The Roles of L-Selectin, β₇ Integrins, and P-Selectin in Leukocyte Rolling and Adhesion in High Endothelial Venules of Peyer’s Patches¹

Peyer’s patches, one component of the gut-associated lymphoid tissue, are located on the wall of the small intestine and play an important role in the mucosal immune response (particularly an IgA response). Lymphocytes continuously recirculate through Peyer’s patches to sample sequestered Ag and further differentiate into mucosal effector T and B lymphocytes. The first step in lymphocyte recirculation through Peyer’s patches is the movement of these cells from the Peyer’s patch high endothelial venules (HEV)³ into the surrounding lymphoid tissue. As in many other lymphoid and nonlymphoid tissues, lymphocyte recruitment into Peyer’s patches is thought to require several steps: initial attachment to the endothelium from the blood (capture), rolling along the endothelium, firm adhesion to the endothelium in response to activating substances, and subsequent emigration out of the venule (reviewed in 1 .

Several molecules have been implicated in mediating lymphocyte recruitment into Peyer’s patches. L-selectin is a Ca²⁺-dependent mammalian lectin expressed constitutively on granulocytes and monocytes as well as on most circulating lymphocytes, including naive and certain effector T cells (2). Although L-selectin-deficient (L^−/−) mice have morphologically normal Peyer’s patches, they exhibit a significant reduction in short term homing of exogenous lymphocytes to Peyer’s patches (3). Subsequent data have shown that this reduction in homing is most likely the result of a deficit in leukocyte rolling when L-selectin function is blocked, because lymphocytes treated with an L-selectin mAb and injected i.v. fail to roll in Peyer’s patch HEV (4), and functional blockade of L-selectin with a mAb in vivo reduces the number of rolling leukocytes in Peyer’s patch HEV by up to 90% (5). However, the normal size and cellularity of Peyer’s patches in L^−/− mice (3) suggest that other important adhesion molecules can bypass the requirement for L-selectin in lymphocyte recruitment to Peyer’s patches. In fact, the absence of L-selectin delays, but does not prevent, lymphocyte homing to Peyer’s patches (3).

The most important adhesion molecule in lymphocyte recruitment to Peyer’s patches is α₄β₇, an integrin expressed at low levels on naive T and B cells and at high levels on effector and memory T cells within the gut (6). This integrin can mediate rolling interactions in vitro independent of L-selectin (7). Treatment of lymphocytes with a mAb specific for the α₄ chain of this integrin results in a twofold increase in the rolling velocity of exogenous lymphocytes passing through HEV of wild-type Peyer’s patches in vivo (4). Recently, β₇ integrin-deficient (β₇^−/−) mice have been generated (5) to address the role of α₄β₇ under physiologic conditions. These mice have severely reduced Peyer’s patch cellularity and decreased lymphocyte adhesion to Peyer’s patch HEV. However, the number of leukocytes rolling in Peyer’s patch HEV of β₇^−/− mice was normal, although the leukocyte rolling velocity was increased. Lymphocyte rolling in Peyer’s patch HEV of both wild-type and β₇^−/− mice was inhibited by >90% with an L-selectin mAb (5). Here, we examine β₇ integrin/L-selectin double mutant (β₇/L^−/−) mice that lack both adhesion molecules known to be important in lymphocyte homing to Peyer’s patches. We hypothesize that in the absence of these two adhesion molecules, other, previously unknown, adhesion mechanisms in Peyer’s patch HEV may be unmasked.

Mice deficient for L-selectin (3) or β₇ integrin (5) expression were generated as described previously. Mice deficient for both L-selectin and β₇ integrin were generated by breeding homozygous L^−/− mice and homozygous β₇^−/− mice and testing the progeny for both mutated genes by Southern blotting. The absence of cell surface expression of L-selectin and β₇ integrins was confirmed using two-color immunofluorescence flow cytometry^4,5. β₇^−/− mice were generated on a C57BL/6 background, and L^−/− mice were at least a fifth generation backcross onto a C57BL/6 background. Control experiments were performed in age-matched C57BL/6 mice bred at the University of Virginia (breeders purchased from Hilltop Farms, Scottdale, PA). All experiments were performed on healthy mice that were at least 8 wk of age. All mice were housed in a conventional facility. All animal experiments were approved by the institutional animal care and use committee.

