Lymphocyte Migration in Lymphocyte Function-associated Antigen (LFA)-1–deficient Mice

A 7-kb PCR product amplified from murine 129/Sv genomic DNA and containing exon 2 of the Lfa-1 gene was subcloned into pSP72. A selection cassette containing stop codons in all three frames, an independent ribosomal entry sequence (IRES) followed by the LacZ gene with an SV40 polyadenylation signal, and a neomycin phosphotransferase gene was introduced into exon 2. Culture and transfection of GK129 embryonic stem (ES) cells was done as described previously (27). EcoRI-digested ES cell DNA was analyzed by Southern blot analysis using probe A, and homologous recombination was confirmed using probe B (see Fig. 1 a). Blastocyst injection and breeding have been described previously (27). Mice were genotyped from EcoRI-digested tail DNA by Southern blot (data not shown) or PCR (Fig. 1 b). A multiplex PCR using primers on either side of the selection cassette and one within the neomycin resistance gene yielded fragments of 90 bp (wild-type) and 660 bp (LFA-1–deficient), respectively.

All mice were kept under specific pathogen–free conditions and in accordance with United Kingdom Home Office regulations. The mice were monitored on a six-monthly basis for a wide selection of parasites, bacteria, and fungi and were regularly found to be pathogen free. LN sections from sentinel mice belonging to litters used in the reported experiments were inspected for the presence of germinal centers by hematoxylin and eosin staining and for positive MAdCAM-1 staining in pLNs (28). These latter indicators of an inflammatory response were routinely negative for mice housed in London and Hamburg.

The following purified rat mAbs were used in this study: LFA-1 α subunit mAbs H35.89.9 (IgG2b) and H68 (IgG2a), both obtained from Dr. Michel Pierres (Centre d'Immunologie, INSERM-CNRS de Marseille-Luminy, Marseille, France); α4 subunit mAb PS/2 (IgG2b); anti–VCAM-1 mAb MK2.7 (IgG1); anti–MAdCAM-1 mAb MECA-367 (IgG2a), obtained from American Type Culture Collection; anti-PNAd mAb MECA-79 (IgM; obtained from Drs. Mark Singer and Steve Rosen, University of California, San Francisco, CA); MECA-325 (IgG1), obtained from Dr. A. Duijvestijn (University of Limburg, Maastricht, Netherlands); CD54 mAb YN1/1.7 (IgG2a; obtained from Dr. Fumio Takei, Terry Fox Laboratory, BC Cancer Research Centre, Vancouver, Canada); CD4, CD8, and B220 mAbs (PharMingen); and control rat IgG1 mAb PyLT-1 (IgG1; Imperial Cancer Research Fund) and control null rat IgG1 and IgG2a mAbs (PharMingen). Secondary detection antibodies were FITC-conjugated goat anti–rat Ig (Jackson ImmunoResearch Laboratories) and biotinylated rabbit anti–rat Ig (Vector Laboratories). Fab fragments were prepared by papain digestion using either the ImmunoPure^® Fab kit (Pierce Chemical Co.) or HPLC to remove any residual intact IgG (7).

Single cell suspensions from 8–12-wk-old mice were prepared by mincing or rubbing between glass slides with frequent rinsing with 5% FCS/RPMI or 0.2% BSA/PBS. Bone marrow cells were harvested by flushing with 5% FCS/RPMI via cut ends of tibias and femurs, followed by disaggregation and filtering through nylon gauze. Cells were washed and incubated at 5 × 10⁶ cells/ml with specific primary mAbs, then FITC-conjugated goat anti–rat Ig (1:200; Jackson ImmunoResearch Laboratories) at 4°C for 30 min. Cells were fixed in 1% HCHO/PBS before analysis on a FACScan™ instrument (Becton Dickinson).

