Antibody Binding to a Conformation-Dependent Epitope Induces L-Selectin Association with the Detergent-Resistant Cytoskeleton¹

Recruitment of leukocytes from the blood into tissues is controlled by a variety of adhesion molecules on the surface of the endothelium and circulating leukocytes, one of which is L-selectin (1). L-Selectin is important in the normal migration and circulation of both memory and naive lymphocytes (2, 3, 4) and is central to leukocyte-endothelial cell interactions, including neutrophil rolling along the vessel wall (5). We and others have also shown that L-selectin and corresponding ligands are important in leukocyte-on-leukocyte rolling interactions that may amplify further recruitment of leukocytes from the blood into tissues (6, 7, 8, 9, 10). A common feature of all adhesive interactions mediated by L-selectin is that they occur under flow in the bloodstream. Indeed, threshold levels of shear promote adhesion through L-selectin (11, 12).

L-Selectin is constitutively expressed on most circulating leukocytes, but is uniquely regulated when the cell becomes activated. After activation of lymphocytes or myeloid cells with chemotactic factors or phorbol esters, there is a rapid increase in functional avidity (13) which, under most circumstances, is followed by proteolytic cleavage of the protein from the cell surface (14, 15, 16). The basis for the increased functional activity in L-selectin immediately after activation of the leukocyte is poorly understood. It may be due to phosphorylation of L-selectin or other proteins, dimerization, hyperthermic conditions, cytoskeletal association of L-selectin through its cytoplasmic tail, or conformational changes in the protein (17, 18, 19, 20). Endoproteolytic release of L-selectin from the surface of leukocytes is regulated by structural features of the L-selectin protein (21, 22). Calmodulin, an intracellular calcium-regulatory protein, specifically coprecipitates with L-selectin through a direct association with the cytoplasmic tail, and calmodulin inhibitors disrupt L-selectin-dependent adhesion by inducing proteolytic release of L-selectin from the cell surface (23).

Many reports suggest that L-selectin can also function as a signal transduction molecule. Cross-linking of human L-selectin with mAbs leads to neutrophil activation as measured by Ca²⁺ flux, superoxide generation, increased adhesiveness, and activation of intracellular protein pathways, such as tyrosine phosphorylation and mitogen-activated protein (MAP)⁴ kinase production (24, 25, 26, 27, 28, 29). Cross-linking of L-selectin also potentiates the response of neutrophils to formyl peptides (30). In most studies, cross-linking of L-selectin by primary anti-L-selectin mAb followed by a secondary reagent is requisite for signaling, although important exceptions exist. For example, signaling through L-selectin can also be induced by sulfatides, which bind L-selectin (31). Further, a mAb directed against a highly conserved region of L-selectin can signal and cause increased adhesion of lymphoid cells transfected with human L-selectin cDNA, neutrophils, and lymphocytes, in the absence of a cross-linking secondary Ab (32). Therefore, it is clear that L-selectin can act as a signaling molecule under certain conditions.

The cytoplasmic tail of L-selectin is vital to its function during leukocyte rolling and adhesion (33). Deletion mutants lacking the carboxyl-terminal 11 amino acids of the cytoplasmic tail of L-selectin do not bind to high endothelial venules (HEV) and do not establish rolling interactions in vivo (33). Subsequent work demonstrated a direct link between the cytoplasmic tail of L-selectin and the cytoskeletal proteins α-actinin and vinculin (34); recently, L-selectin was shown to associate dynamically with the cytoskeleton (20). Ab cross-linking, hyperthermic treatment, and ligand binding studies demonstrate that the cytoplasmic tail is important to the function of L-selectin by regulating linkage to the actin cytoskeleton through direct binding of the cytoplasmic tail (20). However, the nature of the linkage has not been clearly shown, and it is not known what predisposes L-selectin to associate with the cytoskeleton.

