ICAMs Redistributed by Chemokines to Cellular Uropods as a Mechanism for Recruitment of T Lymphocytes

Recombinant human (rh) RANTES (sp. act. 2–5 × 10³ U/mg, purity >97%, endotoxin level <0.1 ng/μg cytokine) and rhMIP-1β (sp. act. 1.6– 2.5 × 10⁴ U/mg, purity >97%, endotoxin level <0.1 ng/μg cytokine) were purchased from R&D Systems (Minneapolis, MN). rhIL-8 (purity >98%, endotoxin level <0.1 ng/μg cytokine), rhMCP-1 (purity >99%), and rhMIP-1α (purity >99%, endotoxin level <0.1 ng/μg cytokine) were obtained from PeproTech EC Ltd. (London, UK). Butanedione monoxime, colchicine, and PMA were purchased from Sigma Chemical Co. (St. Louis, MO).

The anti–ICAM-3 HP2/19, TP1/24, TP1/25, 140.11, ICR2.1, anti–ICAM-1 Hu5/3, anti–CD18 Lia3/2, anti–CD11a YTH81.5, anti–CD43 TP1/36, anti– CD44 HP2/9, anti–CD45 D3/9, anti–CD45RA RP1/11, anti–CD45RO UCHL.1, and anti–l-selectin (CD62L) LAM1/3 mAb used have been described (Spertini et al., 1991; Campanero et al., 1993; Arroyo et al., 1994; del Pozo et al., 1995; Juan et al., 1994; Landis et al., 1994; Luscinskas et al., 1994; Sadhu et al., 1994). P3X63 myeloma protein (IgG₁, κ) was used as negative control.

Recombinant chimeric ICAM-1–Fc and VCAM-1-4D–Fc, consisting of the total extracellular domains fused to IgG₁ Fc fragment, were obtained as described (del Pozo et al., 1995). Briefly, COS-7 cells were transiently transfected with pICAM-1–Fc and pVCAM-1-4D–Fc (ICAM-1 and VCAM-1-4D cDNAs cloned in pCD8IgG1). After 4 d, culture supernatants were precipitated with ammonium sulphate, and thereafter, chimeric proteins were isolated by using protein A coupled to Sepharose (Pharmacia Fine Chemicals, Uppsala, Sweden). Fibronectin was purchased from Sigma Chemical Co. BSA was purchased from Boehringer Mannheim GmbH (Mannheim, Germany).

Resting peripheral blood lymphocytes (PBL) were isolated from fresh human blood by Ficoll-Hypaque density gradient centrifugation (Pharmacia Fine Chemicals), followed by two steps of adherence incubation on plastic flasks (Costar Corp., Cambridge, MA) at 37°C for 1 h. CD45RA⁺ and CD45R0⁺ T cells were prepared by exhaustive negative selection from freshly isolated PBL using anti–CD45RO UCHL.1 or anti–CD45RA RP1/ 11 mAb, respectively, and immunomagnetic beads (Dynal Inc., Oslo, Norway) as described (Vartdal et al., 1987). Paired peripheral blood and synovial fluid (SF) lymphocytes were obtained from patients with rheumatoid arthritis (Department of Rheumatology, Hospital de la Princesa, Madrid, Spain). Human T lymphoblasts were prepared from peripheral blood mononuclear cells by treatment with phytohemagglutinin (PHA) 0.5% (Pharmacia Fine Chemicals) for 48 h. Cells were washed and cultured in RPMI 1640 (Flow Laboratories, Irvine, Scotland) containing 10% FCS (Flow Laboratories) and 50 U/ml IL-2 (kindly provided by Eurocetus). T lymphoblasts cultured for 7–12 d were used in all experiments. T lymphoblasts have been extensively used to study T cell adhesion and activation (van Kooyk et al., 1991; Dransfield et al., 1992; Campanero et al., 1993). Melanoma cancer cells and autologous tumor infiltrating lymphocytes (TIL) were isolated from surgical specimens from patients with metastatic melanoma (Department of Pathology, Hospital Gregorio Marañón, Madrid, Spain). Tumor cells were grown in RPMI 1640/10% FCS. TIL were cultured in AIM-V medium (GIBCO BRL, Life Technologies, Ltd., Paisley, Scotland) containing 4% Ultroser HY (Sepracor Inc., SA, Villeneuve la Garenne, France) and 500 U/ml IL-2. The human dermal microvascular endothelial cell line HMEC-1 has previously been described (Ades et al., 1992) and was kindly provided by Drs. Ades and Lawley (Centers for Disease Control and Prevention and Emory University School of Medicine, Atlanta, GA). HMEC-1 were grown in MCDB 131 medium (GIBCO BRL) supplemented with 20% FCS, 10 ng/ml EGF (Promega Biotech, Madison, WI), 1 μg/ml hydrocortisone (Sigma Chemical Co.), 20 mM Hepes, and 10 mM l-glutamine. Primary cultures of high endothelial cells (HEC) were established from cervical lymph nodes of F₁ hybrid rats, as previously described (Ager, 1987). HEC were grown in RPMI 1640 containing 20% FCS, and confluent cultures were serially passaged and split 1:2. For the experiments described herein, three different HEC cell lines between passage numbers 10 and 20 were used. Previous studies have shown that the interaction between lymphocytes and HEC is independent of the passage number of cell cultures (Ager and Mistry, 1988).

