The possible involvement of the Rho-p160ROCK (Rho coiled-coil kinase) pathway in the signaling induced by the chemokine Stromal cell-derived factor (SDF)-1α has been studied in human PBL. SDF-1α induced activation of RhoA, but not that of Rac. RhoA activation was followed by p160ROCK activation mediated by RhoA, which led to myosin light chain (MLC) phosphorylation, which was dependent on RhoA and p160ROCK activities. The kinetics of MLC activation was similar to that of RhoA and p160ROCK. The role of this cascade in overall cell morphology and functional responses to the chemokine was examined employing different chemical inhibitors. Inhibition of either RhoA or p160ROCK did not block SDF-1α-induced short-term actin polymerization, but induced the formation of long spikes arising from the cell body, which were found to be microtubule based. This morphological change was associated with an increase in microtubule instability, which argues for an active microtubule polymerization in the formation of these spikes. Inhibition of the Rho-p160ROCK-MLC kinase signaling cascade at different steps blocked lymphocyte migration and the chemotaxis induced by SDF-1α. Our results indicate that the Rho-p160ROCK axis plays a pivotal role in the control of the cell shape as a step before lymphocyte migration toward a chemotactic gradient.
Sensitivity of cells of the immune system to chemoattractant gradients is a key feature for cellular responses in a variety of physiological and physiopathological conditions. There is growing evidence on the molecules capable of inducing or regulating the formation of chemoattractant gradients and the responses of the cells. During inflammation, activated endothelial cells expose proteins of the chemokine family on their surface. Chemokines are chemotactic cytokines that play an active role in regulating integrin-mediated interactions of neutrophils, monocytes, and lymphocytes with the endothelium at the area of injury, thereby favoring firm adhesion and extravasation of these cells (1). Chemokines are thought to guide the navigation of the cells during migration to the precise site of inflammation, where they exert their effector roles (2, 3). Also, chemokines have been shown to be important in noninflammatory conditions. In fact, there seems to be a tight balance between inflammatory and systemic chemokines (4), which are involved in the organogenesis and architecture of lymph nodes, and maturation and trafficking of lymphocytes through them.
Among systemic chemokines, SDF-1α3 has attracted major attention due to its role in preventing HIV-1 infection by T-tropic strains of the virus, which are known to employ the specific SDF-1α receptor, CXCR4 (5, 6). Knockout mice of both SDF-1α and its receptor show acute defects in lymphopoiesis and architecture of other nonlymphoid organs, which suggests a pleiotropic role for this chemokine in organogenesis (7, 8, 9). SDF-1α is the chemokine with the widest number of responding cells. CD34+ progenitors, T and B lymphocytes, NK cells and monocytes, as well as endothelium have been shown to bear functional CXCR4 receptors and migrate in response to SDF-1α (10, 11, 12, 13). CXCR4 belongs to the family of seven-transmembrane domain receptors. Binding of SDF-1α to CXCR4 has been shown to induce activation of Gi proteins, the Janus kinase/STAT pathway (14), phosphatidylinositol 3-kinase (15), and mitogen-activated protein kinase pathways (16). However, little is known of the effect of chemokines on the pathways that regulate the morphology of migrating cells. Among these pathways, small GTPases of the Rho subfamily rank as the most widely studied. Rho was originally described to be the main regulator of actin stress fiber and focal adhesion formation (17), and later was shown to be involved in integrin-mediated cell adhesion (18), as well as in the endothelial response to thrombin and stimulation of the contractility pathway in these cells (19). Rac was shown to regulate platelet-derived growth factor-induced lamellipodia formation and spreading (20), and these results have been extended to cells other than fibroblasts and other stimuli aside from platelet-derived growth factor (21, 22). In addition, it has also been involved in phagocytosis (23). Cdc42, the third member of this family, was involved in filopodia formation, and one of its downstream effectors (24), the Wiskott-Aldrich syndrome protein, has been shown to be an important effector in actin-based motility (25). In addition, Cdc42 has been involved in leukocyte chemotaxis and activation of Wiskott-Aldrich syndrome protein (26, 27).
Previously, we have shown that Cdc42, Rac1, and RhoA are involved in the control of polarity and chemotactic response to SDF-1α in constitutively polarized T cell lines (28). Activation of these GTPases was shown to inhibit cell polarization by exerting different effects on the lymphocyte morphology. Hence, RhoA activation induced contraction of the lymphocyte cytoskeleton, leading to rounding of the cell; Rac1 induced a dramatic spreading of the cells with lack of defined rear-front polarity; and Cdc42 induced filopodia formation, also leading to an inhibition of lymphocyte migration.