Here, we have examined the effect of treating leukocytes with anti-L-selectin mAbs that recognize highly conserved and functionally important epitopes on L-selectin expression. We show that one mAb (LAM1-116), which binds an epitope in the lectin domain (32), causes a structural change in human, bovine, and ovine L-selectin in the absence of cellular activation that is detected by increased staining of a second anti-L-selectin mAb, EL-246. The induced conformation predisposes L-selectin to associate with the detergent-resistant cytoskeleton when the EL-246 epitope is engaged. Because both mAbs bind functional epitopes on L-selectin, this type of structural regulation may be important in L-selectin function.

Holstein calves, 1 wk to 3 mo old, and 18-mo-old sheep, housed in the Montana State University large animal facility, and healthy human donors were used as sources of peripheral blood, which was collected by venipuncture into citrate or heparin anticoagulant tubes as previously described (35). Total leukocytes were harvested using a hypotonic solution for 10 s followed by rapid dilution in either HBSS or PBS (Sigma, St. Louis, MO) and centrifuged at 200 × g for 5 min. The process was repeated as necessary to rid preparations of RBC.

The following mAbs were used. Anti-L-selectin mAbs included DREG 55, DREG 56, DREG 110, DREG 152, and DREG 200 (36); LAM1-1, LAM1-5, LAM1-101, LAM1-102, LAM1-104, LAM1-108, LAM1-110, LAM1-115, LAM1-116, LAM1-118, LAM1-119, LAM1-120, LAM1-126 (32); EL-246, and GD 4.22 (7, 35); and Leu-8 (Becton Dickinson, Mountain View, CA). Anti-CD18 mAbs included R15.7 (gift from R. Rothlein, Boehringer Ingelheim, Ridgefield, CT), IB4 (American Type Culture Collection (ATCC), Manassas, VA: ATCC HB-10164), and MHM-23 (Dako, Carpinteria, CA). Anti-CD11b mAb Leu-15 (Becton Dickinson) was also used. Other mAbs included HECA 452 (anti-CLA (37)), Hermes-3 {anti-CD44 (38)), GD 3.5 (anti-γδ T cell (unknown γδ T cell-specific marker (39)), GD 3.8 (anti-γδ TCR (40)), and EL-112 (anti-E-selectin (41)).

The mouse pre-B 300.19 cells transfected with either full length human L-selectin cDNA (300.19/L-selectin) or a deletion mutant lacking the carboxyl-terminal 11 amino acid residues of the cytoplasmic domain (300.19/LΔcyto) have been previously described (33). The selectin transfectants used in the Ab mapping studies were described elsewhere (41).

Isolation of leukocytes and flow cytometric analysis were as described (35, 42). Briefly, 1 × 10⁶ leukocytes were incubated with 1 μg LAM1-116 or other mAbs at 37°C or 4°C, or in the presence of 10–100 mM sodium azide (Sigma), 1–100 mM herbimycin A (Calbiochem, La Jolla, CA), genistein (Calbiochem), or calpeptin (Calbiochem) for 15 min, or buffer alone. After incubation with mAb, the cells were placed on ice and FITC-conjugated (Molecular Probes, Eugene, OR) EL-246, biotin-conjugated (Pierce, Rockford IL) EL-246, or other FITC- or PE-conjugated mAbs were added. Cells were incubated with mAbs for 30 min on ice and washed in PBS with horse serum (FACS buffer), and staining measured on either a FACScan or FACSCalibur (Becton Dickinson, Mountain View, CA). Data were collected from 10,000 cells, and mode fluorescence staining values were reported in table form or as representative histograms. L-Selectin levels were also measured by an indirect stain on the 300.19 L-selectin transfectants, as described (35). LAM1-116 and DREG 56 Fab treatment of leukocytes were performed as above.

