Microtubule-disruption-induced and chemotactic-peptide-induced migration of human neutrophils: implications for differential sets of signalling pathways

Directed migration of neutrophils is essential for the fulfillment of their physiological role, that is, defending the host against invading bacteria. Neutrophils develop polarity and migrate not only when exposed to a gradient of chemoattractant (chemotaxis) but also in uniform concentrations of stimulus(stimulated random migration, known as chemokinesis). Reversible actin polymerization is crucial for neutrophil development of polarity and migration(Chung et al., 2001). Less is known about the role of microtubules. The latter have been extensively studied in motile events in fibroblasts and growth cones. There, disassembly of microtubules attenuates polarity and migration(Vasiliev et al., 1970; Tanaka et al., 1995) and increases contractility and myosin light chain phosphorylation(Kolodney and Elson, 1995). In neutrophils, by contrast, an intact microtubule network does not appear to be required for migration. Microtubule-disrupting drugs such as colchicine,vinblastine or nocodazole even induce development of polarity and chemokinesis(Dziezanowski et al., 1980; Keller et al., 1984). Microtubule depolymerization does not impair chemotaxis induced by activated serum (Dzienzanowski et al., 1980) or by the chemotactic peptide f-Met-Leu-Phe(fMLP), and neutrophils exposed to colchicine still orient in a gradient of chemotactic peptide (Zigmond,1977; Keller et al.,1984). Colchicine itself does not act as a chemoattractant. Interestingly, colchicine-induced motile responses can be abolished by cytochalasin B, indicating a role for actin filaments(Keller et al., 1984). A crosstalk between microtubules and actin filaments is further substantiated by the observation that colchicine treatment induces an increase in cyoskeleton-associated actin and α-actinin as well as enrichment of actin, α-actinin and tubulin in the front and myosin II at the tail of the polarized cells (Keller and Niggli,1993). Microtubule depolymerization thus induces changes in the actin network that are required for the increased motility of neutrophils. One possible interpretation of these earlier findings is based on interactions between the different filament systems, where the modification of one system affects the others [`tensegrity' (for details, see Ingber, 1993)]. Recent evidence obtained in fibroblasts suggests a different explanation. Enomoto showed that in 3T3 fibroblasts microtubule depolymerization by colcemid,colchicine or nocodazole induces formation of stress fibers and vinculin-containing focal adhesions(Enomoto, 1996). These effects are inhibited by cytochalasin D, which prevents addition of actin monomers to the barbed ends of actin filaments, and taxol, which stabilizes microtubules. Interestingly, these responses to microtubule depolymerization were inhibited by exoenzyme C3. This protein selectively ADP-ribosylates and inhibits the small GTP-binding protein Rho without affecting rac or cdc42. An inhibitor of phosphatidylinositol 3-kinase (PI 3-kinase) did not suppress the responses(Enomoto, 1996; Liu et al., 1998).

Microtubule disruption thus may activate Rho, at least in 3T3 cells(Ren et al., 1999). Another small GTP-binding protein, Rac1, in contrast, is activated in those cells by microtubule growth, resulting in increased lamellipodial protrusions in fibroblasts (Waterman-Storer et al.,1999). Microtubule growth at the leading edge and shortening at the rear could thus locally activate Rho or Rac and result in local contraction or ruffling in migrating fibroblasts(Wittmann and Waterman-Storer,2001). In contrast to the data obtained in neutrophils,microtubule disruption reduces polarization, ruffling and the speed of migration in fibroblasts (Wittmann and Waterman-Storer, 2001), which suggests that different mechanisms operate in the different cell types.

We describe here the signalling from microtubule disruption to migration in human neutrophils. We provide evidence for the activation in this process of Rho-kinase but not of heterotrimeric Gi proteins, the PI 3-kinase pathway or the p42/44 mitogen-activated protein kinase (MAPK). This differs from chemotactic peptide-induced migration where Gi proteins are activated as well as PI 3-kinase and the MAPK cascade in addition to Rho family proteins and their downstream targets.