In this study, we examined in detail the activation and functional involvement of the Rho and Rho-dependent effector p160ROCK (Rho coiled-coil kinase) in the pathways elicited by the chemokine SDF-1α binding to the CXCR4 receptor in normal PBL. Stimulation of lymphocytes with SDF-1α induced activation of RhoA and p160ROCK, and phosphorylation of myosin light chain (MLC). Inhibition of either Rho or p160ROCK caused a dramatic change in the lymphocyte morphology and the blocking of chemotaxis induced by SDF-1α. MLC kinase (MLCK) inhibition resulted in the rounding of the cells, and also impaired chemotaxis. Together these results indicate that Rho, p160ROCK, and MLCK are required for the establishment of a normal polarized lymphocyte shape, and for the chemotactic response.
Materials and Methods
29), was a kind gift of Yoshitomi (Saitama, Japan). PMA was from Sigma (St. Louis, MO). Mouse mAb against Rac (clone 102, IgG2b) was from Transduction Laboratories (Lexington, KY), and, according to the manufacturer’s intructions, recognizes both Rac1 and Rac2. Mouse monoclonal anti-RhoA (clone 26C4, IgG1) was from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal sera against p160ROCK and phosphorylated MLC (P-MLC20) have been previously described (30, 31). mAb against total MLC (clone MY21, IgM) was from Sigma. FITC-coupled and Alexa568-coupled phalloidin were from Molecular Probes (Eugene, OR). [γ-32P]ATP (3 × 106 Ci/mol) was from Amersham Pharmacia Biotech (Uppsala, Sweden). mAbs against acetylated, tyrosinated, and α-tubulin were from Sigma.
Peer T cells have been previously described (15). PBL were obtained as described (15). Briefly, mononuclear cells were isolated from freshly prepared buffy coats using a Lymphoprep density gradient, followed by two rounds of adherence to plastic to deplete monocytes. A typical population comprises 70–75% T lymphocytes and 5–10% B lymphocytes, both being responsive to SDF-1 (11, 12).
Rho small GTPase activity assays
GST-C21, which recognizes active RhoA, and GST-PAK-CD, which recognizes active Rac and Cdc42, were kindly donated by J. Collard (The Netherlands Cancer Institute, Amsterdam, The Netherlands), and were prepared as described (32). Pull-down experiments were performed as follows: 15 × 106 Peer T cells or 25 ×106 PBL were resuspended in RPMI 1640 medium (Flow Laboratories, Irvine, U.K.) containing 0.1% BSA, and were stimulated at different times with 10 nM SDF-1α. Following incubation, the cells were washed twice with ice-cold HBSS and lysed at 4°C in buffer containing 50 mM Tris, pH 7.4, 100 mM NaCl, 10% glycerol, 1% Nonidet P-40, 2 mM MgCl2, 1 mM PMSF, 2 mM benzamidine, and COMPLETE cocktail inhibitor tablets (Roche Boehringer Mannheim, Indianapolis, IN). Cell lysates were clarified by centrifugation at 20,000 × g for 15 min at 4°C. The protein content in the cell lysates was measured before the pull-down using a protein detection kit (Bio-Rad, Hercules, CA). Equal amounts of protein were incubated with beads coupled to either GST-C21 or GST-PAK-CD for 60 min at 4°C, then washed four times in lysis buffer, resuspended in Laemmli buffer, separated in 15% SDS-PAGE, and transferred to nitrocellulose membranes. Western blot was performed using Abs against RhoA or Rac, followed by a HRP-conjugated anti-mouse serum. Detection of chemiluminescence was performed using SuperSignal Pico detection kit from Pierce (Rockford, IL). Protein loading was controlled using the protein detection kit and by Western blot of one-tenth of the sample loaded on a separate blot. For statistical purposes, gels of active and total GTPase were subjected to densitometric analysis and normalized with respect to the loading control. Arbitrary units obtained for each time point of SDF-1α stimulation were then referred to the value of the untreated cells point (t = 0) to obtain fold induction.
