Dynamitin Controls β2 Integrin Avidity by Modulating Cytoskeletal Constraint on Integrin Molecules*

Dynamitin, a subunit of the microtubule-dependent motor complex, was implicated in cell adhesion by binding to MacMARCKS (Macrophage-enrichedmyristoylated alanine-riceC kinase substrate). However, how dynamitin is involved in cell adhesion is unclear despite the fact that both MacMARCKS and microtubules regulate β2integrin activation. We report that dynamitin regulates β2 integrin avidity toward iC3b by modulating the lateral mobility of β2 integrin molecules. Using the single particle tracking method, we found that integrin molecular mobility in cells expressing the fusion protein CFP (cyan fluorescent protein)-dynamitin or CFP-MB (the MacMARCKS binding domain peptide of dynamitin) increased 6-fold over the control cells, suggesting that disturbing dynamitin function dramatically altered the cytoskeletal constraint on β2 integrin molecules. Further mechanistic studies revealed that overexpression of dynamitin stimulated the phosphorylation of endogenous MacMARCKS protein, which lead to the enhanced tyrosine phosphorylation of paxillin. This effect of dynamitin correlates with the observation that higher concentration of PKC inhibitor is required to block β2 integrin mobility in dynamitin-expressing cells. Although dynamitin acts at the point of MacMARCKS phosphorylation, it is upstream of RhoA, because its effect was blocked by RhoA inhibitor. Thus, we conclude that dynamitin is a part of the cytoskeletal constraint that locks β2integrin in the inactive form.

The activation of ␤ 2 integrin, i.e. its increased affinity and avidity to its ligand, is essential in many physiological and pathological events such as leukocyte diapedesis (1,2), wound healing (3), and angiogenesis (4,5). Although lacking ␤ 2 integrin-mediated adhesion causes leukocyte adhesion deficiency syndrome (6), many autoimmune diseases are associated with excessive integrin-mediated cell adhesion (7,8). Thus, integrin activation is tightly regulated. In the case of ␤ 2 integrin, its ligand-binding activity on the extracellular surface of the plasma membrane is tightly regulated by a number of intracellular signals originating from other membrane receptors (for review see Ref. 9) such as T cell receptors or chemokine receptors (10). This phenomenon is termed "inside-out" signaling (11). Although the mechanism of inside-out signaling is not clearly understood, both protein kinase C (PKC) 1 and cytoskeleton are involved (for review, see Refs. 9 and 12).
In resting leukocytes, ␤ 2 integrin molecules are immobilized on the cell membrane in a diffused distribution, because they are constrained by cytoskeletal complex (13)(14)(15). Activating PKC causes the cytoskeleton to relax, which sets integrin molecules into a faster and freer diffusion mode, close to completely free diffusion (13)(14)(15). Enhanced mobility of integrin and the presence of ligands facilitate integrin molecules to form clusters (14,16,17). Patches of clustered integrin can be clearly observed on the activated cell membrane (16,17). Such clustering further enhances the avidity of integrin toward its ligands. Thus, the enhancement in the mobility of integrin molecules on the membrane is regarded as the initial event of integrin activation. A number of cytoskeleton components are involved in regulating integrin activation. The most apparent of these is actin microfilament. Its depolymerization caused by low dose cytochalasin D promotes lateral mobility of integrin molecules as well as ␤ 2 integrin-mediated adhesion (13)(14)(15). The PKC signal, via the phosphorylation of its substrate Mac-MARCKS, is also a key regulator of the cytoskeletal constraint on integrin molecules (13,15). Mutation of MacMARCKS or prevention of MacMARCKS phosphorylation blocks PKC-mediated integrin mobility as well as integrin-dependent adhesion (15, 18 -20). Other actin microfilament-associated cytoskeletal proteins such as talin may also be involved (21)(22)(23)(24).
