Leukosialin (CD43, sialophorin) redistribution in uropods of polarized neutrophils is induced by CD43 cross-linking by antibodies, by colchicine or by chemotactic peptides

Neutrophil responses to activating stimuli, such as adhesion, diapedesis, locomotion, chemotaxis and phagocytosis, require changes in cell morphology (Keller et al., 1991; Stossel, 1989). These involve membrane projections and cellular deformation, which are generated and maintained by the cytoskeleton (Stossel, 1988).

In dynamic cells such as neutrophils, little is known about connections between the plasma membrane proteins and the underlying cytoskeleton, because of the transient nature of cell surface movements and shape changes (Bourguignon, 1989; Luna and Hitt, 1992; Sheterline and Hopkins, 1981). The main neutrophil transmembrane receptors, which were reported to be involved in cytoskeleton reorganization, are adhesion molecules, such as integrins (Hughes et al., 1992; Sharma et al., 1995) and selectins (Pavalko et al., 1995), and the receptor for chemotactic formylpeptides (fMLP-R) (Howard and Oresajo, 1985; Jesiatis et al., 1993). During neutrophil adhesion, interactions of integrins or selectins with actinbinding molecule result in outside-in and inside-out signalling together with a reorganization of the submembranous actinbased cytoskeleton into focal adhesions (Clark and Brugge, 1995). Bacterial chemotactic peptides such as fMLP induce multiple signalling events, among which a transient association of the fMLP-R with the cytoskeleton and a subsequent cell polarization with topographic reorganization of the receptor (Jesiatis et al., 1984; Sullivan et al., 1984).

We here investigated the possibility that leukosialin (sialophorin, CD43) could be involved in neutrophil cytoskeleton-mediated cellular deformations. Leukosialin is a major mucin-like transmembrane glycoprotein of leukocytes (Remold-O’Donnell and Rosen, 1990; Remold-O’Donnell et al., 1987). Although CD43 is generally described as an antiadhesive molecule due to its negative charge and elongated structure (Cyster et al., 1991; Remold-O’Donnell and Rosen, 1990), several data suggest an active participation of CD43 in leukocyte functions. Indeed, it appears as an accessory molecule in T-lymphocyte proliferation induced by anti-CD3 or lectins (Sperling et al., 1995) and anti-CD43 mAbs trigger energydependent leukocyte homotypic aggregation (Kuijpers et al., 1992; Rosenkranz et al., 1993) and enhance hydrogen peroxide production by monocytes (Nong et al., 1989). Furthermore, CD43 cytoplasmic domain is highly conserved among various species, which suggests an important function for this molecule, common to all leukocytes (Shelley et al., 1989).

Interactions of CD43 with the cytoskeleton were initially suggested by: (i) an unusual insolubility in Triton X-100 of a 140 kDa glycoprotein of human neutrophils, most probably identical to CD43, following its cross-linking by wheat germ agglutinin (Suchard and Boxer, 1989) and of a rat thymocyte sialoglycoprotein LSGP, later identified as CD43 (Turner et al., 1988); (ii) the redistribution, during thymocyte mitosis, of CD43 in the cleavage furrow together with the actin-binding proteins ezrin/radixin/moezin (Yonemura et al., 1993). Finally, while this work was in progress, Sanchez-Mateos et al. (1995) reported a redistribution of the CD43 molecule into uropods during T-cell adhesion in the presence of anti-CD43 mAbs, further supporting the hypothesis of an association of CD43 with the cytoskeleton.

Cytoskeleton movements also participate in transduction mechanisms (Forgacs, 1995; Rosette and Karin, 1995) and it is worth noting that anti-CD43 mAb cross-linking, in mononuclear leukocytes, results in intracellular Ca²⁺ mobilization, PKC translocation and phosphoinositide hydrolysis (Silverman et al., 1989; Wong et al., 1990). Furthermore, CD43 intracellular portion is constitutively phosphorylated and is hyperphosphorylated during PKC activation by phorbol esters (Axelsson and Perlmann, 1989; Chatila and Geha, 1988; Piller et al., 1989).