Flow cytometric analysis of L-selectin association with the detergent-insoluble cytoskeleton was as described (43, 44), with minor exceptions. Specifically, leukocytes were harvested and isolated as described above and incubated with LAM1-116 or other anti-L-selectin mAbs at 37°C for 15 min. The cells were washed in FACS buffer and either treated with another anti-L-selectin mAb for 15 min at 37°C, or treated directly with 0.5% Nonidet P-40 lysis buffer (150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, 10 mM HEPES, 2% goat serum, 0.5% Nonidet P-40) for 15 min at room temperature in the absence of Ab cross-linking, as described (44). Mock buffer without Nonidet P-40 was used as a control. The cells were washed in FACS buffer and incubated with PE-labeled goat anti-mouse F(ab′)₂ (Jackson ImmunoResearch Laboratories, West Grove, PA) for 30 min on ice. Staining was measured by gating on the detergent-insoluble cytoskeletal fraction using forward and side scatter. The gated population was confirmed by phalloidin-FITC (Sigma), which specifically stains the detergent-insoluble cytoskeleton (44).

Monovalent Fab were generated from whole IgG molecules by papain digestion with the ImmunoPure Fab Preparation Kit from Pierce, per manufacturer’s instructions. To confirm Fab production, 8% SDS-PAGE gels were run under nonreducing and reducing conditions, and molecular mass was determined by Coomassie stain. In some cases, the Fab preparation was also incubated with protein G beads (Boehringer Mannheim, Mannheim, Germany) to reduce the levels of contaminating Fc fragments.

Bovine lymphocytes were isolated as described (39), treated with mAbs, incubated with 0.5% Nonidet P-40 lysis buffer for 10 min at room temperature, and stained with PE-labeled goat anti-mouse F(ab′)₂ (Jackson ImmunoResearch) secondary Ab on ice for 30 min. Cells were washed in FACS buffer and incubated with phalloidin-FITC (Sigma) for 20 min on ice. Cells were washed a final time in FACS buffer and placed in 16-well, LabTek glass chamber slides (Nunc, Naperville, IL) for microscopic examination. Fluorescent microscopy was performed using a super high pressure mercury lamp power supply (Nikon, Melville, NY) model HB-1013AF, linked to a Nikon inverted microscope (Eclipse TE300), and digital data captured using a Spot digital imaging system (Diagnostic Instruments, Sterling Heights, MI). Results were visualized at ×400 magnification. The fluorescence micrographs were representative examples of four separate experiments and are sorted into phalloidin-FITC positive (cytoskeletal) and PE-positive (L-selectin) micrographs.

mAb EL-246 was covalently linked to activated cyanogen bromide (CNBr) Sepharose 4B beads (Pharmacia, Uppsala, Sweden), which were then blocked with 1 M glycine at 25°C for 3 h (45). Bovine lymphocytes were surface labeled with biotin (Pierce) as described (39), and detergent lysates were prepared from these cells. The preparations were incubated with unlabeled CNBr beads for 2 h at 25°C as a preclearing step (39). The lysates were drained from the CNBr columns and incubated without mAb, or with LAM1-116, other anti-L-selectin mAbs, or irrelevant mAbs at 37°C for 15 min. After incubation, equal quantities of EL-246-labeled CNBr beads were added to each lysate, and L-selectin was immunoprecipitated either at 4°C overnight or at 37°C for 2.5 h. The precipitated beads were washed three times with wash buffer (45), mixed with reducing or nonreducing buffer, boiled for 3 min, and loaded directly onto an 8% polyacrylamide gel. Gels were electrophoresed, and the proteins were transferred to a polyvinylidine difluoride membrane (Bio-Rad, Hercules, CA) overnight at 4°C. Proteins were visualized using a streptavidin HRP (Amersham, Little Chalfont, U.K.) reaction and ECL detection system (Amersham) and developed on X-OMAT (Kodak, Rochester, NY) film. Densitometric analysis was performed on the intensity of the immunoprecipitated L-selectin bands, which were identical with the L-selectin positive control (GD4.22) using 1-D MULTI on an Alpha Innotech IS-1000 Digital Imaging System. Protein G (Boehringer Mannheim) immunoprecipitation using anti-L-selectin mAb GD 4.22 was used as a positive control for L-selectin immunoprecipitation, as described (39). A comparison of bands that resulted from the EL-246 precipitation of the various treated lysates, which were developed under identical exposure times, is shown.