In addition to microfilament and its associated proteins, a recent report has also placed microtubule in the cytoskeletal constraint of integrin (25). It has long been observed that disrupting microtubule affects cell spreading (26 -28). Particularly in macrophages, depolymerization of microtubules with nocodazole causes the spreading macrophages to retract quickly into a spherical shape (29). Kinesin, the plus enddirected microtubule-dependent motor protein, has also been implicated in cell spreading (28). Recently, observations showed that growing microtubules are targeted toward the focal contacts and promote the dissociation of these complexes (30 -32). Microtubules affect not only cell spreading via focal contacts but also the molecular activation of integrin (25,33). Our previous study showed that microtubule is a major player in the cytoskeletal constraint on integrin molecules (25). Disturbing microtubule dynamics, either by depolymerizing microtubules with nocodazole or by polymerizing microtubules with taxol, induces enhanced integrin molecular mobility on the cell membrane and leads to enhanced integrin-ligand interaction (25).
There are many possible mechanisms for microtubules to be involved in regulating the cytoskeletal constraint on ␤ 2 integrin molecules. Small GTP-binding proteins are clearly involved in the microtubule-dependent integrin activation and cell spreading (34 -37). However, less is known about the involvement of microtubule-associated proteins, particularly the role of microtubule-motor proteins. One such protein is dynamitin, a subunit of dynactin complex. The dynactin complex is an important regulator of dynein, a microtubule-dependent minus end-directed motor protein (reviewed in Refs. 38 -41). This complex consists of as many as 10 subunits: p150 Glued / p135 Glued doublet, p62, dynamitin (p50), Arp1 (actin-related protein 1), actin, actin-capping protein ␣ subunit, actin-capping protein ␤ subunit, p27, and p24 (for review see ref. 41). It is believed to be involved in the progressivity of motor protein (42) and in bridging microtubule motor protein to cargo membrane (38). Through its N-terminal domain, dynamitin protein interacts with MacMARCKS (43,44), and the latter has been shown to regulate integrin activation (15, 18 -20). Such interaction is disrupted by PKC-mediated phosphorylation of MacMARCKS (43). Disrupting dynamitin-MacMARCKS interaction promotes cell spreading (43), which coincides with the known function of MacMARCKS in regulating integrin functions (15, 18 -20) and the reported involvement of p62, the other subunit of dynactin, in cell adhesion (45).
However, cell spreading is a complicated process involving many steps starting with integrin activation, extracellular matrix organization, and ending with focal adhesion remodeling. It is not known at which step during cell spreading that dynamitin regulates cell spreading. Considering that (1) dynamitin interacts with MacMARCKS (43), (2) dynamitin is part of microtubuledependent motor protein (46), and (3) both microtubule (25) and MacMARCKS (15) are parts of the cytoskeletal constraint on ␤ 2 integrin molecules that regulate integrin molecular mobility and integrin activation, we examined the effect of dynamitin on the ligand binding activity (avidity) of integrin. Here we report that disruption of dynamitin function indeed directly stimulates integrin avidity as measured by its ligand binding. Furthermore, we found that the increased avidity is likely the result of the disruption of cytoskeletal constraint on integrin molecules and subsequent enhancement of integrin molecular mobility, which is linked to PKC and Rho GTPase in the signal transduction pathway leading to integrin activation.

EXPERIMENTAL PROCEDURES
Materials-RAW 264.7 cells and 293 cells were purchased from ATCC (MD). Hybridoma HB 226, which produces hamster anti-mouse ␤ 2 integrin antibody (2E6), and hybridoma TIB 112, which produces mouse IgM anti-sheep erythrocytes (S-S.3), were also purchased from ATCC and were grown in serum-free media from Invitrogen. Monoclonal anti-MHC-II antibody was kindly provided by Dr. R. Steinman (Rockefeller University, New York, NY). DMEM and other cell culture media were also purchased from Invitrogen. Anti-paxillin antibody was purchase from BD Bioscience. Anti-phosphotyrosine antibody 4G10 was purchased from Upstate Biotechnology Inc. cDNA-encoding dynamitin was obtained as described previously (43). Plasmid pECFP-C1 was purchased from CLONTECH. Rabbit anti-green fluorescence protein polyclonal antibody was purchased from Invitrogen. Goat anti-dynamitin antiserum was generated using purified glutathione S-transferase fusion protein of dynamitin at Ferrell Farms (Oklahoma City, OK). Sheep erythrocytes were also purchased from Ferrell Farms. The carboxylated fluorescent latex beads were purchased from Molecular Probe (Cat no. F-8811). PMA (phorbol 12-myristate 13-acetate), C5-deficient human serum, neuraminidase X, fluorescein isothiocyanate-conjugated monoclonal anti-␣-tubulin Clone DM 1A, and other routine chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). Exoenzyme C3 was purchased from Calbiochem. Other secondary antibodies were purchased from Jackson Immunological Laboratory. SuperFect trans-fection reagents were purchased from Qiagen.