Lymphocytes from pLNs and mLNs of LFA-1^+/+ and LFA-1^−/− mice were prepared and adjusted to 5 × 10⁶ cells/ml in 5% FCS/RPMI for labeling as described (29). LFA-1^+/+ lymphocytes were incubated with 0.2 mM CellTracker™ (CT) Green 5-chloromethylfluorescein diacetate (CMFDA; Molecular Probes) and LFA-1^−/− lymphocytes with 5 mM CT Orange 5-chloromethyl tetramethylrhodamine (CMTMR) at 37°C for 45 min. The two cell preparations were washed and counted, then mixed together to achieve a 1:1 ratio of differentially labeled LFA-1^+/+ and LFA-1^−/− lymphocytes for injection into recipient 8–12-wk-old C57BL/6 mice for a period of 1 h. A correction factor was applied to the subsequent data if the lymphocyte ratio was not precisely 1:1. Reversing the CT dyes had no effect on results. The lymphocytes were frequently tested for activation status before and after the labeling procedure by measuring the level of L-selectin that can be enzymatically cleaved after an activating stimulus (30). In general, a total of 2 × 10⁷ cells in a volume of 200 μl was injected per tail vein. In experiments involving antibody blocking, the cells and mAbs were coinjected. Fab fragments were used in amounts of 250–400 μg per mouse. Irrelevant control Fab fragments were without effect.

Lymphocytes obtained as above were labeled with 20 μCi ⁵¹Cr/ml as described previously (29), and dead cells removed by centrifugation on a 17% Nycodenz cushion (Nycomed). 5 × 10⁶ lymphocytes in 200 μl were injected into the tail vein of recipient 8–12-wk-old C57BL/6 mice. mAbs were coinjected as Fab fragments and at 300 μg/200 μl. Sample groups of four mice were used in each condition. The mice were killed after 1 h, and the distribution of radioactivity was determined for the different organs and for the residual body mass. PBLs were calculated for a blood volume of 1.5 ml.

LN cells were labeled by dissolving 1 mg digoxigenin in 0.5 ml DMSO and incubating the lymphocytes for 15 min at a final concentration of 5 × 10⁷ cells/ml with 60 μg digoxigenin. After washing, 0.8 × 10⁷ cells were injected intravenously, and pLNs were removed 30 min later and frozen in liquid nitrogen. To localize the donor cells in the HEVs, two antigens were revealed simultaneously as described (31). In brief, cryostat sections were incubated with the anti-HEV mAb MECA-325, which was revealed by the alkaline phosphatase anti-alkaline phosphatase (APAAP) technique (blue [31]). The donor cells were identified by a peroxidase-conjugated antidigoxigenin antibody (brown) as described previously (31). The area of the HEVs (excluding the lumen) together with the area within 4-cell diameters surrounding the vessel was then determined (31), and the number of injected cells adhering to and within the endothelium, and within the tissue was analyzed.

Sections of 6 μm were cut from snap-frozen tissues of pLN, mLN, and PP and placed on silane-coated slides which were fixed in acetone for 10 min and allowed to air dry. Slides were washed in standard Tris-buffered saline (TBS), pH 7.6, and primary mAb was added for 1 h at room temperature. mAbs were used at optimal dilution: 10 μg/ml for anti– VCAM-1 mAb MK2.7 (IgG1) and IgG1 isotype control mAb PyLT-1, MECA-79 at 4 μg/ml, and MECA-367 as tissue culture supernatant. Sections were also tested with null IgG1 and IgG2a mAbs (PharMingen; data not shown). Biotinylated rabbit anti–rat Ig (1:100; Vector Laboratories) was added for 35 min. A tertiary stage was carried out using the StreptABComplex/HRP kit following the manufacturer's instructions (K377; Dako), and the slides were incubated again for 35 min before exposure to 1 mg/ml 3,3′diaminobenzidine.

Lfa-1, the gene for the LFA-1 α subunit, was mutated in ES cells using a replacement-type targeting vector and the strategy shown in Fig. 1 a. G418-resistant colonies were identified by Southern blot analysis of EcoR1-digested ES cell genomic DNA using the probe shown in Fig. 1 a. Of 400 G418-resistant ES cell colonies, 10 correctly targeted colonies were identified. Three different homologously targeted ES cell clones were injected into C57BL/6 blastocysts, and all of them were transmitted through the germline. Southern blot analysis of mouse tail DNA enabled identification of LFA-1 wild-type (+/+), heterozygous (+/−), and gene-deleted status (−/−) (data not shown). This was confirmed by PCR analysis (Fig. 1 b).