In vitro kinase assay
To analyze p160ROCK activity, 25 × 106 PBL were resuspended in RPMI 1640 medium containing 0.1% BSA at 37°C. When indicated, the cells were incubated with 50 μg/ml C3 exoenzyme for 12 h, 20 μM Y-27632, or 0.75 μg/ml PT for 30 min. Cells were then stimulated with 10 nM SDF-1α for the times indicated, rinsed twice in ice-cold HBSS, and lysed in buffer containing 50 mM Tris, pH 7.4, 100 mM NaCl, 10% glycerol, 0.5% Triton X-100, 10 mM MgCl2, 1 mM Na3VO4, 10 mM NaF, 1 mM PMSF, and COMPLETE cocktail inhibitor tablets (Roche Boehringer Mannheim). Cell lysates were clarified by centrifugation at 20,000 × g for 15 min at 4°C, and protein content in the cell lysates was measured before the pull-down using a protein detection kit (Bio-Rad). Protein extracts were precleared by incubation with 20 μl Pansorbin cells (Calbiochem), centrifuged (15,000 × g, 1 min), and immunoprecipitated with the 34490 pAb against p160ROCK, followed by protein A-Sepharose (50 μl/sample) for 60 min. Samples were centrifuged (15,000 × g, 15 min, 4°C); the Sepharose pellet was washed twice with lysis buffer and twice with kinase buffer containing 50 mM Tris, pH 7.4, 100 mM NaCl, 10% glycerol, 0.05% Triton X-100, 2 mM MgCl2, 2, mM MnCl2, 1 mM Na3VO4, and 10 mM NaF. p160ROCK immunoprecipitates were washed twice with lysis buffer and with kinase buffer, and then incubated with histone 2B (H2B; Roche Boehringer Mannheim) in reaction buffer kinase buffer plus 10 μM ATP and 10 μCi [γ-32P]ATP (Amersham Pharmacia Biotech) for 30 min at 37°C. The phosphorylation reaction was stopped by addition of 3× Laemmli buffer. Samples were resolved by SDS/PAGE, and radioactivity was analyzed by autoradiography. Protein loading was controlled by Western blotting of one-third of the sample loaded on a separate blot with an anti-p160ROCK pAb. For statistical purposes, in vitro kinase gel and Western blot showing loading control were subjected to densitometric analysis and normalized with respect to the loading control. Arbitrary units obtained for each time point of SDF-1α stimulation were then referred to the value of the untreated cells point (t = 0) to obtain fold induction with respect to the untreated control.
To analyze MLC phosphorylation, 25 × 106 PBL were resuspended in RPMI 1640 medium containing 0.1% BSA at 37°C. When indicated, the cells were preincubated with 50 μg/ml C3 exoenzyme for 12 h, and 20 μM Y-27632, 20 μM ML-7, or 0.75 μg/ml PT for 30 min. Cells were then stimulated with 10 nM SDF-1α for the times indicated and rinsed twice in ice-cold HBSS. Cellular proteins were then precipitated with 500 μl/sample of ice-cold 10% TCA for 30 min. Precipitates were rinsed twice in acetone containing 10 mM DTT to remove traces of TCA. Proteins were solubilized in Laemmli buffer and subjected to 15% SDS/PAGE. To detect P-MLC20, a rabbit pAb against P-MLC20 was employed (31). Phosphorylated myosin was correlated to total myosin by comparison with a duplicate blot in which total MLC was evaluated by Western blot. For statistical purposes, Western blots for P-MLC20 and total MLC (loading control) were subjected to densitometric analysis and P-MLC20 signal normalized with respect to the loading control. Arbitrary units obtained for each time point of SDF-1α stimulation were referred to the value of the untreated cells point (t = 0) to obtain fold induction with respect to the untreated control.
Actin polymerization assay
To evaluate actin polymerization induced by SDF-1α, 1.5 × 106 PBL in 100 μl RPMI 1640 were treated with 50 μg/ml C3 exoenzyme for 12 h, and 20 μM Y-27632, 20 μM ML-7, 20 μM PD98059, or vehicle alone for 30 min at 37°C, and stimulated with 10 nM SDF-1α under discontinuous stirring conditions for the times indicated. Intracellular actin polymerization was stopped by addition of an equal volume of a solution containing 4% formaldehyde in PBS, 1% Triton X-100, and 5 μg/ml FITC-conjugated phalloidin. The cells were incubated for 15 min at 37°C and washed once in PBS, and the intracellular polymerized actin was then determined in a FACScan flow cytometer (BD Biosciences, Mountain View, CA) using CellQuest software. Statistics shown are performed on normalized data in which every time point value was referred to the value of untreated cells (t = 0), thus showing fold induction of actin polymerization.