Construction of Expression Plasmids and Transfection-The coding region of dynamitin was amplified by polymerase chain reaction with the 5Ј-primer, 5Ј-GGA CTC GGA TCC CCG GAA TTC-3Ј, and the 3Ј-primer, 5Ј-CTG TTC TCC GGA TCC CAA ATG-3Ј, using wild type dynamitin in pGEX plasmid (43) as template. The primers contained a BamHI site for easy cloning into pECFP-C1 downstream of CFP. "CFPdynamitin" indicates that the CFP is at the N terminus of dynamitin. cDNA-coding MacMARCKS binding domain (MB) of dynamitin was generated in the same way with primer pair 5Ј-primer, 5Ј-ATA AGC TTG GAT CCA TGG-3Ј, and the 3Ј-primer, 5Ј-TAG GAT CCA GAT TCA TAT CCT GTC-3Ј. The sequences of all polymerase chain reaction products were confirmed before use.
The dynamitin plasmids (2 g) described above were transfected into 5 ϫ 10 5 RAW cells or into 5 ϫ 10 5 293 cells using the SuperFect liposome method as described in the manufacturer's instructions (Qiagen). The RAW cells were analyzed by single particle tracking and Western blot 16 h after transfection.
Cell Preparation-RAW cells were cultured in DMEM with 10% fetal bovine serum (FBS). Before the experiment, 5 ϫ 10 5 cells in a 35-mm dish were washed with phosphate-buffered saline (PBS) once and then treated with neuraminidase X (1 milliunit in 3 ml of buffer containing 0.13 M NaCl, 0.05 M NaAc, pH 6.5) for 20 min. Then 0.5 ml of the cell suspension was added to 2.5 ml of Hanks' solution and plated onto an acid-washed coverslip coated with poly-L-lysine (13,47).
Conjugation of Antibodies to the Fluorescent Latex Beads-Antibody against ␤ 2 integrin (2E6) either in the form of whole antibody or Fab fragments at a concentration of 3 g/l was conjugated to the carboxylated beads according to the manufacturer's instructions (Molecular Probes). Because no difference between Fab-conjugated and whole antibody-conjugated beads was observed in single particle tracking (15), the whole antibody-coated beads were used in most cases. In a glass tube, 100 l of antibody (3 g/l) or BSA was added to the reaction mixture containing 50 l of MES (200 mM), 50 l of H 2 O, and 200 l of carboxylated latex beads and left at room temperature for 15 min. Then 2 mg of 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide was added to the tube, which was vortexed for 2 h at room temperature. At the end of 2 h, 450 l of glycine (1 M, pH 6.25) was added, and the tube was vortexed for an additional 30 min to terminate the reaction. The excess proteins were removed by dialyzing the reaction mixture in a dialyzing bag (molecular weight cutoff, 300,000) against 1 liter of 50 mM MES (pH 6.0) overnight. The buffer was then changed to PBS by dialyzing the beads against 1 liter of PBS for another 4 h. The beads were stored at 4°C in 1.5 ml of PBS containing 1% BSA and 0.02% azide.
The size of the beads, 200 nm, is below the resolution of a light microscope. However, with video contrast enhancement, a shadow at the size of 800 nm, similar to most intracellular organelles, can be seen. Because the microbeads we used have autofluorescence, the beads were illuminated at the end of the experiments for their fluorescence so that they would not be confused with the subcellular organelles.
Video Microscopy-Cells were plated on poly-L-lysine-coated coverslips in a steel chamber and placed on a water-jacked heating stage on a Zeiss Axiovert 100 microscope with minimum light to prevent "burning" the cells. On poly-L-lysine-coated surfaces in the serum-free medium, all types of cells we used adhered and spread equally well (data not shown). After the cells spread and clearly showed their lamellipodia, the anti-␤ 2 integrin antibody (2E6)-conjugated beads were added at a concentration of four to five beads per cell. Recording started and continued for 30 s as soon as the 2E6-conjugated beads fell onto and bound to the cells. Beads coated with BSA or antibody against MHC II were used as controls. Bead motion was observed using video-enhanced differential interference contrast microscopy. A charge-coupled device camera system (Optronic DE750) with digital contrast enhancement and a Pentium IV 1.4-GHz computer with 512 MB of RAM recorded the beads' motion directly into RAM at a rate of 30 frames per second. For each track, a 36-s segment was recorded using PIXCI software (Epix Inc., IL) and transferred later to a hard drive. The Location of the beads was determined automatically in each video frame for 15 s, i.e. 450 frames, using in-house-generated software.
