Protein Kinase C-regulated Dynamitin-Macrophage-enriched Myristoylated Alanine-Rice C Kinase Substrate Interaction Is Involved in Macrophage Cell Spreading*

Macrophage spreading requires the microtubule cytoskeleton and protein kinase C (PKC). The mechanism of involvement of the microtubules and PKC in this event is not fully understood. Dynamitin is a subunit of dynactin, which is important for linking the microtubule-dependent motor protein dynein to vesicle membranes. We report that dynamitin is a Ca2+/calmodulin-binding protein and that dynamitin binds directly to macrophage-enriched myristoylated alanine-rice C kinase substrate (MacMARCKS), a membrane-associated PKC substrate involved in macrophage spreading and integrin activation. Dynamitin was found to copurify with MacMARCKS both during MacMARCKS purification with conventional chromatography and during the immunoabsorption of MacMARCKS using anti-MacMARCKS antibody. Vice versa, MacMARCKS was also found to cosediment with the 20 S dynactin complex. We determined that the effector domain of MacMARCKS is required to interact with the N-terminal domain of dynamitin. MacMARCKS and dynamitin also partially colocalized at peripheral regions of macrophages and in the cell-cell border of 293 epithelial cells. Treatment with phorbol esters abolished this colocalization. Disrupting the interaction with a short peptide derived from the MacMARCKS-binding domain of dynamitin caused macrophages to spread and flatten. These data suggest that the dynamitin-MacMARCKS interaction is involved in cell spreading. Furthermore, the regulation of this interaction by PKC and Ca2+/calmodulin provides a possible regulatory mechanism for cell adhesion and spreading.

Cell adhesion and spreading are essential for many physiological processes, including immune cell activation, angiogenesis, and cell proliferation and differentiation (1)(2)(3)(4)(5). Many cytoskeletal proteins (6,7), including the microtubules (8 -10), are involved in these processes. Particularly in macrophages, depolymerization of microtubules with nocodazole causes the spreading macrophages to retract quickly into a spherical shape (11). In fibroblasts, cell spreading is impaired when microtubules are disassembled (8,9). Kinesin, the plus end-directed microtubule-dependent motor protein, has also been implicated in cell spreading, because anti-kinesin antibody decreases cell spreading when injected into cells (10). However, it is not clear whether the microtubule cytoskeleton provides merely a structural support or plays a more active role such as regulating focal adhesion assembly. Recently, studies have revealed that cell adhesion and spreading closely correlates with microtubule function; the integrity of microtubules can influence both the growth of actin stress fibers and the assembly of focal adhesion points (12,13). The growing microtubule filaments are targeted toward the focal adhesion point (13). The effects of microtubules on cell adhesion and on actin cytoskeleton are mediated by the GTP-binding protein Rho (14). All the data suggest an active role for microtubules in cell adhesion and spreading.
The dynactin complex was proposed to be a "molecular bridge" connecting dynein to vesicle membranes (15,27) and was also proposed to enhance the processivity of dynein (28). This bridging function is based on observations that the p150 Glued subunits of dynactin bind to the dynein complex (24) and Arp (25), whereas the Arp subunit binds to the spectrin network (18,22,25,29) that lies underneath the cargo membranes (30). Dynamitin, when overexpressed, disrupts Golgi and lysosome distribution (31) and spindle organization (32). Overexpression of wild type dynamitin (32) or addition of excess amounts of recombinant full-length dynamitin (33) breaks the dynactin complex, whereas overexpression of the conserved N-terminal of dynamitin is not sufficient to break the dynactin complex but is sufficient to disturb the Golgi and endosomes (34). Because dynamitin overexpression dissociated dynein from the mitotic kinetochore (32) and Golgi membrane (35), it has been proposed to link dynein to its cargo.
However, interactions between motor proteins and membranes are usually reversible, and one or more regulatory mechanisms may exist. Our data demonstrate a direct binding between dynamitin and MacMARCKS. 1 This interaction may regulate the association of dynactin to membrane, because the association of MacMARCKS to the membrane is regulated by PKC (36 -38). MacMARCKS, also known as MARCKS-related protein, is a member of the MARCKS family of PKC substrate (38 -40). MacMARCKS was so named because of its enrichment in endotoxin-activated macrophages (38). This myristoylated calmodulin-binding protein associates with both plasma and vesicle membrane, and its calmodulin-binding and membrane association are regulated by PKC-mediated phosphorylation on the two serine residues in its effector domain (36 -38).