To establish that the capacity for LFA-1 synthesis had been completely ablated, leukocyte populations were examined for LFA-1 expression and function. Thymocytes from LFA-1^−/− mice were compared with LFA-1^+/+ littermates and shown to be devoid of LFA-1 expression (Fig. 1 c). The lack of LFA-1 surface expression was also demonstrated for leukocytes from other lymphoid tissue, including pLN, mLN, PP, spleen, bone marrow, and blood (data not shown). To test for absence of LFA-1 function, control phorbol ester–treated LFA-1^+/+ thymocytes were adhered to murine ICAM-1–transfected COS-7 cells, and this adhesion could be blocked with anti–LFA-1 mAb H68 (data not shown). In contrast, no adhesion was evident with LFA-1^−/− thymocytes. Therefore, testing for both expression and function of the LFA-1 receptor provided further proof that the LFA-1^−/− mice were totally deficient in this β2 integrin.

Analysis of LFA-1^−/− compared with LFA-1^+/+ mice revealed an increase in spleen size and a decrease in size of the pLNs, as observed previously (25). For 30-g male mice (n = 29), the LFA-1^−/− spleens were 1.7 times larger than those of LFA-1^+/+ mice (182.0 ± 56.4 compared with 107.8 ± 30.0 mg). For 25-g female mice (n = 37), the same comparison yielded a 1.2-fold increase in weight (137.6 ± 33.0 compared with 118.5 ± 42.3 mg). Second, the pLNs from LFA-1^−/− mice were smaller than those from LFA-1^+/+ mice, with an average decreased lymphocyte number for LFA-1^−/− mice of ∼30% that of the LFA-1^+/+ littermates (3.54 ± 0.74 × 10⁶ compared with 12.15 ± 2.90 × 10⁶; n = 9).

To discover whether alteration had occurred in lymphocyte numbers in general or in a particular subtype, we next analyzed CD4 and CD8 T cell subsets and used mAb B220 to detect B cells. In the pLNs, substantial loss in numbers of CD4 and CD8 T cells as well as a deficiency in B cells was observed (Fig. 2). Furthermore, there was no difference in naive versus memory phenotype in LFA-1^−/− and LFA-1^+/+ mice as indicated by expression levels of L-selectin, CD44, and CD45RB antigens (data not shown). In the other LNs, wild-type and LFA-1–deficient mice were similar in terms of lymphocyte numbers and subsets, as was expected from the fact that these LNs were comparable in size between the two types of mice. In spleen, there was a significant increase in CD4 T cells and CD8 cells (Fig. 2).

To gain further information about the diminished cellularity of the pLNs and to investigate trafficking capabilities of LFA-1–deficient lymphocytes, short-term migration studies were performed. The approach taken was to label LFA-1^−/− and LFA-1^+/+ lymphocytes with two distinguishable CT orange and green fluorescent dyes and inject equivalent numbers into the tail vein of C57BL/6 recipients. This allowed direct comparison of the homing activity of the lymphocytes within each LN setting (29). An example of the methodology is illustrated in Fig. 3 a. When the pLN lymphocytes were examined for proportions of CT orange to green after 1 h of homing, it was observed that ∼13% of LFA-1^−/− cells compared with LFA-1^+/+ cells had migrated into the pLNs (Fig. 3 a). We then examined the relative ability of LFA-1^−/− cells to gain entry into other secondary lymphoid tissues (Fig. 3 b). When compared with LFA-1^+/+, the LFA-1^−/− cells were most poorly represented in pLNs (0.21 ± 0.01), followed by mLNs (0.51 ± 0.02) and PPs (0.68 ± 0.03). In contrast, there was a marked increase in the LFA-1^−/− cells in the spleen (1.29 ± 0.04) which is likely due to redistribution of lymphocytes from LN to spleen. These findings on the effect of LFA-1 absence on the trafficking of lymphocytes were confirmed using the technique of tracking ⁵¹Cr-labeled lymphocytes to compare homing in LFA-1^−/− and LFA-1^+/+ mice (data not shown). When CD4, CD8, and B220⁺ lymphocyte subsets were analyzed separately, the ratio of migrated to injected lymphocytes was identical among the subsets whether derived from wild-type or LFA-1^−/− mice, indicating that the lack of LFA-1 caused equal difficulty for all types of lymphocytes to gain entry into the LNs (data not shown).