The P-selectin mAb RB40.34 (rat IgG1, 30 μg/mouse) was a gift from Dr. Dietmar Vestweber at the University of Munster (Munster, Germany). This Ab blocks P-selectin-dependent adhesion in vitro (8), P-selectin-dependent leukocyte rolling in vivo (9), and P-selectin-dependent leukocyte recruitment in vivo (8). The L-selectin mAb MEL-14 (rat IgG2a; 100 μg/mouse) was purified from hybridoma supernatant (American Type Culture Collection, Manassas, VA). This Ab blocks L-selectin-dependent lymphocyte homing to peripheral lymph nodes (10) and L-selectin-dependent leukocyte rolling in vivo (9, 11). The mAb PS/2 specific for the α₄ integrin chain (rat IgG2b; 100 μg/mouse) was also purified from hybridoma supernatant (American Type Culture Collection). This Ab blocks α₄β₇-dependent binding of lymphocytes to recombinant MadCAM-1 in vitro (7) and lymphocyte homing to Peyer’s patches in vivo (12).

Mice were premedicated with an i.p. injection of a mixture of 30 mg/kg sodium pentobarbital (Nembutal, Abbott Laboratories, North Chicago, IL) and 0.1 mg/kg atropine (Elkins-Sinn, Cherry Hill, NJ) in isotonic saline. Mice were then anesthetized with an i.p. injection of 100 mg/kg ketamine hydrochloride (Ketalar, Parke-Davis, Detroit, MI) in isotonic saline. The trachea, carotid artery, and jugular vein were cannulated, and mice were thermocontrolled at 37°C using a small animal heating pad (model 50-7503, Harvard, South Natik, MA). Blood pressure was monitored continuously (model BPMT-2, Stemtech, Menomonee Falls, WI).

A Peyer’s patch was prepared for intravital microscopy as described previously (4, 5). Briefly, a 1.5-cm incision was made along the linea alba to open the peritoneal cavity and expose the cecum and small intestine. The mouse was then turned on its side, and the intestines were pushed out of the peritoneal cavity by gently pressing on the back. The intestines were superfused with thermocontrolled (37°C) bicarbonate-buffered saline as previously described (13). To immobilize the small intestine and allow for viewing of the Peyer’s patch microvasculature, a plastic coverslip (Baxter, Deerfield, IL) cut to approximately 1 cm² was overlaid with two parallel strips of clear silicone, high vacuum grease (Dow Corning, Midland, MI). This coverslip was then laid over the preparation, with the small intestine lying parallel between the two rows of grease and the Peyer’s patch flattened against the coverslip to allow better microscopic viewing. To visualize intravascular leukocytes by epifluorescent microscopy, each mouse was given a 0.15-ml venous injection of 1 mg/ml acridine red (Chroma, Stuttgart, Germany) 30 min before intravital observation.

Microscopic observations were made using an intravital microscope (Axioskop, Carl Zeiss, Thornwood, NY) modified for stroboscopic (60/s; Strobex 236, Chadwick Helmuth, Mountain View, CA) epifluorescence illumination (filter block Zeiss 9) with a saline immersion objective (SW 40, 0.75 numerical aperture). HEV were observed and recorded (S-VHS recorder, Panasonic, Osaka, Japan) through a CCD camera system (model VE-1000CD, Dage-MTI, Michigan City, IN). Throughout the experiment, 10-μl blood samples were withdrawn at 45-min intervals from the carotid catheter, diluted 1/9 with Kimura (11 ml of 5% (w/w) toludine blue, 0.8 ml of 0.03% light green SF yellowish, 0.5 ml of saturated saponin, and 5 ml of 0.07 M phosphate buffer, pH 6.4; all reagents were obtained from Sigma, St. Louis, MO), and analyzed for leukocyte concentration (expressed as number of leukocytes per microliter of whole blood). Additional blood samples were taken, using the above procedure, before and after administration of mAb and at the termination of the experiment.

