Intraepithelial lymphocytes (IEL) are a phenotypically distinct population of lymphocytes that reside in mucosal epithelia, below the intercellular tight junctions. Although adhesive functions of this population have been previously studied, relatively little is known about IEL migration from the microvasculature into the epithelium. We demonstrated that cultured human IEL were capable of migration into polarized epithelial cells in vitro, where they assumed a subjunctional position, identical to that observed in vivo. The migration was rapid and efficient and was directionally polarized, such that IEL migrated into epithelial monolayers from the basolateral, but not the apical, aspect. After a 4-h period of residence, up to one-half of the IEL then exited the monolayer basolaterally. Migration was partially inhibited by pertussis toxin, suggesting a potential mechanism for IEL migration by chemokine receptor-mediated signaling. The conditions and ligand pairs used in IEL migration were different from those for neutrophils, another cell type known to migrate through epithelia. This system may serve as a model for microenvironmental homing of IEL into the epithelium.

  • mucosa
  • homing
  • lymphocyte recirculation

the intestinal mucosa is routinely exposed to a range of microorganisms and foreign substances but provides both a physical and immunological barrier to such challenges from the exterior environment. In the intestine, the mucosal immune system consists of organized secondary lymphoid organs, such as mesenteric lymph nodes and Peyer’s patches, as well as dispersed leukocytes throughout the intestinal wall and particularly in the mucosa. Mucosal lymphocytes that are located between epithelial cells, below the intercellular tight junctions, are known as intraepithelial lymphocytes or IEL. IEL express a different set of surface receptors than do peripheral blood lymphocytes (PBL) and comprise a phenotypically distinct population. In humans, ∼95% of intestinal IEL (iIEL) are T lymphocytes, whereas ∼80% of PBL are T cells (31). iIEL bear an oligoclonal repertoire of T cell antigen receptors, but PBL are polyclonal (13, 37, 38). Up to 90% of iIEL are CD4−CD8+, and about 60% of these express the CD8 homodimer (CD8αα) (6, 19), in contrast to PBL and lamina propria lymphocytes (LPL), which are predominantly CD4+CD8−. Additionally, those PBL which are CD4−CD8+ usually express the CD8 heterodimer (CD8αβ). A higher proportion of iIEL express the CD45 RO and RB isoforms than do PBL (5, 17), suggesting a previously activated or memory phenotype. Approximately 95% of iIEL and 40% of LPL express the integrin αEβ7 (also known as the human mucosal lymphocyte antigen-1 or HML-1), but <5% of resting PBL bear this heterodimer (10).

The trafficking of mucosal lymphocytes is still poorly understood. Much of what is known concerns lymphocyte attachment to the endothelium or the epithelium. Such studies have highlighted ligand pairs, both lymphocyte-endothelial (α4β7/L-selectin and MAdCAM-1) (3, 4) and lymphocyte-epithelial (αEβ7 and E-cadherin) (9, 22). Although some iIEL may develop within the epithelium (31), it is likely that many iIEL traffic from blood vessels present in the lamina propria to the epithelium. After diapedesis from the microvasculature, and preceding adhesive interactions with epithelial cells, iIEL must migrate across a complex matrix. Presumably, like migration of other cells in the immune system, iIEL movement is not random, but targeted. However, directed migration of lymphocytes in general, and iIEL in particular, is not well understood. For example, whereas many members of the chemokine family induce directed movement of lymphocytes, the observed response is often restricted to specific subpopulations, and the magnitude of the response is typically weak in comparison with other migrating cells (polymorphonuclear cells or PMN, fibroblasts, macrophages) (7, 18). Interestingly, iIEL have been demonstrated to have in vitro chemotactic activity toward interleukin-8 (IL-8), RANTES, growth-related genes, macrophage inflammatory protein, and monocyte chemotactic protein (32). However, it is not known which of these chemokines is relevant in vivo.