Pretreatment of human neutrophils or bovine lymphocytes with mAb LAM1-116 at 37°C for 15 min enhanced the subsequent staining of L-selectin by FITC-labeled EL-246 mAb, as measured by flow cytometric analysis (Fig. 1). The same effect was seen with all leukocyte cell types tested from humans, cattle, and sheep (data not shown). The increased staining by EL-246 after leukocyte treatment with the LAM1-116 mAb was dose dependent with 1 μg/L LAM1-116 mAb giving optimal results (data not shown). Importantly, the increase in EL-246 staining was not due to increased expression of L-selectin protein, because the staining of four other FITC- or PE-labeled anti-L-selectin mAbs was unaffected (Fig. 2,A). This effect was also detected by EL-246 labeled with biotin (Table I). Moving human or bovine leukocytes from 4°C to 37°C for 15 min greatly reduced the staining of L-selectin by EL-246 in the absence of L-selectin shedding (Table I, and data not shown). In contrast, staining by other anti-L-selectin mAbs was not altered by this treatment (data not shown). Binding studies with biotinylated EL-246 demonstrated that EL-246 did not bind to LAM1-116 (data not shown).

To determine the specificity of the LAM1-116 mAb-induced up-regulation of the EL-246 epitope, human and bovine leukocytes were pretreated with 19 other anti-L-selectin mAbs, including at least 6 reactive with the same domain as LAM1-116, as well as mAbs directed to other surface Ags, including R15.7 (anti-CD18), HECA 452 (anti-CLA), Hermes-3 (anti-CD44), GD 3.5 (unknown ligand on γδ T cells) and GD 3.8 (anti-γδ TCR). None of the other anti-L-selectin mAbs or other mAbs listed above induced an increase in EL-246 staining (Fig. 2,B and data not shown). In similar experiments, LAM1-116 pretreatment had no effect on the expression level of other anti-L-selectin mAbs or of other surface Ags, such as CD18 and TCR (Fig. 2,A and data not shown). To be confident that staining of L-selectin by FITC-labeled EL-246 was specific, excess (10×) unlabeled EL-246 or excess isotype-matched (10×) GD 3.8 (anti-γδ TCR) Abs were added along with FITC-labeled EL-246 as above. The addition of excess unlabeled EL-246 mAb effectively blocked the staining of FITC-labeled EL-246 (Table I). However, treatment with excess GD 3.8 mAb did not inhibit staining of FITC-labeled EL-246 to an appreciable extent (Table I).

To determine whether intact LAM1-116 was requisite for enhanced EL-246 reactivity, a Fab of LAM1-116 was generated by papain digestion, as described in Materials and Methods. The Fab of LAM1-116 was able to enhance EL-246 epitope expression, although to a lesser extent than the whole IgG Ab (Fig. 3). These data suggested that LAM1-116 did not induce cross-linking of adjacent L-selectin molecules given that monovalent Fab do not have a second binding site needed for effective cross-linking. The larger increase in EL-246 staining seen with the IgG was likely due to the ability of dual binding sites to act on L-selectin and enhance the EL-246 epitope more effectively. To ensure the specificity of the LAM1-116 Fab interaction, Fab of DREG 56 mAb were generated; and although the Ab preparations stained L-selectin, they did not increase EL-246 reactivity in parallel assays (data not shown).

Tyrosine kinase activity may play a role in the function of L-selectin (25). To address whether tyrosine kinases are involved in up-regulation of the EL-246 epitope, we tested the effects of two tyrosine kinase inhibitors, herbimycin A and genistein, to see whether they could block the increased presentation of the EL-246 epitope in the presence of the LAM1-116 mAb. As shown in Table I, neither herbimycin A or genistein blocked the LAM1-116 enhancement of the EL-246 epitope. No inhibition of EL-246 epitope enhancement was seen over a concentration range of 1–100 mM of each inhibitor. Thus, tyrosine kinase activity seems to have no role in the conformational change induced in L-selectin by the LAM1-116 mAb. To further investigate whether other signaling events in the cell were responsible for the up-regulation of the EL-246 epitope, studies were undertaken at 4°C, and in the presence of sodium azide, which blocks electron transport and eventual ATP production. Both treatments had no inhibitory effect on the ability of LAM1-116 to induce the conformation that increases the expression of the EL-246 epitope (Table I). Therefore, enhancement of the EL-246 epitope was not dependent on common signaling events.