The above findings showed that LFA-1^−/− cells migrated less effectively, particularly to the pLNs, within the 1-h period but did not reveal the stage at which the lack of LFA-1 had its effect. To discover where LFA-1 deficiency caused difficulty, we turned to histochemistry and examined pLN tissue sections from mice injected with LFA-1^−/− and LFA-1^+/+ lymphocytes for 30 min. As shown in Fig. 4, the LFA-1–deficient lymphocytes bound less well to HEVs than the LFA-1-expressing lymphocytes. The total numbers of lymphocytes at the HEV level and surrounding 4-cell diameters were 285 ± 70 cells/mm² for LFA-1^+/+ (Fig. 4, a and b) and 85 ± 10 cells/mm² for LFA-1^−/− lymphocytes (Fig. 4, c and d) (n = 3). These data suggest that loss of LFA-1 expression reduces lymphocyte adherence to and transmigration across HEVs by 70%, thereby causing substantially diminished recruitment into pLNs. When the cells were divided between those adhering to and within the HEVs and those found within 4-cell diameters beyond HEVs, the following proportions were observed: for LFA-1^+/+ lymphocytes, 73 ± 7 and 27± 2%; for LFA-1^−/− lymphocytes, 78 ± 1 and 22 ± 1%, respectively. These data suggest that the major block is at the level of the HEVs. If the deficiency had been acting selectively at the level of the transmigration step, a larger accumulation of lymphocytes at the HEV level would have been expected.

Although LFA-1 was obviously playing a critical role at the HEV level, a proportion of lymphocytes remained capable of migrating into lymphoid tissue (Fig. 3 b, and Fig. 4). Of other adhesion receptors that might substitute for LFA-1, the α4 integrins were attractive candidates, as they were active in migration in other situations. In the next series of experiments, the migratory behavior of CT-labeled LFA-1^−/− and LFA-1^+/+ cells was compared in host animals simultaneously injected with Fab fragments of α4 mAb or α4β7-specific mAb. The α4 mAb completely prevented the migration of the residual numbers of LFA-1^−/− lymphocytes homing to pLNs, mLNs, and PPs (Fig. 5). In addition, the α4β7-specific mAb blocked lymphocyte entry into pLNs to ∼22% and completely prevented migration into mLNs and PPs. There was no significant effect of α4 or α4β7 mAbs on migration into spleen.

These results were duplicated in a ⁵¹Cr lymphocyte labeling experiment in which α4 mAb abolished all entry into the pLNs, mLNs, and PPs of LFA-1^−/− lymphocytes (data not shown). The α4β7 mAb blocked entry into mLNs and PPs and inhibited entry into pLNs to ∼35% (data not shown). It was possible that the α4-dependent LFA-1^−/− cells migrating into pLNs were a specific subset of lymphocytes expanded in the LFA-1–deficient environment. However, phenotyping of these cells using a third fluorescent tag of Tricolor-conjugated anti–rat Ig showed the migrated LFA-1^−/− and LFA-1^+/+ lymphocytes to have identical phenotypic profiles with regard to their levels of α4 and α4β7 integrins and L-selectin (data not shown).

The suggestion that α4 integrins, α4β7 acting together with α4β1, have a role in migration to the pLNs has not previously been recognized. To further confirm the findings and to gain information about tissues other than secondary LNs, ⁵¹Cr-labeled lymphocytes from normal BALB/c mice were coinjected with 300 μg of Fab fragments from either α4 mAb, LFA-1 mAb, or both into host BALB/c mice. The findings with the single mAbs were as reported previously (7, 10; Fig. 6), but the combination of α4 and LFA-1 mAbs totally prevented migration into pLNs as well as mLNs and PPs (Fig. 6). There was also an appreciable decrease in migration into intestinal tissue and into the “body” (see below). This mAb blockade provoked an increase in circulating blood cells as well as an increase in migration into the spleen. Put together, these observations provide evidence that the α4 integrins operating together with LFA-1 have an essential role in migration to pLNs, as found previously for other secondary LNs.

The identification of the ligand(s) recognized by the α4 integrins, particularly in pLNs, was next explored. Using a MAdCAM-1–specific mAb, homing to both PPs and mLNs was substantially prevented, as described previously (16, 17; Fig. 5). However, the mAb had no significant effect on migration of LFA-1^−/− lymphocytes into pLNs and spleen. VCAM-1 is another ligand recognized by both α4β1 and α4β7 integrins (32, 33), and in the present experiments, the anti–VCAM-1 mAb MK2.7 completely eliminated homing of LFA-1^−/− lymphocytes to pLNs and also substantially blocked entry into both mLNs and PPs although not into spleen (Fig. 5).