We hypothesized that iIEL, if positioned adjacent to the basolateral surface of their target tissue in vitro, would migrate into the monolayer, as they are thought to do in vivo. Furthermore, this migration might be more substantial than lymphocyte adherence or migration into endothelium, since in our case the migrating population was normally resident in the monolayer. To test these hypotheses, we devised a migration system in which a human iIEL line was allowed to migrate into the basolateral aspect of the model polarized human intestinal cell line T84. This assay is likely to describe a useful tool for dissection of the molecular events involved in microenvironmental homing of mucosal lymphocytes from the lamina propria into the epithelium. Data presented suggest that when an appropriate target tissue is available, lymphocyte migration is efficient and as rapid as that which occurs with “professionally” migrating cells such as PMN. Finally, the conditions and ligand pairs involved in this migration are distinctly different from those involved in migration of PMN into epithelium.


T84 epithelial cell culture.

T84 cells were grown in a 1:1 mixture of DMEM and Ham’s F-12 medium, supplemented with 15 mM HEPES, 14 mM NaHCO3, 40 μg/ml penicillin, 8 μg/ml streptomycin, and 5% newborn calf serum (14). Cells were passaged every 6–8 days using 0.1% (wt/vol) trypsin and 1 mM EDTA in PBS. For migration experiments, T84 cells were grown on collagen-coated 0.33 cm2 permeable polycarbonate filters with 0.5 μm pore size (Costar, Cambridge, MA) as previously described (23). Cells were plated either on the upper surface of the filter, resulting in a right-side-up monolayer (for evaluating apical to basolateral migration, Fig.1,inserts) or on the lower surface of the filter, resulting in an inverted monolayer (for evaluating basolateral to apical migration, Fig. 1,inverts).

Fig. 1.

Monolayers of epithelial cells may be grown on underside of porous polycarbonate filters (inverts) or on upper side (inserts). When lymphocytes are added to upper chamber of inverts they are in contact with collagen-coated filter and basally secreted extracellular matrix. In contrast, lymphocytes in upper chamber of inserts are in contact with epithelial apical surfaces.

iIEL culture.

A CD8+ T lymphocyte line, 032891, derived from normal human small intestine has been previously described (8, 20). This line closely resembles iIEL in phenotype, but absolute markers for iIEL have not been identified. The established line was grown in the medium described by Yssel et al. (39) containing 4% fetal bovine serum (Hyclone, Logan, UT) 50 μM 2-mercaptoethanol, and 20 U/ml recombinant human IL-2 (Boehringer Mannheim, Indianapolis, IN). The line was periodically restimulated with phytohemagglutinin-P (Difco, Detroit, MI) and irradiated feeder cells consisting of human peripheral blood mononuclear cells (PBMC) and JY B lymphoblastoid cells (ATCC). The iIEL were used in migration assays typically 7–25 days after stimulation. Fresh iIEL were isolated from patients undergoing gastric bypass surgery for morbid obesity (32). Jejunal mucosa was minced and incubated in 1 mM dithiothreitol for 30 min at 37°C. This was followed by shaking in 0.75 mM EDTA at 37°C three times for 45 min each. The supernatant cells were purified over a Percoll density gradient, with a typical yield of 94% CD2+ and 89% CD8+ cells.


The parent line, JY′, is an Epstein-Barr virus-transformed B lymphoblastoid variant line that expresses high levels of integrin α4β7 (11). The J5 and JV transfectant lines have been described in detail elsewhere (36). Briefly, the J5 line was created by transfection with the full-length cDNA for αE, and expresses both αEβ7 and α4β7 on the cell surface by flow cytometry and immunoprecipitation. J5 has been demonstrated to bind E-cadherin-expressing nonpolarized epithelial cells in static adhesion assays. This binding was blocked by monoclonal antibodies (MAb) against αEβ7 and against E-cadherin (36). The JV line was transfected with a vector control and expresses α4β7 (but not αEβ7).

Lymphocyte migration assays.