EL-246 mAb recognizes a conserved epitope on both E- and L-selectin. Initial mapping studies using chimeric selectins comprised of L- and P-selectin extracellular domains suggested that the short consensus repeat (SCR) domain of L-selectin was important for optimal EL-246 mAb binding (35). Additional chimeric selectin molecules were generated, expressed, and analyzed for EL-246 staining to gain further insight into the nature of the EL-246 epitope. As expected, EL-246 bound recombinant L and E-selectin (Table II, LLL and EEE), but not P-selectin (data not shown). However, staining of multiple chimeric molecules, generated by swapping different domains of each selectin, revealed that the EL-246 epitope on L-selectin is unlikely to be primarily dependent on protein sequences provided by its respective domains. For example, EL-246 failed to stain a transfectant expressing the lectin domain of P-selectin, and epidermal growth factor (EGF) and SCR domains of L-selectin, as well as other chimeras with different selectin domains (Table II, PLL, LLP, and LEE). EL-246 was preferentially reactive with E-selectin over L-selectin, as originally reported (35, 42), because EL-246 reactivity with EEL was 7-fold higher than reactivity with LLL, whereas EEL and LLL were bound similarly by a mAb that binds the SCR domains of L-selectin (Table II). Reactivity with E-selectin followed expression of the lectin domain, although the constructs also contained the EGF and SCR domains of either L- or E-selectin. For example, all three E-selectin transfectants were stained brightly by EL-246 (ELL, EEL, and EEE). Highest reactivity was seen when the L-selectin SCR domain was combined with the E-selectin lectin and EGF domains (EEL). However, because of the structural conservation between all of the selectins, it was not possible to rule out the prospect that sequences within domains other than the lectin domain also contributed to mAb specificity. Nonetheless, these results are consistent with the EL-246 mAb requiring an appropriate conformation for binding to either E- or L-selectin.

Because the EL-246 epitope appears to require a unique conformation of L-selectin, we tested whether an intact cytoplasmic tail domain, which can link to the cytoskeleton, was required for optimal staining. L-Selectin transfectants that lack the carboxyl-terminal 11 amino acids of the cytoplasmic tail domain (LΔCyto) of L-selectin do not bind to HEV in vitro, nor do they exhibit rolling interactions in vivo in exteriorized rat mesenteric venules, suggesting that the cytoplasmic domain of L-selectin regulates leukocyte adhesion by controlling cytoskeletal interactions and/or receptor avidity (33). Staining of the EL-246 epitope on L-selectin in the LΔCyto vs the wild-type L-selectin transfectants was reduced (Fig. 4A). Indeed, EL-246 staining of the majority of the LΔCyto mutant cells fell below the upper threshold of background staining. In comparison, LAM1-116 staining of the LΔCyto mutants remained high (Fig. 4,B). To test whether or not LAM1-116 could enhance the expression of the EL-246 epitope on the cytoplasmic tail mutants, the mutants were pretreated with LAM1-116 parallel to the studies described above. Interestingly, EL-246 epitope expression could be enhanced by LAM1-116 pretreatment on both the wild-type and the LΔCyto mutants (Fig. 5). There was a dramatic up-regulation of the EL-246 epitope on the cytoplasmic tail mutants as the EL-246 epitope expression increased well beyond the minimal staining levels seen in Fig. 4. These data suggest that an intact cytoplasmic tail of L-selectin is not needed for the LAM1-116 induction of the conformational change that increases EL-246 epitope expression.