In the present study, we analysed neutrophil responses to CD43 cross-linking by mAbs, supposed to mimic the unknown ligands of this membrane glycoprotein, to colchicine, a microtubule-disrupting agent eliciting cell polarity and chemokinesis, or to a chemotactic agent such as fNLPNTL. Our results indicate that CD43 interacts either directly or indirectly with microfilaments and redistributes in a cap structure. Interestingly, this redistribution of CD43 molecules occurs with or determines cell polarization, since capping by cross-linked anti-CD43 mAbs was associated with cell polarization and, conversely, cell polarization by colchicine or by fNLPNTL resulted in the clustering of CD43 molecules into caps, always located in the cell uropod.

Anti-CD43 mAbs MEM59 and L60 were from STC Diagnostics (Bethlehem, PA) and Becton Dickinson (San Jose, CA), respectively, while L10 was a generous gift from Eileen Remold O’Donnell (Center for Blood Research, Boston). F(ab′)₂ fragments of these antibodies were obtained by digestion with 2%, w/w, pepsin (Worthington, NJ) followed by the absorption of undigested IgGs and Fc fragments with Protein A-Sepharose (Sigma Chemical Co, St Louis, MO). The purity of F(ab′)₂ fragments was assessed by PHAST gel chromatography (Pharmacia, Uppsala, Sweden). Anti-CD11b (Bear 1) and anti-CD16 (3G8) were from Immunotech (Marseille, France), anti-CD18 was a generous gift from Samuel Wright (Rockfeller University). Control mouse IgG1 was from Immunotech (Marseille, France). FITC- and TRITC-conjugated affinity-purified goat F(ab′)₂ and Fab′ anti-mouse IgGs were from Caltag Laboratories (San Fransisco, CA). TRITC-labelled phalloidin was from Molecular probes (Eugene, OR). Streptavidin TRITC-conjugated was from Harlan Sear-Lab (Sussex, England), while anti-α-actinin clone BM-75.2, rabbit polyclonal anti-myosin, deoxyribonuclease I type II-S from bovine pancreas (DNase I), cytochalasin B, colchicine, benzamidine, trypsin inhibitor type I-S from soybean, leupeptin, AEBSF, were all purchased from Sigma. fNLPNTL (For-Nle-Leu-Phe-Nle-Tyr-Lys-OH) was purchased from Bachem (Bubendorf, Switzerland).

Neutrophils were prepared at room temperature from EDTA-anticoagulated blood from healthy adult volunteers. After centrifugation for 20 minutes at 120 g, the platelet-rich plasma was discarded and neutrophils were isolated by a one-step density gradient centrifugation, on Polymorphprep (Nycomed, Oslo, Norway) according to the manufacturer’s instructions. Residual erythrocytes were lysed in 0.2% NaCl for 1 minute and the osmolarity of the medium then equilibrated by the addition of an equal volume of 1.6% NaCl. Cells were washed and resuspended in Hanks’ balanced salt solution (HBSS) without Ca²⁺/Mg²⁺ (Gibco, Paisley, Scotland).

Cells were lysed in extraction buffer containing 0.5% Triton X-100 (TX-100), 100 mM NaF, 50 mM KCl, 2 mM Mg²⁺, 20 mM Pipes, pH 6.8 (Lacy and Underhill, 1987) and protease inhibitors (2 mM AEBSF, 10 mM benzamidine, 100 i.u./ml aprotinin, 1 mg/ml soybean trypsin inhibitor). After 10 minutes of incubation on ice, lysates were fractionated by centrifugation at 11,600 g for 10 minutes at 4°C. The TX-100 soluble supernatant (soluble fraction) was collected and the pellet (insoluble fraction containing cytoskeletal proteins), washed once with the extraction buffer, was finally resuspended in boiling solubilization buffer (50 mM Tris-HCl, pH 8.8, 5 mM EDTA, 1% SDS) and sonicated. When mentioned, 4 mg/ml DNase I was included in the TX-100 extraction buffer and, in that case, detergent extraction was performed for 45 minutes at room temperature.