Reagents and suppliers were N-formyl-L-norleucyl-L-leucyl-L-phenylalanyl-L-norleucyl-L-tyrosyl-L-lysine(fNLPNTL), Bachem, Bubendorf, Switzerland; human serum albumin (HSA), ZLB Bioplasma AG, Bern, Switzerland; neutrophil isolation medium (NIM), Cardinal Associates, Santa Fe, NM, USA; OptiPrep, Axis-Shield PoC As, Oslo, Norway;dimethylsulfoxide (DMSO), enhanced chemiluminescence (ECL) western blotting detection reagents, Pierce, Rockford, IL, USA; diisopropylfluorophosphate(DFP), Fluka AG, Buchs, Switzerland; pertussis toxin, colchicine,lumicolchicine, taxol (paclitaxel), wortmannin, Sigma, Buchs, Switzerland. Gey's medium contained 138 mM NaCl, 6 mM KCl, 1 mM MgSO₄, 1.1 mM CaCl₂, 0.1 mM EGTA, 1 mM Na₂HPO₄, 5 mM NaHCO₃, 5.5 mM glucose, 20 mM HEPES, pH 7.2, 0.1% HSA (wt/vol). In some experiments (see Results) MgSO₄ and CaCl₂ were omitted to prevent cell aggregation. HSA was omitted for determination of membrane association of proteins and for assays of protein phosphorylation. Y-27632 was kindly provided by Welfide Corporation (Osaka, Japan). Stock solutions of Y-27632 (10 mM) were prepared in H₂O and aliquots were stored at 4°C. Wortmannin and fNLPNTL were added as stock solutions in DMSO. DMSO alone, up to 0.4%, had no effect in our experiments.

A monoclonal antibody specific for RhoA was obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and a monoclonal antibody directed against Rokα was obtained from Transduction Laboratories (Lexington, KY). A polyclonal antibody specific for p42/44 MAPK phosphorylated on threonine and tyrosine was obtained from New England Biolabs, Beverly, MA. A polyclonal antibody specific for Akt phosphorylated on serine 473 and a polyclonal antibody reacting with Akt independent of its phosphorylation state were obtained from Cell Signaling Technology Inc. (Beverly, MA). A polyclonal antibody produced in rabbits against a peptide corresponding to myosin light chain (MLC) phosphorylated on Thr-16 and Ser-19 (PPMLC)(Ratcliffe et al., 1999) was kindly provided by J. M. Staddon (Eisai London Research Laboratories, London,UK). Goat anti-mouse and goat anti-rabbit antibodies conjugated to alkaline phosphatase were obtained from Bio-Rad (Richmond, CA).

Neutrophils were isolated from heparinized human blood (10 units/ml). In a first step, red blood cells were removed with a solution containing 12% (w/v)OptiPrep and 16.6 g/l methocel in 130 mM NaCl(Böyum, 1968). Subsequently, mononuclear cells were removed using NIM, which contains Hypaque and Ficoll (Ferrante and Thong,1980). Approximately 95% of the leukocytes were neutrophils.

Neutrophils (3×10⁶ cells/ml) were incubated in Gey's medium containing 0.1% HSA but lacking divalent cations in a reciprocating waterbath at 37°C without or with inhibitors, colchicine, taxol or fNLPNTL, as indicated in Results. The reaction was stopped by fixing the cells in 1% glutaraldehyde (final concentration) for 30 minutes. The cells were washed and resuspended in 0.9% NaCl containing NaN₃ (1 mg/ml). Cell shape was determined using differential interference contrast microscopy(Nomarski optics) using a Zeiss IM 35 microscope with a 100× objective(NA 1.25). The shapes of neutrophils were classified into the following categories as previously described (Keller and Niggli, 1993): spherical (for examples, see Fig. 1, control), spherical with unifocal projections, polarized (for examples, see Fig. 1, fNLPNTL, 10 minutes)and non-polar cells with surface projections (for examples, see Fig. 1, fNLPNTL, 1 minute).