Immunofluorescence experiments were performed essentially as described (del Pozo et al., 1995). Briefly, 0.5–1 × 10⁶ T cells (PBL, SF lymphocytes, TIL, or lymphoblasts) were incubated in flat-bottomed, 24-well plates (Costar Corp.) in a final volume of 500 μl complete medium on coverslips coated with different protein substrata or melanoma tumor cells. Chemokines (10 ng/ml), PMA (20 ng/ml), or 5 μg/ml HP2/19 mAb was added, and cells were allowed to settle in a cell incubator at 37°C and 5% CO₂ atmosphere. After different periods of time, cells were fixed with 3.7% formaldehyde in PBS for 10 min at room temperature and rinsed in TBS (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.1% NaN₃). To visualize different membrane adhesion molecules, cells were stained with specific mAb. After washing, cells were incubated with a 1:50 dilution of an FITClabeled rabbit F(ab′)₂ anti–mouse IgG (Pierce, Rockford, IL) or a 1:500 dilution of a Cy3-goat anti–mouse IgG (Amersham Life Science, Pittsburgh, PA). The cells were observed using a Nikon Labophot-2 photomicroscope with 40, 60, and 100× oil immersion objectives. The proportion of uropod-bearing cells was calculated by random choice of 10 different fields (60× objective) of each condition and direct counting of total cells (400–500) and uropod-bearing cells. Preparations were photographed on either Ektachrome 400 (color pictures) or TMAX 400 (black and white) film (Kodak Co., Rochester, NY). The latter was processed to 800–1,600 ASA with TMAX developer (Kodak Co.).

Samples for confocal microscopy were prepared as for standard immunofluorescence studies. T lymphoblasts were cytoplasmically labeled with the fluorescent probe CFDA-SE, following manufacturer's instructions (Molecular Probes Europe BV, Leiden, The Netherlands). Labeled T lymphoblasts were incubated in flat-bottomed, 24-well plates (Costar Corp.) in a final volume of 500 μl complete medium on coverslips coated with 10 μg/ml ICAM-1–Fc or a confluent monolayer of EC. Chemokines, control antibody (P3X63), or the anti–ICAM-3 TP1/24 or HP2/19 mAb (5 μg/ml) was added, and cells were allowed to settle in a cell incubator at 37°C and 5% CO₂ for 1 h. Cells were then fixed with 3.7% formaldehyde in PBS for 10 min at room temperature and rinsed in TBS (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.1% NaN₃). After washing, cells were incubated with a 1:50 dilution of a Texas red–conjugated goat F(ab′)₂ anti–mouse IgG (Caltag Laboratories, So. San Francisco, CA). Confocal microscopy analysis was performed using the confocal laser scanning system (MRC1000; Bio-Rad Laboratories Ltd., Hertfordshire, U.K.) and a Nikon Diaphot 200 inverted microscope. Images of serial cellular sections were acquired with the BioRad CoMos graphical user-interface and subsequently processed to obtain three-dimentional reconstructions using the VoxelView (Vital Images, Inc., Fairfield, IA) software on an Indigo computer workstation.