Protein contents of the TX-100 soluble and insoluble fractions were measured by the Bio-Rad Detergent Compatible Assay Kit (Bio-Rad, CA) according to the manufacturer’s instructions.

Soluble and insoluble fractions, equivalent to 1.5×10⁵ cells per well, were analysed by SDS-PAGE on a 7.5% polyacrylamide gel under reducing conditions and analysed by western blotting as described (Remold-O’Donnell et al., 1987) with anti-CD43 mAb clone L60, followed by biotinylated goat anti-mouse IgG secondary antibody and alkaline phosphatase-labelled biotinylated streptavidin (Amersham, Buckinghamshire, UK) and revealed with Bio-Rad phosphatase substrates (Bio-Rad, CA). To enhance the signals of insoluble fractions, a horseradish peroxidase-labelled secondary antibody was used, revealed with a chemiluminescent detection ECL kit (Amersham, England).

PMN were incubated for 30 minutes at 4°C in HBSS without Ca²⁺/Mg²⁺ containing 1% bovine serum albumin and 10 μg/ml of either anti-CD43 mAbs or of control mouse IgG1. After two washes in HBSS without Ca²⁺/Mg²⁺, FITC-labelled goat F(ab′)₂ anti-mouse IgG or, when mentioned, Fab′ fragments, were added for 30 minutes at 4°C. In capping experiments, 10⁶/ml labelled cells were then resuspended in HBSS with Ca²⁺/Mg²⁺ and incubated at 37°C for various times. When mentioned, either 2×10⁻⁵ M cytochalasin B, 2×10⁻⁵ M colchicine or 10⁻⁶ M fMLP (for the 10 last minutes of incubation at 37°C) were added. Cells were then fixed for 10 minutes at room temperature with 3.7% formaldehyde, cytocentrifuged for 2 minutes on a slide and mounted in Fluoprep (BioMerieux, France).

The intracellular labelling of F-actin was achieved as described (Keller and Niggli, 1995). After incubation with antibodies to CD43 or colchicine, neutrophils were fixed by the addition of an equal volume of 8% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4, for 10 minutes at 37°C. Cells were then centrifuged and washed in Tris-buffered saline (TBS), pH 7.8, containing 0.04% human serum albumin (HSA). After centrifugation, cells were mixed with a solution of TBS containing rhodamine-phalloidin 5 units/ml (from a stock solution in methanol, after evaporation), 100 μg/ml of lysolecithin and 0.04% HSA for 10 minutes, followed by the addition of an equal volume of 4% HSA in TBS, pH 7.8, and centrifugation. Cells were finally washed in TBS.

Myosin labelling was performed according to the method of Keller (Keller and Niggli, 1993) with minor modifications. Briefly, 10⁶ cells/ml were fixed in methanol containing 5 mM DFP for 15 minutes, washed and cytocentrifuged on a slide. Incubation with 1:10 dilution of normal goat serum (NGS) (Sigma) and 5% skimmed milk, was carried out for 30 minutes at room temperature. Slides were washed with PBS and reacted with a dilution of rabbit polyclonal anti-myosin (Sigma) in PBS containing 0.5% skimmed milk and 1:10 NGS for an additional 30 minutes. After washes in PBS, secondary FITC-conjugated goat anti-rabbit antibodies, diluted in PBS, 1:10 NGS and 0.5% skimmed milk, were added for 30 minutes. Cells were finally washed, mounted in fluoroprep and examined with a fluorescent Leica DMRB microscope (Leica mycroscopy, Heerbrugg, Switzerland) using narrow range filters (FITC selective L4, rhodamine filter N2.1, Blue/green/red: B/G/R, Leica microscopy).