PKC-mediated signal transduction pathway is one of the major pathways that regulate cell spreading (41)(42)(43)(44)(45) and integrin activation (reviewed in Ref. 46). Both MARCKS and Mac-MARCKS participate in PKC-dependent cell spreading (47,48). Other studies have shown that the role of MacMARCKS in cell spreading may be mediated by its regulation of the activation of ␤ 2 integrin (49 -51). MacMARCKS is also involved in vesicle redistribution (52). Therefore, the interaction between dynamitin and MacMARCKS may explain the involvement of microtubules in cell adhesion and spreading and may provide a regulatory mechanism for the link between dynactin and the membrane cargoes.

EXPERIMENTAL PROCEDURES
Materials-J774 macrophage cells, 293 epithelial cells, and COS-7 cells were purchased from American Type Culture Collection. Human brain cDNA library inserted in pJG4-5 plasmid was kindly provided by Dr. R. Brent (Harvard University). Human brain cDNA library inserted in pcDNA I plasmid was purchased from Invitrogen. Calmodulin-Sepharose 4B and glutathione-Sepharose 4B were purchased from Amersham Pharmacia Biotech. Purified rat brain PKC was kindly provided by Dr. Y. Liu (University of Oklahoma Health Science Center). Rabbit anti-MacMARCKS antibody was generated in collaboration with Dr. S. Slivka (Tenabe Laboratory, La Jolla, CA) against His-MacMARCKS fusion protein and was affinity purified as described (53). Goat antidynamitin antiserum was generated using purified GST fusion protein of dynamitin in Ferrell Farms (Oklahoma City, OK). Other secondary antibodies were purchased from Jackson Immunological Laboratory (West Grove, PA). All other routine chemicals were purchased from Sigma.
Yeast Two-hybrid System and Dynamitin Cloning-Yeast two-hybrid system was constructed as described (54,55). In brief, cDNA encoding full-length MacMARCKS was inserted in pEG202 in frame at the Cterminal of LexA using the 5Ј EcoRI-BamHI 3Ј sites. This construct was used as bait to screen an expression human brain cDNA library constructed in pJG4-5 plasmid. Using a previously described method (56), the bait and the library cDNA constructs were cotransformed into yeast EGY199 strain containing a reporter gene (plasmid pSH18-34). The transformants were subjected to the selection in His Ϫ , Ura Ϫ , Trp Ϫ , and Leu Ϫ plate. ␤-Galactosidase activity of the yeast colonies was detected using a filter assay as described (56). The pJG4-5 plasmids from the positive colonies were isolated and cotransformed with the bait again. The positive colonies from the second round selection were then isolated, and their pJG4-5 plasmids were sequenced. A positive clone contained a cDNA sequence nearly identical to dynamitin. A human brain cDNA library in pcDNA I plasmid was also screened using the nucleotide sequence of dynamitin. The screening of the library was carried out as described (38). Among the positive clones were the fulllength originally reported dynamitin (32) (Clone C5, referred to in this paper as dynamitin-1), the new dynamitin isoform (Clone C19, referred to as dynamitin-2) (see Fig. 1A), and partial dynamitin (clone C16, lacking N-terminal 71 amino acid residues).
RT-PCR of the Fragments of Dynamitin-1 and Dynamitin-2-Because dynamitin-2 is identical to dynamitin-1 except in the 15-nucleotide stretch where the dynamitin-2 is nine nucleotides shorter, we used primers flanking this short stretch to amplify their RNAs to determine the expression patterns. The following pair of primers was used to amplify the fragment from bp 155 to bp 256 of dynamitin-1 (a 102-bp fragment) and bp 155 to bp 247 of dynamtin-2 (a 93-bp fragment): 5Ј primer, TATGAAACTAGCGACCTA; 3Ј primer, GACAATGATGTGT-TCCAC. RNAs from J774 macrophages, Jurkat T cells, and 293 epithelial cells were isolated using the purification kit from Qiagen. RNAs from human brain tissue were purchased from CLONTECH. The RT-PCR was carried out according to the instructions from the RT-PCR kit (Tetra-Link, Buffalo, NY). Reverse transcription was carried out at 50°C for 30 min and was then terminated by heating the sample at 94°C for 2 min. PCR reaction was carried out under the following conditions: 10 cycles at 94°C for 20 s, 61°C for 30 s, and 68°C for 30 s, followed by 25 cycles at 94°C for 20 s, 61°C for 30 s, and 68°C for 35 s. The amplified products were analyzed on a 10% polyacrylamide gel. The amplified DNAs were cloned using PCR cloning kit from Promega, and their sequences were confirmed by DNA sequencing.