Recently, L-selectin was shown to associate dynamically with the actin cytoskeleton under a variety of conditions (20). Using a similar approach as these authors, we examined the effect of EL-246 and LAM1-116 on the association of L-selectin with the detergent (0.5% Nonidet P-40)-resistant cytoskeleton. Specifically, cells were treated with primary anti-L-selectin mAbs as described in Materials and Methods, subjected to 0.5% Nonidet P-40 buffer and L-selectin staining of the detergent-insoluble cytoskeleton visualized by goat F(ab′)₂ anti-mouse Ab labeled with PE. Thus, if cytoskeletal association of L-selectin was detected, it was due to a direct effect of the mAbs on L-selectin and not to cross-linking because the cells were treated with detergent before the addition of the secondary Abs. This is in contrast to the recent report by Evans and colleagues in which cross-linking by second stage Ab was a prerequisite for Ab-induced cytoskeletal association of L-selectin (20).

Treatment of human and bovine lymphocytes with the LAM1-116 or the EL-246 mAbs alone did not induce association of L-selectin with the detergent-resistant cytoskeleton (Fig. 6,A). However, in the presence of LAM1-116, EL-246 induced L-selectin association with the cytoskeleton without subsequent cross-linking of the surface protein (Fig. 6 B). Other anti-L-selectin mAbs tested did not induce L-selectin association in tandem with LAM1-116 mAb treatment, suggesting that EL-246 mAb-induced cytoskeletal association was specific to the LAM1-116/EL-246 interaction with L-selectin (data not shown). Cross-linking studies parallel to those done by Evans et al. did show L-selectin cytoskeletal association, but the association was no more dramatic than LAM1-116/EL-246 Ab treatment alone (data not shown and Ref. 20).

Cytoskeletal association of L-selectin was also visualized by two-color fluorescence microscopy by PE-labeled secondary Abs and phalloidin-FITC, which specifically stains the actin cytoskeleton (44). Parallel to our flow cytometry results, engagement of the EL-246 epitope in the presence of the LAM1-116 mAb caused L-selectin to associate with the detergent-resistant cytoskeleton (Fig. 7,B; note red staining). The micrographs are representative of four separate experiments in which red staining represents the presence of L-selectin associated with the cytoskeleton (Fig. 7, B, D, and F) and green staining represents phalloidin-FITC staining of the actin cytoskeleton (Fig. 7, A, C, and E). Phalloidin-FITC staining was specific because it did not stain viable, non-detergent-treated cells (data not shown). Also, non-detergent-treated lymphocytes exhibited a normal, punctate L-selectin staining pattern (data not shown). As seen with our flow cytometry data (Fig. 6), neither the LAM1-116 mAb nor the EL-246 mAb alone induced L-selectin association with the cytoskeleton (Fig. 7, D and F; note lack of red staining), although detergent-resistant cytoskeletons were generated (Fig. 7, C and E; green phalloidin-FITC staining). These data suggested that the conformational change in L-selectin induced by the LAM1-116 mAb predisposes L-selectin to cytoskeletal association.

To test whether or not the intact leukocyte was needed for a conformational change to take place in the L-selectin protein, bovine lymphocytes were surface labeled with biotin, and detergent lysates were prepared. The biotin-labeled lysates were divided into fractions, which were treated with the LAM1-116 mAb, DREG 56 mAb, a negative control mAb Ab, EL-112, or no Ab at all for 15 min at 37°C. The treated lysates were then immunoprecipitated with EL-246 covalently attached to Sepharose 4B beads at either 4°C or 37°C and electrophoresed under reducing and nonreducing conditions. As shown in Fig. 8a, lane 8, EL-246 weakly immunoprecipitated L-selectin from control lysates analyzed under nonreducing conditions. In contrast, treatment of lysates with LAM1-116 increased the amount of L-selectin subsequently immunoprecipitated by EL-246 (lane 7) when compared with another anti-L-selectin mAb DREG 56 and the negative control, an irrelevant, isotype-matched mAb, EL-112 (Fig. 8,A, lanes 5–7). Similar results were seen when the immunoprecipitated proteins were analyzed under reducing conditions, although the background was higher under these conditions (lanes 1–4). Protein G bead immunoprecipitation of L-selectin with anti-L-selectin mAb GD 4.22 (anti-bovine L-selectin) was used as a control for L-selectin precipitation (Fig. 8,A, lane 9). All lanes are from a single gel at the same exposure. EL-246 precipitation of L-selectin also resulted in other bands, seemingly distinct from L-selectin. Similar bands were also precipitated with DREG 56 (Fig. 8 A, lane 6).