Tissue distribution of ICAM-3–bearing lymphocytes was determined using double immunofluorescence staining. Surgical biopsy specimens were snap frozen in liquid nitrogen. 4 μm frozen sections were cut, air dried, and acetone fixed. Sections were overlaid with mAb supernatants and incubated for 30 min at room temperature in a humidified chamber. After two washings in TBS, a 1:50 dilution of Cy3-goat anti–mouse IgG was added. After washing, the sections were saturated with 10% nonspecific mouse serum in TBS. To stain T lymphocytes, FITC-conjugated Leu 4 anti-CD3 mAb was used (Becton Dickinson, San Jose, CA). After washing, the sections were mounted in Mowiol (Calbiochem-Novabiochem, La Jolla, CA), and confocal microscopy analysis of serial tissue sections was carried out.

Video microscopy analysis was performed using a Nikon Diaphot 300 inverted microscope equipped with a Sony SSC-M350CE CCD black and white videocamera coupled to a Sony SVT-5000P time lapse videocassette recorder and a Sony PVM-1453MD video monitor. T cells (PBL, SF lymphocytes, TIL, or lymphoblasts) were incubated for 30 min at 37°C in 35-mm plastic petri dishes (Costar Corp.) previously coated with VCAM-1, ICAM-1 (10 μg/ml), or melanoma tumor cells from a patient, in the presence of chemokines, PMA, or anti–ICAM-3 TP1/24 or HP2/19 mAb. After addition of a second cohort of T cells, cell–cell interactions were filmed for 1 h under phase contrast using a 20× objective. Where indicated, cells of either the first or the second cohort were previously incubated for additional 15 min at 37°C with different blocking mAbs. Images were acquired every 3.2 s, and sequential frames were photographed. In each experimental condition, the number of cells that were attached, spread, and moving on the substrate (phase-dark cells), as well as the number of cell–cell interactions between the uropod of adhered cells and lymphocytes from the second cohort (phase-bright cells), was counted. The recruitment index was expressed as number of cells of the second cohort being captured per number of cells of the first layer adhered to the substrate.

T lymphoblast migration through a confluent monolayer (HMEC-1) was assayed in a Transwell cell culture chamber (Costar Corp.). The chemotaxis chambers are separated by a polycarbonate membrane (6.5 mm diam, 10 μm thickness and 8-μm-diam pore size). Filters were coated on their upper surface with fibronectin and layered with HMEC-1. The lower well of the chamber was filled with HMEC-1 culture medium. The integrity of the endothelial barrier was assessed by diffusion of a trypan blue– albumin complex across the endothelial monolayer as described (Rotrosen and Gallin, 1986). Only cell monolayers with permeability ⩽0.5% were used for the transendothelial migration assays.

All assays were performed in triplicate or quadruplicate chambers, with T cells from a single donor and HMEC-1–covered Transwell filters from the same batch. T lymphoblasts were poured at different concentrations to the upper compartment of the chamber in the presence of either anti– ICAM-3 mAbs TP1/24 (IgG_2a) or HP2/19 (IgG_2a) at 5 μg/ml for 30 min at 37°C in a 5% CO₂ atmosphere. Unbound T cells were removed, and a second cohort of 10^{6 51}Cr-labeled T lymphoblasts was added to the upper compartment of the chamber. After different periods of time (5 min–6 h), the percentage of cells that migrated across the endothelial cell layer was calculated related to the total input of ⁵¹Cr-labeled T cells. In some experiments, after the induction of the uropod formation, cells of the first cohort were incubated for additional 15 min with different blocking mAbs. Where indicated, lymphocytes were pretreated either with 10 mM butanedione monoxime or with 20 μM colchicine for 30 min at 37°C, extensively washed, and used as described above.