Expression of Fusion Proteins-All cDNA constructs encoding dynamitin-2 and its mutant were generated by PCR using clone C19 as template. All cDNA constructs of MacMARCKS and its mutants were generated as described (49). cDNAs encoding deletion mutants were generated by ligating two PCR fragments flanking the deleted domains using a previously described method (57). For point mutations, the oligonucleotide primers containing the mutant codon were used to amplify the two half-molecules, and then the half-molecules were ligated to form a full-length molecule containing the desired point mutations. The sequences of all PCR products were confirmed by DNA sequencing after being cloned into corresponding plasmids. The following GST fusion proteins of dynamitin and its mutants were constructed by inserting the coding cDNA into pGEX-4T-1 expression vector (Amersham Pharmacia Biotech) (see The following His 6 -tagged MacMARCKS and its mutants were constructed by inserting the coding cDNAs into pET28a expression vector (Novagen, CA) (see Fig. 1B): F (wild type MacMARCKS), SA (serines 93 and 104 of phosphorylation sites changed to alanine), SD (serines 93 and 104 of phosphorylation sites changed to aspartic acid), ND (Nterminal 18-amino acid deletion), CD (C-terminal 34-amino acid deletion), and ED (effector domain deletion, i.e. deletion of amino acids 85 to 110). The GST-dynamitin fusion proteins of dynamitin-2 and His 6 fusion proteins of MacMARCKS were purified as described in the manufacturer's manuals.
Calmodulin Binding Assay-Purified GST-dynamitin-2 (5 g) or its mutants were applied to a 2-ml calmodulin-Sepharose 4B column packed in a 3-ml syringe, which was preequilibrated with calmodulin loading buffer (20 mM Tris, pH 8, 2 mM CaCl 2 ). The column was washed with five column volumes of loading buffer followed by five column volumes of the same buffer containing 200 mM NaCl. Dynamitin-2 was eluted with elution buffer containing 20 mM Tris, pH 8.0, 200 mM NaCl, and 4 mM EGTA. Fractions were collected at 0.5 ml/fraction, were applied to SDS-PAGE and detected by Coomassie Blue staining.
Dynamitin Binding with MacMARCKS-The same amount of His 6 fusion proteins of MacMARCKS or its mutants (2 g each) were incubated with glutathione-Sepharose (50 l of 50% slurry) conjugated with GST fusion proteins of dynamitin-2 or its mutant, C16, in 1 ml of phosphate-buffered saline in the Eppendorf tubes for 1 h at 4°C. After three washes, dynamitin-2-Sepharose beads were transferred to fresh tubes and were washed again. The GST-dynamitin and the dynamitinbound MacMARCKS were then subjected to SDS-PAGE (9%) and transferred to the Immobilon membrane. MacMARCKS and its mutants were detected by immunoblotting with the anti-MacMARCKS antibody. The same amounts of GST-dynamitin fusion proteins on the Sepharose were used in each tube as indicated by Coomassie Blue staining of the same membrane.
To test the effect of PKC phosphorylation, the same procedure was used except phosphorylated MacMARCKS (2 g) was incubated with dynamitin. The phosphorylated MacMARCKS was prepared as described (38).
To test the competitive effect of MB on dynamitin-MacMARCKS interaction, GST-free MB was generated. To 1 ml of GST-MB at a concentration of 50 g/ml, 50 units of bovine thrombin were added, and the sample was digested at 37°C for 15 min. The free GST and undigested GST-MB were removed by passing the sample through the glutathione-Sepharose columns. MacMARCKS (2 g) was incubated with or without GST-free MB (8 g) at 4°C for 30 min and then allowed to bind to 50 l of dynamitin-Sepharose as described above.