Immunofluorescence microscopy and polarization assay
Immunofluorescence experiments were performed essentially as described (15). Briefly, 1–1.5 × 106 PBL were incubated in flat-bottom, 24-well plates (Costar, Cambridge, MA) in a final volume of 500 μl complete medium on coverslips coated with human fibronectin (FN) at 50 μg/ml. When indicated, cells were pretreated with C3 exoenzyme for 12 h or other inhibitors for 30 min at 37°C. Then, 10 nM SDF-1α was added, and cells were allowed to adhere for 30 min at 37°C and 5% CO2 atmosphere. Cells were then fixed in 3.7% formaldehyde and permeabilized with 0.5% Triton X-100 for 5 min at room temperature. MLC and α-tubulin were visualized by staining the cells with the anti-MLC MY-21 (Sigma) mAb or an anti-α-tubulin (Sigma), respectively, plus a 1/500 dilution of rhodamine-X-labeled goat F(ab′)2 anti-mouse Ig (Molecular Probes). For actin visualization, the cells were stained with a 1/50 dilution of Alexa568-labeled phalloidin. Cells were observed using a Nikon Labophot-2 photomicroscope with ×40, ×60, and ×100 oil immersion objectives. Images were acquired with a COHU high performance CCD camera (Cohu, Tokyo, Japan) coupled to the microscope and connected to a LEICA Q550CW workstation (Leica Imaging Systems, Cambridge, U.K.). Images were visualized, processed, and stored by using the LEICA QFISH software version V1.01 (Leica). For quantification experiments, the proportion of spike-bearing cells was calculated by random choice of 10 different fields (×60) and direct counting of total cells 400–500(400–500) and cells in which at least one α-tubulin-positive spike could be observed.
Microtubule stability assay
PBL (20 × 106) were pretreated with C3 exoenzyme for 12 h or other inhibitors for 30 min at 37°C. Cells were then lysed in buffer containing 0.1% SDS, 1% Nonidet P-40, and 0.5% deoxycholate in TBS for 30 min at 4°C. Lysates were then centrifuged for 15 min at 14,000× rpm at 4°C in a laptop centrifuge, and 3× Laemmli buffer was added to the supernatants. Protein concentration was measured before loading into the gel using a protein detection kit (Bio-Rad). Proteins were resolved in a 10% SDS/PAGE under reducing conditions, and Western blot was performed with Abs specific to acetylated, tyrosinated, or total α-tubulin in the same blot. Stripping of the membranes was performed when required by incubating twice the membrane with 0.1 M Tris-glycine at pH 2.5 for 15 min, followed by washing of the membrane with TBS-0.1% Tween 20. Complete stripping was evaluated by incubation of the membrane with the HRP-conjugated rabbit anti-mouse Ab employed as a secondary Ab for the blots. Samples were then quantified by densitometry, as described above, and both acetylated and tyrosinated tubulin correlated to total α-tubulin.
Assays were performed in polycarbonate membranes, 6.5 mm diameter, 10 μm thickness, 3-μm-diameter pore size Transwell cell culture chambers (Costar). Human PBL (100 μl at 10 × 106/ml) suspended in RPMI 1640/0.1% human serum albumin were added to the upper chamber, and SDF-1α was added to the lower well. When indicated, cells were pretreated with 20 μM Y-27632, 20 μM ML-7, 10 μM butanedione monoxime, or 20 μM PD98059 for 30 min, or C3 exoenzyme for 12 h at 37°C. Cells were allowed to migrate for 3 h at 37°C in 5% CO2 atmosphere. Migrated cells were recovered from the lower part of the chemotaxis chamber and counted by flow cytometry. Briefly, cells were stained with propidium iodide and were counted for 1.5 min, calibrating the flow rate of the FACScan with Trucount tubes (BD Biosciences). Cell chemotaxis was expressed as the migration index, which was calculated with the following formula: (number of cells in the lower well/number of cells in the lower well + number of cells in the upper chamber) × 100.
SDF-1α binding to CXCR4 induces RhoA, but not Rac activation
SDF-1α has been described to activate multiple signaling pathways in different cell types via the receptor CXCR4 (16). To investigate whether SDF-1α activated the small GTPase RhoA in lymphocytes, Peer T cells, which express high levels of the CXCR4 on their surface (15), were stimulated with 10 nM SDF-1α, a concentration that has been previously reported as optimal for T and B cell migration (12, 15). Cell lysates were incubated with GST-C21 as previously described (32), and precipitates were subjected to SDS-PAGE, followed by Western blot with mAbs against RhoA (Fig. 1⇓, A and C). Bound RhoA fraction showed that RhoA undergoes sustained activation through time (up to 30 min), reaching an activation plateau at 15 min. In contrast, GST-PAK-CD coupled to glutation beads, which has been shown to bind to active Rac (32), showed no significant Rac activation. As a positive control, lymphocytes were stimulated with PMA, causing a 2-fold Rac activation, which has been described for neutrophils (33) (Fig. 1⇓B). Similar experiments performed with human PBL revealed that upon treatment of the cells with SDF-1α in these cells, RhoA undergoes a rapid activation (2 min), reaching a plateau between 5 and 15 min, declining thereafter (Fig. 2⇓A). Quantitative estimation of the data presented showed a 4-fold induction in RhoA activation when compared with unstimulated cells (Fig. 2⇓C). In contrast, no significant activation of Rac was detected in PBL upon SDF-1α treatment (Fig. 2⇓, B and C), as determined by pull-down assays performed with glutation beads coupled to GST-PAK-CD. Again, treatment of PBL with PMA caused a significant increase in the activity of Rac (Fig. 2⇓B).