Data Analysis-For each track of a bead, the mean square displacement (MSD) for each time interval was calculated from the xy coordinates of the particles according to Equations 1-3 (for review see Ref. 48) using in-house-generated software.
The obtained MSD is the sum of the random and directed motion, Thus, By fitting the MSD calculated from the experimental data to the quadratic equation (Equation 7) using SigmaPlot software, the diffusion coefficient D can be extracted.
RAW cells were plated on six-well plates with 5 ϫ 10 5 cells and cultured overnight, then transfected with CFP-dynamitin, CFP-MB, and CFP separately. Sixteen hours after transfection, the cells were incubated with E iC3b (2 ϫ 10 8 ) for 30 min. For positive and negative control, untransfected cells were treated with or without PMA, respectively, for 25 min while incubating with E ic3b . The cells were either fixed for photography or lysed for further analysis of hemoglobin content.
C3 Treatment-Raw cells were transfected as described above. Four hours into transfection, the transfection media were replaced with fresh DMEM containing 10% FBS. At this time, the cells were split into two parts: one medium contained 1 g/ml C3 exoenzyme and the other contained none. After 16 h of further incubation, the cells were subjected to single particle tracking analysis.
Tyrosine Phosphorylation of Paxillin-Cells in 100-mm dishes were transfected as described in the previous section. After 16 h, the cells were lysed in 500 l of ice-cold Tris-HCl lysis buffer (150 mM NaCl, 10 mM Tris-HCl, pH 8, 0.5% Triton X-100, and a mixture of protease inhibitors). Paxillin was immunoprecipitated with 2.5 g of anti-paxillin antibody for 1 h at 4°C, followed by a 1:1 mixture of 50% slurry of protein A-and protein G-Sepharose for 30 min at 4°C. After SDS-PAGE, proteins were transferred to Immobilon membrane and probed with 4G10, an anti-phosphotyrosine antibody, followed by peroxidaseconjugated secondary antibody. Later, the membrane was stripped and probed again for paxillin to show the amount of paxillin in each lane.
MacMARCKS Phosphorylation-Sixteen hours after transfection, cells were labeled with [ 32 P]P i (500 Ci/60-mm dish) for 2 h in phosphate-free DMEM containing 10% dialyzed FBS. The cells were then lysed in the lysis buffer (49), and MacMARCKS protein was immunoprecipitated with polyclonal anti-MacMARCKS antibody (50). The immunoprecipitant was then subjected to SDS-PAGE and radioautography.

Expression of CFP-dynamitin and Its
Mutant-To visualize cells expressing transfected dynamitin, we tagged wild type dynamitin with CFP at its N-terminal. However, because CFP is a 266-amino acid protein that might interfere with the natural function of the tagged protein, we first tested whether the fusion protein was capable of disrupting endogenous dynamitin function. We did this based on the observation that the overexpression of Myc-tagged dynamitin causes accumulation of cells at the metaphase and generation of multispindles in cells (46). Because this effect was observed only in epithelial cells, not in macrophages, we tested CFP-dynamitin in 293 epithelial cells. Twenty hours after transfection with cDNA-encoding CFP-dynamitin, the 293 cells were stained for microtubules to determine how many cells were blocked in metaphase and whether multispindles were accumulated. We observed ϳ12% of CFP-dynamitin-expressing cells in metaphase, ϳ2-fold higher than the control cells (5%). Also, we saw an increased number of cells with multispindles, similar to a previous report (46). A similar result was obtained with CFP-MB-transfected cells. Thus, we concluded that CFP-dynamitin maintained its ability to disrupt dynactin functions (Fig. 1).