Capping of CD43 was performed as described above, except that the secondary antibody was an immunogold-conjugated (10 nm) goat anti-mouse IgG (British BioCell International, Cardiff, UK). After 30 minutes of incubation at 37°C, neutrophils were fixed by 1.25% glutaraldehyde in 0.1 M phosphate buffer for 30 minutes at 4°C. After three washes, cells were fixed with acid osmic, then alcohol dehydrated and embedded in Epoxy resin. Ultrathin sections were counterstained with uranyl acetate and lead citrate. Cells were observed with a Philips CM10 electron microscope.

Cell solubilization in TX-100 buffer results in a soluble fraction containing cytosolic and solubilized membrane constituents and an insoluble fraction containing the nucleus, cytoskeletal insoluble components and associated membrane molecules.

To analyse a possible interaction of CD43 with the cytoskeleton, neutrophils were extracted in 0.5% TX-100 buffer in the presence or absence of the F-actin depolymerizing agent DNase I, to further assess the role of the F-actin network in the TX-100-insolubility. As shown by western blot analysis (Fig. 1a), CD43 of control neutrophils (C) was mainly TX-100 soluble (s), except for a small proportion which was retained in the detergent-insoluble fraction (i); this TX-100 insoluble CD43 fraction was more readily detected with a chemiluminescent amplifying system (Fig. 1b). TX-100 insolubility was F-actin dependent, since it was prevented by the addition of DNase I in the extraction buffer (Fig. 1b) or by cell pre-incubation with cytochalasin B (an F-actin depolymerizing drug), prior to TX-100-extraction (Fig. 1c). These results show that a small fraction of CD43 is directly or indirectly associated with microfilaments.

Cell incubation with anti-CD43 mAbs, for 30 minutes at 37°C before extraction, enhanced the proportion of TX-100 insoluble CD43 (Fig. 1a,c). There again, addition of DNase I in the extraction buffer or cell preincubation with cytochalasin B (Fig. 1c) decreased the CD43 insoluble fraction. Cell interaction with anti-CD18 (IB4), anti-CD11b or anti-CD11a and with a secondary antibody did not modify the proportion of insoluble CD43 (data not shown).

When anti-CD43 mAbs were cross-linked with secondary antibodies before extraction, all CD43 molecules became immediately insoluble in TX-100 independently of a cytoskeleton association, since it still occurred at 4°C (data not shown) and in the presence of cytochalasin B or DNase I (Fig. 1d).

CD43 distribution at the cell surface was analysed by immunofluorescence microscopy. As shown in Fig. 2A,A′, CD43 was uniformely distributed on the surface of resting neutrophils, either immediately fixed and labelled with mAb to CD43 and secondary FITC-labelled antibodies, or labelled at 4°C before fixation (data not shown). However, CD43 redistribution in a cap occurred when cells, labelled at 4°C, were incubated at 37°C for 30 minutes (Fig. 2B,B′). This redistribution did not involve interactions of antibodies with Fcγ-receptors, since a similar capping was observed with F(ab′)₂ fragments of anti-CD43 mAbs (MEM59 or L60 clones) cross-linked with F(ab′)₂ secondary antibodies (data not shown). All tested anti-CD43 mAbs (MEM59, L60, L10) induced the capping of CD43, when cross-linked by secondary antibodies at 37°C (data not shown). CD43 molecules were first redistributed in small patches, which were then clustered into a single cap. This capping was time-dependent and reached 50±10% (mean ± s.d.) of capped cells after 30 minutes incubation at 37°C. The percentage of capped cells increased to 95% if two anti-CD43 mAbs were simultaneously cross-linked (data not shown). This was observed with various combinations of either MEM59 and L60 mAbs, which recognize carbohydrate-dependent epitopes, or L10, which binds a sialic acid-independent epitope, all three antigenic sites being located in the amino-terminal portion of the molecule (Remold-O’Donnell and Parent, 1994).

This clustering of CD43 molecules was energy-dependent since it did not occur at 4°C (Fig. 2C,C′) and was inhibited by 20 mM D-deoxyglucose (data not shown).