Lymphocytes were centrifuged and resuspended in DMEM at 5 × 106 cells/ml. An intravital fluorescent dye, 5-chloromethylfluorescein diacetate (CMFDA; Molecular Probes, Eugene, OR), was added to a final concentration of 1 μM, and the cells were incubated for 30 min at 37°C. After this labeling step, the cells were washed and resuspended in a final volume of 2.2 × 106 cells/ml in T84 medium. Fluorescently labeled lymphocytes (150 μl) were added to the upper chamber of Transwell filters containing inverted 5- to 14-day-old T84 monolayers, and incubated at 37°C for varying periods of time. The Transwell filter was then manually removed, rinsed by dipping in Hanks’ balanced salt solution (HBSS), and transferred to a fresh well containing 300 μl HBSS in the bottom chamber; 150 μl HBSS were added to the upper chamber. Fluorescence in the monolayer was then measured using a fluorescence plate reader (CytoFluor 2350; Millipore, Bedford, MA). Assays were performed in triplicate and reported as means ± SD. A background reading of T84 monolayers alone was subtracted from each data set. Control experiments indicated that migration of unlabeled lymphocytes did not alter the background fluorescence of T84 monolayers (data not shown). To measure remaining fluorescence in the upper chamber (corresponding to the remaining or nonmigrated lymphocytes), the cell suspension was agitated by repeated pipetting, transferred to a 96-well plate, and read in the fluorescence plate reader. Because of different geometries, fluorescence levels in 96-well plates may not be directly compared with fluorescence levels in monolayers in 24-well plates. To normalize fluorescence between different cell types, 150 μl of labeled lymphocytes were added to a 96-well plate, along with a negative control consisting of medium alone. The fluorescence data from this plate were used to measure relative fluorescence between different cell types, using the formulaNormalized fluorescence =(invert readinginvert negative control)×100150μlcells in96­well plate150μlmedium in96­well plate Figures are representative of at least three independent experiments unless otherwise indicated.

Pertussis toxin treatment of lymphocytes.

Fluorescently labeled lymphocytes were pretreated with 1 μg/ml of pertussis toxin (Biomol, Plymouth Meeting, PA) in medium for 2 h at 37°C. After treatment, lymphocytes were washed three times by centrifugation and then added to monolayers as usual. Controls for these experiments were similarly manipulated in the absence of pertussis toxin.

Confocal microscopy.

CMFDA-labeled iIEL were allowed to migrate into T84 monolayers, and then the monolayers were subjected to fixation with 3.7% paraformaldehyde in PBS for 30 min at 4°C. After fixation, monolayers were acetone permeabilized and stained with rhodamine-phalloidin as previously described (25) and visualized using dual-channel fluorescence imaging to selectively identify iIEL and the actin cytoskeleton.

Salmonella treatment of T84 monolayers.

This was performed as described previously (26). Briefly,Salmonella typhimurium strain χ3306 was grown overnight in 10 ml LB medium, under microaerobic conditions. The overnight culture was washed twice in HBSS by centrifugation at 8,000 rpm for 5 min at 4°C. This pellet was resuspended in 300 μl HBSS, and 25 μl were used to colonize the apical membrane of T84 monolayers. After 45 min of treatment at 37°C, in a moist chamber, the colonized monolayers were washed in medium to remove nonadherent bacteria and placed into fresh wells containing T84 medium. Fluorescently labeled iIEL were added in subsaturating amounts (75 μl at 2.2 × 106 cells/ml) and migration was measured as usual.

PMN migration assays.

This assay was performed as previously described (23). Briefly, PMN were isolated from whole blood, using a gelatin-sedimentation technique, followed by isotonic ammonium chloride lysis of red blood cells to a final concentration of 5 × 107 cells/ml. To the upper chamber of T84 inverts were added 106 PMN at 37°C for 2 h. A positive control of 1 μMN-formyl-methionyl-leucyl-phenylalanine (fMLP) was used as a chemoattractant in the lower well. Transmigration was quantified using a standardized assay for myeloperoxidase, present in PMN azurophilic granules. PMN cell equivalents (CE) were calculated from a standard curve.


Cultured iIEL migrate into an epithelial monolayer.