Two Peyer’s patches per mouse (one being the observed Peyer’s patch) were excised from the small intestine and fixed in 4% (w/w) paraformaldehyde (Sigma) for at least 4 h but not >24 h. The tissues were then dehydrated in increasing concentrations of ethanol (75% (v/v), 95% (v/v), and absolute), cleared in xylene, and soaked in warm (56°C) liquid paraffin for several hours to allow saturation. The tissues were embedded by cooling until the paraffin hardened. Ten-micron serial transverse sections were cut (Leitz microtome, Wetzlar, Germany), floated out in a water bath onto poly-l-lysine (Sigma)-coated slides, and dried at room temperature. Sections were cleared in xylene, rehydrated with decreasing concentrations of ethanol (absolute, 95% (v/v), 70% (v/v), and distilled water), stained with hematoxylin and eosin (Sigma), cleared in absolute ethanol and xylene, and mounted using Permount (Fisher Scientific, Pittsburgh, PA).

Initial attempts at staining Peyer’s patch HEV with indirect immunofluorescence of peroxidase-conjugated secondary Abs yielded unsatisfactory results, most likely because of low expression of P-selectin (data not shown). Therefore, we developed a new staining technique employing fluorescent beads. Recombinant protein G (Sigma) was covalently coupled to fluorescent-carboxylated (2-μm diameter) YG microspheres (Polysciences, Warrington, PA) using a carbodiimide coupling kit according to the manufacturer’s directions (Polysciences). Beads were stored in the storage buffer provided and used within 1 wk. Thirty minutes before observation of the Peyer’s patch, 200 μl of protein G-coupled beads were mixed with 200 μg of either a nonbinding rat IgG (Pierce, Rockford, IL) or the P-selectin mAb RB40.34. The mixture was allowed to incubate for 30 min before the beads were washed in isotonic saline with 1% BSA three times. The beads were resuspended in isotonic saline with 1% BSA and vortexed to break up any aggregates. Animals were injected with three sequential 50-μl boluses of the rat IgG-coupled beads over 5 min followed by a 10-min resting period to allow for removal of any circulating beads. Subsequently, the RB40.34-coupled beads were injected in the same manner, and any bound beads were observed by intravital fluorescence microscopy.

Indirect immunofluorescence flow cytometry was used to examine binding of a recombinant human P-selectin-IgG chimera (gift from S. R. Watson, Genentech, South San Francisco, CA) to suspensions of Peyer’s patch lymphocytes. Human P-selectin can bind to mouse PSGL-1 (14). Peyer’s patches were isolated from wild-type, β₇^−/−, L^−/−, and β₇/L^−/− mice and suspended in PBS, pH 7.4, with 2 mM CaCl₂, 0.02% sodium azide, and 5% goat serum. Lymphocytes from Peyer’s patches were then released by gentle syringe aspiration. A multivalent P-selectin complex (15) was preformed by mixing recombinant P-selectin-IgG chimera and FITC-conjugated goat F(ab′)₂ anti-human IgG (Fc-specific) Ab (Caltag, Burlingame, CA) in a 1/2 (w/w) ratio followed by incubation at 4°C for 30 min. Aliquots of the P-selectin-IgG FITC complex (2 μg P-selectin-IgG/4 μg goat anti-human IgG FITC) were added to 1 × 10⁶ cells and incubated for 20 min on ice. Lymphocytes in the Peyer’s patch preparation were identified by coincubation with 1 μg each of phycoerythrin-conjugated mAb against the CD3 complex on T cells (clone 17A2, rat anti-mouse IgG2b) and CD45R/B220 on B cells (clone RA3-6B2; rat anti-mouse IgG2a; PharMingen, San Diego, CA). Selectin-specific adhesion was confirmed by inhibition of P-selectin complex binding to Peyer’s patch lymphocytes by treating cell suspensions with 5 mM EDTA. Cell suspensions were analyzed on a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ) using CellQuest software (version 1.2).

