Mucosal T Lymphocyte Numbers Are Selectively Reduced in Integrin α_E (CD103)-Deficient Mice¹ Free

Mucosal T lymphocytes appear to be functionally distinct from those in peripheral blood. Indeed, intestinal intraepithelial lymphocytes (IEL)⁴ have been found to differ from PBL in their Ag-recognition specificity (1, 2, 3), accessory costimulatory molecule expression (4), TCR αβ and γδ type, and thymus-dependent vs -independent T cell development (reviewed in 5). There are estimated to be as many T lymphocytes in the intestinal immune system as in the spleen (6). Furthermore, many infectious agents invade via mucosal epithelia, emphasizing the importance of mucosal T lymphocytes for immune surveillance and immune responses to mucosal pathogens under normal conditions. In addition, intestinal T lymphocytes have been implicated in the pathogenesis of inflammatory bowel diseases, based upon the development of intestinal inflammation in animals with defects in T lymphocyte regulation (7). Thus, it will be important to understand the mechanisms whereby mucosal lymphocytes selectively localize and function.

Following primary stimulation in organized lymphoid tissues, such as mesenteric lymph nodes and Peyer’s patches, some activated intestine-derived lymphocytes recirculate and then preferentially return to the intestinal tract. The selective expression of chemokine receptors and adhesion molecules are thought to contribute to T cell homing (8). Once in the intestine, lymphocyte subpopulations localize to particular microenvironments. For example, the CD8⁺ T cells are found preferentially within the epithelium, where they comprise 90% of the resident population, whereas CD4⁺ T cells predominate in the lamina propria, where they constitute more than half of the T lymphocytes (9). Targeted migration along chemokine gradients and selective adhesion to extracellular matrix or to cellular counterreceptors may account for the localization and retention of T cell subsets within mucosal microenvironments.

One candidate to mediate the selective localization or retention of intraepithelial T lymphocytes is the integrin α_E(CD103)β₇. This integrin is expressed selectively on >90% of intestinal IEL and on 45–50% of lamina propria T lymphocytes (9, 10, 11) in both mice and humans. It is also found on T lymphocytes in some other mucosal epithelia, such as the genitourinary epithelium (12), on ∼40% of bronchioalveolar lavage T cells obtained from normal humans (13) and on some cells of dendritic morphology in rats (14). Furthermore, α_Eβ₇ expression can be induced on T lymphocytes and murine mast cells by culture in the presence of TGF-β1 (15), a cytokine produced by intestinal epithelial cells (16) as well as other cell types. In contrast, α_Eβ₇ is expressed on <5% of PBL in humans (10), on only 15% of splenic T lymphocytes in mice (17), and has not been found on B lymphocytes, underscoring its selective expression on mucosal T cells.

The α_Eβ₇ integrin mediates T cell adhesion to epithelial cells (18) through its binding to E-cadherin (19, 20, 21), a member of the cadherin family of adhesion molecules that is expressed selectively on epithelial cells. Cadherins are characterized by their tissue-specific distribution and are known to mediate homophilic adhesion of cells within tissues (22). In addition, evidence has suggested that α_Eβ₇-dependent adhesion is regulated by inside-out signals, based upon the observation that α_Eβ₇ function is enhanced following stimulation through the TCR (21). Integrin α_Eβ₇ also appears to transmit an intracellular signal, as anti-human α_E mAbs enhance T cell proliferation in response to suboptimal anti-CD3 stimulation (9, 23), and as an anti-murine α_E mAb induces T cell-mediated lysis of FcR-bearing target cells in the absence of a signal from the TCR (17). Thus, the α_Eβ₇ integrin may be important in the localization or function of T cells, dendritic cells, and/or mast cells.

To determine the role of α_Eβ₇ in in vivo immune responses, the murine integrin α_E-encoding gene (Itgae) was cloned and localized to chromosome 11. Then, integrin α_E-deficient mice were generated. Study of these animals revealed an altered distribution of T lymphocytes within epithelia and in the intestinal lamina propria, in the absence of experimentally induced infection or inflammation. Thus, these studies demonstrate an important role for α_Eβ₇ in modulating the homeostasis of T lymphocyte numbers in selected tissue sites.

A murine 129/Sv-derived λ-Fix II library (Stratagene, La Jolla, CA) was screened by hybridization with an α_E cDNA probe incorporating nucleotides 1088–1253 of the murine α_E cDNA sequence (15). Inserts were subcloned and partial sequence determined by deletion cloning, followed by dideoxy chain termination analysis. The nucleotide sequence within identified exons was identical to nucleotides 14–2771 of the murine α_E cDNA reported previously with the exception of only 4/2757 bases (nucleotide changes t411c, g668a, c1057g, and g1058c; amino acid changes I207E and R337A).