To study the role of dynamitin in ␤ 2 integrin activation, we then transfected RAW macrophage cells with cDNA-encoding CFP-dynamitin) and CFP-MB. We used RAW macrophage cells instead of 293 epithelial cells, because ␤ 2 integrin expression is restricted to leukocytes. Fig. 2A shows a Western blot with anti-dynamitin antibody on total cell lysate of cells transfected with CFP-dynamitin, CFP-MB, CFP alone, and non-transfected cells. In cells transfected with CFP alone, the endogenous wild type dynamitin is recognized by anti-dynamitin antibody. In addition to the endogenous dynamitin, cells transfected with full-length CFP-dynamitin also show a band at 76 kDa, equal to the sum of the molecular weight of dynamitin and CFP. Because the additive molecular weight of the CFP and MB fragment roughly equals that of wild type dynamitin, the expected migration of this fusion protein is superimposed with wild type endogenous dynamitin. To further clarify this point, the same filter was stripped and blotted again with antibody against green fluorescent protein. In CFP-transfected cells, a band appeared around 26 kDa (Fig. 2B). In CFP-dynamitin cells, the CFP-dynamitin band appeared again at the same position as anti-dynamitin antibody (Fig. 2B). A band of protein equal to the additive molecular weight of CFP and MB showed in CFP-MB-transfected cells, suggesting that CFP-MB is also expressed in these cells (Fig. 2B).

Transfection of Dynamitin and MB Induced ␤ 2 Integrin
Binding to iC3b-coated Sheep Erythrocytes in RAW Macrophages-By definition, integrin activation is its enhanced binding to its ligand. Although dynamitin is implicated in cell spreading (43), it is unclear whether it causes integrin activation or affects focal adhesion, because cell spreading is a sum of multiple cellular events. Therefore, we tested whether overexpression of dynamitin or its mutant would directly activate ␤ 2 integrin to bind to iC3b, a ligand that binds only to activated ␣ m ␤ 2 integrin in macrophages stimulated by chemokines, phorbol esters, or other PKC-mediated stimulus (47). In this experiment, sheep erythrocytes were coated with iC3b and were used as a ligand. In non-transfected cells and CFP-transfected cells only when activated by external addition of PMA, ␤ 2 integrin on macrophages bound to iC3b-coated erythrocytes and formed rosettes (Fig. 3A). However, in cells transfected with CFPdynamitin or CFP-MB, rosettes were formed even without PMA stimulation (Fig. 3A). This effect was further quantified by lysing the bound erythrocytes in water and measuring the concentration of hemoglobin in the solution. A 13-and 12-fold increase in erythrocyte binding was observed in the dishes containing CFP-dynamitin and CFP-MB-transfected cells with about 25% transfection efficiency, respectively (Fig. 3B). These data suggest that dynamitin is indeed involved in the direct activation of ␤ 2 integrin and that the disruption of dynamitin function actually bypasses the requirement of external stimulation of PKC.

Dynamitin Expression-promoted ␤ 2 Integrin Activation Is a Result of the Relaxation of Cytoskeletal Constraint on Individual ␤ 2 Integrin Molecule-The obvious questions are (a) at which step of integrin activation is dynamitin involved and (b)
where is dynamitin in the regulatory signal pathway of integrin activation? To address the first question, we tested whether dynamitin played a role in the cytoskeletal constraint on integrin molecules, because dynamitin is a cytoskeletal protein and binds to the MacMARCKS protein (43) and the latter is a part of the cytoskeletal complex restraining integrin in an inactive state (15). After transient transfection of RAW cells with CFP-dynamitin, CFP-MB, and CFP control, the molecular mobility of ␤ 2 integrin was monitored on these transfected cells (which was identified by CFP fluorescence). The molecular mobility of integrin molecules was reflected by the particles conjugated to each integrin molecule, which in turn reflected the cytoskeletal constraint on the integrin molecules at the cytoplasmic domain. We found that the lateral molecular mobility of b 2 integrin at the surface of CFP-dynamitin or CFP-MB expressing cells increased by 6.31-and 6.13-fold, respectively (Fig. 4A), to a similar level as PMA stimulation and artificial depolymerization of microfilaments and microtubules (15,25). Randomness analysis of each track revealed that, after expression of dynamitin or its mutant, the motion of ␤ 2 integrin was non-directed (random) and close to free diffusion (data not shown), suggesting that the cytoskeletal restriction is released. This effect is not generalized to all membrane proteins, because the molecular mobility of MHC II was not changed in dyna- mitin-transfected cells compared with control cells (Fig. 4B). Thus, we conclude that dynamitin is a part of the cytoskeletal constraint on ␤ 2 integrin molecules. Its effect on integrin activation is exerted at least partly at the level of initial relaxation of cytoskeletal constraint on integrin molecules.