The foregoing experiments showed that VCAM-1 has a role in the trafficking into LNs, suggesting that this ligand is indeed expressed on the HEVs. To further address this question, immunohistochemical staining was performed using fresh frozen tissue sections of pLNs, mLNs, and PPs from LFA-1^−/− and LFA-1^+/+ mice and normal C57BL/6 mice. In pLNs, anti–VCAM-1 mAb MK2.7 (IgG1) was identified to label the same HEVs as PNAd-specific mAb MECA-79 (Fig. 7, a and b). Of interest was the observation that the MAdCAM-1 mAb MECA-367 was negative, as was an IgG1 isotype control mAb, PyLT-1 (Fig. 7, c and d). In mLNs where both MAdCAM-1 and PNAd are chiefly coexpressed on HEVs (34), VCAM-1 is also present, whereas the isotype control mAb was negative (Fig. 7, e–h). In a preliminary quantitative study of mLNs, complete overlap was observed in HEV staining of VCAM-1 with one other HEV ligand, MAdCAM-1 (data not shown). The level of VCAM-1 expression was comparable on HEVs from all LNs and from LFA-1^+/+, LFA-1^−/−, and C57BL/6 mice (data not shown). Therefore the absence of LFA-1 did not induce a compensatory increase in expression of VCAM-1.

To confirm that VCAM-1 is expressed on the luminal surface of the endothelium and therefore accessible to recirculating lymphocytes, C57BL/6 mice were intravenously injected for 10 min with rat anti–VCAM-1 mAb MK2.7 or anti-HEV mAb MECA-325. On tissue sections stained with anti–rat Ig, both the pLN and mLN HEVs scored positive to a similar degree for both the anti– VCAM-1 and the anti-HEV mAbs (data not shown). This experiment provided a second type of histochemical proof that VCAM-1 is luminally expressed on the HEVs of the LNs.

In the migration experiments with ⁵¹Cr-labeled lymphocytes, significantly fewer lymphocytes distributed into the mouse carcass after treatment with a combination of LFA-1 and α4 mAbs (see Fig. 6). To discover the identity of the relevant tissue compartment, individual body parts of muscle, heart, kidney, thymus, and bone were isolated and counted separately. The target tissue with diminished lymphocyte migration was determined to be the limb bones, indicating that lymphocyte recirculation was occurring in the bone marrow (data not shown). Within 1 h, a single femur recruited ∼1% of ⁵¹Cr-labeled lymphocytes, suggesting that the total bone marrow compartment attracted lymphocytes at a rate comparable to the 5–10% lymphocyte entry into the combined LNs. Which lymphocytes were migrating into bone marrow was tested by using CT-labeled lymphocytes and subsequent staining with subset-specific mAbs detected with Tricolor-conjugated anti–rat Ig. The ratio of migrated to injected CD4/CD8/B220 subsets was ∼0.5/ 1.0/2.0. This advantage for B cell and disadvantage for CD4 T cells were independent of LFA-1 expression status.

LFA-1^+/+ and LFA-1^−/− cells did not significantly differ in their trafficking to bone marrow (ratio of LFA-1^−/− to LFA-1^+/+ was 1.1; data not shown), suggesting that the migration was not solely reliant on LFA-1. However, when LFA-1^−/− lymphocytes were coinjected with α4 mAb, the trafficking to bone marrow was completely abolished (Fig. 8). These results suggest that migration into bone marrow can be accomplished either by LFA-1 or in its absence by an α4 integrin. As the α4β7 mAb had a partial effect on migration, whereas α4 mAb blocked migration completely, we conclude that in bone marrow, in contrast to the migration into LNs, α4β1 dominates α4β7. Finally, the VCAM-1 mAb, but not MAdCAM-1 mAb, had an inhibitory effect. Considered together, these results demonstrate that migration of recirculating lymphocytes into bone marrow can be mediated by either LFA-1 or the α4 integrins, and, as in pLNs, that VCAM-1 serves as the chief ligand for the latter receptors.

This study shows that LFA-1 has a key role in migration to the pLNs, other LNs, and bone marrow but not into the spleen. Also revealed is a hitherto unrecognized ability of the α4 integrins, α4β7 and α4β1, to compensate for the lack of LFA-1 in lymphocyte trafficking to pLNs and other lymphoid tissues, including bone marrow. In general, these findings highlight common features between lymphocyte homing and the response to inflammatory stimuli and extend the validity of the multistep model of adhesion and transmigration.