Primers were designed to amplify a region corresponding to intronic sequence of Itgae to test for single-strand conformation polymorphisms (SSCPs) between mouse strains. These were analyzed as previously described (24). Briefly, oligonucleotides were radiolabeled with [³²P]ATP using polynucleotide kinase, and genomic DNAs from a series of mouse strains were amplified using standard protocols (anneal at 55°C for 1 min, extend at 72°C for 2 min, and denature at 94°C for 1 min for 40 cycles, with a final extension at 72°C). A total of 2 μl of the amplified reaction was added to 8.5 ml USB (United States Biochemical) stop solution, denatured at 94°C for 5 min, and immediately placed onto ice. A total of 2 μl of each reaction is loaded on a 6% nondenaturing acrylamide sequencing gel and electrophoresed in 0.5× TBE buffer for 2–3 h at 40 W in a 4°-cold room. A primer pair based upon sequence in the intron between exons 2 and 3 (5-AAGGTCAGATGAGCAATATGT-3′ (forward) and 5′-GCCAGCAGACTCAGCATTACT-3′ (reverse)) identified a polymorphism between C57BL/6J and Mus spretus. This primer pair was used to analyze DNA prepared from the BSS ((C57BL/6JEi × SPRET/Ei)F₁ × SPRET/Ei) backcross (25). The strain distribution pattern was analyzed using the Map Manager Program (26).

A targeting construct was generated (Fig. 1,a) and transfected by electroporation into the 129/Sv-derived embryonic stem cell line ES-D3. Transfected cells were selected for G418 and gancyclovir resistance, as reported previously (27), and cloned. Clones that incorporated a single copy of the construct into the genome by homologous recombination were identified by Southern blot analysis of genomic DNA using a 5′ probe (Fig. 1 a), a 3′ probe, and the neomycin resistance gene. Such clones were injected into BALB/c blastocysts to generate chimeric animals. The α_E^+/− progeny of matings between chimeric animals and BALB/c mice were intercrossed to generate F₂ (129/Sv × BALB/cJ) offspring, which were utilized in these experiments except where otherwise indicated. Two independent clones were used to generate chimeric animals that expressed similar phenotypes. All animals were housed under specific pathogen-free (SPF) conditions.

Genomic DNA was isolated from tail biopsies by proteinase K digestion, followed by phenol extraction. Southern blot analysis was performed using 10 μg genomic DNA, Hybond-N (Amersham, Arlington Heights, IL) filters and digoxigenin-labeled probes (Nonradioactive DNA labeling and detection kit; Boehringer Mannheim, Indianapolis, IN) according to standard procedures.

PCR was performed on genomic DNA using primer sequences flanking the inserted neomycin resistance gene to yield products of 1100 bp from disrupted and 900 bp from wild-type alleles (primers used: 5′-GCA ACA ACG CAT CGT TCA TAT GGA-3′ and 5′-GTG CTC TGT CTA TTG TTC CCC TCC TT-3′; conditions: anneal at 62°C for 1 min, extend at 72°C for 2 min, and denature at 94°C for 1 min for 40 cycles, with a final extension at 72°C).

A single cell suspension of splenic leukocytes was stimulated with 5 μg/ml Con A (Sigma, St. Louis, MO) and cultured for 5 days. A total of 5 ng/ml human recombinant TGF-β1 was added to the medium, and the cells were grown for an additional 5 days to induce α_E expression.

Intestinal IEL were isolated as described (28) with minor modifications. Briefly, small intestines were flushed with HBSS, trimmed to remove Peyer’s patches, opened longitudinally, and cut into 3- to 5-mm fragments. The fragments were incubated in a shaking water bath at a rate of 120 shakes per minute in medium (RPMI 1640 containing 10⁴ U/ml penicillin/streptomycin, 20 μg/ml gentamicin, 2% bovine calf serum, and 20 mM HEPES buffer) at 37°C for 20–30 min. The intestinal pieces were shaken manually for three cycles of 15 s each to release intestinal IEL. The resulting cell suspensions were collected, pooled, passed through a glass wool column, and the IEL separated from epithelial cells using 44/66% Percoll density gradient centrifugation. To estimate the size of Peyer’s patches, their length and width were multiplied to estimate their relative cross-sectional area in mm². To isolate Peyer’s patch lymphocytes, the Peyer’s patches were mechanically disrupted by rubbing the patches between two frosted glass slides.