T84 is a polarized human epithelial cell line that appropriately models several regulated events, including Cl secretion, neutrophil epithelial interactions, and barrier function (14, 15, 23, 24). T84 cells were allowed to form a monolayer on the underside of collagen-coated permeable supports (Fig. 1). Fluorescently labeled iIEL were placed on the basolateral side of the monolayer (i.e., the upper chamber) and examined for their ability to migrate into and across monolayers. As can be seen in Fig. 2,A andB, confocal microscopy showed that iIEL were present in the monolayer in as little as 15 min, where they were positioned below the intercellular tight junctions and within the paracellular space. This direct visualization correlated with fluorescent plate reader data, allowing quantification of migration (Fig. 2 C). Thus cultured iIEL migrated into a T84 monolayer in a time-dependent manner. This migration was rapid, peaking at between 3 and 4 h, involved a large proportion (30–60%) of the lymphocytes, and did not result in movement of iIEL through the monolayer and into the “luminal” (apical or lower) compartment. Similar results were observed with other epithelial cells [HT-29, a colon carcinoma cell line, and 16E6.A5, a transformed mammary epithelial cell line (1)].

Fig. 2.

A: intestinal intraepithelial lymphocytes (iIEL) migrate into T84 monolayers in a time-dependent manner. iIEL were fluorescently labeled with 5-chloromethylfluorescein diacetate (CMFDA) and allowed to migrate into T84 monolayers (15 min, 30 min, 1 h, and 2 h), after which they were visualized by en face confocal fluorescent microscopy. B: iIEL are positioned in epithelial monolayer below tight junctions, in paracellular space. In thisx-zconfocal section, CMFDA- labeled iIEL were allowed to migrate into T84 monolayer for 2 h, after which the monolayer was counterstained with rhodamine-phalloidin to visualize the actin cytoskeleton (right). CMFDA-labeled iIEL are visualized in left panel (arrows) between cells and beneath apically situated tight junctions. C: iIEL migrate into T84 monolayers in a time-dependent manner. CMFDA-labeled iIEL were allowed to migrate into T84 monolayer for time periods shown, after which they were scanned using fluorescent plate reader. Negative control (no IEL) consisting of T84 monolayers alone (with no lymphocytes added) was subtracted from all data. Only background fluorescence was detected in lower chamber (corresponding to apical or luminal side of T84 monolayers), indicating that lymphocyte migration was into but not through the monolayer (not shown).

iIEL migration into T84 monolayers exhibits pronounced polarity.

In contrast with basally applied iIEL (Fig.3, inverts), iIEL were unable to migrate into T84 monolayers from the apical aspect (Fig. 3, inserts). Although a small degree of fluorescence was associated with the epithelial monolayers in the inserts, it did not increase in a time-dependent manner and was probably due to surface adhesion of lymphocytes. Such data indicate that the efficient and rapid migration of basolaterally applied iIEL is not simply a reflection of random baseline movement but rather is directed.

Fig. 3.

iIEL migrate into polarized T84 monolayers from basolateral aspect (inverts, basolateral → apical), but not from apical side (inserts, apical → basolateral). A small degree of adherence is observed with inserts, but it is not time dependent, indicating that it is probably nonspecific.

Cultured iIEL migrate to a greater degree than do fresh iIEL or fresh or cultured PBMC.

Cultured iIEL were compared with freshly isolated iIEL in their ability to migrate into T84 monolayers (Fig.4 A). Fresh iIEL were not capable of substantial migration, although independently derived cultured iIEL lines migrated efficiently. Similarly, cultured iIEL were compared with freshly isolated and stimulated PBMC in their ability to migrate into T84 monolayers (Fig.4B). Fresh PBMC were not capable of substantial migration, although stimulation with phytohemagglutinin and IL-2 resulted in a modest increase in migratory ability. After repeated stimulation over a period of months, the resulting T lymphocyte lines showed an improved ability to migrate (data not shown). This finding suggests the possibility that chronic stimulation results in the upregulation of the adhesion molecules and/or chemokine receptors necessary for migration or, alternatively, that a subpopulation of migratory cells was expanded after stimulation.

Fig. 4.

A: 2 independently derived iIEL lines (iIEL lines 1 and 2) migrate efficiently into T84 monolayers, but freshly isolated iIEL show only weak migration. These data represent 1 of 2 experiments, with different fresh iIEL donors.B: stimulated iIEL migrate efficiently into T84 monolayers (iIEL line), stimulated peripheral blood lymphocytes (PBL) migrate somewhat less efficiently (PBL line), but fresh PBL are incapable of migrating. Fluorescence was normalized to compensate for differences in labeling efficiency between different cell types (see materials and methods).

iIEL residence in epithelial monolayer is followed by basolateral exit.