Neutrophils (3×10⁶/ml) were preincubated in Iscove's medium with 10% fetal calf serum without or with pertussis toxin (400 ng/ml)for 2 hours at 37°C in a humidified atmosphere. After 2 hours, the cells were centrifuged and the cells resuspended in Gey's medium with divalent cations, 0.5% HSA with or without pertussis toxin (450 ng/ml). Cell shape was then determined after incubation in the presence or absence of fNLPNTL or colchicine.

Neutrophils (4×10⁶/ml, 0.45 ml per assay) were incubated in Gey's medium containing divalent cations and 2% (experiments with colchicine) or 10% (experiments with fNLPNTL) HSA with inhibitors and/or stimuli at 37°C in a reciprocating waterbath as indicated in Results. The cells were centrifuged at 300 g for 5 minutes and resuspended in 50 μl of the supernatant. Neutrophils (5 μl containing 0.18×10⁶ cells) were placed between a slide and a round coverslip (25 mm diameter). The slide-coverslip preparation was sealed with paraffin and placed on the heated stage (37°C) of a Zeiss IM 35 microscope with a 63× objective. The locomotor behavior of the cells (7-33 cells for each condition) was recorded for 10 minutes using video microscopy. The outline of the cell at the initial and final position and the path traveled during the 10 minutes was drawn on a transparency. Cells remaining totally or partially within the outline of the initial position are defined as stationary; cells found outside of the outline after 10 minutes are defined as locomoting. The length of the path of individual cells was measured using a KS-300 analysis system (Kontron, Eching, Germany).

After treatment with DFP to block endogenous proteases(Amrein and Stossel, 1980),neutrophils (12×10⁶ cells/ml) were resuspended in Gey's medium without divalent cations and HSA. Aliquots (450 μl) of this suspension were incubated with inhibitors and/or stimuli as described in Results. The reaction was stopped by centrifugation (300 g, 5 minutes, 4°C). Cells were then resuspended in 500 μl relaxation buffer(Dusi et al., 1996) containing 100 mM KCl, 3 mM NaCl, 3.5 mM MgCl₂, 1 mM EGTA, 10 mM Pipes, pH 7.3, 1 mM EDTA, 10 mM NaF. This suspension was incubated for 5 minutes on ice followed by sonication (Branson tip sonifier B-10, output 3.5, 1×10 seconds). After addition of 4 mM DFP to the cell lysate and a low-speed centrifugation for 5 minutes at 700 g and 4°C to remove unbroken cells and nuclei, the supernatants were centrifuged at 100,000 g for 30 minutes at 4°C in a Beckmann table top ultracentrifuge. The resulting pellets (corresponding to the membrane fraction) were dissolved in 100 μl of Laemmli sample buffer(Laemmli, 1970) by incubation at 95°C with vortexing for 10 minutes. The solubilized pellets, together with aliquots of the high-speed supernatants (soluble cytosolic fractions),were subjected to electrophoresis on 5-10% (for Rokα) or 5-20% (for RhoA) SDS-polyacrylamide gradient gels and transblotted to nitrocellulose using a mini-genie blotter from Idea Scientific (Minneapolis, MN) for 70 minutes, 24 Volts. Blots were exposed to antibodies to RhoA (diluted 1:1000)or to Rokα (diluted 1:500). For visualization of bound antibody an anti-mouse antibody conjugated to alkaline phosphatase, diluted 1:7000,followed by ECL detection of the second antibody, was used. The bands were scanned with a Camag TLC scanner at 590 nm and the area of the peaks was taken as a relative measure for the amount of protein present in the bands.