For FACS analysis, 10⁵ cells were stained with saturating concentrations of mAb (buffer: 2% BSA, 0.05% NaN₃ in PBS; blocking reagents: 10% mouse serum (for rat Abs) or 10% goat serum and 20 μg/ml mAb 2.4G2 (for hamster Abs)). In some experiments, dead cells were excluded by propidium iodide staining. Samples were analyzed using a FACScan and FACScalibur for four-color analysis and a FACSort and the Cell Quest software (Becton Dickinson, San Jose, CA) for one color analysis.

A total of 4 × 10⁷ TGF-β1-stimulated splenocytes was surface iodinated and then solubilized in TBS/0.5% Triton X-100 (Sigma). Then, 1 × 10⁷ cell equivalents were precleared with protein G-Sepharose resin (Pharmacia, Piscataway, NJ), immunoprecipitated with 5 μg of purified mAb followed by protein G-Sepharose, washed, and the immunoprecipitated proteins were analyzed by SDS-PAGE using 7% gels under reducing conditions (29).

The following mAbs were used as controls: rat IgG1 (R59-40; PharMingen, San Diego, CA), rat IgG2A (R35-95; PharMingen), rat IgG2B (SFR3-DR5; American Type Culture Collection (ATCC), Manassas, VA), and hamster IgG (UC8-4B3; PharMingen). The following Ags were detected by mAbs: CD3ε (500A2; PharMingen, and 145-2C11; ATCC), CD4 (RM4-5; PharMingen), CD8α (53-6.72; ATCC), CD8β (53-5.8; ATCC), CD11b (α_M integrin, Mac-1, M1/70; ATCC), CD18 (β₂ integrin, 2E6; ATCC), CD25 (IL-2R α-chain, 3C7; PharMingen), CD29 (β₁ integrin, Ha2/5; PharMingen), CD45R/B220 (RA3-6B2; PharMingen), CD45RB (MB23G2; ATCC, and 16A; PharMingen), CD49a (α₁ integrin, Ha31/8; PharMingen), CD49b (α₂ integrin, HMα2; PharMingen), CD49d (α₄ integrin, P/S2, or R1-2; ATCC), CD49e (α₅ integrin, HMα5; PharMingen), CD49f (α₆ integrin, GoH3; Dianova, Hamburg, Germany), CD49d/β₇ (α₄β₇, DATK32 (30); LeukoSite, Cambridge, MA), CD51 (α_v integrin, H9.2B8; PharMingen), CD54 (ICAM-1, YN1/1.7.4; ATCC), CD62L (L selectin, MEL-14; ATCC), CD90 (Thy-1, AT15.E; R. MacDonald Ludwig Institute for Cancer Research, Epalinges, Switzerland), CD103 (α_E integrin, M290 (11); P. Kilshaw, Department of Immunology, AFRC Babraham Institute, Cambridge, U.K. or 2E7 (17); L. Lefrancois, Department of Medicine, University of Connecticut Health Center, Farmington, CT), β₇ (M293; P. Kilshaw), CD106 (VCAM-1, M/K-2.7; ATCC), CD32/CD16 (Fc-γII/IIIR, 2.4G2; ATCC), MHC class I (M1/42.3, rat IgG2A; ATCC), anti-TCR αβ (H57-597; PharMingen), and anti-TCR γδ (GL3; PharMingen). FITC-conjugated goat anti-hamster and mouse anti-rat secondary Abs used in FACS or direct immunofluorescent staining of intestinal sections were purchased from Jackson ImmunoResearch (West Grove, PA), and biotinylated goat-anti-hamster and mouse adsorbed rabbit-anti-rat serum were obtained from Vector Laboratories (Burlingame, CA).

For histochemistry, tissue samples were embedded in JB-4 plastic resin (Polysciences, Warrington, PA) and 3-μm sections stained with hematoxylin-eosin. For immunohistochemistry, tissue samples were frozen in OCT and 5- to 10-μm cryostat-cut sections stained using the ABC (avidin/biotin complex)-immunoperoxidase kit according to the manufacturer’s instructions (Vector Laboratories). For analysis of lamina propria T cell numbers, a cell was counted as a lamina propria lymphocyte if it did not overlap the basement membrane and was contained within villi rather than crypts. This criteria may account for the unusually high ratio of CD4⁺/CD8⁺ T cells observed within the lamina propria in these studies, as some cells were counted as IEL that were largely within the lamina propria. For analysis of T cell numbers within vaginal tissue sections, the number of intraepithelial T lymphocytes within an entire tissue section derived from the middle third of the vagina was determined and expressed per mm basement membrane. For evaluation of T lymphocyte numbers within lung tissue, the left lung was frozen in OCT, the average number of anti-CD3 mAb stained cells in at least seven randomly selected high power (40×) fields was determined and used to calculate the number of T lymphocytes/mm². In addition, the average number of CD3⁺ cells per bronchus was determined evaluating all of the bronchi within these randomly located tissue sections.