The LFA-1^−/− mouse described in this report has, as its key phenotypic characteristic, pLNs that contain ∼30% normal lymphocyte numbers, as also noted previously (25). The decreased trafficking of LFA-1^−/− lymphocytes to the pLNs to ∼20–30% of wild-type lymphocytes suggested that LFA-1^−/− lymphocytes, irrespective of which subset, had difficulty gaining entry to the LNs. This was confirmed by microscopic studies showing that circulating LFA-1^−/− lymphocytes bound poorly to HEVs. Migration into mLNs and PPs was also depressed, but LN cell counts were normal, suggesting compensatory measures were in operation in these tissues but not in pLNs. The decrease in migration of lymphocytes to pLNs is consistent with previous work using function-blocking LFA-1 mAbs (10) and has recently been reported in another study using LFA-1–deficient mice (35). Thus, the general importance of LFA-1 in mature lymphocyte trafficking to secondary lymphoid tissue is confirmed.

In our studies, LFA-1^−/− lymphocytes were able to gain entry into the pLNs, albeit at a lower level, suggesting involvement of further receptors. We here demonstrate these additional receptors to be the α4 integrins. Skewed receptor usage towards increased expression of α4 integrins in LFA-1^−/− mice as a possible cause of experimental bias was excluded (data not shown). The fact that the presence of mAbs to both LFA-1 and α4 integrins caused complete blockade of lymphocyte entry into normal pLNs and other LNs strongly implies that α4 has a critical role in migration of normal lymphocytes into pLNs. That this role for α4 integrin has not previously been observed might be because LFA-1 has a larger role than α4 integrin in adherence to pLN HEVs than HEVs of other organs. This is in good agreement with intravital microscopy studies which show LFA-1 and L-selectin are the essential codependent pLN-specific adhesion pair for lymphocyte adherence to pLN vessels (36). Put together, the data suggest that α4 integrins have a less prominent role in adherence to pLN HEVs, but potentially a larger role in the transmigration step.

Our data also suggest unexpectedly that α4β7 acts as the major α4 receptor in conjunction with α4β1 in lymphocyte recirculation into pLNs. There has previously been no evidence to link α4β7 with pLN migration, although its dominant role in trafficking into mucosal tissue of the mLNs and PPs is well documented (3, 17) and is confirmed here. The integrin α4β1 has usually been associated with inflammatory responses (18–20) rather than with lymphocyte recirculation. However, as these two α4 integrins might be acting either in synergy or in sequence, the extent of the contribution of α4β1 must await a potent murine CD29 blocking mAb. The results imply that the succession of adhesive events in lymphocyte recirculation is similar to that in inflammatory responses, and that the α4 integrins in cooperation with L-selectin and LFA-1 have a more specific and necessary role than previously perceived.

The data presented here identify VCAM-1 and not MAdCAM-1 as ligand on pLN HEVs for α4 integrins in spite of α4β7 involvement. The result was unexpected, as α4β1 has been identified as the major VCAM-1 binding integrin in inflammatory responses (37) in studies using transfectant (38) and cell lines (17). VCAM-1 also made a significant contribution to trafficking into mLNs and, to a lesser extent, PPs. This cooperative activity with MAdCAM-1 was also unexpected, as it has previously been considered that entry into these LNs was regulated only by MAdCAM-1. The findings presented here suggest that potential roles for α4β7 in addition to α4β1 should be sought in other circumstances where VCAM-1 is the major ligand.

One explanation for having overlooked the importance of VCAM-1 as an HEV ligand is its reported absence from normal lymphoid tissue (22, 23, 39), with the exception of a report on its expression on rat HEVs (40). In the present study, expression occurred at comparable levels on HEVs in pLNs, mLNs, and PP LNs and without obvious difference between LFA-1^−/− mice, their wild-type littermates, C57BL/6, or BALB/c mice. Such broad and constant presence of VCAM-1 implies it is constitutively expressed rather than as a consequence of an inflammatory signal or a compensatory mechanism in LFA-1^−/− mice.