Neutrophils in 0.45 ml Gey's medium lacking divalent cations and HSA(0.5×10⁶ cells for analysis of p42/44 MAPK,1.5×10⁶ cells for analysis of Akt and 2×10⁶cells for analysis of PPMLC) were exposed to stimuli as described in Results. The reaction was stopped by addition of 0.5 ml of a solution containing 20%(w/v) trichloroacetic acid (TCA), 40 mM NaF and 10 mM Na₂HPO₄. After a 20 minute incubation on ice, the precipitates were collected by centrifugation and washed once with 0.5 ml of 5% TCA and once with 0.5% TCA. The pellets were solubilized in sample buffer(see above), followed by separation of proteins on a 5-10% SDS-polyacrylamide gradient gel for analysis of p42/44 MAPK, a 10% gel for Akt and a 15% gel for analysis of PPMLC. Phosphorylated proteins were detected using a polyclonal antibody specific for p42/44 MAPK phosphorylated on threonine and tyrosine,diluted 1:1000, or a polyclonal antibody specific for Akt phosphorylated on serine 473, diluted 1:1000, or a polyclonal antibody specifically reacting with PPMLC, diluted 1:500. For visualization of PPMLC, blots were blocked in PBS containing 2% BSA and 0.05% Tween, followed by incubation with the primary antibody overnight at 4°C and a 1 hour incubation with the goat anti-rabbit IgG antibody prior to ELC detection. Detection of phosphorylated Akt was carried out as recommended by the producer.

Differences between data were analyzed with the Student's t-test for paired data with a P value of <0.05 considered significant. Data correspond to the mean±s.e.m.

First, we tested whether stabilization of microtubules prevents the effects of colchicine on cell shape and migration. As shown in Fig. 1, incubation of neutrophils with 10 μM colchicine for 30 minutes induces polarity in up to 80% of the cells. The morphology of colchicine-stimulated cells is comparable to that of cells stimulated with chemotactic peptide, although in colchicine-induced cells ruffling and formation of veils in the leading edge is reduced (Fig. 1)(Keller et al., 1984). Preincubation of cells with 1 μM taxol completely suppressed colchicine-induced development of polarity (Figs 1 and 2). The cells assumed a spherical shape with a smooth surface that is indistinguishable from that of the controls. Up to 20 μM taxol did not affect morphology of resting cells and also did not significantly affect morphology of cells stimulated with chemotactic peptide (Fig. 2). Inhibition of colchicine-induced polarity correlates with a 96±4%inhibition of colchicine-induced migration by 1 μM taxol(Table 1A). Although taxol did not affect polarity of chemotactic-peptide-activated cells(Fig. 2), this drug significantly reduced the fraction of cells migrating in the presence of fNLPNTL by 81±6% (n=4)(Table 1B). Cells preincubated with taxol prior to exposure to chemotactic peptide tended to be more elongated, with the rear of the cell adhering to the substrate (data not shown).

On the basis of the data obtained in fibroblasts where microtubule disruption activates Rho (Ren et al.,1999), we investigated a possible role for the Rho-effector Rho-kinase in colchicine-induced development of polarity and migration in human neutrophils. Preincubation of cells with the Rho-kinase inhibitor Y-27632 suppressed development of polarity in a concentration-dependent manner(Figs 1 and 3), which is comparable to effects on chemotactic-peptide-stimulated cells(Niggli, 1999). Maximal inhibition was obtained at 10 μM of the inhibitor (95±3% inhibition, n=5). Half-maximal values were obtained at 2.6±0.8 μM(n=5), which is comparable to the concentration necessary for inhibition of the purified enzyme (Uehata et al., 1997). As shown in Fig. 1, most of the cells preincubated with 10 μM Y-27632 before addition of colchicine were spherical, although a few cells still showed small projections. This is different from chemotactic-peptide-stimulated cells pretreated with Y-27632, which still exhibit large F-actin-rich ruffles(Niggli, 1999). We also assessed the effects of Y-27632 on colchicine-induced chemokinesis and observed a highly significant (P<0.0025) reduction of 76±9%of the fraction of migrating cells (Table 2). The speed of migration of the remaining motile cells was also significantly (P<0.005) decreased by 66-86%. These findings suggest that microtubule disassembly activates the Rho/Rho-kinase cascade in neutrophils and that this process is required for colchicine-induced neutrophil migration.

We assessed the effect of inhibiting activity of heterotrimeric Gi proteins with pertussis toxin on chemotactic-peptide- and colchicine-induced development of polarity. As shown in Fig. 4, preincubation with 400 ng pertussis toxin/ml completely abolished chemotactic peptide-induced polarity. The cells assumed the spherical morphology typical of resting cells. By contrast, colchicine-induced polarity was not affected, suggesting that microtubule disassembly does not activate motile processes in neutrophils via Gi.