Statistical analysis was performed using an unpaired, two-tailed Student’s t test when n = 3 and the Mann-Whitney nonparametric U test when n > 3, unless otherwise indicated. Analysis was performed using the Instat software (GraphPad Software, San Diego, CA).

Genomic clones encoding murine α_E were identified by hybridization with PCR products encoding fragments of the murine α_E cDNA (GenBank accession nos. AF133070-AF133085). Based upon comparison with the murine α_E cDNA sequence (15) and identification of the conserved consensus splice sites (Ref. 31, and data not shown), a partial map of the murine α_E intron/exon structure was generated (Fig. 1 a). To determine the chromosomal location of Itgae, SSCP analysis was used as previously described (24). Primers corresponding to intronic sequence between exons 2 and 3 of Itgae were analyzed and found to identify an SSCP between inbred mouse strains (see Materials and Methods). The BSS-interspecific backcross was genotyped and the allele distribution pattern analyzed using the Map Manager program. Itgae was found to map to chromosome 11 with a logarithm of odds likelihood score of 27.4. No recombinants were found in 91 progeny between Itgae and the marker D11Abb1. The position of Itgae with respect to flanking microsatellite markers was: D11 Mit4 − 5.5 ± 2.4 cm − Itgae, D11Abb1 − 1.1 ± 1.1 cM − D11 Mit7, D11 Mit32, D11 Mit34 (mapping data submitted to the Mouse Genome Database).

To generate integrin α_E-deficient mice, exon 10 within the integrin α_E-encoding gene was replaced with a neomycin resistance gene by homologous recombination (Fig. 1, a–c). FACS analysis confirmed that intestinal IEL isolated from α_E^−/− mice lacked α_E expression (Fig. 1,d). In addition, splenocytes from α_E^−/− mice were cultured in the presence of TGF-β1, a cytokine that up-regulates α_E expression on wild-type T cells. Integrin α_E was not detected on the surface of TGF-β1-treated splenocytes derived from α_E^−/− mice by immunoprecipitation or FACS analysis. In contrast, it was readily detected on the surface of cells derived from α_E^+/+ animals, and was detected at reduced levels on cells from α_E^+/− animals (Fig. 1,e, and data not shown). Thus, α_E^−/− mice lacked expression of the α_E polypeptide, while heterozygous animals expressed intermediate levels of α_Eβ₇ both in immunoprecipitation (Fig. 1 e) and in FACS analysis (data not shown). Integrin α_E deficiency did not alter fecundity, morphogenesis, or overall weight gain, and most animals survived for >18 mo in a SPF facility (data not shown).

Because α_E is selectively expressed on leukocytes, studies were performed to determine the impact of α_E deficiency upon the histologic appearance of organized lymphoid tissues. In this initial evaluation, α_E deficiency had no apparent effect on the size of the thymus or on the immunohistologic appearance of the thymus, peripheral lymph nodes, or spleen after staining with anti-CD3, anti-CD4, and anti-CD8 mAbs. In addition, there were no changes in the serum levels of IgM, IgG isoforms, or IgA in α_E^−/− as compared with α_E^+/+ mice (data not shown).

To further evaluate the impact of α_Eβ₇ expression upon the number and subset composition of splenic T lymphocytes, additional studies were performed. First, the overall number of leukocytes/spleen after RBC lysis was similar in α_E^−/− and α_E^+/+ mice three to four generations backcrossed toward the C57BL/6 strain and then intercrossed (N_3–4) (α_E^−/− mice: 1.07 ± 0.15 × 10⁸ cells/spleen; α_E^+/+ mice: 1.14 ± 0.16 × 10⁸ cells/spleen; n = 3; p = 0.55). In addition, flow cytometry was used to determine the total number of splenocytes expressing CD3, CD4, or CD8 using the formula: (the total number of splenocytes) × (the proportion of the total splenocyte population that expressed each marker). In this analysis, the number of CD3⁺, CD4⁺, and CD8⁺ splenocytes was not altered significantly by α_E deficiency. In contrast, when mice N₁₀ to the BALB/c strain were evaluated, the total number of splenic leukocytes was increased by 24% in α_E-deficient mice (α_E^−/− mice: 1.07 × 10⁸ ± 0.11 cells/spleen; α_E^+/+ mice: 0.86 × 10⁸ ± 0.04 cells/spleen; n = 3; p = 0.04). In addition, the numbers of CD3⁺, CD4⁺, and CD8⁺ splenocytes were increased by 14%, 28%, and 13%, respectively in α_E^−/− mice (Fig. 2), while CD4⁺CD8⁺ double positive splenic T cells were not detected in either α_E^+/+ or α_E^−/− mice (n = 1, pooling cells isolated from three mice of each genotype, data not shown). Overall, peripheral lymphocyte compartments appeared similar in α_E^−/− and α_E^+/+ mice, with a trend toward increased numbers of CD4⁺ and CD8⁺ splenic T cell numbers in α_E^−/− animals on the BALB/c genetic background.