Microvessel diameters were measured using a digital image processing system (16). Centerline blood flow velocity was determined after i.v. injection of 2-μm diameter fluorescent YG microspheres (Polysciences, Warrington, PA) by measuring frame-to-frame displacement (three microspheres per venule). Average blood flow velocity was determined by dividing the measured centerline velocity by a factor of 2. The rolling leukocyte flux fraction was determined as the number of rolling leukocytes expressed as a percentage of all leukocytes passing through the venule per unit time. In Peyer’s patch HEV, the number of rolling leukocytes was consistently higher (by ∼60%) than the product of the flow rate and the systemic leukocyte concentration, a parameter used in previous studies (9, 11) as the denominator for the rolling flux fraction. An increased concentration of leukocytes in microvessels can occur as a consequence of multiple flow partition at bifurcations in the microvascular network (17, 18), leading to leukocyte accumulation in the most distal branches of the arteriolar tree. Consequently, venules fed by these branches also carry blood containing a higher leukocyte concentration. Therefore, the flux fraction in wild-type mice was set at 100%, and the flux fractions in the other genotypes were adjusted accordingly. The leukocyte rolling flux fraction was also corrected for variations in centerline velocity and venular diameter as described previously (9). Individual leukocyte rolling velocities were measured from video recordings by analyzing 5 to 15 leukocytes/venule and measuring the time necessary to travel a fixed distance (∼50–80 μm) using a digital image-processing system (16).

All statistical comparisons were conducted using an analysis of variance with a post-hoc Student-Newman-Keuls multiple comparison procedure. SPSS software (SPSS, Chicago, IL) was used for all statistical analyses. Statistical significance was set at p < 0.05.

All mice used in this work appeared healthy and of normal size and weight for their ages. During the intubation surgery, no enlarged cervical lymph nodes were observed in any mice. Similar to what has been previously reported, the systemic leukocyte counts in L^−/− mice (3) were slightly higher than those in wild-type mice (Table I). Although previous work has not identified an increase in circulating counts in β₇^−/− mice (5), we found that circulating counts in these mice were also higher than those in wild-type mice (Table I). Similarly, the circulating leukocyte counts in β₇/L^−/− mice were elevated (Table I). The percentages of granulocytes and mononuclear leukocytes were similar in each genotype (Table I).

We first examined the morphology of Peyer’s patches from wild-type, L^−/−, and β₇/L^−/− mice (Fig. 1). The number of Peyer’s patches visible along the small intestine was reduced from ∼8 to 10 in wild-type mice to ∼1 to 2 in both the β₇^−/− and β₇/L^−/− mice (data not shown). The Peyer’s patches that could be located using a dissecting microscope on the wall of the small intestine in β₇/L^−/− mice were small and hypocellular (Fig. 1,C) compared with wild-type Peyer’s patches (Fig. 1,A). The number of Peyer’s patches in L^−/− mice was not different from that in wild-type mice, and they were similar in size and cellularity to those in their wild-type counterparts (Fig. 1,B). Concomitant with reduced Peyer’s patch size in β₇^−/− mice, the number of HEV observed through intravital observation was reduced compared with that in wild-type mice (data not shown) and the average diameter of those seen was slightly smaller (Table II).