HEC were grown to confluence in polystyrene tissue culture chamber slides (Nunc A/S, Roskilde, Denmark). Aliquots of T lymphoblasts were incubated with different mAb (5 μg/ml) during 15 min at 37°C, washed, and poured onto HEC layers. After incubating for different periods of time, unbound lymphocytes were removed by aspiration, and the HEC layers were washed four times in prewarmed HBSS containing 1% FBS, fixed (1% glutaraldehyde in PBS for 20 min), and stained with 1% toluidine blue. Slides were examined by two independent observers using a Nikon Labophot-2 photomicroscope with a 60× objective. Previous studies have shown that surface-bound lymphocytes (small, round, and darkblue cells) and transmigrated cells (large, flattened, and light-blue cells) can be easily distinguished in these slides (Ager and Mistry, 1988). All assays were run in triplicate, and the number of lymphocytes of each type was counted in three separate fields. The migration index was calculated as follows: number of transmigrated cells/number of cells. Preparations were photographed using Ektachrome 400 color film (Kodak Co.).

T cells were collected before and 5 h after transendothelial migration, incubated at 4°C for 30 min in PBS with the primary mAb, and washed two times, and then FITC-conjugated goat anti–mouse F(ab′)₂ (Caltag Laboratories) was used as secondary antibody. The fluorescence of 2,000 cells was measured on a FACS^®can (Becton Dickinson) using Lysis II software.

To assess the spatial disposition of the cellular uropod and the membrane distribution of ICAM-3, confocal microscopy analysis of fluorescent-labeled (green fluorescence) T lymphoblasts, adhering to immobilized ICAM-1 was carried out in the presence of the uropod-inducing anti– ICAM-3 HP2/19 mAb. This mAb was employed since it is able to mimic the effect caused by chemokines on uropod formation (Campanero et al., 1994). We found that no ICAM-3 signal (red fluorescence) was detected in the area of contact between cells and protein substrate (Fig. 1,a, A and B). Interestingly, ICAM-3 fluorescence gradually increased on upper planes of uropods, displaying the highest intensity at a distance of 7–10 μm from the plane of the coverslip, coincident with the tips of uropods (Fig. 1,a, C–F). Similar results were obtained in experiments in which T cells were adhered to EC instead of ICAM-1 (data not shown). Moreover, the same morphological changes and ICAM-3 surface redistribution effect were observed by using chemokines. Unstimulated lymphoblasts were also stained as controls (Fig. 1 b). Hence, these results show that uropods extend away from the large flattened area of cell contact with the substratum, being well exposed to the outer milieu.

The projection of uropod towards outer milieu and its high content of several adhesion molecules further suggested its involvement in the recruitment of free lymphocytes. To address this point we carried out a confocal microscopy analysis of the contacts that occur between adhered, uropod-bearing lymphocytes and other T cells. Interestingly, we found that the anchorage between those cells was located 5 μm above of the plate level and that it was mediated by the ICAM-3–bearing uropod of one cell (Fig. 2, A–D). Three-dimensional reconstruction of all different optical sections of the specimen and rotation of the image clearly confirmed that the intercellular contact was mediated by the uropod (Fig. 2,E). Similar cell–cell contacts were observed when the uropod was induced by the relevant physiological stimuli, the chemokines (Fig. 1 b).

Since confocal microscopy analysis only provided a static view of the possible functional role of uropod, we decided to undertake a dynamic assessment of this issue through a time-lapse phase contrast microscopy analysis. A first layer of T cells was allowed to attach to and spread on an ICAM-1–coated surface, and then these cells were induced to develop uropods with the HP2/19 anti–ICAM-3 mAb or different chemokines. A second cohort of T cells was then added, and cell–cell interactions were filmed. Attached lymphocytes in the first cohort adopted a phasedark morphology. Under conditions of uropod induction, a significant proportion of cells of the second cohort, which exhibited a phase-bright appearance readily distinguishable from the first cohort, were contacted, trapped, and finally engaged by cells of the first cohort (Fig. 3, A and B). Interestingly, the adhered lymphocytes of the first cohort were moving on the ICAM-1–coated surface, displaying the uropod in the antipodean pole relative to the advancing front of the cell, and were transporting other T cells engaged through the uropod (Fig. 3,B). By counting the number of cells of the second cohort that were recruited by cells of the first layer, it was estimated that the induction of the uropod resulted in a 5–10-fold increase in cell recruitment (Fig. 3,C). In contrast, other stimulating agents, such as phorbol esters, that promote integrin activation but are unable to trigger uropod formation, failed to stimulate T cell recruitment (Fig. 3 C), thus underlining the participation of uropods in this phenomenon.