Overexpression of Dynamitin Modulates Protein Phosphorylation in the Integrin Activation Pathways-To investigate the mechanism of how dynamitin is involved in regulating the cytoskeletal constraint on integrin molecules, we examined the effect of dynamitin expression on protein phosphorylation related to integrin activation. MacMARCKS phosphorylation has been shown be a key requirement in the relaxation of the cytoskeletal constraint on integrin molecules (13), thus Mac-MARCKS phosphorylation was first examined. Cells transfected with CFP-dynamitin were metabolically labeled with [ 32 P]P i , and MacMARCKS protein was immunoprecipitated with anti-MacMARCKS antibody. MacMARCKS protein was specifically immunoprecipitated, and a dramatic increase in MacMARCKS phosphorylation was observed in cells expressing CFP-dynamitin compared with vector-transfected cells (Fig.  5). Similarly, MacMARCKS phosphorylation was also increased to the same extent in CFP-MB-transfected cells (Fig. 5). The increased MacMARCKS phosphorylation in these cells agrees with the previously reported role of MacMARCKS in ␤ 2 integrin activation (15, 18 -20), which suggested that the point of action for dynamitin could be at the MacMARCKS phosphorylation.
Another important hallmark of MacMARCKS phosphorylation and ␤ 2 integrin activation is the increased tyrosine phosphorylation of paxillin. Previous reports show that tyrosine phosphorylation of paxillin is MacMARCKS-and ␤ 2 integrindependent (18,51,52). To further confirm that dynamitinstimulated MacMARCKS phosphorylation indeed results in the activation of ␤ 2 integrin signaling pathways, we also examined the phosphorylation of paxillin. Indeed the tyrosine phosphorylation was obviously increased in cells transfected with either CFP-dynamitin or CFP-MB compared with the CFP vector-transfected control cells, whereas the total paxillin loaded on the gel is roughly the same (Fig. 6). The phosphorylation of paxillin correlates with the activation of ␤ 2 integrin in these cells, as observed in the iC3b binding experiment in the previous section, again emphasizes the role of dynamitin in the activation signal transduction pathways of ␤ 2 integrin initiated by promoting MacMARCKS phosphorylation.
MacMARCKS Phosphorylation and Rho A Are Required in Dynamitin-stimulated Cytoskeletal Relaxation-Based on the above observations, dynamitin overexpression-induced cytoskeletal relaxation is related to MacMARCKS phosphorylation. Thus, inhibition of MacMARCKS phosphorylation should block the dynamitin-promoted increase of ␤ 2 integrin mobility. Therefore, we tested the effect of inhibiting MacMARCKS phosphorylation in dynamitin-transfected cells. Staurosporin was used here, because it is known to inhibit MacMARCKS phosphorylation under many conditions. 2 As shown in Fig. 5, staurosporin dramatically reduced MacMARCKS phosphorylation in dynamitin-transfected cells (at 200 nM). At the same time, staurosporin also blocked dynamitin expression-induced integrin mobility (Fig. 7), suggesting that MacMARCKS phosphorylation is required in dynamitin's effect on ␤ 2 integrin mobility.
On the other hand, small GTP binding proteins have been widely implicated in cell spreading (34 -37). Rho is reported to be part of the signal transduction pathway in microtubule-dependent integrin mobility (25). We observed that inhibition of Rho function by C3 dramatically reduced dynamitin-induced increase in integrin mobility (Fig. 7), suggesting that dynamitin is upstream of the Rho signal in regulating the cytoskeletal constraint on integrin molecules. DISCUSSION In this report, we showed that dynamitin, a subunit of the dynactin/dynein motor complex of microtubules, regulated ␤ 2 integrin activation. Disruption of dynactin function induced integrin binding to its ligand, iC3b. Most importantly, unlike normal macrophages where the enhanced avidity toward iC3b is PKC-dependent, the expression of dynamitin has the same effect as activating PKC by PMA.