We used blocking mAb to determine which adhesion molecules were mediating the leukocyte rolling in each genotype (Fig. 2). As shown previously (4, 5), leukocyte rolling in Peyer’s patches from wild-type mice was predominantly mediated by L-selectin, since the L-selectin function-blocking mAb MEL-14 blocked most of the observed rolling (Fig. 2,A). However, injection of the mAb RB40.34 significantly reduced the rolling flux fraction by about 29% in wild-type mice (Fig. 2,A). Interestingly, the percentage of granulocytes in the wild-type mice investigated here was about 29 to 38% of the total circulating leukocyte count (Table I), suggesting that many of the cells rolling using P-selectin may be granulocytes. Neutrophils are known to express a constitutively functional ligand for P-selectin called PSGL-1 (19). However, effector T cells also express functional PSGL-1 (20, 21) and may be rolling via P-selectin in Peyer’s patches.

Consistent with the mAb blocking data, the leukocyte rolling flux fraction in L^−/− mice was only about one-third of that found in wild-type mice (Fig. 2,B). Residual rolling in these mice was largely mediated by P-selectin (the mAb RB40.34 to P-selectin blocked almost all leukocyte rolling), but the rolling flux was unaffected by the mAb PS/2 to the α₄ integrin chain (Fig. 2,B). Similar to wild-type mice and consistent with previous results (5), leukocyte rolling in β₇^−/− mice was mediated by L-selectin (Fig. 2,C). P-selectin blockade with the mAb RB40.34 in β₇^−/− mice led to a slight (∼10%), but not significant, reduction in the rolling flux fraction (Fig. 2,C). β₇/L^−/− mice showed a low rolling flux fraction similar to that found in L^−/− mice. This residual leukocyte rolling was totally P-selectin dependent (Fig. 2 D).

Leukocyte rolling in vivo is known to be affected by geometric and hemodynamic conditions such as venular surface-to-volume ratio and average blood flow velocity (9, 22). Therefore, we measured several parameters that could affect leukocyte rolling in these experiments (Table II). Fluorescent beads were used to measure the centerline blood flow velocity in each observed venule. The data from all genotypes were stratified for venules with similar average wall shear rates.

Adhesion molecules involved in leukocyte rolling may change the total number of rolling leukocytes, their rolling velocity, or both parameters (11). Therefore, we also analyzed the velocities of rolling leukocytes in each genotype (Fig. 3). In HEV of wild-type mice, leukocyte rolling velocities ranged from <10 to >100 μm/s (Fig. 3,A) with a median velocity of 43 μm/s (Fig. 3,E and Table III). In L^−/− mice, the population of faster rolling leukocytes was missing (Fig. 3,B), which significantly reduced (p < 0.05 vs wild-type mice) the median rolling velocity to 30 μm/s (Fig. 3,E and Table III). Conversely, slow rolling leukocytes were missing in β₇^−/− mice (Fig. 3,C), which significantly increased (p < 0.05 vs wild-type and L^−/− mice) the median velocity to 85 μm/s (Fig. 3,E and Table III). Taken together, these findings indicate that the leukocyte rolling velocity histogram in wild-type mice is composed of two distinct populations: slower β₇ integrin-dependent rolling and more rapid L-selectin-dependent rolling. It should be noted, however, that the rolling flux fraction in L^−/− mice is much lower (Fig. 2). This suggests that L-selectin may be important in capturing leukocytes from the flowing blood in Peyer’s patches in addition to mediating rolling. It therefore appears that efficient leukocyte attachment requires L-selectin even in the presence of β₇ integrin. In β₇/L^−/− mice (Fig. 3,D), only a residual population of rolling leukocytes was detected (Fig. 2), which rolled at an intermediate velocity with a median of 56 μm/s (Fig. 3,E and Table III), not significantly different from the velocity distribution in wild-type mice. Rolling in β₇/L^−/− mice was mediated by P-selectin (Fig. 2 D). The observed velocity distribution is similar to that seen in venules of the mouse cremaster muscle where rolling is dominated by P-selectin (23).