Although T lymphoblasts likely represent the memoryactivated subset of T cells, the number of circulating lymphoblasts is very low. Therefore, we explored whether the mechanism of cell recruitment mediated by uropods could be observed in lymphocytes directly isolated from the blood. Although these cells adhered nicely onto the ICAM-1– coated surface, they moved less than activated T lymphoblasts, and chemokines only weakly induced lymphocyte polarization and recruitment of a second layer of unstimulated peripheral blood lymphocytes (Fig. 4,A, and data not shown). Therefore, we decided to explore whether the recruitment was selectively exerted by a subset of the cells. In this regard, it has recently been reported that most of the chemokine receptors are upregulated on the CD45R0⁺ memory-activated phenotype cells (Loetscher et al., 1996; Mackay, 1996; Qin et al., 1996). Thus, we isolated CD45RA⁺ naive and CD45R0⁺ memory peripheral blood lymphocytes and compared their ability to contact, engage, and drive other lymphocytes. Chemokine-mediated T cell recruitment was much more efficiently promoted by CD45R0⁺ cells than by CD45RA⁺ naive cells (Fig. 4,B). Accordingly, CD45R0⁺ showed a higher polarized morphology, both in basal conditions and after chemokine treatment (Fig. 4,C, b and d). CD45R0⁺ cells developed a proper uropod, and they actively locomote on the ICAM-1–coated surface, whereas the CD45RA⁺ cells only displayed a capping of the ICAM-3 molecule and migrated slower (Fig. 4 C and data not shown). These data indicate that CD45R0⁺ memory T cells responded more efficiently to chemokines likely due to the higher expression of chemokine receptors on these cells and therefore can account for the T cell recruitment observed in the total population of freshly isolated lymphocytes.

We also studied lymphocyte recruitment with physiologically in vivo TIL. These lymphocytes, which infiltrate different tumors, display an in vivo activated phenotype (CD45R0⁺ CD69⁺) and high levels of cytotoxicity against autologous tumor cells (Itoh et al., 1988). Freshly prepared TIL adhered to autologous melanoma tumor cells displayed a polarized morphology with ICAM-3 redistribution to the cell uropod (Fig. 5,a). Moreover, TIL were found to be polarized, and ICAM-3 was found to be concentrated in the area of cell contact in the small aggregates of T cells that infiltrate the melanoma tissue sections (Fig. 5,b). TIL were able to induce T cell recruitment when adhered to either ICAM-1– or FN80-coated surfaces (Fig. 6,a); this phenomenon was also observed when TIL adhered to a monolayer of the tumor cells that these lymphocytes infiltrated in vivo (Fig. 6 b). Additionally, we also explored the in vivo relevance of this phenomenon in the inflammatory disease rheumatoid arthritis. Lymphocytes isolated from either peripheral blood or synovial fluid of these patients were polarized when adhered to ICAM-1 (20 and 77%, respectively) and consequently, were able to recruit peripheral blood lymphocytes of the same patient (recruitment indexes: 0.13 and 0.37, respectively).

A panel of blocking mAbs was used to dissect the adhesion molecules involved in the lymphocyte recruitment mediated through uropods. We found that mAb against ICAM-1 and -3, two molecules concentrated in the uropod, efficiently inhibited lymphocyte recruitment (Fig. 7,A). In addition, the blockade of LFA-1, the counterreceptor for ICAM-1 and -3 (Campanero et al., 1993), also prevented these cell– cell interactions. Moreover, the YTH81.5 antibody that reacts with the I domain of the LFA-1α chain subunit and that selectively blocks the interaction with ICAM-3 but not with ICAM-1 (Landis et al., 1994) only produced a partial inhibition in cell recruitment. Interestingly, antibodies to CD43 and CD44, which are also concentrated in the uropod, as well as to CD45 that it is not redistributed (del Pozo et al., 1995) did not affect lymphocyte recruitment. Although T lymphoblasts express L-selectin (90% positive cells; mean fluorescence intensity, 188, versus negative control 5%; mean fluorescence intensity, 15), a blocking mAb directed to L-selectin did not exert any effect on cell recruitment as measured in the time-lapse videomicroscopy assay (Fig. 7,A). On the other hand, cell recruitment elicited by the HP2/19 anti–ICAM-3 mAb was abolished by coincubation with either of two different blocking anti–ICAM-3 mAbs but not with an anti-CD45 mAb (Fig. 7 B). Together, these data demonstrate that the interaction of LFA-1 integrin with its ligands ICAM-1 and -3, located in the uropod, is critical for the process of lymphocyte recruitment mediated through chemokine-induced uropods.