At various intervals over a 24-h period, cultured iIEL were added to inverted T84 monolayers for 2 h. After 2 h the medium in which they were suspended was removed from the upper (basolateral) chamber and replaced with fresh medium without iIEL. At the end of the 24-h period the upper (basolateral) chamber of all inverts was sampled for the presence of fluorescently labeled lymphocytes that had exited the monolayer. The monolayer was also scanned for lymphocytes remained in it. As shown in Fig. 5, there were approximately one-half as many lymphocytes present in the monolayer at the 23.5-h time point as there were at the 4-h time point. Furthermore, sampling of the basolateral aspect shows increasing amounts of lymphocytes present in that compartment over the time course. Thus, following migration, lymphocytes reside in the monolayer for 4 h, after which approximately one-half of them exit basolaterally.

Fig. 5.

iIEL migrate into T84 monolayer maximally at 4 h, after which one-half of them exit basolaterally. CMFDA-labeled iIEL decrease in monolayer from 4 to 23.5 h (monolayer), but their presence in basolateral (upper) chamber increases over this time period (exited lymphocytes). Because the basolateral fluid is transferred to a 96-well plate to read levels of exited lymphocytes, they may not be compared on same scale with monolayer fluorescence (which is read in 24-well plates).

Migration of iIEL into epithelial monolayer is inhibited by pertussis toxin.

Pretreatment of the lymphocytes with 1,000 ng/ml pertussis toxin resulted in partial inhibition of migration (30–60%) compared with a mock-treated control (Fig. 6, PTX and CTRL), indicating that a G protein-mediated pathway may be responsible for migration.

Fig. 6.

Pertussis toxin treatment blocks migration of iIEL. CMFDA-labeled iIEL were treated with pertussis toxin for 2 h (PTX), after which they were allowed to migrate into T84 monolayer. A control sample, treated with buffer alone (CTRL), migrated more efficiently than did the toxin-treated cells.

Washing epithelial monolayers does not affect migration of IEL.

T84 monolayers were rinsed apically and basolaterally with medium before addition of iIEL in an attempt to wash away any soluble factors that may induce migration. This rinsing of the monolayers did not result in a change in migration, compared with untreated monolayers (not shown). This indicates that in the time frame of the assay, it was not possible to remove any soluble factors affecting migration by simple rinsing.

Salmonella treatment of epithelium did not alter lymphocyte migration.

Salmonella typhimurium colonization results in the release of a variety of chemoattractants/cytokines, both from T84 cells and from epithelium in vivo, and greatly increases the ability of PMN to migrate across the monolayer (27). As shown in Fig.7 A,Salmonella colonization resulted in an increase in PMN transmigration (Salmonella) compared with untreated monolayers (CTRL). A positive control chemoattractant of fMLP is included for comparison. T84 monolayers similarly colonized bySalmonella (Fig.7 B) did not stimulate or inhibit iIEL migration compared with similarly treated controls (CTRL). Thus, in marked contrast to PMN movement, the signals that might stimulate iIEL movement into the monolayer are not altered bySalmonella colonization.

Fig. 7.

A: Salmonella colonization of T84 monolayers resulted in an increase in polymorphonuclear cell (PMN) transmigration. Monolayers were colonized withSalmonella, and then PMN were added to basolateral reservoir and allowed to migrate for 2 h. Control monolayers treated with medium alone (CTRL) show only a low level of transmigration. Positive control ofN-formyl-methionyl-leucyl-phenylalanine (fMLP) is shown in comparison. B:Salmonella treatment of T84 monolayers did not alter iIEL migration. Monolayers treated with Salmonella are compared with mock-treated control monolayers (CTRL), over a range of times from 30 min to 2 h. Similar experiments extended to as long as 24 h also did not show reproducible differences between Salmonella and mock-treated monolayers (not shown).

Transfectants expressing αEβ7 and α4β7 are not capable of migrating into T84 monolayers.

