acknowledgments 16. Ledbetter, J., R. Rouse, S. Micklem, and L.Herzenberg. 1980. The authors thanktheir colleagues, Drs. Pirooz Eghtesady and Hilda HolT cell subsets defined by expression of Lyt- 1,2,3 and Thy- 1 combe for advice and discussions; Drs.Hilde Cheroutre, Robert Modlin, antigens. J. Exp. Med. 152:280. and Stephen Targan for reviewof the manuscript; Dr. David Camerini for advice on phase contrast and fluorescence microscopy; Hong Jun Peng for17. Ledbetter, J., and L. Herzenberg. 1979. Xenogeneic monoclonal antibodies to mouse lymphoid differentiation antigens. help with paraffin embeddedsections; Teresa McHugh for help with prepImmunol. Rev. 47:63. aration of the frozen sections; Katherine Williamsfor figure layouts and FACScan data acquisition; the UCLA Flow Cytometry Core Facility for 18. Kubo, R. T., W. Born, J. W. Kappler, P. Marrack, and M. FACScan data acquisition and analysis; and David Ng for help with prepPigeon. 1989. Characterization of a monoclonal antibody aration of the manuscript. which detects all murine ap T cell receptors. J. Immunol. 142:2736. 19. Goodman, T., and L. Lefrancois. 1989. Intraepithelial lymReferences phocytes. Anatomical site, not T cell receptor form, dictates 1. Fichtelius, K. E. 1968. The gut epithelium-a first level lymphenotype and function. J. Exp. Med. 170:1569. phoid organ. Exp. Cell Res. 49:87. 20. Bottomly, K., M. Luqman, L. Greenbaum, S. Darding, J. 2. Emst, P. B., A. D. Befus, and J. Bienenstock. 1985. LeukoWest,T. Pasqualini, and D. Murphy. 1989. A monoclonal cytes in the intestinal epithelium: an unusual immunological antibody to murine CD45R distinguishes CD4 populations compartment. Immunol. Today 2:50. that produce different cytokines. Eul: J. Immunol. 19:617. 3. Lefrancois, L. 1991. Intraepithelial lymphocytes of the in21. Coffman, R. L., and I. L. Weissman. 1981. A membrane antestinal mucosa: curiouser and curiouser. Semin. Immunol. tibody that recognizes B cells and B cell precursors in mice. 3: 99. J. Exp. Med. 133.1026. 4. Trejdosiewicz, L. K. Intestinal intraepithelial lymphocytes 22. Dialynas, D. P., Z. S. Quan, K. A. Wal1,A.Pierres, J. Quintans, and lymphoepithelial interactions in the human gastrointesM. R. Loken, M. Pierres, and F.W. Fitch. 1983. Charactertinal mucosae. 1992. Immunol. Lett. 32:13. ization of the murine T cell surface molecule, designated 5. Parrot, D. M. V., C. Tart, S. MacKenzie, A. Mowat, M. D. J. L3T4, identified by monoclonal antibody GK1.5: Similarity Davies, and H. S. Micklem. 1983. Analysis of the effector of L3T4 to the human Leu31T4 molecule. J. Immunol. 131: functions of different populations of mucosal lymphocytes. 2445. Ann. N.Z Acad. Sci. 409:307. 23. Ortega, G., R. Robb, E. Shevach, and T. Malek. 1984. The 6. Goodman, T., and L. Lefrancois. 1988. Expression of the y6 murine IL-2 receptor. J. Immunol. 133:1970. T cell receptor on intestinal CD8+ intraepithelial lympho24. Gallatin, W. M., I. L. Weissman, and E. C. Butcher. 1983. A cytes. Nature 333355. cell surface molecule involved in organ specific homing of 7. Bonneville, M., C. A. Janeway, Jr., K. Ito, W. Haas, I. Ishida, lymphocytes. Nature 303:30. N. Nakanishi, and S . Tonegawa. 1988. Intestinal intraepithe25. Kilshaw, P. J., and K. C. Baker. 1988. A unique surface anlial lymphocytes are a distinct set of y6 T cells. Nature 336: tigen on intraepithelial lymphocytes in the mouse. Immunol. 4 79. Lett. 18:149. 8. Lefrancois, L., and T. Goodman. 1989. In vivo modulation of 26. Farr, A., J. Nelson, J. Truex, and S . Hoiser. 1991. Epithelial cytolytic activity and Thy-1 expression in TCR-y6+ intraheterogeneity in murine thymus: a cell surface glycoprotein epithelial lymphocytes. Science 243:1716. 9. Lefrancois, L. 1991. Extrathymic differentiation of intraexpressed by subcapsular and medullary epithelium. J. Hisepithelial lymphocytes: generation of a separate and unequal tochem. Cytochem. 39:645. repertoire. Immunol. Today 12:436. 27. Gajewski, T. E , and F. W. Fitch. 1987. Antiproliferative effect 10. Itohara, S., A. G. Farr, J. J. Lafaille, M. Bonneville, Y. of IFN-y in immune regulation. I. IFN-y inhibits the prolifTakagaki, W. Haas, and S . Tonegawa. 1990. Homing of a y6 eration of Th2 but not Thl murine helper T lymphocyte thymocyte subset with homogeneous T-cell receptors to muclones. J. Immunol. 140:4245. cosal epithelia. Nature 343:754. 28. Koren, H.S . , B. S . Handwerger, and J. R. Wunderlich. 1975. 11. Davies, M. D. J., and D. M. V. Parrott. 1981. Preparation and Identification of macrophage-like characteristics in a cultured purification of lymphocytes from epithelium and lamina promurine tumor cell line. J. Immunol. 114:894.