To explore whether the enhancement in lymphocyte interactions mediated by uropods could result in an increase of the transendothelial migration of the recruited cells, we carried out migration assays using a modified Boyden chamber (Transwell^®; Costar Corp.) and confluent monolayers of dermal microvascular endothelial cells (HMEC-1). We found that the induction of uropod formation in the first layer of T cells, adhered to EC, resulted in a significant enhancement of transmigration of ⁵¹Cr-labeled T cells of the second cohort (Fig. 8,A); this effect was maximal at very short time periods (45–60 min) and was detectable as early as after 15 min (Fig. 8,B). Since lymphocyte migration was similar at 4–6 h, irrespective of the addition of uropod-inducing mAb, it seems that the presence of uropods accelerates transendothelial migration rather than causes an enhancement of its overall magnitude. As expected, the enhancement of transmigration was dependent on the number of cells in the first layer of lymphocytes. Thus, 1.8 × 10⁴ cells/ mm² in the first layer under conditions of uropod induction, produced about a threefold enhancement in migration, compared to 4.5 × 10³ lymphocytes/mm² (Fig. 8 C).

When the role of uropod in lymphocyte migration was studied by the migration assay described by Ager and Mistry (1988), using HEC, similar results were found. In this assay, T cells were coincubated with HEC, and after different periods of time, the number of lymphocytes bound to the apical surface or underneath the HEC monolayer were counted. We have also found that the induction of the uropod caused an important increment in the migration index in short periods of time, and that this effect vanished at longer times (Fig. 9,b). Interestingly, we observed that in these assays the interaction between migrated lymphocytes and bound T cells was also mediated through ICAM-3–bearing uropod structures (Fig. 9, a and c), indicating that this interaction takes place during transendothelial migration.

The role of uropod in lymphocyte migration was confirmed with blocking experiments. We found that the increase in the migration was virtually abrogated by inhibition of uropod function in the first layer of T cells by using blocking anti–ICAM-3 mAb (Fig. 10,A). In addition, the prevention of uropod formation with the myosin-disrupting drug butanedione monoxime also blocked the increment in transendothelial migration of the second cohort of T cells (Fig. 10 B). This drug has previously been reported to prevent lymphocyte polarization and redistribution of adhesion molecules to the uropod (Campanero et al., 1994; del Pozo et al., 1995).

When we analyzed the expression of the adhesion molecules that redistribute to uropods before and after migration across HMEC-1, we found that they did not show any quantitative variation in their level of expression after lymphocyte transmigration; moreover, migrated cells exhibited a higher polarized morphology (Table I).

Cell polarization is a complex phenomenon that results in asymmetrical cell shape common to many cell types and has a key role in cellular processes as diverse as activation, differentiation, localized membrane growth, cell migration, vectorial transport of molecules across cell layers, antigen presentation, and target cell recognition and killing (Rodriguez-Boulan and Powell, 1992; Huttenlocher et al., 1995; Drubin and Nelson, 1996; Helander et al., 1996; Lauffenburger et al., 1996; Negulescu et al., 1996). Lymphocytes are among the diverse cellular types that undergo polarization; activated and migrating lymphocytes change from a spherical to a polarized cell shape, displaying a cytoplasmic projection termed uropod due to its similarity with amoeboid morphology (Haston et al., 1982; Wilkinson, 1986; Wilkinson and Higgins, 1987; Shutt et al., 1995). We have previously described that chemotactic and proinflammatory chemokines are responsible for the induction of the uropod on T cells adhered to endothelial cells, as well as for the redistribution of certain adhesion molecules to this structure (del Pozo et al., 1995). In this regard, it has been reported that these cytokines induce leukocyte chemotaxis and regulate adhesive properties of leukocyte integrins (Tanaka et al., 1993a,b; Taub et al., 1993a,b; Gilat et al., 1994; Schall and Bacon, 1994; Carr et al., 1996); however, the mechanisms responsible for these effects are poorly understood. We have found herein that chemokine-mediated cell polarization and adhesion molecule redistribution to the cellular uropod confers T lymphocytes the ability to capture additional cells through this structure. This phenomenon may constitute a cooperative mechanism that can significantly contribute to the directional guidance of leukocytes to inflammatory foci induced by chemokines.