CD43 capping was found to require microfilament integrity and acto-myosin contraction. Indeed, the capping was inhibited by the microfilament depolymerizing drug cytochalasin B (Fig. 2D,D′) or by the myosin disrupting drug butanedione monoxime (Fig. 2E,E′), although some CD43 clustering into small patches still occurred. These results suggest that the capping of CD43 is mediated through a direct or indirect link between CD43 and the underlying cytoskeleton. The role of microtubules was then investigated: colchicine, at a dose (100 nM) which has no effect by itself on CD43 redistribution, did not disturb the antibody-induced capping (data not shown). At a concentration of 2×10⁻⁵ M, colchicine increased the number of antibody-induced capped cells to 73±20%. Although our experimental protocol involves a cell preincubation step with antibodies at 4°C, a temperature which is known to dissociate microtubules (Tilney and Porter, 1967), a rapid repolymerization of microtubules upon warming has been observed on thymocytes, concomitant with the capping of membrane glycoproteins (Turner et al., 1988). Furthermore, in all our experiments, control cells were preincubated at 4°C in medium or with irrelevant IgGs, then re-warmed at 37°C. This did not result either in a spontaneous CD43 capping or an increased percentage of polarized cells (1%), unless cells were pretreated with anti-CD43 mAbs or reacted with 2.10⁻⁵ M colchicine.

Costimulation of the cells by 10⁻⁶ M fMLP enhanced to 68±10% the capping induced by cross-linked antibodies.

In the absence of cross-linking by secondary antibodies or with Fab′ fragments of secondary antibodies, anti-CD43 mAbs did not result in capping and cells appeared homogeneously labelled.

Finally, we occasionally observed, independently of the presence of antibodies, cells which ‘spontaneously’ developed blebs (Fig. 2F,F′). Those blebs were devoided of membrane CD43, while other membrane molecules such as β2 integrins were uniformly present on those cells (data not shown).

As shown in Fig. 3, pre-embedding immunogold labelling for CD43 of control neutrophils showed that gold particles were regularily disposed along the plasma membrane, underlining the entire cell surface. On polarized neutrophils, following CD43 mAbs cross-linking at 37°C, the immunolabelling appeared to be clustered at one pole of the cell where all the secretion granules were concentrated (uropod). In contrast, the opposite pole of the cell, which appeared to be devoid of any organelles, was not labelled. The total amount of gold particles located on the cell surface was lower in polarized cells than in control neutrophils but some degree of endocytosis of antibody-coated gold particles had occurred during the incubation at 37°C before glutaraldehyde fixation (this could occasionally be seen in some sections, associated with the membrane of endocytosis vacuoles). The labelling control consisted of cells incubated with irrelevant IgG instead of anti-CD43 mAb. In that case, no gold particles were observed on the cell surface (data not shown).

Colchicine is known to induce a neutrophil polarization similar to what is observed on motile cells (Keller and Niggli, 1993). As mentioned above, CD43 was uniformly distributed on cells incubated at 37°C in buffer (Fig. 4A,A′). Stimulation with 2×10⁻⁵ M cochicine induced CD43 molecules clustering to the cell uropod of front/tail polarized cells (Fig. 4B,B′-F,F′), while β2-integrin (CD11b) and FcγRIII (CD16) were uniformly distributed on the whole cell surface (Fig. 4E,E′-G,G′-H,H′). When cells were incubated at 37°C with both colchicine and antibodies, CD43 molecules were highly concentrated in a single cap in the uropod (Fig. 4C,C′). Cells displayed the same pattern of CD43 distribution whether they were preincubated or not at 4°C before warming at 37°C. Cell incubation at 4°C or at 37°C with cytochalasin B (Fig. 4D,D′) or butanedione monoxime (data not shown), completly inhibited the colchicine-induced cell deformation and CD43 redistribution. Furthermore, at lower colchicine concentrations (10⁻⁶ M), leading to the polarisation of a limited number of neutrophils, CD43 was redistributed exclusively in polarized cells (data not shown).