PI 3-kinase plays an important role in chemotactic-factor-induced neutrophil migration (Hirsch et al.,2000; Stephens et al.,2002). As shown in Fig. 5A and in our previous work(Niggli and Keller, 1997), the PI 3-kinase inhibitor wortmannin markedly reduces chemotactic-peptide-induced development of polarity and migration in human neutrophils, correlating with inhibition of phosphatidylinositol (3,4,5)-trisphosphate[PtdIns(3,4,5)P₃] production. Half-maximal effects of wortmannin on cell polarization were obtained in the range of 14-32 nM of the drug. Development of polarity was inhibited by 58±8% (n=5) by 100 nM wortmannin (Fig. 5A)(Niggli and Keller, 1997). We assessed whether the incomplete inhibition of chemotactic-peptide-induced cell polarization by wortmannin is due to incomplete inhibition of PI 3-kinase by monitoring the effect of wortmannin on activation of a major target of PI 3-kinase, the serine-threonine protein kinase Akt (or protein kinase B; PKB). Hirsch et al. showed that chemotactic-peptide-induced increased phosphorylation of Akt is completely abolished in neutrophils of mice lacking PI 3-kinase γ, whereas chemotaxis of neutrophils from these mice is only partially impaired (Hirsch et al.,2000). We now show that wortmannin (100 nM) also completely suppressed fNLPNTL-induced Akt phosphorylation in human neutrophils(Fig. 5B). By contrast, this concentration of wortmannin did not affect colchicine-induced development of polarity (Fig. 5A). Moreover we could not detect any Akt phosphorylation in cells treated with colchicine for 5 to 30 minutes, whereas exposure of cells to 1 nM fNLPNTL for 5 minutes induced a marked signal (Fig. 5B).

Previous data suggested that microtubule disruption activates p42/44 MAPK rapidly in human pro-monocytic cells(Schmid-Alliana et al., 1998). However, we could not observe a significant increase in phosphorylation of p42/44 MAPK in neutrophils exposed to colchicine for 5-30 minutes, whereas activation with fNLPNTL for 5 minutes had a very marked effect(Fig. 6).

Translocation of Rho-kinase to the plasma membrane may correlate with its activation (Taggart et al.,1999). Therefore we studied the effect of colchicine on membrane association of Rokα (Rho-kinase II) in human neutrophils and compared it with the effects of chemotactic peptide. In resting cells, 12±7%(n=3) of total enzyme was recovered in the membrane-containing fraction. Incubation of cells with colchicine (5 minutes) induced a marked,99±20% increase (n=7, P<0.0025) in membrane-associated Rokα, which is comparable with that induced by chemotactic peptide (171±50% increase, n=3), as shown in Fig. 7A,B. Membrane association of Rokα was maximal 5 minutes after addition of colchicine and decreased to a lower level, which was still significantly higher than control levels 10 minutes after addition of the drug (data not shown). This response is thus more transient than the effect of colchicine on cell polarity where 93% of the maximal polarization response was observed 5 minutes after addition of colchicine and 100% 10 minutes after addition of the drug. Similar to effects on cell shape and migration (Fig. 2, Table 1),preincubation with taxol also completely prevented colchicine-induced membrane association of Rokα (Fig. 7A,B), whereas that induced by chemotactic peptide was only partially affected (63±16% inhibition, n=6, Fig. 7C,D). In the latter case the extent of inhibition was quite variable in different experiments. Lumicolchicine, which is inactive on microtubules, did not induce membrane association of Rokα (Fig. 7A,B).

Activation of neutrophils with chemotactic peptide induces recruitment of a fraction of Rho and Rac from the cytosol to the membrane(Dusi et al., 1996; Marcil et al., 1999). Incubation of cells with 10 μM colchicine for 5 minutes also induced a significant shift of RhoA to the membrane (50-230% increase). This increase was comparable to that induced by a 30 minutes incubation with 1 nM fNLPNTL(102±47% increase) (data not shown).