To explain the potential mechanism underneath dynamitin involvement, we checked the involvement of dynamitin in the cytoskeletal constraint on integrin molecules, because dynamitin is a part of the microtubule cytoskeleton system. As previously recognized (13), the integrin molecules are immobilized by a cytoskeleton complex in resting leukocytes, which can be termed "pre-activation complex." So far, it is known that actin and microtubule are part of this complex. PKC, Mac-MARCKS (15), Ca 2ϩ , and calmodulin 3 participate in regulating this complex. Due to the involvement of MacMARCKS in regulating integrin mobility on membrane surface, it is reasonable that dynamitin, which interacts with MacMARCKS, is involved. Furthermore, dynamitin is a part of the microtubule 2 T. Jin and J. Li, personal observation. . 3  . Although incubation of CFP-dynamitin expressing cells with staurosporin 100 nM (2.49 Ϯ 1.14 ϫ 10 Ϫ10 ) had no effect on dynamitin-stimulated ␤ 2 integrin mobility, higher concentrations of staurosporin at 200 and 250 nM indeed blocked dynamitin-stimulated ␤ 2 integrin mobility. cytoskeleton system, and microtubule is also involved in regulating integrin mobility. So it again agrees with the fact that dynamitin's role in integrin activation was through its regulation on integrin mobility. Our data support the fact that an enhanced ␤ 2 integrin molecular mobility associates with the disruption of dynamitin function, demonstrating the relaxation of cytoskeletal constraint on integrin molecules.
The relationship between dynamitin and MacMARCKS in integrin activation has been illustrated here. Our data clearly showed that overexpression of dynamitin substitutes the requirement of external PKC activation signal. This is due to dynamitin-stimulated MacMARCKS phosphorylation. Inhibition of MacMARCKS phosphorylation blocks dynamitin-stimulated ␤ 2 integrin mobility. In addition, expression of dynamitin further activates the ␤ 2 integrin-dependent signal transduction pathways that lead to the tyrosine phosphorylation of Paxillin. Therefore, dynamitin acts at the point of Mac-MARCKS phosphorylation and thus regulates cytoskeletal constraint on ␤ 2 integrin.
However, the relationship between dynamitin and microtubules' involvement in regulating the ␤ 2 integrin activation is still a puzzle. It is not unreasonable to speculate that dynamitin and microtubules are on the same pathway. We found in this study that the dynamitin effect is upstream of Rho GTPase. The same is true for ␤ 2 integrin mobility induced by depolymerizing microtubules (25), suggesting that dynamitin and microtubules could be on the same pathway. However, there is no direct evidence yet to suggest whether dynamitin actually mediates microtubules' role in integrin activation or vice versa.
Our data show that microtubule disturbance induces integrin mobility and integrin ligand binding (25). However, by constraining ␤ 2 integrin, microtubules may play dual roles in integrin activation. On one hand, growing microtubule is seen to extend to focal contact, which destabilizes focal contact (32). On the other hand, microtubule seems involved in stabilizing the proper localization of LFA-1 in cell-cell contacts between killer T cells and its target (53). We believe that in either way, dynamitin could mediate these microtubules' effects on integrin. As a component of dynein motor complex, dynamitin could be required for transporting regulatory components such as PKC to the site of integrin activation, which might be an avenue for dynamitin to stimulate MacMARCKS phosphorylation. A similar example is that PKC␤ is associated with and transportedtotheleadingedgeofcrawlingTcellsinamicrotubuledependent manner (54). Another possibility is that dynamitin, similar to p62, could function as a bridge connecting microtubule to somewhere near the cytoskeletal complex restraining the integrin via its interaction with MacMARCKS protein.
Based on the facts that (a) disruption of dynamitin or microtubules releases their restraint on integrin and (b) once the restraint is released, the activation leads to ligand binding, we believe that dynamitin and microtubules play restrictive roles in keeping integrin molecules diffused. Once the constraint is released, two factors could determine where the integrin molecules should go and whether they should form clusters. One factor could be the presence of clustered ligands at the extracellular matrices or on the other cells. The second factor could be the chemical property of the membrane lipids and proteinlipid organization, such as membrane rafts.
In conclusion, dynamitin is an essential player in regulating integrin activation by interacting with PKC substrate Mac-MARCKS on one hand, and being a part of the microtubule dependent motor proteins on the other.