To investigate if other polarizing agents also induce CD43 redistribution, we analysed CD43 localization upon fNLPNTL stimulation. Cells were first labelled with anti-CD43 mAb and the FITC-Fab′ fragment of secondary antibody, to avoid crosslinking, then incubated for 30 minutes with or without 10⁻⁹ M fNLPNTL. Cell polarization was observed with fNLPNTL, together with a redistribution of CD43 in neutrophil uropods, while β2-integrin (CD11b) and FcγRIII (CD16) did not redistribute (Fig. 5) Similar results were obtained with 10⁻⁹ M fMLP (data not shown).

Control neutrophils incubated with irrelevant antibodies at 4°C and then warmed to 37°C were spherical for 80% to 94%, containing less than 1% of front tail polarized cells. One should note that in the 1% polarized cells, CD43 was ‘spontaneously’ concentrated in the uropod (data not shown). During antibody cross-linking, cell shape changes with membrane ruffles were observed on 50% to 70% of neutrophils, 15% to 25% of cells showing a clear front-tail polarization with uropod formation (Fig. 6: 1). In that case, the CD43 cap was constantly located in the uropod. Spherical cells were uniformly labelled for F-actin (Fig. 6: 1), while polarized cells displayed heterogenous F-actin labelling with F-actin concentrated at the front of the cell, the CD43 cap being at the tail (Fig. 6: 2). Colchicine (2×10⁻⁵ M), with or without antibodies, induced similar neutrophil shape changes, with again an F-actin redistribution mainly at the front of the cell and CD43 localization in the uropod at the tail of the cell (Fig. 6: 3). Those results are in the line of previously described polarized and/or motile cells (Keller and Niggli, 1993), where F-actin was shown to be concentrated at the leading front and to a lesser extent in the uropod at the rear of the cells. This F-actin distribution is a sign of polarization leading to cell motility. Myosin was colocalized with CD43 caps independently of cell shape during mAb crosslinking (Fig. 7: 1) and/or colchicine treatment (Fig. 7: 2). When cells were polarized, myosin was located with CD43 in the uropod, in the same way as for motile neutrophils.

The present report demonstrates that leukosialin (CD43) binds directly or indirectly to cytoskeleton microfilaments, on the basis of its detergent insolubility and its membrane redistribution, in response to cell stimulation by anti-CD43 antibodies supposed to mimic a putative ligand of the CD43 molecule or by colchicine.

TX-100 insolubility of membrane proteins may result from an association with the cytoskeleton and also from properties of the molecules such as homotypic aggregability or the ability to interact with TX-100-insoluble glycosphingolipid (Sargiacomo et al., 1993). The use of a microfilament disrupting agent, such as cytochalasin B or DNase I, allowed us to distinguish cytoskeleton dependent and independent phenomenons in our studies of CD43 insolubilization in TX-100: (i) a small percentage of membrane CD43 molecules is ‘constitutively’ detergent-insoluble in resting neutrophils, a percentage that is increased after cell incubation at 37°C with a single anti-CD43 mAb, in the absence of cross-linking. These CD43 molecules precipitate in TX-100 in association with insoluble actin filaments, as shown by the inhibitory effect of cytochalasin B or of DNAse I. This clearly shows that at least a proportion of CD43 molecules on the neutrophil membrane is naturally or becomes associated with the cytoskeleton. (ii) On the other hand, CD43 cross-linking, by anti-CD43 mAbs followed by secondary antibodies or by wheat germ agglutinin (data not shown), results in an immediate precipitation of most CD43 molecules in Triton, by a mechanism which does not require microfilament integrity, since it occurs at 4°C or in the presence of cytochalasin B or of DNase I. We thus conclude that in the case of CD43 cross-linking, Triton insolubility is not a suitable assay to demonstrate a CD43 cytoskeletal association.