Our data strongly suggest that microtubule disassembly in neutrophils activates RhoA and its effector Rokα.

We wanted to obtain information on downstream targets of Rho-kinase activated in cells by colchicine. As shown previously, exposure to chemotactic peptide induces a sustained increase in myosin light chain phosphorylation in human neutrophils that can be abolished by the Rho-kinase inhibitor Y-27632(Niggli, 1999). As assessed with an antibody that specifically reacts with myosin light chain phosphorylated on Thr-16 and Ser-19 (PPMLC)(Ratcliffe et al., 1999),treatment of neutrophils with 10 μM colchicine induced a time-dependent increase in the amount of PPMLC; this increase was maximal 20-30 minutes after addition of the drug (a 223±65% increase, n=5 at 20 minutes, P<0.025; Fig. 8). 34 to 57% of the response was attained 5 minutes after addition of the drug to the cells. As shown in Fig. 8,the maximal increase obtained 20 minutes after addition of colchicine was even higher than that induced by a 5 minute incubation with chemotactic peptide and could be completely abolished by preincubation of the cells with 10 μM Y-27632. Preincubation with 10 μM taxol also completely abolished the colchicine-induced increase in PPMLC (data not shown).

Disruption of the neutrophil microtubule network induces development of polarity and chemokinesis in human neutrophils comparable to the effects of chemotactic peptide. We confirm here that these effects are due to microtubule disruption and not to non-specific effects of colchicine as they can be prevented by the addition of taxol, which stabilizes microtubules, and lumicolchicine, which is inactive on microtubules, did not affect neutrophil shape and migration (Keller et al.,1984). We provide now data on the signalling pathways involved in microtubule-induced development of polarity and migration in neutrophils and on a downstream target.

We demonstrate here that neutrophil motility induced by microtubule disruption is probably not due to direct effects on actin organization;instead it is probably due to activation of specific signalling pathways. We studied four different signalling pathways that could conceivably be affected by microtubule disruption: heterotrimeric G-proteins, the Rho-Rho-kinase pathway, the PI 3-kinase pathway and activation of p42/44 MAPK. Concerning heterotrimeric G-proteins, membrane-bound tubulin has been reported to directly activate Gαs, Gα1 and Gαq(Popova et al., 1997). However, Gi inhibition in neutrophils by pertussis toxin did not suppress colchicine-induced cell polarization in our system, which excludes G_i proteins from having a role in this response. Chemotactic-peptide-induced shape changes in contrast were completely prevented by pertussis toxin.

We have shown previously that inhibition of Rho-kinase activity results in almost complete suppression of chemotactic-peptide-induced polarization and migration of human neutrophils (Niggli,1999). We now show that Rho-kinase inhibition also almost completely suppresses colchicine-induced development of polarity and markedly inhibits colchicine-induced chemokinesis in neutrophils. This finding correlates with reports showing that microtubule disassembly activates the Rho-Rho-kinase pathway in other cell types (Witman and Waterman-Storer, 2001). Indeed we could show that colchicine treatment results in increased membrane association of Rho-kinase, which can be prevented by taxol-based stabilization of microtubules (Fig. 6). Previous evidence obtained in smooth muscle cells suggests that membrane association of Rho-kinase correlates with its activation. There, carbachol stimulates association of Rho-kinase with the plasma membrane as demonstrated using immunofluorescence staining (Taggart et al., 1999). We now show that this also occurs in stimulated neutrophils exposed to chemotactic peptide or colchicine. We also provide evidence that Rho-kinase, activated by microtubule disassembly, induces a marked sustained increase in myosin light chain phosphorylation, comparable to that induced by exposure of cells to chemotactic peptide, which then leads to tail contraction and is crucial for colchicine-induced migration. Interestingly, the time course of colchicine-induced membrane association of Rho-kinase is somewhat faster than that of myosin light chain phosphorylation and development of polarity. Moreover the first event is transient, whereas the two other events are sustained for up to at least 30 minutes after addition of the stimulus. Transient activation of Rho-kinase thus results in sustained myosin light chain phosphorylation and polarization. Rho-kinase is thought to act by phosphorylating the myosin-binding subunit of myosin light chain phosphatase, thereby inhibiting this enzyme(Somlyo and Somlyo, 2000). A transient activation of Rho-kinase may be sufficient to inactivate the phosphatase for longer. This suggests that phosphatases acting on the myosin light chain phosphatase are inactive in the migrating cells. Interestingly,the phosphorylated myosin-binding subunit of type 1 protein phosphatase is highly resistant to dephosphorylation(Takizawa et al., 2002). Alternatively, activation of Rho-kinase may be maintained even after its dissociation from the membrane.