We previously reported that uropods induced by both chemokines and activating mAb appear to emerge away from the flattened cell body that is adhered to EC or ECM substrate (Campanero et al., 1994; del Pozo et al., 1995; Sánchez-Mateos et al., 1995;). This point was inferred from conventional fluorescence microscopy studies in which we moved the plane of focus through the specimen. Very interestingly, we have confirmed our early observations by using confocal laser scanning fluorescence microscopy. This analysis allowed us to definitively conclude that the uropod is directed to the outer milieu, being projected from the flattened cell body, and furthermore, that this structure is able to mediate contacts among lymphocytes.

However, since confocal microscopy analysis only provided us a static approach to the understanding of the functional role of the uropod, several key questions remained undetermined. Therefore, we decided to undertake a dynamic assessment of this issue by using a time lapse videocassette recorder coupled to a phase contrast microscope. Interestingly, this analysis showed that attached lymphocytes are able to move over the substrate displaying the uropod at the back of the cell, emerging from the most distal (antipodean) pole to the leading edge of the cell. Hence, this cellular projection corresponds exactly to the clasically defined lymphocyte uropod (Lewis, 1931; Haston et al., 1982). It is worth mentioning that we clearly observed that the cells captured through uropods were tightly engaged through these cellular structures and therefore migrated together with the uropod-bearing cell. Very importantly, we have found that this lymphocyte recruitment seems to be indeed relevant in vivo since it was also observed when uropods were promoted by physiologic stimuli such as chemokines. This phenomenon was strictly dependent on the presence of the uropod, since it did not occur upon cell stimulation with other agents that induce cell and integrin activation like phorbol esters, but was unable to induce uropod formation. These data strongly suggest that lymphocyte–lymphocyte interactions are cooperative for cell recruitment to sites where EC have been activated, bearing chemokines on their cell surface. On the other hand, and although there is no precedent for capture and transport of one leukocyte by another during transendothelial migration, there are several recent papers reporting on the synergism between leukocyte–leukocyte interactions as a novel regulatory mechanism in recruitment of leukocytes in inflammation (Bargatze et al., 1994; Alon et al., 1996; Jutila and Kurk, 1996; Finger et al., 1996; Fuhlbrigge et al., 1996). L-selectin plays an important role in these cell–cell interactions as determined by blocking studies with specific antibodies. However, in our experiments the blockade of L-selectin did not interfere with lymphocyte recruitment, since we have looked at the tight adhesion between lymphocytes (β₂ integrin/ICAM-1,-3–dependent) rather than at the rolling phase observed under flow conditions (selectin-dependent; Springer, 1995; Butcher and Picker, 1996). Therefore, our results do not rule out the possible involvement of L-selectin and its carbohydrate ligands in the process of lymphocyte recruitment under flow conditions. In this regard, it is worth mentioning that the ligand of P-selectin PSGL-1, which also binds L-selectin, has been detected on the uropods of activated, polarized neutrophils (Lorant el al., 1995).