The number of intestinal IEL in adult α_E^−/− and α_E^+/+ littermates also was compared by immunohistology, with the expectation that intestinal IEL numbers would be reduced by α_E deficiency due to the loss of the α_Eβ₇/E-cadherin-mediated adhesion. Indeed, the number of jejunal IEL was reduced by 54% in the α_E^−/− progeny of F₂ (BALB/c × 129/Sv) mice as compared with α_E^+/+ mice of similar genetic background (p < 0.002, Mann-Whitney U test; n = 5 α_E^+/+ and 8 α_E^−/− mice) (Figs. 3 and 4a). In these studies, it was apparent that the number of intestinal IEL was influenced by genetic background and/or environment, as there were more intestinal IEL in α_E^+/+ mice on the (129/Sv × BALB/c) or the C57BL/c backgrounds than on the BALB/c background. However, α_E^−/− mice had reduced intestinal IEL numbers in all genetic backgrounds examined, including groups of mice housed in two independent animal facilities. Thus, reduced intestinal IEL numbers represents a consistent feature of α_E-deficient mice.

Because integrin α_Eβ₇ is also expressed on T cells in or adjacent to other mucosal epithelia, such as the lung and genitourinary tract, the number of T lymphocytes in these other sites was compared in α_E^−/− and α_E^+/+ mice. While vaginal IEL numbers were reduced by 60% (p < 0.01; n = 3) (Fig. 4,b) in α_E^−/− mice of mixed (129/Sv × BALB/c) genetic background, the number of T cells within the lung parenchyma and adjacent to the bronchi was not affected by α_E deficiency (Fig. 4, c and d). Thus, under conditions where there was not an experimentally induced disease process, α_E^−/− mice had reduced numbers of T cells within some epithelia, including the intestine and vagina, but not in others, such as in the peribronchial regions in the lung.

It appeared that the number of lamina propria T lymphocytes was reduced in immunohistochemical analysis of tissue sections from (129/Sv × BALB/c) mice. This was not an expected finding because E-cadherin, the known α_Eβ₇ counterreceptor, is not expressed in the lamina propria (32). To quantitate this difference, the number of T lymphocytes within the villus lamina propria that did not appear to be in contact with either the basement membrane or epithelium was determined per 0.5-mm villus length. In this analysis, the number of lamina propria lymphocytes was significantly diminished by α_E deficiency in animals of the (129/Sv × BALB/c) background (CD3⁺ lamina propria lymphocytes/0.5-mm villus length in α_E^+/+ mice: 76 ± 6 vs 39 ± 9 in α_E^−/− mice; p < 0.01; n = 3). In a second group of mice partially backcrossed toward the BALB/c strain, integrin α_E deficiency had a similar impact upon lamina propria T lymphocyte numbers (in α_E^+/+ mice, 70 ± 7 T cells/0.5-mm villus length vs 30 ± 6 in α_E^−/− mice; p = 0.002; n = 3). Thus, α_E^−/− mice had significantly fewer lamina propria T lymphocyte numbers than α_E^+/+ mice in these two groups. However, when additional groups of mice were evaluated, α_E deficiency did not appear to have an impact upon lamina propria T lymphocyte numbers (Fig. 5,a). In these two additional groups, it was notable that the α_E^+/+ mice had fewer lamina propria T lymphocytes than those in the groups initially studied (Fig. 5 a). This could be due to differences in genetic background, as these later groups of mice were backcrossed toward either the C57BL/6 or BALB/c strains. However, it seems more likely that environmental factors may have had an effect. Of note with respect to this possibility, the two more fully backcrossed groups of mice were housed in an animal facility where the cages were changed twice/wk, and the microisolator cages were vented with an air circulation system. These animals had relatively few lamina propria T lymphocytes and α_E deficiency had no effect on lamina propria T cell numbers. In contrast, the other groups of mice were housed in a facility where the cages were changed once/wk, and the microisolator cages were not vented. In the setting of this less-stringent environment, there were more lamina propria T lymphocytes in α_E^+/+ mice, and α_E deficiency had a significant impact upon lamina propria T lymphocyte numbers.