The involvement of VCAM-1 raises the issue as to whether lymphocytes can use this ligand for migration across HEVs into the LN as well as adhesion to the luminal surface of HEVs. In this study, the lack of LFA-1 decreased migration across HEVs by ∼25% only, suggesting involvement of other receptor/ligand pair(s) (data not shown). For HUVECs, VCAM-1 is reported to be restricted to an apical location and therefore available for adhesion but not for transendothelial migration (41). However, monocytes can transmigrate HUVECs, making use of α4/VCAM-1 independently of β2 integrins (42). In mice, two major forms of VCAM-1 have been identified, the common seven-domain form and also a three-domain glycosylphosphatidylinositol (GPI)-linked splice variant which is induced by inflammation (43, 44). It is of interest that in polarized epithelial cells, the GPI-linked VCAM-1 was found apically, whereas the seven-domain VCAM-1 was localized to the basolateral surface, where it could theoretically serve as a ligand for transmigration (45). The issue of where VCAM-1 is expressed, particularly in murine HEVs, warrants further investigation.

Migration of recirculating lymphocytes to bone marrow, a primary lymphoid tissue, has been noted in the past (46). Bone marrow serves as a site of hematopoiesis and for B cell maturation in the mammal. It can also act as a reservoir for a primary immune response in the absence of a spleen and presence of L-selectin mAb (47). In this study, we find an unexpectedly high degree of mature lymphocyte recirculation in which B cells dominate into bone marrow. We suggest here that bone marrow, but not the thymus (data not shown), should be considered as a major part of the lymphocyte recirculation network.

Migration to bone marrow is restricted by adhesion mechanisms. Both human (48) and murine (49) hematopoietic progenitors make use of α4 integrin to lodge in the bone marrow. The α4 integrins are also reported to be necessary for correct lymphocyte development (50). We show that normal lymphocyte recirculation through the bone marrow is regulated by LFA-1 and the α4 integrins and, in contrast to the LNs, α4β1 substantially dominates α4β7. In this context, it is again VCAM-1 and not MAdCAM-1 which acts as ligand for the α4 integrins. VCAM-1 is expressed constitutively by bone marrow sinusoidal vasculature (51, 52), and as progenitor cells have been demonstrated to roll on bone marrow endothelial VCAM-1 (48), it can be speculated that other lymphocytes might behave similarly. However VCAM-1 is also expressed constitutively by stromal reticular cells (53, 54), and it is possible that lymphocytes may be retarded not only at the level of the endothelium but also within the bone marrow matrix. Factors produced by the stromal cells might be beneficial for the maintenance of the recirculating lymphocytes. For example, the interaction with VCAM-1 might suppress lymphocyte apoptosis (55). On the other hand, the incoming cells might contribute to the regulation of hematopoiesis.

In summary, the observations presented here extend and validate the concept of a multistage adhesion response requiring selectins and integrins. Rather than exclusive LN-specific use of receptors such as LFA-1 and L-selectin in trafficking to pLNs or α4β7 to mucosal sites, we have presented evidence that LFA-1 and the α4 integrins can operate in migration of unstimulated lymphocytes to the pLNs as well as other secondary LNs and bone marrow. These general features, including the use of VCAM-1 as an α4 integrin ligand, resemble the response to an inflammatory signal. Our findings suggest a unifying hypothesis for migration of lymphocytes across HEVs into secondary lymphoid tissue and also establish further parallels between the activities of adhesion receptors in the “homing” context and the mechanisms used in the more sporadic responses to tissue injury and inflammation.

We thank Ian Rosewell for blastocyst injection, and Tracy Crafton and Peter Hagger for mouse breeding. For carrying out the histochemical staining procedures, we are indebted to Phillipa Munson and Keith Miller, Department of Histopathology, University College London Medical School, and George Elia and Richard Poulsom, Histopathology Laboratory, Imperial Cancer Research Fund. We also gratefully acknowledge the statistical help of Henry Potts, Imperial Cancer Research Fund Medical Statistics Group, Oxford. We thank Britta Engelhardt for the murine ICAM-1 construct; and Mark Singer, Steve Rosen, Fumio Takei, and Michel Pierres for mAbs.

CT

CellTracker™

ES

embryonic stem

HEV

high endothelial venule

HUVEC

human umbilical vein endothelial cell

ICAM

intercellular adhesion molecule

MAdCAM-1

mucosal addressin cell adhesion molecule 1

pLN

peripheral lymph node

PNAd

peripheral node addressin

PP

Peyer's patch

VCAM

vascular cell adhesion molecule