SDF-1α induces p160ROCK activation
p160ROCK (Rho-associated kinase), is a well-known downstream effector of RhoA, whose activation has been described in V14RhoA-transfected cells (34). To investigate whether SDF-1α induced p160ROCK activation, PBL were stimulated with 10 nM SDF-1α, lysed, and immunoprecipitated with an Ab against p160ROCK (Fig. 3⇓, A and B). Then, in vitro kinase assays were performed on the immunoprecipitates, employing H2B as a high affinity substrate for this kinase (35). A significant activation of p160ROCK was observed in SDF-1α-treated cells after 2–5 min, declining thereafter (Fig. 3⇓, A and B), which correlates with SDF-1α-induced RhoA activation (Fig. 2⇑, A and C). To assess the dependence of p160ROCK activation on RhoA activation, PBL were either treated overnight with C3 exoenzyme, a well-described and specific Rho inhibitor (36, 37), or treated for 30 min with the p160ROCK inhibitor Y-27632. The blockade of Rho by C3 exoenzyme or p160ROCK by Y-27632 inhibited SDF-1α-induced p160ROCK activation (Fig. 3⇓, C and D), demonstrating the dependence of RhoA-mediated signaling for p160ROCK activation. In addition, the inhibition of Gαi by pertussis toxin (PT) also impaired SDF-1α-induced p160ROCK activation (Fig. 3⇓C).
SDF-1α induces MLC phosphorylation
Myosin light chain phosphorylation is a key event in the association of myosin to actin in the formation of actomyosin motors, essential for cell migration. A mechanism that involves MLCK activation induced by Rho and p160ROCK has been recently proposed in highly invasive rat hepatoma cells (38). To investigate whether activation of p160ROCK by SDF-1α results in myosin phosphorylation, precipitated proteins from SDF-1α-treated peripheral lymphocytes were subjected to Western blot using phosphorylation-specific anti-MLC pAb. Interestingly, SDF-1α-treated lymphocytes showed increased levels of P-MLC20, with a rapid activation kinetics that correlated with those of RhoA and p160ROCK activation when the cells were stimulated with SDF-1α (Fig. 4⇓, A and B), thus indicating the activation of the different components in this signaling cascade. To further assess the involvement of the RhoA-p160ROCK axis in MLC phosphorylation, cells were pretreated with ML-7, a specific inhibitor of the MLCK, or with C3 exoenzyme or Y-27632 and stimulated with SDF-1α. A complete dependence of RhoA, p160ROCK, and MLCK for MLC phosphorylation was observed, thus confirming the role of the RhoA-p160ROCK-MLCK axis in MLC phosphorylation (Fig. 4⇓, C and D). Also, inhibition of Gαi proteins by PT blocked SDF-1α-mediated MLC phosphorylation (Fig. 4⇓, C and D).
SDF-1α activates actin polymerization by a p160ROCK-independent pathway
SDF-1α has been shown to induce a very rapid and transient actin polymerization in different cell types, including lymphocytes (11). As p160ROCK has been suggested to play a pivotal role in the actin cytoskeleton remodelling after stimulation of cells with different stimuli, we have investigated whether p160ROCK is involved in early actin polymerization induced by SDF-1α. For this purpose, PBL were pretreated with 50 μg/ml C3 exoenzyme or 20 μM of the specific p160ROCK inhibitor Y-27632, doses that completely abolish lymphocyte migration to SDF-1α (see below). As a control, cells were pretreated with 20 μM PD98059, a specific MEK1 inhibitor that impaired neither lymphocyte polarization nor chemotaxis induced by SDF-1α (Ref. 15 and this study). This inhibitor was found to abolish activation of MEK1 by SDF-1α, as determined by employment of a phosphospecific anti-MEK1 Ab (Ref. 16 and data not shown). As shown in Fig. 5⇓, there was no difference among C3 exoenzyme-, Y-27632-, and vehicle-treated cells, arguing against the involvement of p160ROCK in actin polymerization induced by chemokines. Conversely, pretreatment of the cells with 0.75 μg/ml PT completely blocked SDF-1α-induced actin polymerization, confirming the dependence of Gαi proteins in actin-related SDF-1α-induced responses. In addition, cells were treated with ML-7, and this treatment was found to partially inhibit SDF-1α-induced actin polymerization (Fig. 5⇓).