In contrast to the observed decrease in intestinal IEL and lamina propria T lymphocyte numbers in α_E^−/− mice, the size of Peyer’s patches was increased in the first cohort of mice examined. However, Peyer’s patch size was not altered in three additional groups of mice of different genetic backgrounds, and the proportion of Peyer’s patch cells expressing CD3, CD4, or CD8 was not altered (Fig. 5, b and c). Finally, the organization of Peyer’s patches appeared normal in α_E^−/− as compared with α_E^+/+ mice, based upon immunohistochemistry after staining with anti-CD3 mAbs. These findings demonstrate that Peyer’s patch size and T lymphocyte composition are not, on average, altered by α_E deficiency. However, it is possible that α_E deficiency results in enlarged Peyer’s patch size when present in a given genetic background or under some environmental conditions.

The γδ T cells populate the intestinal epithelium at an earlier age than αβ T cells (33, 34) and the recruitment/expansion of TCR αβ⁺ intestinal IEL depends upon microbial colonization of the intestinal tract (33). Thus, studies were performed to determine at what age reduced numbers of intestinal IEL were detected in α_E-deficient animals. In immunohistochemical analyses, intestinal IEL were not observed until 5 days of age in either α_E^+/+ or α_E^−/− mice. By 3 wk of age, the intestinal IEL number increased to 40–50 intestinal IEL/1000 epithelial cells in α_E^+/+ mice, with only a 20% reduced number of intestinal IEL in α_E^−/− mice. Finally, the intestinal IEL number reached a plateau in animals of both genotypes at 4–5 wk of age. In mice older than 5 wk of age, the number of proximal jejunal IEL in α_E^−/− mice was only 46% of that seen in α_E^+/+ mice (Fig. 6). Overall, this kinetic analysis suggested that α_E might be more important in the localization of αβ than of γδ T cells and that the impact of α_E deficiency upon intestinal IEL numbers becomes apparent at a time when TCR αβ cells are increasing in number, presumably due to bacterial colonization.

If integrin α_E is more important in the localization of TCR αβ than of γδ T cells, then one would predict that adult α_E-deficient mice would have an intestinal IEL TCR repertoire skewed toward a lower proportion of TCR αβ⁺ cells. To test this possibility, the proportion of IEL expressing each TCR was evaluated by flow cytometry in populations isolated from α_E^+/+, α_E^+/−, or α_E^−/− mice. Mice N_3–4 to the C57BL/6 strain were used for this study because α_E deficiency had a greater impact upon IEL number in the C57BL/6 than in the BALB/c genetic background and because the (129/Sv × BALB/c) mice were no longer available. In these studies, TCR gene segment usage was similar in α_E^+/− and α_E^+/+ mice (data not shown), despite the reduced levels of α_E expressed on α_E^+/− T lymphocytes (Fig. 1,e). In contrast, while the proportion of TCR αβ⁺ intestinal IEL was 54% in α_E^+/− or α_E^+/+ mice, it was only 35% in α_E^−/− mice (p < 0.01; n = 4 α_E^−/− and 6 α_E^+/+ or α_E^+/−) (Fig. 7 a). The decreased proportion of TCR αβ⁺ IEL α_E^−/− mice was accompanied by a corresponding increase in the relative proportion of IEL that expressed the γδ TCR. Thus, in α_E^−/− mice, the number of intestinal IEL that expressed the TCR-αβ was reduced more than that of γδ TCR-expressing cells.

Within the intestinal epithelium, α_Eβ₇ is expressed by >90% of the CD8⁺ T cells but by only half of the CD4⁺ T cells (9, 11). Thus, it seemed likely that α_Eβ₇ disruption might affect the number of CD8⁺ IEL more than the number of CD4⁺ IEL. Thus, three-color FACS analysis was performed to evaluate the subsets of TCR αβ⁺ intestinal IEL in α_E^−/− as compared with α_E^+/− or α_E^+/+ mice backcrossed toward the C57BL/6 strain, in which the TCR γδ/TCR αβ cell ratio was altered (Fig. 7,a). In this analysis, the proportions of intestinal TCR αβ⁺ IEL that expressed CD8αα, CD8αβ, or both CD4 and CD8 were not consistently altered in α_E^−/− as compared with α_E^+/− or α_E^+/+ mice (Fig. 7, b and c). However, the proportion of TCR αβ⁺ IEL that were CD4⁺CD8⁻ was greater in α_E^−/− than in α_E^+/+ or α_E^+/− mice (7.3 ± 0.9% of the IEL in α_E^+/+ or α_E^+/− mice vs 12.9 ± 1.0% of the IEL in α_E^−/− mice; n = 4; p = 0.028). As the total number of IEL was reduced to ∼50% of normal, an apparent doubling of the CD4⁺CD8⁻ subset suggests that CD4⁺CD8⁻ T cell numbers were not increased but rather remained unchanged, while the other subsets were reduced in number in α_E-deficient mice. In contrast, in the BALB/c strain, where α_E deficiency had relatively little impact upon IEL numbers (Fig. 4,a), the proportion of IEL that expressed CD4 was not altered in α_E-deficient mice (Fig. 8). Finally, within the lamina propria of mice backcrossed toward the C57BL/6 strain, the CD4⁺ and CD8⁺ populations both were reduced in number with a slightly greater impact upon the number of CD4⁺ cells (Fig. 5 a). Overall, α_E deficiency resulted in diminished numbers of CD8⁺ IEL but not of CD4⁺ IEL in backgrounds where α_E deficiency had a dramatic impact upon IEL numbers, but resulted in reduced numbers of both CD4⁺ and CD8⁺ lamina propria subpopulations.