Regulatory subunits of PI 3-kinase interact with tubulin(Inukai et al., 2000). Activity of this enzyme could thus be affected by microtubule disassembly. However, colchicine-induced polarization was not sensitive to treatment with concentrations of the PI 3-kinase inhibitor wortmannin, which completely suppress chemotactic-peptide-induced phosphorylation of a major target of PI 3-kinase, Akt (Fig. 5B), and which markedly inhibit neutrophil migration induced by exposure to chemotactic peptide or to membrane-permeable PtdIns(3,4,5)P₃(Niggli and Keller, 1997; Niggli, 2000). This strongly suggests that colchicine does not act via PI 3-kinase. In line with this notion, stimulation of cells with colchicine did not induce any detectable Akt phosphorylation (Fig. 5B).

A substantial fraction of p42/44 MAPK is associated with microtubules, and microtubule disruption activates these enzymes within minutes of addition of colchicine in human pro-monocytic cells(Schmid-Alliana et al., 1998). However, this was not the case in human neutrophils, suggesting differential regulation of different signalling systems depending on the cell type.

Our data strongly suggest that microtubule disassembly induces a selective activation of the Rho-Rho-kinase pathway but not of PI 3-kinase or of p42/44 MAPK. This activation of Rho-kinase, resulting in increased phosphorylation of myosin light chain on Thr-18 and Ser-19, is required for the colchicine-induced development of polarity and the subsequent migratory response in neutrophils. Concerning the speed of migration and the fraction of migrating cells, this response compares well with that induced by chemotactic peptides. However, in colchicine-induced cells, less ruffling occurs at the leading edge (Fig. 1)(Keller et al., 1984). This may be due to the lack of activation of PI 3-kinase and lack of local accumulation of PtdIns(3,4,5)P₃, which is thought to be crucial for chemotactic-peptide-induced polarity in neutrophils(Knall et al., 1997; Niggli and Keller, 1997; Niggli, 2000; Hirsch et al., 2000; Weiner, 2002; Stephens et al., 2002). Migration is thus not obligatorily coupled to localized production of PtdIns(3,4,5)P₃.

Concerning the mechanism of microtubule-disassembly-induced Rho activation,evidence has recently been provided for a direct regulation of the nucleotide exchange factor GEF-H1 by interaction with microtubules. This factor appears to be inactive when bound to microtubules. It is activated by microtubule disassembly and selectively activates Rho but not Rac1 or Cdc42(Krendel et al., 2002).

Our data thus show that different stimuli activate migration in human neutrophils involving differential sets of signalling proteins depending on the initial stimulus.

Microtubule disassembly in neutrophils induces development of a polarized shape. The front of these polarized cells, as assessed by videomicroscopy, is defined as that part of the cell where protrusions are extended in the direction of migration, whereas pseudopod protrusion is suppressed in the rear of the cell, which is characterized by a narrow, contracted tail(Fig. 1)(Keller et al., 1984). Interestingly, polarization of cells induced by microtubule disassembly is accompanied by accumulation of F-actin and α-actinin in the front and myosin II in the tail; these results are similar to findings in chemotactic-peptide-stimulated cells. Also, a small increase in the fraction of stable actin resistant to Triton X-100 could be detected in colchicine-treated cells (Keller and Niggli, 1993), which is similar to previous findings by another group (Tsai et al., 1998). Relatively small changes in the F-actin:G-actin ratio are thus sufficient to sustain rapid migration. The mechanism of this cytoskeletal reorganization remains to be clarified. We observed previously that Rho-kinase inhibition does not affect actin reorganization in human neutrophils(Niggli, 1999). Moreover, Rho inhibition in neutrophils did not suppress chemotactic-peptide-induced actin polymerization (Ehrengruber et al.,1995). However, we observed that in neutrophils pretreated with Y-27632 prior to addition of colchicine, F-actin was diffusely distributed in the cytosol, which is similar to results for untreated controls, suggesting a role for Rho-kinase in regulating the polarized accumulation of F-actin in colchicine-treated cells (data not shown).