Role of p160ROCK in the acquisition of a polarized morphology in PBL
We have previously described the drastic changes in PBL morphology induced by the chemokine SDF-1α, and suggested a key role for the Rho-regulated signaling pathway in the maintenance of cell morphology (28). To elucidate the role of RhoA and p160ROCK in the morphology of PBL, freshly isolated cells were incubated with 50 μg/ml C3 exoenzyme, 20 μM Y-27632, or stimulated with 10 nM SDF-1α, as previously described (15), and allowed to adhere to FN-coated dishes. Immunofluorescence analyses revealed that SDF-1α induced the change from a flat, round shape to a bell-like, polarized morphology. MLC and F-actin concentrated at leading edge of the cells as revealed by Alexa568-conjugated phalloidin staining and indirect immunofluorescence with anti-MLC Ab (Fig. 6⇓, E and F). Cell treatment with C3 exoenzyme induced a dramatic change in the morphology of the lymphocytes, promoting the formation of one or two long, needle-like spikes from the cell membrane with high efficiency (see below) (Fig. 6⇓, G–I). Actin localization was severely altered by pretreatment with the inhibitor, with no actin-based lamellipodia being detected (Fig. 6⇓I). MLC staining was also affected, and it was found mainly at the tip of the protrusions (Fig. 7⇓H, hollow arrowheads). Y-27632 induced a similar phenotype, but some differences could be observed. First, the protrusions were somewhat shorter than C3-induced spikes (note in Fig. 6⇓L), and second, Y-27632 is less efficient in inducing spikes in lymphocytes (Fig. 7⇓B, see below). However, the addition of a specific MLCK inhibitor, ML-7, induced a different cell shape, which returned to a round morphology similar to untreated PBL (Fig. 6⇓, M–O) or butanedione monoxime-treated T lymphoblasts (39). These changes were found to be independent of the addition of SDF-1α (data not shown), which suggests a pivotal role for both Rho and p160ROCK in regulating the basal morphology of lymphocytes.
Spike projection in C3- and Y-27632-treated lymphocytes is dependent on microtubule dynamics
The dramatic changes induced in lymphocytes by Rho and p160ROCK inhibitors C3 and Y-27632 pointed to a disruption on the actin cytoskeleton. However, the existence of a cross talk between the actin and tubulin cytoskeletons (40) prompted us to investigate whether microtubules played a role in these phenomena. PBL treated with the inhibitors C3 and Y-27632 exhibited strong staining for tubulin alongside the spikes, thus involving microtubule elongation in spike projection (Fig. 7⇑A). This effect was quantitatively evaluated in α-tubulin-stained cells, in which the projections were more identifiable, and the results showed that C3 is about 4 times more efficient than Y-27632 in promoting spike projection (Fig. 7⇑B). In contrast, either untreated, SDF-1α-, or ML-7-treated lymphocytes showed no significant spike projection (Fig. 7⇑, A and B). It is interesting to note that when the cells were left in suspension or allowed to adhere to the integrin-independent substrate poly-l-lysine, the phenotype induced by either C3 or Y-27632 was less accentuated with the projection of shorter spikes, suggesting that integrin-dependent adhesion to the substrate may also contribute to the regulation of the phenotype induced by Rho inhibition (data not shown).
The existence of a correlation between posttranslational modifications and the degree of microtubular stability has been demonstrated; in this regard, acetylated and detyrosinated modifications are found in pools of highly stable tubulin (41, 42), whereas tyrosinated tubulin reflects a more unstable, dynamic state (42). To investigate the degree of stability of the microtubules involved in spike projection, lymphocytes treated with the inhibitors were lysed and Western blot was performed with Abs specific to either acetylated, tyrosinated, or total α-tubulin. Notably, we found a decrease in the levels of acetylated tubulin in C3-treated cells that correlated well with an increase in tyrosinated, dynamic tubulin (Fig. 7⇑, C and D). However, Y-27632 exerted a much lower effect in tubulin stabilization, which correlated with its lower efficiency in promoting spike extension in a significant proportion of lymphocytes. As a control, we employed paclitaxel, a well-described cytostatic with a clear-cut effect on tubulin stabilization (Fig. 7⇑, C and D). These results support a role for dynamic tubulin in spike extension in C3-treated cells, as well as suggest a pivotal role for RhoA and, to a lesser extent, p160ROCK in the stability and dynamics of lymphocyte microtubules.
Effect of the disruption of the actomyosin system and p160ROCK in SDF-1α-mediated chemotaxis
SDF-1α has been described as a potent chemoattractant for T, B lymphocytes, and cells of the monocytic lineage (11, 12, 13, 43). To investigate the requirement of an intact RhoA-p160Rho-myosin system in lymphocyte chemotaxis, PBL were pretreated with the different inhibitors used in this study and allowed to migrate in response to 10 nM SDF-1α in Boyden-modified chemotaxis chambers that were either uncoated or coated with 20 μg/ml FN. Interestingly, treatment of the cells with C3 exoenzyme, Y-27632, or ML-7 completely abolished SDF-1α-induced migration and spontaneous motility (Fig. 8⇓, A and B), in contrast with the myosin ATPase inhibitor butanodione monoxime, which blocked these effects only partially. These effects were found to be extracellular matrix independent, as coating of the transmigration chambers with 20 μg/ml FN affected neither SDF-1α-induced migration nor the effect of the inhibitors (Fig. 8⇓).