Given the expression of α_Eβ₇ on >80% of intestinal IEL, it was surprising that the intestinal IEL numbers were reduced by only 30% in mice partially backcrossed to the BALB/c strain. Thus, additional FACS analyses were performed to determine whether α_E^−/− IEL derived from mice N_3–4 to BALB/c express higher levels of an adhesion molecule that might functionally compensate for α_E deficiency in IEL localization, and thus reduce the impact of α_E deficiency on intestinal IEL number. Importantly, the expression of α₄β₇, the other β₇ integrin, was not altered on intestinal IEL or Peyer’s patch lymphocytes (Fig. 8, and data not shown). While there were some interindividual variations in cell surface expression of other adhesion receptors, many adhesion molecules were expressed, on average, at similar levels on intestinal IEL isolated from α_E^−/− or α_E^+/+ mice. These included CD18 (LFA-1 β₂-chain), CD54 (ICAM-1), CD62L (L-selectin), CD29 (β₁ integrin subunit), CD49d (α₄), and CD49a (α₁ integrin subunit) (Fig. 8). Other adhesion molecules were not detected on intestinal IEL isolated from either α_E^−/− or α_E^+/+ mice, including CD49b (α₂), CD49e (α₅), CD49f (α₆), and CD51 (α_V) (data not shown). Overall, these studies emphasize the relatively limited expression of other adhesion molecules, such as the β₁ integrins, on intestinal IEL. On IEL derived from α_E^−/− mice, the only consistent shift in adhesion molecule expression was increased expression of CD44 (average mean fluorescence intensity (MFI) 118 on cells derived from α_E^−/− mice vs 78 on cells derived from α_E^+/+ mice; n = 4; p < 0.03), a hyaluronate receptor (35, 36). This molecule could participate in mucosal T cell adhesion within the intestinal mucosa, as hyaluronate is expressed within the lamina propria (37). In addition, CD45RB was expressed a slightly higher levels on IEL derived from α_E^−/− than on IEL derived from α_E^+/+ mice (average MFI = 890 vs 780; p = 0.0005; n = 3, not visible in the profile due to the logarithmic scale). This finding, while highly reproducible, was of unclear functional significance.

Overall, the organization of the α_E-encoding gene was strikingly similar to that of other I-domain containing integrins, α_M and α_X (38, 39), with a high degree of conservation in the intron/exon boundaries and exon spacing. In addition, an exon was identified between exons 5 and 6 that was not homologous to exons within other integrins, suggesting that it may have resulted from a gene-insertion event. This exon encoded 44 amino acids, corresponding to the X-(extra)-domain identified within the predicted human α_E (40) and murine α_E subunits (15) (Fig. 1, a and b). The exons were numbered based upon homology with the other I domain-containing integrins, and the exon unique to integrin α_E was designated E (Fig. 1). SSCP analysis was used to localize Itgae to mouse chromosome 11, tightly linked to Abr, and was consistent with the localization of sequence-tagged sites derived from the human homologues of both Itgae (SHGC-12572, WI-22591) and Abr (WI-19743) to 17p13 on the human transcript map (http://www.ncbi.nlm.nih.gov/genemap98 (41)).