We thus turned to immunofluorescence investigations and we show here that CD43 cross-linking with anti-CD43 mAbs and a secondary antibody at 37°C results in a single very concentrated cap on the cell. This redistribution is energydependent, since it does not occur at 4°C or in the presence of D-deoxyglucose, it is microfilament-dependent and involves acto-myosin contractions, since it is inhibited by cytochalasin and butanedione monoxime (Cramer and Mitchison, 1995). The occurrence of small CD43 patches in spite of the presence of cytochalasin or butanedione monoxime, could reflect an initial cytoskeleton-independent but energy-requiring aggregation of CD43 into small patches, which would then be clustered into a single cap via an acto-myosin contraction process. In contrast, colchicine does not prevent the capping, showing that microtubule integrity is not required for cap formation. Moreover, it enhances the number of capped cells, suggesting a regulatory role for microtubules in CD43 redistribution. The recently reported modulation by colchicine of L-selectin expression on neutrophils indeed suggests that microtubules may be involved in surface molecule regulation (Cronstein et al., 1995).

Redistributions into caps of membrane molecules has been described, mainly on lymphocytes (Jack and Fearon, 1984; Schreiner et al., 1977; Turner et al., 1988), either as the result of antibody cross-linking or ‘spontaneously’ occurring on locomoting cells (Braun et al., 1978). Those capping phenomena require energy and involve a link between surface molecules and the cell cytoskeleton (Bourguignon and Bourguignon, 1984). In our hands, upon antibody cross-linking, neutrophil receptors such as FcγRIII receptors (CD16) or β2 integrins (CD11b) redistribute into more or less numerous patches with rare caps, contrasting with the striking clustering of CD43 into single caps (Fig. 8).

In the absence of an identified ligand for CD43, antibody cross-linking has previously been used to analyse possible cell responses involving CD43. Anti-CD43 mAbs were shown to induce leukocyte homotypic aggregation (Kuijpers et al., 1992; Nong et al., 1989; Rosenkranz et al., 1993) and to trigger, in mononuclear leukocytes, signalling events resulting in Ca²⁺ mobilization (Silverman et al., 1989; Wong et al., 1990). Data are controversial in neutrophils, where cell stimulation by anti-CD43 antibodies has been shown either to have no effect on calcium mobilization and respiratory burst (Kuijpers et al., 1992) or to increase the intracellular Ca²⁺ concentration and trigger the oxidative burst (Rosenkranz et al., 1993; Rothwell and Wright, 1994). Our data suggest that neutrophil responses to putative CD43 ligands most probably include cytoskeleton movements and membrane redistribution of CD43, which could be functionally important for adhesion and motility.

During chemokine-induced T cell adhesion to various substrates, the formation of cell uropods was recently described by del Pozo et al. (1995, 1996). Those uropods contained myosin and concentrated mainly ICAM-3, but also CD43 and CD44 on 15-18% of cells. On the basis of the ability of ICAM-3 to bind β2 integrins, these authors proposed an interesting model, in which unactivated leukocytes in the blood flow would interact via β2 integrins with ICAM-3 enriched uropods protruding from leukocytes adherent to the endothelium. Similarly, one could speculate that CD43, since it is a mucin and expresses the sialyl-Lewis^x structure, could be an L-selectin ligand and as such could participate in the rolling of passing leukocytes on adherent neutrophils handing up their CD43 at the tip of uropods. One should note that we here exclusively analysed suspended neutrophils and that in our experiments we observed CD43 redistribution independently of adhesion.

In addition, it is tempting to relate CD43 redistribution to cell motility. Indeed, we propose that CD43 might be involved in cell polarization and motility on the basis of three observations: (i) CD43 is concentrated in the uropod of spontaneously front/tail polarized cells, which are considered as motile cells (Keller et al., 1991); (ii) colchicine and fNLPNTL, which are known to induce neutrophil polarization and locomotion, also result in CD43 redistribution in the uropod; (iii) finally, CD43 cross-linking by specific antibodies promotes cell shape changes and polarization, with the CD43 cap located again in the uropod.

We occasionally observed that some cells spontaneously displayed blebs, which expressed β2 integrins (data not shown) but not CD43 on their surface. This phenomenon was observed on suspended but not on cytocentrifuged cells, and still occurred when the erythrocyte lysis step was omitted and was independent of cell fixation. Blebs were also observed at the leading edge of colchicine-polarized neutrophils, and may be similar to pinocytosis-related blebs previously described on locomoting neutrophils (Keller and Niggli, 1993). One should point out that, in all cases, these blebs were completely devoided of CD43. It is tempting to speculate that this phenomenon represents an initial step in cell polarization and that the bleb will become the leading edge of the cell.