The velocities of the residual rolling leukocytes was also examined after several blocking mAb treatments. Functional blockade of P-selectin with the mAb RB40.34 in wild-type mice caused a significant increase (p < 0.05) in the median leukocyte rolling velocity from about 43 to about 96 μm/s (Table III). Since blocking P-selectin in wild-type mice also causes an ∼30% reduction in the rolling flux fraction (Fig. 2,A), the velocity increase may be due to the lack of a population of leukocytes that rolls more slowly using P-selectin. In β₇^−/− mice, addition of RB40.34 did not significantly alter the rolling flux fraction (Fig. 2,B) and also did not affect the median rolling velocity (Table III). As shown above, P-selectin mediated almost all the rolling in L^−/− mice (Fig. 2,C). Interestingly, the median velocity of the few remaining rolling leukocytes in these mice was very low (Table III).

mAb blocking data (Fig. 2) and rolling velocity data (Fig. 3) showed that P-selectin contributes to leukocyte rolling in Peyer’s patches. To directly visualize P-selectin expression, we attempted to use standard immunoperoxidase staining using biotinylated secondary Ab and streptavidin-peroxidase conjugate as described previously (24). This method was not sensitive enough to detect P-selectin in Peyer’s patch HEV. To increase the sensitivity, we developed a bead-based assay. Fluorescent beads were coated with the P-selectin mAb RB40.34 and injected i.v. (Fig. 4). Injection of control beads coated with an irrelevant rat IgG Ab showed low, but detectable, background binding to the Peyer’s patch venular endothelium of wild-type mice (Fig. 4,A), whereas the RB40.34-coated beads bound in much greater quantities (Fig. 4,B). Binding was specific for P-selectin on the endothelium because i.v. pretreatment of the mouse with RB40.34 reduced the binding of the RB40.34-coated beads to levels comparable to that of the control beads (Fig. 4,C). As a positive control, the binding of RB40.34-coated beads to the venular endothelium of the acutely exteriorized mouse cremaster muscle was investigated. Under these conditions, cremaster muscle venules, but not arterioles, express P-selectin (24). RB40.34-coated beads bound avidly to the venular (Fig. 4,D), but not arteriolar (data not shown), endothelium in the mouse cremaster muscle. HEV in Peyer’s patches of L^−/− and β₇/L^−/− mice also bound RB40.34-coated beads (data not shown), consistent with the observed P-selectin-dependent rolling in these vessels (Fig. 2).

P-selectin expression is known to be induced by surgical trauma in several models of inflammation (9, 24, 25, 26). We injected both control and RB40.34-coated beads into intact mice before surgery. After waiting 20 min (the time necessary after an injection to clear any circulating beads), the Peyer’s patch was exteriorized and observed with intravital microscopy. Very few beads were bound to the endothelium (data not shown). In contrast, when the same number of beads was injected i.v. 5 to 10 min later (after exteriorization), extensive binding was observed. These data suggest that the P-selectin expression was increased as a consequence of the manipulation of the small intestine during intravital microscopic viewing.

Because we observed that P-selectin was involved in leukocyte rolling in all genotypes (Figs. 2 and 3) and was expressed on the endothelium of Peyer’s patch HEV (Fig. 4), we determined whether purified Peyer’s patch lymphocytes from each genotype exhibited P-selectin ligand activity. A multivalent P-selectin complex was formed by cross-linking a P-selectin-IgG chimera with a FITC-conjugated secondary Ab (15) and was used to investigate P-selectin binding activity of purified Peyer’s patch lymphocytes from wild-type, L^−/−, β₇^−/−, and β₇/L^−/− mice by flow cytometry (Fig. 5). Approximately 26% of wild-type Peyer’s patch lymphocytes bound multivalent P-selectin (Fig. 5,A). A similar number (∼32%) of Peyer’s patch lymphocytes from L^−/− mice also stained positively (Fig. 5,C). In contrast, 45% of lymphocytes from β₇^−/− mice (Fig. 5,B) and 57% of lymphocytes from β₇/L^−/−mice (Fig. 5,D) stained positively for P-selectin ligand activity. Binding was selectin specific, as shown by complete inhibition in the presence of 5 mM EDTA (dotted lines in Fig. 5). As a positive control, we used HL-60 cells, a myeloid precursor cell line known to bind to P-selectin (27, 28). All HL-60 cells stained positively with the P-selectin complex (data not shown).