we observed. Thus, LI-IEL and SI-IEL mayhave sitespecific functions in both mucosal immune protection and tolerance induction. In the large intestine, it is possible that primary sensitization may occur in the epithelial compartment. In contrast, in the small intestine the immune system may be designed to prevent such sensitization and perhaps avoid deleterious responses to food or autologous Ag. If similar differences are found in humans, the differences between IEL in the two sites could account for the sitespecific localization and pathophysiology of human inflammatory bowel diseases, such as ulcerative colitis and Crohn’s disease.
1776 29. Perry, G. A., and J. G. Sharp. 1988. Characterization of proximal colonic lymphoid tissue in the mouse. Anat. Rec. 220: 305. 30. Goodman, T., R. LeCorre, and L. Lefrancois. 1992. A T-cell receptor y6-specific monoclonal antibody detects a Vy5 region polymorphism. Immunogenetics 35:65. 31. Kilshaw, P. J., and S. J. Murant. 1991. Expression and regulation of p7 (bp) integrin on lymphocytes: relevance to the mucosal immune system. ELK J. Immunol. 21:2591. 32. Rocha, B., P. Vassalli, and D. Guy-Grand. 1991. The V p repertoire of mouse gut homodimeric a CD8+ intraepithelial T cell receptor apt lymphocytes reveals a major extrathymic pathway of T cell differentiation. J. Exp. Med. 173:483. 33. Mosley, R. L., M. Whetsell, and J. R. Klein. 1991. Proliferative properties of murine intestinal intraepithelial lymphocytes (IEL): IEL expressing TCR cup or TCR y6 are largely unresponsive to proliferative signals mediated via conventional stimulation of the CD3-TCR complex. lnt. Immunol. 3:563. 34. Sydora, B. C., P. E Mixter, H. Holcombe, P. Eghtesady, K. Williams, M. C. Amaral, A. Nel, and M. Kronenberg. 1993. Intestinal intraepithelial lymphocytes are activated and cytolytic but do not proliferate as well as other T cells in re-
INTRAEPITHELIAL LYMPHOCYTES OF LARGE INTESTINE sponse to mitogenic signals. J. Immunol. 150:2179. 35. Bierer, B. E., B. P. Sleckman. S. E. Ratnofsky, and S. J. Burakoff. 1989. The biologic roles of CD2, CD4,and CD8 in T-cell activation. Annu. Rev. Immunol. 7:579. 36. Hershberg, R., P. Eghtesady, B. C. Sydora, K. Brorson, H. Cheroutre, R. Modlin, and M. Kronenberg. 1990. Expression of the thymus leukemia antigen in mouse intestinal epithelium. Proc. Natl. Acad. Sci. USA 87:9727. 37. Bleicher. P. A., S. P. Balk, S. J. Hagen, R. S. Blumberg, T. J. Flotte, and C. Terhorst. 1990. Expression of murine CD 1 on gastrointestinal epithelium. Science 250:679. 38. Teitell, M., M. Mescher, C. Olson, D. Littman, and M. Kronenberg. 1991. The thymus leukemia antigen binds mouse and human CD8. J. Exp. Med. 174:1131. 39. Poussier, P.,P. Edouard, C. Lee, M. Binnie, and M. Julius. 1992. Thymus-independent development and negative selection of T Cells expressing T cell receptor ap in the intestinal epithelium: evidence for distinct circulation patterns of gutand thymus-derived T lymphocytes. J. Exp. Med. 176:187. 40. Guy-Grand, D. M., M. Malassis-Seris, C. Briottet, and P. Vassalli. 199I . Cytotoxic differentiation of mouse gut thymodependent and independent intraepithelial T lymphocytes is induced locally. J. Exp. Med. 173:1549.