Altogether our findings on cell recruitment suggested that T cell interactions through the uropod can significantly increase migration through an endothelial barrier and underlying ECM to inflammatory foci. To further support this point we performed specific transendothelial migration assays. Under conditions of uropod formation, lymphocyte migration through dermal microvascular endothelial cells was clearly enhanced. It is worth mentioning that most of the previously reported assays of lymphocyte chemotaxis have been performed using inert polycarbonate filters. At best, endothelial cells derived from the umbilical cord vein (HUVEC) have been used by some authors (Roth et al., 1995; Randolph and Furie, 1996) in an attempt to model the blood vessel wall. The use of HUVEC increases the signal to background ratio of chemotaxis and provides a more physiologic model of lymphocyte emigration from the bloodstream than the conventional in vitro chemotaxis assays lacking endothelium (Roth et al., 1995). However, it is evident that lymphocyte extravasation occurs primarily through postcapillary venules rather than through large vessels such as the umbilical vein. To our knowledge, this is one of the first reports employing microvascular endothelium specialized in diapedesis in chemotaxis assays. Furthermore, we have also studied lymphocyte migration through high endothelial venules. The endothelial cells from high endothelial venules stem from specialized postcapillary vascular sites in lymphoid organs that are able to support high levels of lymphocyte extravasation (Ager and Mistry, 1988; Girard and Springer, 1995). Interestingly, we found that in both systems, uropods significantly enhance T cell transmigration at very short time periods. In this regard, TNF-α stimulation of HMEC-1 also rendered an initial increment in lymphocyte migration, whereas at longer periods of time the transmigration level was equal to nonstimulated EC (not shown). Therefore, different phenomena such as the generation of lymphocyte uropods or activation of HMEC-1 by TNF-α rendered a similar effect in transendothelial migration. It is worth mentioning the existence of contacts through ICAM-3–bearing uropods during cell transmigration. Unfortunately, we could not fully assess the effect of exogenously applied chemokines on uropod formation in our migration assays, since the presence of chemokines in the upper compartment of the chamber would impair lymphocyte transendothelial migration towards the lower compartment by an inversion of the chemotactic gradient generated by the dermal microvascular endothelial cells. However, our clearcut results on the effect of chemokines in T cell recruitment strongly suggest their involvement in transendothelial migration.

Lymphocytes that are recruited into inflamed tissues bear preferentially the memory/activated CD45R0⁺ phenotype (Carr et al., 1994). It is now becoming clear that within the T lineage, most of the chemokine receptors are restricted to CD45R0⁺ cells (Loetscher et al., 1996; Mackay, 1996). T cells that migrate in response to RANTES, MCP-1, and MIP-1α are mainly CD45RA⁻ CD45R0⁺ memory T cells (Qin et al., 1996). In this context we have found that these chemokines act preferentially on freshly isolated CD45R0⁺ cells inducing uropod formation, and therefore, these cells are able to recruit other T cells through the uropod. Furthermore, physiologically in vivo activated CD45R0⁺ T lymphocytes isolated from inflammatory infiltrates of rheumatoid arthritis patients and TIL displaying a highly polarized morphology were able to recruit T cells very efficiently. Altogether, these results provide evidence that the mechanism of uropod-mediated lymphocyte recruitment has in vivo physiologic relevance.

In additional experiments of cell recruitment, we have observed that this phenomenon is not only induced when lymphocytes are attached to adhesion molecules expressed by inflamed endothelium (VCAM-1 or ICAM-1) but also to extracellular matrix proteins such as fibronectin, as shown for TIL. In this regard, it is well known that the chemotactic gradients produced by chemokines occur underneath the endothelial barrier, and we have recently found that uropods are efficiently promoted by chemokines on lymphocytes adhered to several ECM substrates (del Pozo et al., 1995). Moreover, the adhesion molecules that redistribute to uropods do not show any quantitative variation in their level of expression after lymphocyte migration across HMEC-1, and migrated cells exhibit a higher polarized cell morphology. Therefore, it is conceivable that uropods could also play a role in leukocyte recruitment once cells have extravasated and are migrating towards the inflammatory foci. In this regard, this issue is underscored by the observations that TIL adhering to a monolayer of the original melanoma tumor cells are able to recruit other lymphocytes through the uropod. Lymphocyte recruitment mediated by uropods could therefore represent a general mechanism of major importance during a wide variety of immunological processes in which cell migration is involved.

This work was supported by a grant SAF96-0039 from Plan Nacional de Investigación y Desarrollo, 07/44/96 from Comunidad Autónomia de Madrid, and from Asociación de la Lucha contra el Cancer to F. SánchezMadrid and by fellowships from Fondo de Investigaciones Sanitarias BAE 96/5357 (to M.A. del Pozo) and Comunidad Autónoma de Madrid (to M.C. Montoya).