We have analyzed leukocyte-endothelial interactions in mice deficient for two molecules deemed very important for lymphocyte recruitment to Peyer’s patches: L-selectin and β₇ integrin. Using epifluorescence intravital microscopy of Peyer’s patches in these mice, we have determined that in the absence of L-selectin and β₇ integrin, endothelial P-selectin can mediate a low level of leukocyte rolling. We show expression of P-selectin on the endothelium of Peyer’s patch HEV after surgical manipulation and P-selectin binding activity of lymphocytes isolated from Peyer’s patches. We hypothesize that although P-selectin is probably not involved in recruitment of naive lymphocytes to Peyer’s patches, this molecule may be very important for recruitment of effector T cells to sites of mucosal inflammation.

Although we observed P-selectin-dependent leukocyte rolling in all mouse genotypes studied in this work (i.e., wild-type, L^−/−, β₇^−/−, and β₇/L^−/−), P-selectin has never before been implicated in participating in the recruitment of lymphocytes to Peyer’s patches. In fact, no evidence exists that suggests that P-selectin is even expressed on Peyer’s patch HEV. P-selectin has, however, been shown to be expressed on HEV in inflamed as well as noninflamed palantine and adenoid tonsils (29, 30). Additionally, P-selectin expression has been detected in the small intestine using ¹²⁵I-labeled RB40.34 (31), but was not localized to any specific region (e.g., Peyer’s patches or lamina propria). By using fluorescent beads coupled to the P-selectin mAb RB40.34, we show that P-selectin is indeed expressed on the vascular endothelium of Peyer’s patch HEV under the conditions studied. Our data represent the first evidence for P-selectin expression in Peyer’s patch HEV. More importantly, we also show that P-selectin is functional in these vessels and can mediate rolling after minimal trauma (exteriorization). P-selectin expression is known to occur in response to mediators such as histamine (26, 32) or during inflammation in response to cytokines such as TNF-α and IL-1 (33, 34). Since RB40.34-coated beads did not bind to the vascular endothelium in Peyer’s patches before surgery was initiated, we conclude that this expression is due to up-regulation of surface expression of this molecule after surgical manipulation of the small intestine. Therefore, P-selectin may not be involved in the recruitment of lymphocytes to Peyer’s patches under in situ conditions, but may play a role in trafficking of effector T lymphocytes during trauma or inflammation.

Many types of leukocytes express P-selectin ligand activity. Neutrophils (19), monocytes (35), and eosinophils (36) express a constitutively functional P-selectin ligand (PSGL-1) that mediates at least 90% of the P-selectin binding activity during inflammation in vivo (37). Although all lymphocytes express PSGL-1 (38), only certain subsets, including both effector αβ and γδ T-cells (20) and memory T cells (39), can bind to P-selectin. PSGL-1 functionality is regulated by an inducible fucosyl transferase called FucT-VII (40). Thus, for a T cell to bind to P-selectin, PSGL-1 must have the proper post-translational modifications (41). P-selectin binding to PSGL-1 appears to mediate Th1 cell recruitment into inflamed skin in vivo (42). Of the leukocyte subsets present in the circulation under normal conditions, neutrophils, eosinophils, and effector T cells all can bind to and roll on P-selectin. We therefore suspect that neutrophils, eosinophils, monocytes, and effector T cells are rolling through Peyer’s patch HEV using P-selectin expressed on the HEV endothelium after surgical manipulation.