Coimmunoabsorption of Dynamitin-2 with MacMARCKS-The affinity purified anti-MacMARCKS antibody, rabbit IgG, or bovine serum albumin was coupled to the Hi-Trap Sepharose column (1 ml; Amersham Pharmacia Biotech) as described by the manufacturer's instructions. Human 293 cells (4 ϫ 10 7 ) were incubated with or without 100 nM PMA for 1 h. The cells were lysed in 2 ml of PEM lysis buffer (0.1 M PIPES, pH 6.9, 1% Nonidet P-40, 0.5 mM MgCl 2 , and 1 mM EGTA). The lysate was centrifuged at 18,000 ϫ g for 10 min, and the supernatant was passed through the antibody column or control columns. After washing four times with 5 ml of phosphate-buffered saline, the columnbound proteins were eluted with 3 ml of 0.2 M glycine at pH 3. The eluted proteins were concentrated to 100 l of volume and subjected to SDS-PAGE followed by immunoblotting with anti-dynamitin antibody.
Copurification of Dynamitin with MacMARCKS-The purification procedure for MacMARCKS was described previously (38). In brief, approximately 3 ml of packed 293 cells were homogenized in 5 ml of lysis buffer containing 20 mM Tris-HCl at pH 7.6, 1 mM EDTA, 10 mM ␤-mecaptoethanol, 0.2% Triton X-100, and protease inhibitors as described (38). The lysate was centrifuged at 4°C for 40 min at 10,000 ϫ g. The supernatant was heated to 80°C for 5 min and immediately cooled in ice water. The denatured proteins were removed by centrifugation at 18,000 ϫ g for 10 min at 4°C. The supernatant was then passed through a Q-Sepharose column (1 cm ϫ 5 cm) and eluted with 200 ml of linear NaCl gradient from 0 to 1 M in the same lysis buffer without Triton X-100. The fractions containing MacMARCKS were concentrated in a Centriprep filter (Amicon) to 1 ml, loaded onto a Sephadex G-100 column (1 cm x 20 cm), and eluted with the calmodulin loading buffer (20 mM Tris-HCl, pH 8.0, 1 mM CaCl 2 ). Fractions containing MacMARCKS were combined and loaded onto a calmodulin-Sepharose column (1 cm ϫ 3 cm). After washing with the same buffer containing 200 mM NaCl and without CaCl 2 , MacMARCKS was then eluted with 2 mM EGTA. Fractions (1 ml) were collected, and fractions 2-4 were concentrated and loaded onto SDS-PAGE followed by immunoblotting with anti-MacMARCKS antibody and anti-dynamitin antibody.
Cosedimentation of MacMARCKS with Dynamitin-COS-7 and bovine brain extracts were prepared by high speed centrifugation as described by Paschal et al. (58). The extracts were fractionated by layering a 1-ml volume of each extract onto a separate 11-ml, 5-20% sucrose gradient. The sucrose gradient was prepared in Tris-KCl buffer consisting of 20 mM Tris-HCl, pH 8.0, 50 mM KCl, 5 mM MgSO 4 , and 0.5 mM EDTA. The samples were then centrifuged at 125,000 ϫ g for 16 h at 4°C in a SWTi4l rotor (Beckman Instruments). The gradients were collected at 0.5 ml/fraction starting from the bottom of the tube. The fractions were then subjected to SDS-PAGE and followed by immunoblotting with anti-MacMARCKS and anti-dynamitin antibody.
Microinjection of Dynamitin-2 Peptides-J774 macrophages were cultured overnight on a coverslip. The coverslip was then placed in a steel vessel on a water-jacked heated stage of a microscope (Zeiss Axiovert 100). Minimum light intensity was used to avoid "burning" the cells. The temperature of the jacked stage was set to 37°C. Under these conditions, J774 macrophages appeared spherical. Injected fusion proteins included: GST-MB (fusion peptide of the N-terminal Mac-MARCKS-binding peptide of dynamitin, i.e. amino acids 1-83), GST-C (fusion peptide of a C-terminal peptide of dynamitin from amino acids 291 to 335), GST, and GST-MB depleted solution prepared by repeatedly passing GST-MB through a Sepharose 4B column conjugated with anti-dynamitin antibody. In addition to wild type J774 cells, J774 cells expressing an effector domain deletion mutant of MacMARCKS (J774-ED) (57) were also injected. All injected fusion proteins were dialyzed against the injection buffer (5 mM Na 2 HPO 4 , 80 mM KCl, pH 7.4) and adjusted to a final concentration at 5 g/l before being injected into J774 cells using an Eppendorf microinjector under the following conditions: compensation pressure of 80 hectopascals, injecting pressure of 100 hectopascals, and injection duration of 0.3 s.