JY transfectants expressing αEβ7 and α4β7 integrins (Fig.8, J5), or α4β7 alone (JV), were incapable of migrating into T84 monolayers. At all time points measured, cultured iIEL migrated more efficiently than either J5 or JV. Control experiments performed in parallel confirmed the ability of J5 (but not JV) to bind to nonpolarized 16E6.A5 epithelium (data not shown), indicating that they expressed functional αEβ7 on the cell surface, at least in adhesion assays. Because the JY transfectants have not been shown to have functional migrational machinery, these experiments are only suggestive of a lack of involvement of the β7 integrins in migration. However, MAb known to block αEβ7 and α4 integrins in adhesion assays were unable to alter either epithelial migration or retention of iIEL in vitro (data not shown), further strengthening the potential lack of involvement. Similarly, treatment of iIEL with transforming growth factor-β1 (TGF-β1), which upregulates expression of αEβ7, failed to substantially increase migration or retention of iIEL in monolayers (not shown).

Fig. 8.

Transfectants expressing αEβ7 and/or α4β7 were not capable of migration into T84 monolayers. J5 line expresses both αEβ7 and α4β7 integrins and JV line expresses α4β7. Positive control of iIEL was included for comparison. Control experiments performed at same time indicated the ability of J5 (and not JV) to adhere to E-cadherin- expressing nonpolarized epithelial cells (not shown), indicating expression of functional integrins.


Previous studies of mucosal lymphocyte homing have focused mainly on adhesion to either the epithelium or the endothelium (8, 9, 21, 33,34). However, relatively little is known concerning the movement of iIEL or their precursors toward and into the epithelial monolayer. To study lymphocyte migration in greater detail, we devised an in vitro assay system in which lymphocytes that were originally resident in the intestinal epithelium (cultured iIEL) were allowed to colonize a polarized epithelial monolayer from the basolateral aspect, where they were retained. We observed rapid migration that involved 50–70% of the lymphocytes. Confocal microscopy confirmed that lymphocytes which had migrated into the monolayer occupied a position analogous to that of IEL within the intestinal epithelium in vivo, below the intercellular tight junctions, in the paracellular space. In contrast to PMN, which in vivo and in vitro migrate across epithelia in response to pathogens, iIEL did not cross the epithelium but remained within it and migrated toward the target tissue in the basal state, as is thought to occur in vivo. Thus the iIEL migration described here may represent a physiological model of microenvironmental homing of lymphocytes from the lamina propria into the epithelium.

Lymphocytes were unable to migrate into the monolayer from the apical (luminal) aspect. This suggests that the adhesion molecule(s) or the chemotactic factor(s) that were responsible for iIEL migration were expressed differentially between the apical and basolateral surfaces. Consistent with this hypothesis, migration of iIEL into the epithelium was reduced by pertussis toxin, indicating the involvement of a Gαi-mediated signal, common to chemokine receptor-mediated migration pathways. For example, after rolling of lymphocytes on endothelium, the integrin-mediated arrest is thought to be due to chemokine-mediated activation, and chemokines are known to act via a G protein- mediated, pertussis toxin-sensitive pathway. Thus arrest of lymphocyte rolling in Peyer’s patch endothelium is inhibited by pertussis toxin (2), as is lymphocyte migration from the spleen into the splenic white pulp (12). Interestingly, washing the monolayers immediately before addition of lymphocytes did not disrupt migration, suggesting that any chemokine activity is either quickly replaced in the supernatant over the time course of the assay or is imprinted on the extracellular matrix laid down by the epithelial cells. Experiments with epithelial supernatants suggested a secreted chemokine activity toward iIEL but were poorly reproducible (data not shown).

Comparison of different types of lymphocytes indicated a hierarchy of migratory ability, in which stimulated iIEL migrated more efficiently than did stimulated PBL, which in turn migrated more efficiently than did freshly isolated PBL or iIEL. Thus migration of lymphocytes exhibits specificity as to lymphocyte origin and activation state and may represent a difference in surface receptors between the cell types. Mitogen activation of lymphocytes resulted in an increase in migratory ability, with repeated stimulation of PBL resulting in a migratory ability comparable to that of iIEL. This may reflect the in vivo situation in which mucosal epithelium is an effector site to which lymphocytes migrate after activation in a lymph node, such as the Peyer’s patches. The failure of freshly isolated iIEL to migrate may therefore be representative of their having undergone differentiation or quiescence following activation and migration. Future studies may identify a lymphocyte population capable of migration without exogenous stimulation. Intestinal LPL and Peyer’s patch lymphocytes may represent such populations.