Localized activation of signalling proteins and generation of diffusable inhibitors may be prerequisites for development of polarity and chemotaxis(Weiner, 2002; Stephens et al., 2002). Neutrophils develop polarity not only in gradients of chemoattractants but also in uniform concentrations of stimuli. In the latter case, stochastic differences in concentration of stimuli are thought to induce polarity. It is not clear how global depolymerization of microtubules induced by colchicine can result in localized activation of Rho/Rho-kinase and localized myosin activation presumably required for development of polarity. One would have to assume a fortuitous local higher activation of the Rho/Rho-kinase system,which would result in generation of a diffusable inhibitor and localized positive feedback. However such factors regulating the Rho/Rho-kinase system have not yet been identified in neutrophils(Stephens et al., 2002).

As outlined by Wittmann and Waterman-Storer, cells such as fibroblasts or macrophages require an intact microtubule network for development of polarity and migration whereas other cells such as neutrophils do not(Wittmann and Waterman-Storer,2001). The exact role of microtubules in maintaining polarity in specific cell types is still under debate. Microtubules may be required for transport of membranes to the leading edge and/or for selectively stabilizing the leading edge and/or for regulating focal adhesion dynamics ensuring disassembly of focal contacts in the rear of the cell(Wittmann and Waterman-Storer,2001). Neutrophils polarize not only when adhering to a substrate but also in suspension. Possibly the need for intact microtubules for fibroblast polarization is related to the role of cell substrate contacts in this process. Adhesion-independent polarization of neutrophils does not require intact microtubules. As microtubule disassembly on its own can induce a migratory response in neutrophils, the question arises whether localized microtubule disassembly induced by chemotactic peptide also plays a role in receptor-linked chemotactic responses. The microtubule-stabilizing drug taxol does not affect chemotactic-peptide-induced cell polarization in suspension but partially inhibits chemotactic-peptide-induced membrane association of Rho-kinase and markedly inhibits chemotactic-peptide-induced chemokinesis(Fig. 7C,D; Table 1B). The third observation is in agreement with earlier data showing abolishment of fMLP-induced chemotaxis in neutrophils by taxol(Roberts et al., 1982). According to these authors, taxol reduces spreading of neutrophils. Localized microtubulin depolymerization occurring during cell migration, which in turn results in localized Rho/Rho-kinase activation, may contribute to regulation of cell adhesion. The partial inhibition of Rho-kinase activation by taxol observed in chemotactic-peptide-stimulated cells may not be sufficient to abolish development of polarity in suspension but may significantly impair cell migration.

In summary, we provide novel data showing that microtubule disassembly in human neutrophils selectively activates the Rho/Rho-kinase pathway but not PI 3-kinase or p42/44 MAPK. Motility ensuing from microtubule disruption is independent of PI 3-kinase but requires Rho-kinase activity. This is in contrast to development of polarity and migration induced by chemotactic peptides or membrane-permeable PtdIns(3,4,5)P₃ where the PI 3-kinase pathway plays a major role. Differential sets of signalling pathways are thus required for migration depending on the initial stimulus.

We thank Kathy Mujynya Ludunge, Patrick Lier and Christoph Oppliger for excellent technical assistance, H. U. Keller and J. Laissue for continuous support, E. Sigel for critical reading of the manuscript and J. M. Staddon for providing the antibody recognizing phosphorylated myosin light chain. This work was supported by the Swiss National Science Foundation (Grant No. 31-58836.99 to V.N.).