Immunofluorescence-Cells on coverslips were treated with lipopolysaccharide (1 g/ml) for 2 h or with PMA for 1 h before being fixed with 10% formalin in phosphate-buffered saline for 15 min at 4°C. The cells were then permeabilized in acetone at Ϫ20°C for 5 min. Both polyclonal rabbit anti-MacMARCKS antibody and polyclonal goat antidynamitin antibody were affinity purified against corresponding pure fusion protein as described previously (57). The affinity purified antibodies did not cross-react with other proteins (data not shown). Mac-MARCKS was visualized with anti-MacMARCKS antibody followed by fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG. Dynamitin was visualized with anti-dynamitin antibody followed by Texas Red-conjugated donkey anti-goat IgG. The subcellular localization of these proteins was documented on a Zeiss 510 confocal microscope.

Dynamitin Is a Ca 2ϩ /Calmodulin-binding Protein and Contains ATPase Motifs in Its Primary
Sequence-In the process of identifying MacMARCKS-binding proteins using the yeast twohybrid system (54), interaction was detected between Mac-MARCKS and a polypeptide that is nearly identical to dynamitin (32) (Fig. 1A), except for a difference of 15 nucleotides, i.e. five amino acid residues (Fig. 2). We further screened a human brain cDNA library using the nucleotide sequence of dynamitin, and both forms of dynamitin were found in the same library. To distinguish them, we refer to the original reported form as dynamitin-1 and to the isoform identified here as dynamitin-2. Because this library was made from one individual, the existence of the two isoforms is not likely due to polymorphism between individuals. To verify the existence of these two isoforms, we performed RT-PCR using the primers flanking the differing sequences of these two isoforms, which should give rise a 93-bp DNA fragment for dynamitin-2 and a 102-bp fragment for dynamitin-1 (Fig. 2). Using the RNAs from three cell lines including Jurkat, J774, and 293 cell, we ampli- fied DNA bands corresponding to the dynamitin-2 (Fig. 2), whereas from human brain RNA, we amplified a band corresponding to the dynamitin-1 (Fig. 2). Sequence analysis confirmed that the amplified DNAs indeed coded for dynamitin-1 and dynamitin-2, respectively (data not shown). Based on these data, we believe that dynamitin-1 is the major form expressed in the brain, and dynamitin-2 is the major form expressed in our experimental cell lines. Therefore, these cell lines provide us with a useful tool to study the cellular functions of the interaction between MacMARCKS and dynamitin-2 without interference from dynamitin-1. Nevertheless, we should note that both forms of dynamitin interact with MacMARCKS equally in our in vitro binding assay.
Three potential domains were identified in dynamitin. First is the N-terminal MacMARCKS binding domain (Fig. 1A, bold type) that will be discussed in detail in the following sections. Second, a Walker A motif and a Walker B ATPase motif (59,60) were identified in dynamitin (Fig. 1A, underlined italic type). However, whether these motifs carry any ATPase activity and the potential function of this sequence are still under investigation. Third, a stretch of basic amino acid residues was detected in both forms of dynamitin between amino acid residues 58 and 80. This domain can be modeled into an amphipathic helix. In the helix, positively charged amino acid residues and hydrophobic amino acid residues each form a cluster adjacent to each other (Fig. 3A). This structure resembles a calmodulinbinding motif (61). Our experimental results show that both forms of dynamitin bind to the calmodulin-Sepharose column in the presence of CaCl 2 . In this report, we show the interaction between dynamitin-2 and calmodulin as an example (Fig. 3). When GST-dynamitin-2 was loaded onto the calmodulin-Sepharose column in the presence of Ca 2ϩ , the protein was retarded on the column. Although a small amount was eluted by a high concentration of salt, the majority of GST-dynamitin-2 was eluted from the column only when EGTA was added to chelate the Ca 2ϩ (Fig. 3B, Dynamitin), indicating that the binding is calcium-dependent. Deletion of this putative calmodulin-binding motif (amino acids 59 -83) abolished the calmodulin-binding activity of dynamitin-2 (Fig. 3B, CamD). Therefore, our data show that dynamitin-2 is a Ca 2ϩ -dependent calmodulin-binding protein.