Finally, as previously described on colchicine treated cells (Keller and Niggli, 1993), we observed that myosin, generally encountered in areas of cell contraction, was concentrated in the CD43-enriched uropod, while F-actin was mainly located at the front. We obtained the same results with anti-CD43 cross-linking, myosin being concentrated under the CD43 cap. These results are reminiscent of the reported localisation of CD43 in the cleavage furrow of dividing lymphocytes, where strong contractions via acto-myosin occur (Yonemura et al., 1993). Major T cell glycoproteins CD45 (T200) and Thy-1 were reported to cap on colchicine-stimulated cells, with an accumulation of myosin under the caps (Bourguignon and Bourguignon, 1984). Similarly, the neutrophil receptor for activated platelets, i.e. presumably the mucin ligand of P-selectin PSGL-1, appears to concentrate in the uropod of polarized neutrophils (Doré et al., 1996). The locomotion process is thought to be carried out by the extension of F-actin, resulting in a driving force at the leading edge of the cell, and retraction of the rear by an acto-myosin contractile system (Keller and Niggli, 1993).

As far as F-actin is concerned, we observed that during cross-linking experiments, CD43 is first redistributed on round cells in an F-actin-rich cap structure (Fig. 6: 1). Cells then become polarized with a rearrangement of F-actin typical of front-tail polarization. Still, some F-actin remained under most CD43 caps, on cells polarized by cross-linking or by colchicine, as shown by the yellow color (as opposed to green) in double fluorescence staining (Fig. 6: 1A, 3Aa,b). On some polarized cells, however, F-actin appeared nearly absent from the uropod (Fig. 6: 2). We propose that F-actin would be mostly involved in the clustering of CD43, via acto-myosin contraction (myosin always stayed under the cap even in non polarized cells) and that a fewer molecules of F actin would be required to maintain the cap. F-actin would then be mainly recruited at the leading edge of the cell for locomotion.

We suggest that a redistribution of the highly negatively charged CD43, but also of CD44 and possibly sialylated glycoproteins such as CD45 and PSGL-1, in the acto-myosin rich uropod, might be a prerequisit for the cell locomotion process. To ensure that CD43 redistribution in neutrophil uropods is not restricted to CD43 cross-linking or to colchicine stimulation, we studied the effect of chemotactic peptides. Chemotactic peptides strongly affect neutrophil shape in a time sequence and in a dose dependent fashion (Keller and Niggli, 1995). We confirm that 10⁻⁶ M chemotactic peptide induces nonpolar membrane projections without front-tail polarization (data not shown), while a concentration of 10⁻⁹ M induces front-tail polarization of neutrophils. CD43 redistribution in uropods occurred only at concentrations leading to cell polarization.

In conclusion, three hypotheses can be put forward, which are now being investigated in our laboratory: (1) CD43 redistribution is important for leukocyte interaction with other cells, as proposed by Del Pozo et al. (1996) and further suggested by platelet interactions with neutrophils uropods. (2) This redistribution may clear negatively charged sialylated molecules from a portion of the membrane, at the leading edge of polarized cells, which would then be available for other functionally important proteins such as integrins and Fcγ-receptors. (3) In parallel, cell polarization and shape changes during cell locomotion would require the anchoring of cytoskeleton structures to membrane molecules such as CD43.

The authors thank Monique Arpin (Institute Curie, Paris), Pierre Bongrand (INSERM U387, Marseille), Hansuli Keller and Verena Niggli (Bern University) for helpful discussions, and Eileen Remold O’Donnell (Blood Center, Boston) and Samuel Wright (Rockefeller University, New York) for providing monoclonal antibodies. This work was partly supported by the Fondation pour la Recherche Médicale (for electron microscopy) and by Guerbet research.