Chemokines have been described as specific mediators involved in physiological and pathological conditions for cells of the immune system. Among them, SDF-1α is a potent, pleiotropic mediator that acts as a systemic chemokine involved in lymphopoiesis and lymphocyte recruitment to the lymph nodes (44). At a cellular level, SDF-1α induces signals following engagement of the chemokine receptor CXCR4, and activates Gi proteins and different signaling pathways involved in cellular responses such as activation of transcription (16). Apart from its chemotactic effect, SDF-1α has been previously shown to enhance adhesion in T cells (45), and modify the morphology of lymphocytes (12, 15). It therefore exerts an effect on the actin cytoskeleton, which has been shown to be an essential requirement for chemokine-induced changes in lymphocyte morphology (39).
In this study, we provide evidence that SDF-1α chemokine induces activation of Rho, its downstream effector p160ROCK, and MLC phosphorylation in normal lymphocytes. Furthermore, this pathway is involved in the regulation of the lymphocyte shape, cell motility, and chemotactic response. It is worth noting that there is a great difference in the kinetics of RhoA activation between T cell lines such as Peer leukemic cells and freshly isolated human PBL. We therefore employed primary cells in this study to get a more clear-cut view of the possible physiological activation of Rho. Notably, Rac is not activated by SDF-1α, whereas Rho is activated. Our results concur with those recently described, showing that a reciprocal balance between Rac and Rho activity exists (32, 46). The failure of SDF-1α to activate Rac is also confirmed by correlation with the change in cellular morphology induced by chemokines, which results in extension of a pseudopod, but not appreciable increase in spreading, opposed to that observed by PMA treatment, which is Rac dependent (21). In this regard, PMA treatment induced Rac activation (Fig. 1⇑B), which confirms both lack of activation of Rac by SDF-1α and PMA activation of Rac by a direct pull-down assay in a leukocyte type different from neutrophils, where it was originally described (33).
Our data demonstrate that SDF-1α activates p160ROCK, also with a rapid kinetics, and this activation was found to be RhoA dependent. SDF-1α also promoted a rapid MLC phosphorylation in a RhoA-, p160ROCK-, and MLCK-dependent fashion. RhoA-dependent p160ROCK activation together with RhoA-, p160ROCK-dependent MLC phosphorylation demonstrate the existence of a linear pathway activated by SDF-1α. Activation of the RhoA-p160ROCK pathway was found to be Gi dependent, as p160ROCK activation and MLC phosphorylation were inhibited by PT. In this regard, activation of the Cdc42-Rac1 pathway by lysophosphatidic acid, which binds to a seven-transmembrane domain receptor homologous to CXCR4, has recently been shown to be pertussis dependent (47).
Inhibition of p160ROCK blocked actin polarization at the leading edge of the cells and impaired migratory capability. Thus, it is likely that SDF-1α triggers a polarized signal that rearranges the actin cytoskeleton favoring actin polymerization at the leading edge of the cell, and thus p160ROCK may act as a regulatory molecule that governs the spatial positioning of the actin fibers at the leading edge rather than de novo actin polymerization. In this regard, Rho has been shown to regulate both de novo actin polymerization and spatial rearrangement of newly formed actin cables through a double-effector system. p140 mDia, a homologue of the Drosophila diaphanous gene (48), is thought to act by regulating de novo actin polymerization, whereas p160ROCK regulates its positioning (49). Our data support these evidences obtained in the fibroblast system and extend them to lymphocytes, as inhibition of p160ROCK in lymphocytes does not exert any effect on the levels of polymerized actin nor failure in the response to SDF-1α in terms of de novo actin polymerization, but it causes a mislocalization of the network of F-actin. On the other hand, MLCK inhibition resulted in partial inhibition of actin polymerization induced by SDF-1α. A likely explanation for this observation would imply that interference with myosin activity and actomyosin function not only impairs cell contractility, but also the mechanisms of actin polymerization, perhaps due to a role of myosin in the proper spatial orientation of actin fibers during polymerization.