In the α_E ^−/− mice, there were reduced numbers of intestinal IEL in three different genetic backgrounds, evaluating mice housed in two different animal facilities. Thus, reduced intestinal IEL numbers was a consistent feature of α_E deficiency, although the magnitude of this effect varied with genetic background and/or environment. IEL numbers were also reduced in the vaginal epithelium, while the number of lymphocytes in lung parenchyma or adjacent to the bronchiolar epithelium was not affected. While lamina propria T lymphocyte numbers also were diminished by α_E deficiency in some strains and/or environmental conditions, the number of T lymphocytes in Peyer’s patches and spleen was not diminished by disruption of the α_E-encoding gene. Thus, α_E deficiency appeared to function selectively in the generation/localization of IEL and lamina propria T lymphocytes but not in the generation of organized lymphoid tissues in the mucosa. While other explanations may exist, these data are most consistent with a role of α_Eβ₇ in modifying T lymphocyte localization to the intestinal mucosa, but suggest that other receptors, such as CD44 and possibly α_Lβ₂ (42), may partially compensate for the loss of α_Eβ₇.

It was notable that the proportion of intestinal IEL that expressed the TCR-αβ was reduced in α_E-deficient mice and that the impact of α_E deficiency upon IEL number was greater in animals that were 5 wk of age or older than in younger mice. Thus, the impact of α_E deficiency upon intestinal T lymphocyte numbers appeared during the time period when microbial colonization results in increased numbers of TCR αβ-expressing intestinal IEL. Finally, the impact of α_E deficiency upon IEL/lamina propria lymphocyte (LPL) numbers may have been influenced by environmental factors, as LPL numbers were reduced in some groups but not others. Overall, these findings suggested that integrin α_Eβ₇ may play a role in the recruitment/expansion of TCR αβ⁺ IEL that is triggered by microbial colonization of the intestine. Based upon the flow cytometry analysis of IEL, it appeared that the αβ T cell subset that was CD4⁺CD8⁻ was not affected by α_E deficiency. In contrast, the CD8αα⁺, CD8αβ⁺, and CD4⁺CD8⁺ subsets were diminished to a similar extent, consistent with the expression of α_Eβ₇ on a larger proportion of CD8⁺ than of CD4⁺ intestinal IEL.

While reduced IEL numbers were consistent with the known function of α_Eβ₇, we were surprised to find reduced numbers of CD3⁺ cells in the intestinal lamina propria of α_E^−/− mice, as E-cadherin is not detected within the lamina propria by immunohistology (Ref. 32, and our unpublished observations). This finding suggested one of two major possibilities, either that intestinal IEL play a role in maintaining lamina propria lymphocyte numbers or that α_Eβ₇ mediates another adhesive interaction that results in the homing or retention of lymphocytes within the lamina propria. While integrin β₇-deficient mice had markedly reduced numbers of lamina propria T lymphocytes, we found that intestinal IEL and lamina propria T lymphocytes were present in normal numbers in RAG-2-deficient mice that were reconstituted with bone marrow from α₄^−/− fetuses (our unpublished observations). Taken together with the reduced lamina propria lymphocyte number observed in α_E^−/− mice, these studies suggest that α_Eβ₇ participates in the localization of lamina propria T lymphocytes. Additional studies will be required to define the cellular and molecular basis for this interaction.

By comparing the impact of targeted integrin gene disruption, insight is gained into the relative roles of the α₄β₇ and α_Eβ₇ integrins. In integrin β₇-deficient mice, there are markedly reduced numbers of T and B cells in Peyer’s patches, intestinal lamina propria, and in the intestinal epithelium (43). In addition, in mice whose T lymphocytes lack the α₄ subunit, there are reduced numbers of T and B lymphocytes in Peyer’s patches, while the numbers of IEL and of lamina propria T lymphocytes were not reduced (Ref. 44, and our unpublished observations). These findings suggest that α₄β₇ is essential for B and T cell localization to Peyer’s patches, but not for T lymphocyte localization to the intestinal epithelium or lamina propria. Finally, in the α_E^−/− mice described in this report, the number of IEL and lamina propria T lymphocytes was partially reduced, without an impact on the localization of B or T cells to Peyer’s patches. Thus, it appears that both α_Eβ₇ and α₄β₇ can function in the localization of T lymphocytes to the intestinal lamina propria and epithelium, while α_Eβ₇ does not mediate T lymphocyte localization to Peyer’s patches. The integrin α_E-deficient mice described herein will provide an important reagent with which to further define the in vivo functions of α_Eβ₇, and of the cells that express it, in health and disease.

We thank T. J. Smith and J. H. Weis for murine α_E cDNA probes; H. B. Warren for histopathologic analyses; P. Kilshaw and R. MacDonald for mAbs; C. Nagler-Anderson for technical advice; X. Hu, E. Meluleni, and J. Connolly for technical assistance; and M. Hemler, V. Hsu, J. Higgins, M. Carr, and G. J. Russell for helpful discussions.