Densitometric analysis was used to quantify the differences in band intensity in the reducing and nonreducing gels. Analysis of the bands showed that LAM1-116 mAb treatment of the detergent lysates increased the amount of L-selectin immunoprecipitated by EL-246 mAb 4- to 5-fold compared with nontreated or other mAb-treated fractions (Fig. 8 B)

In this report, we demonstrate a conformation-induced change in L-selectin mediated by a mAb that recognizes a functionally conserved epitope on L-selectin. The mAb, LAM1-116, generated in L-selectin deficient mice, specifically caused a conformational change in L-selectin as measured by an increase in FITC-labeled EL-246 staining of L-selectin (Fig. 1). This effect was specific to mAb LAM1-116. Nineteen other anti-L-selectin mAbs did not induce the EL-246 epitope, nor did other Abs against distinct cell surface markers, such as the CD18 integrins and CD44 (Fig. 2,A and data not shown). Also, LAM1-116 induction of the EL-246 epitope was specific to the EL-246 mAb. Other mAbs against L-selectin, as well as mAbs directed against other cell surface markers, were not enhanced by treatment with LAM1-116 (data not shown and Fig. 2 B). These data relating to the specificity of this interaction are important because others have shown that L-selectin engagement by some mAbs can activate cellular processes and cause integrin-dependent adhesion of neutrophils under flow (29, 32).

As stated previously, EL-246 recognizes a conserved epitope on both L-selectin and E-selectin (35). Past mapping studies using L-selectin/P-selectin chimeras suggested that the EL-246 epitope requires the SCR domains, as well as the lectin domain, for optimal Ab binding (35). At that time, it was proposed that the EL-246 epitope was possibly contained within the L-selectin SCR domains. These studies were extended here, and we now show that none of the domains of L-selectin can, by themselves, confer the EL-246 epitope. Staining of a number of recombinant selectins containing different domains of L-, E-, and P-selectin show that optimal presentation of the EL-246 epitope requires the correct spatial orientation of the protein, obviously requiring the SCR and lectin domains of E- or L-selectin, and can be enhanced with LAM1-116 binding of L-selectin.

Recently, studies have been done that analyze L-selectin function within the cell. Studies done by Pavalko et al. (34) first described a linkage between the cytoplasmic tail of L-selectin and cytoskeletal proteins, and this observation was shown to be functionally important in intact cells by Evans et al. (20). The latter was the first report of dynamic cytoskeletal association of L-selectin with the actin cytoskeleton, presumably by a linkage of α-actinin with the cytoplasmic tail of L-selectin and the actin cytoskeleton. Our studies show that the conformation-induced change in L-selectin, caused by LAM1-116, predisposes the protein to associate with the detergent-resistant cytoskeleton, which is triggered by EL-246 (Figs. 6 and 7). Thus, we have expanded on the data presented by Evans et al. and demonstrate that a specific conformation of L-selectin predisposes the protein to cytoskeletal association in the absence of cross-linking (Figs. 6 and 7; Ref. 20). We propose that this conformational change regulates L-selectin such that a high affinity binding epitope, the EL-246 epitope, is increased in expression and that when this epitope is engaged dynamic cytoskeletal association takes place. These data parallel other reports of conformation playing a role in L-selectin function. Studies on L-selectin shedding have shown that the cleavage site recognized by the membrane-bound metalloprotease has relaxed sequence specificity, suggesting that the conformation of L-selectin is the most important factor in protease recognition and that this conformation may be regulated by the presence of calmodulin binding to the cytoplasmic tail (21, 22, 23).