Recent work on fibroblast migration points to a role for the microtubule network in different steps of the migratory process (50, 51). Indeed, microtubules seem to play a role in the maintenance of the cell morphology, but to date this role as well as microtubule regulation have not been addressed in lymphocytes. We found the induction of long microtubular spikes arising from the lymphocyte cell body when RhoA and in a lower extent p160ROCK activities are impaired. The establishment of long protrusions had been previously reported in PBL (37), but its nature remained largely unknown. Our data clearly demonstrate that RhoA inhibition greatly reduces the pool of stable, acetylated tubulin, thus increasing the dynamic rate of microtubule polymerization. We conclude that RhoA inhibition, and p160ROCK to a lesser extent, increases tubulin instability, thus favoring the turnover rate, allowing the cell to produce dynamic structures that could account for microtubule-based spikes. These results concur with previous observations made in a mouse fibroblast system, in which both lysophosphatidic acid and transfection of an active mutant of RhoA increased the pool of stable (detyrosinated) tubulin, as determined by indirect immunofluorescence (52). Our results also offer a suitable explanation for the different changes in lymphocyte shape induced by RhoA or p160ROCK and MLCK inhibition. Hence, inhibition of MLCK exerted a negligible influence on microtubule stability, thus lacking long spikes (Fig. 6⇑), whereas more upstream effectors such as RhoA and p160ROCK seem to play a key regulatory role in such a process. The mechanism by which RhoA inhibition would induce tubulin growth is currently unknown. A RhoA activator, p190RhoGEF has been very recently shown to interact with microtubules (53). It has been postulated that microtubules are captured, and their targeting of focal adhesions destabilizes and ultimately dissociates the contact with the substrate (54, 55). An alternative interpretation of these data would point to a restraining role for focal adhesions in microtubule growth. Disruption of the cell contact with the substrate by interference with RhoA activity would allow microtubules to grow indiscriminately, which would account for spike protrusion. Due to the lack of focal adhesions in lymphocytes, we only can speculate with the whole cell substrate contact area playing such a role in microtubule constraint. Nevertheless, this hypothesis is further supported by the observation of C3 exoenzyme being more efficient in inducing spikes in other cell types with a higher degree of adherence than lymphocytes such as blood monocytes (M. Vicente-Manzanares, unpublished observations).
Blockade of Rho by either chemical inhibition in PBL or overexpression of a dominant-negative mutant in T cell lines has been previously shown to induce cell polarization (Refs. 28, 37 , and this study). However, this polarization did not result in increased cell motility or chemotaxis, but instead caused a partial inhibition (28) when a competitive dominant-negative mutant (N19RhoA) was employed, or complete when endogenous RhoA activity was abolished by C3 exoenzyme treatment, as shown in this study. Our data also demonstrate that p160ROCK inhibition completely impaired motility and chemotaxis. The different extent of inhibition caused by C3 exoenzyme and Y-27632 could be explained in terms of a more efficient blockade by the C3 exoenzyme and the p160ROCK inhibitor than by a dominant-negative mutant of RhoA. It is also clear that T cell lines may have alternative pathways to maintain chemotactic responses, which can be inferred from the lack of effect of the MLCK inhibitor, ML-7, on Peer T cell chemotaxis to SDF-1α (M. Vicente-Manzanares, unpublished observations), whereas the same dose of the inhibitor was able to block chemotaxis of normal PBL to the chemokine.
In summary, the data presented in this work provide a mechanistic insight about how chemokines activate the actomyosin cytoskeleton, and establish a requirement for intact Rho-p160ROCK-MLCK activities for the development of the chemotactic response of lymphoid cells.
We are indebted to Dr. John Collard for kindly providing the GST-C21 and GST-PAK-CD constructs, Dr. Fumio Matsumura for the anti-P-MLC20 Ab, and Dr. Shuh Narumiya for providing the anti-p160ROCK Ab. We also acknowledge Dr. B. Alarcón, L. del Peso, and J. L. Rodríguez for critical reading of the manuscript, and V. Centeno for editorial assistance.
↵1 This work was supported by Grants SAF99-0034-CO2-01 and 2FD97-0680-CO2-02 from the Ministerio de Ciencia y Tecnología, and QLTR-1999-01036 from the European Community to F.S.-M.
↵2 Address correspondence and reprint requests to Dr. Francisco Sánchez-Madrid, Servicio de Inmunología, Hospital de la Princesa, c/Diego de León, 62, E-28006 Madrid, Spain. E-mail address:
↵3 Abbreviations used in this paper: SDF, stromal cell-derived factor; FN, fibronectin; H2B, histone 2B; MLC, myosin light chain; MLCK, MLC kinase; P-MLC20, phosphorylated MLC; PT, pertussis toxin; ROCK, Rho coiled-coil kinase.
- Received July 19, 2001.
- Accepted October 30, 2001.
- Copyright © 2002 by The American Association of Immunologists