Neutrophils are another cell type known to migrate into the basolateral aspect of epithelial monolayers (29). However, significant differences exist between neutrophil and lymphocyte migration in vitro, corresponding with the behavior of these two cell types in vivo. Neutrophils migrate across the monolayer (transmigration), with only brief residence times in the monolayer, followed by an apical exit. The lymphocyte migration described here is characterized by prolonged residence in the monolayer, followed by a basolateral exit. In neutrophil migration, the interepithelial tight junctions are impaled by the migrating cells, whereas lymphocytes remain subjunctional. Furthermore, neutrophil migration only occurs in vitro in the presence of exogenously added or agonist-directed endogenous release of chemoattractants, whereas no exogenous chemoattractants or agonists were required to stimulate iIEL migration. Reversal of the chemoattractant gradient causes apical-to-basolateral neutrophil migration, whereas lymphocyte migration in this direction was not measurable. Treatment of monolayers with Salmonella results in an increase in neutrophil transmigration, due to secretion of an apically derived factor (27). However, Salmonella treatment did not alter lymphocyte migration, indicating that different factor(s) may be responsible in this case. Salmonella treatment also results in increased basolateral secretion of IL-8 from epithelial cells (16, 26). Although IL-8 is chemotactic for iIEL in vitro (32), the lack of response to Salmonella in the current study indicates that this factor may not be responsible for their migration into epithelia. Anti-CD47 MAb have been shown to significantly inhibit neutrophil migration (28) but had no effect on iIEL migration (data not shown), suggesting further differences in migration mechanisms between neutrophils and lymphocytes.

Significantly, transfectants expressing the αEβ7 and α4β7 integrins on their surface did not migrate. Likewise, MAb blocking studies were also unable to show involvement of αEβ7, α4β7, or E-cadherin in migration. These MAb are known to block binding of IEL to nonpolarized epithelium in the adhesion assay mentioned previously (8,35). However, the physiological relevance of IEL adhesion to the apical surfaces of nonpolarized epithelial cells is not clear. Furthermore, TGF-β1 treatment of iIEL, which upregulates expression of αEβ7, did not increase their migration or retention within the monolayer. Therefore it is possible that either the αEβ7 and α4β7 integrins are not adequate for the migration observed in this assay system or they are not involved in this function, but instead may have another role in vivo.

After a period of residence in the monolayer (4–8 h), up to one-half the iIEL left the epithelium basolaterally. Experiments using parabiotic mice have shown that iIEL leave the intestine to reseed the partner mouse’s intestine at a very low rate (30). However, in these studies, lymphocytes had not only to leave the intestine of the donor mouse but also to re-enter the epithelium of the recipient, potentially explaining the difference in observed efficiencies between the in vivo studies and the current one.

In conclusion, we have demonstrated an in vivo assay that may model the microenvironmental homing of iIEL. To our knowledge, this is the first demonstration of lymphocyte migration into epithelium in vitro. This system is characterized by migration of a substantial fraction of the lymphocytes and exhibits several features that indicate that it closely parallels the in vivo migration of iIEL. The migration observed is different quantitatively from other models of lymphocyte migration and qualitatively different from models of neutrophil migration. We anticipate that this assay may be used to dissect out the adhesion molecules and signals responsible for migration of iIEL from the lamina propria into the epithelium.


We thank R. A. Kerner and C. J. Roy for technical help, F. Sanchez-Madrid and W. Muller for MAb, and F. W. Luscinskas and D. M. Kozikowski for critical reading of this manuscript.


  • Address for reprint requests: S. K. Shaw, Dept. of Pathology, LMRC-414, Brigham & Women’s Hospital, Boston, MA 02115.

  • These studies were supported by National Institutes of Health Grants 5-F32-DK-09427 (S. K. Shaw), DK-50989 (B. A. McCormick), HL-54229 (C. A. Parkos), DK-35932 and DK-47662 (J. L. Madara).


View Abstract