As stated earlier, others have shown that L-selectin can act as a signal transduction molecule (24, 25, 26, 27, 28, 29, 30, 31, 32). However, a signaling event through L-selectin is not responsible for the increased binding of EL-246 to L-selectin. The event occurs at 4°C and in the presence of sodium azide and tyrosine kinase inhibitors. Furthermore, LAM1-116 increased the expression of the EL-246 epitope on L-selectin in detergent lysates. This latter finding is important because EL-246 does not recognize shed, soluble L-selectin very well and is, generally, a poorly immunoprecipitating Ab (M. A. Jutila, unpublished observation). A previous report by Schleiffenbaum et al. (46) has shown that soluble L-selectin is conformationally distinct from cell surface L-selectin which may be the reason for poor L-selectin immunoprecipitation by EL-246, although EL-246 binding was not tested in their study. LAM1-116 likely alters this distinct conformation of soluble L-selectin through a direct change in or, perhaps, aggregation or clustering of the molecule. If aggregation occurred, increased coprecipitation of nonspecifically associated molecules with the aggregates could have resulted, although it has been shown that LAM1-116 does not cause dimerization of L-selectin and most likely does not cause clustering (18). However, what these observations clearly show is that the cell is not needed for the enhancement of EL-246 reactivity with L-selectin and that a cellular signal is not required for increased presentation of the EL-246 epitope. Thus, if our hypothesis is correct, overt cellular signaling may not be needed for increased functional activity of L-selectin as others have proposed.

The cytoplasmic tail of L-selectin is important for adhesive events including leukocyte rolling, receptor avidity, and cytoskeletal association of L-selectin (20, 33). Thus, we tested whether or not the expression of specific epitopes on L-selectin are regulated by the cytoplasmic tail. In comparison with wild-type L-selectin transfectant staining, EL-246 staining of LΔCyto L-selectin was reduced to minimal levels (Figs. 4 and 5). In contrast, staining of the LAM1-116 epitope on the LΔCyto transfectants remained high (Fig. 4,B). However, in the presence of LAM1-116, the EL-246 epitope could be enhanced on the LΔCyto transfectants (Fig. 5), suggesting that an intact cytoplasmic tail is not required for the conformational change induced in L-selectin and the subsequent increase in the EL-246 epitope expression.

On the basis of our data, we propose that initial engagement with ligand induces a conformational change in L-selectin, leading to exposure of a high avidity binding site (EL-246 epitope) for the original ligand, or possibly a second ligand, to bind and induce stronger tethers and slower rolling. In effect, our data also suggest that LAM1-116 binding to L-selectin may make it more “E-selectin-like.” E-selectin is an inducible member of the selectin family of adhesion molecules and is expressed on the cell surface at its highest levels 4–6 h after an activation signal, such as TNF-α. It is synthesized de novo after activation and is expressed in a functional form that allows immediate capture of leukocytes from the bloodstream. After the primary activation signal on the endothelium, no further activation is needed to mediate E-selectin capture of leukocytes. E-selectin mediates a more avid interaction than L-selectin, because leukocyte rolling is slower on the former in assays done under similar shear forces (47). In vivo, L- and E-selectin can recognize the same naturally occurring ligands on HEV (48, 49). Past studies in our laboratory have demonstrated that moving E-selectin-expressing cells to 37°C does not decrease EL-246 mAb staining. In addition, EL-246 preferentially binds to E-selectin vs L-selectin in competitive binding assays (42). If this model is correct, it could explain why EL-246 is effective as a therapeutic agent during ischemia and reperfusion injury, as well as other shear-dependent events (50, 51, 52, 53, 54, 55). Presently, we are not able to functionally test the hypothesis that LAM1-116 induces higher ligand binding avidity, because both mAbs (LAM1-116 and EL-246) and native ligand block the function of L-selectin. However, studies are under way comparing binding affinities of EL-246 for E-selectin and LAM1-116-bound L-selectin.

Here, we report that a conformational change in L-selectin induced by an artificial ligand may play an important regulatory role in the molecular interaction of L-selectin with its ligands and predisposes the protein to cytoskeletal association upon engagement of a second functionally important epitope. The functional importance of the conformational change described is currently under investigation. Our report provides additional insight into the structural features of L-selectin and may lead to a better understanding of L-selectin/ligand interactions.

We thank Keri Csencsits for her help with the fluorescent microscopy and Ginger Perry and Larissa Turk for their expert technical assistance. We also thank M. Kemal Aydintug for his helpful comments on the manuscript.