In vivo differentiation of leukocyte subsets in Peyer’s patch HEV is not achievable with current intravital microscopic tools, so we are unable to determine which types of leukocytes are rolling via P-selectin in this model. Ex vivo labeling of isolated lymphocytes would allow differentiation of lymphocytes from granulocytes, as shown previously (43). However, excessive handling and manipulation of isolated cells may lead to partial activation and phenotypic changes. We were interested in observing leukocyte-endothelial interactions under relatively unperturbed conditions, so isolated, labeled lymphocytes were not used in this study. Nevertheless, we found that a significant number of isolated Peyer’s patch lymphocytes from all genotypes express P-selectin ligand activity. Even though P-selectin expression in Peyer’s patch HEV is below the detection limit of our current methods under in situ conditions, significant numbers of T and B lymphocytes in the Peyer’s patch can bind to this molecule (26–57%). Naive lymphocytes may enter a Peyer’s patch through the use of L-selectin and β₇ integrin and may acquire P-selectin binding activity after differentiation into Ag-specific effector lymphocytes. These T lymphocytes would then be able to traffic to mucosal inflammatory sites using P-selectin, similar to what has been observed for Th1 cell recruitment to inflamed skin (42). The increased fraction of P-selectin-binding lymphocytes found in Peyer’s patch preparations from β₇^−/− mice and β₇/L^−/− mice may be due to an enrichment in the number of effector T lymphocytes (which can bind to P-selectin) in Peyer’s patches of mice lacking β₇ integrin, since fewer naive lymphocytes can enter this tissue when this molecule is absent. A similar enrichment in effector T lymphocytes has been observed in peripheral lymph nodes of L^−/− mice, since few naive lymphocytes can traffic through this tissue in the absence of L-selectin (44). This interpretation is consistent with the recent observation of a shift in lymphocyte subsets in β₇/L^−/− mice⁴.

Our data support a role for β₇ integrin in mediating leukocyte rolling. This evidence comes from the shift in the rolling velocity distribution observed in the absence of β₇ integrin. An increase in rolling velocity when α₄β₇ integrin has been blocked (4) or when β₇ integrin is knocked out (5) has been reported. Our present analysis confirms and expands these data to show that at least three molecules are involved in modulating leukocyte rolling velocity in Peyer’s patches: L-selectin, β₇ integrin, and P-selectin. The rolling velocity distribution in wild-type mice appears to be a superposition of three distributions: a population of leukocytes that rolls rapidly (∼80 μm/s) mainly via L-selectin (seen in the β₇^−/− mice), a population of leukocytes that rolls more slowly (∼30 μm/s) mainly using β₇ integrin (seen in the L^−/− mice), and a smaller population of leukocytes that rolls with an intermediate range of velocities via P-selectin (seen in the β₇/L^−/− mice). These three distributions superimposed can replicate the wild-type distribution measured here. Pure β₇ integrin-dependent leukocyte rolling can be seen in L^−/− mice treated with a P-selectin mAb. In these mice, the residual rolling flux is only 5% of that in wild-type mice, suggesting that L-selectin is important for efficient leukocyte capture in Peyer’s patch HEV. The median rolling velocity in L^−/− mice treated with a blocking P-selectin mAb is on the order of 20 μm/s. A small rolling flux fraction in conjugation with a low rolling velocity suggests that β₇ integrin mainly functions to strengthen rolling interactions (i.e., reduce the rolling velocity), but is unable to efficiently mediate leukocyte capture and initiate rolling.

In conclusion, we present evidence that P-selectin can be induced on Peyer’s patch venular endothelium but is probably not involved in the recruitment of naive lymphocytes to Peyer’s patches. Our data showing P-selectin ligand activity in lymphocytes isolated from Peyer’s patches suggests that effector lymphocytes can bind to P-selectin and may use this molecule for recruitment to mucosal sites of inflammation. Under physiologic conditions, L-selectin and β₇ integrins appear to account for all lymphocyte rolling and adhesion in Peyer’s patch HEV.

We thank John M. Sanders in the Department of Cardiovascular Medicine, University of Virginia (Charlottesville, VA), for help with the paraffin section histology.