Dynamitin Directly Binds to MacMARCKS-Because a positive clone in the yeast two-hybrid system does not result in a definitive conclusion, it was important to further evaluate the interaction between MacMARCKS and dynamitin-2 by using purified fusion proteins. This evaluation would also allow us to determine whether there is a direct binding between these two proteins. In this experiment, His 6 fusion protein of Mac-MARCKS and a number of its mutants (see Fig. 1B for their structures) were incubated with GST-dynamitin-2-Sepharose beads (Fig. 4A) (Fig. 4A). MacMARCKS or any of its mutants did not bind to control Sepharose beads (Fig. 4A). Thus, we conclude that the effector domain of MacMARCKS is required for its binding to dynamitin-2, whereas its N or C terminus is not. The mutation of the serine residues of the phosphorylation sites in the effector domain to either aspartic acid (SD) or to alanine (SA) also decreased the binding of the mutants to dynamitin-2 (Fig. 4A). This result indicated that these two serine residues are important for MacMARCKS to bind to dynamitin. Because aspartic acid is negatively charged, which could mimic a phosphate group, the SD mutant hinted a possibility of phosphorylation-mediated regulation. On the other hand, because the SA mutant is not necessarily equal to a nonphosphorylated form of MacMARCKS, its failure to bind dynamitin merely suggested that alanine cannot replace serine in this case.
To determine the MacMARCKS-binding domain in the dynamitin-2 molecule, we used clone C16, which encodes a mutant dynamitin-2 lacking the N-terminal 71 amino acid residues. No binding between MacMARCKS and this mutant dynamitin was detected (Fig. 4A), indicating that the N terminus of dynamitin-2 is essential for MacMARCKS binding. To determine the minimum length of the N terminus of dynamitin-2 that is sufficient for MacMARCKS binding, an 83amino acid peptide of the N terminus of dynamitin-2 (MB) was expressed as a GST fusion peptide and conjugated to Sepharose as the affinity matrix. This N-terminal peptide of dynamitin-2 was sufficient to bind MacMARCKS (Fig. 4B), which confirms that the MacMARCKS-binding domain of dynamitin-2 is within the N terminus. In fact, experiments showed that this 83-amino acid peptide is sufficient to inhibit the interaction between wild type MacMARCKS and dynamitin (Fig. 4C). Because the calmodulin-binding domain is also located in the N terminus of dynamitin-2, the calmodulin-binding domain deletion mutant of dynamitin-2 (CamD) was generated to test whether this domain contributes to binding MacMARCKS. Deletion of amino acid residues 59 -83 of dynamitin-2 did not abolish its binding to MacMARCKS (Fig. 4B, CamD). Thus, we conclude that the MacMARCKS-binding domain of dynamitin-2 must be located within the N-terminal 58 amino acid residues. Deletion of amino acids 1-44 of dynamitin-2 abolished its MacMARCKS-binding activity (Fig. 4B, MBD), also supporting our conclusion.

Dynamitin Associates with MacMARCKS in Vivo in Mammalian Cells, and Their Interaction Is Regulated by PKC and
Ca 2ϩ /Calmodulin-Once the direct interaction between these two proteins was established in vitro, we evaluated their in vivo association. Evidence from three different experiments strongly suggests that at least a portion of these two proteins associate with each other in cells. First, dynamitin was found to be coimmunoabsorbed with MacMARCKS on MacMARCKS antibody-conjugated Sepharose column. In this experiment, Sepharose beads conjugated with anti-MacMARCKS antibodies were used to immunoabsorb MacMARCKS from cell lysate. The antibody-bound proteins were then eluted and subjected to immunoblotting with antibodies against MacMARCKS and dynamitin. Rabbit total IgG and bovine serum albumin were conjugated to the Sepharose-4B and used in a binding assay as a negative control. We found that dynamitin was coimmunopurified with MacMARCKS when MacMARCKS was immunoabsorbed by anti-MacMARCKS antibody (Fig. 5A), whereas no dynamitin was absorbed by rabbit IgG-beads or bovine serum albumin-Sepharose beads. Although the anti-dynamitin antibody could not distinguish dynamitin-1 from dynamitin-2, what we observed here was likely to be dynamitin-2, because it is the major form expressed in 293 cells. Second, during the purification of MacMARCKS using the previously described procedure (38), dynamitin-2 was found in the final preparation of MacMARCKS (Fig. 5B). The entire purification procedure underwent heat precipitation, ion exchange, gel filtration, and calmodulin affinity purification. Although the calmodulin affinity column was irrelevant in this case because dynamitin is also a calmodulin-binding protein, the copurification after the first three steps suggested a very strong possibility of their interaction. Finally, sucrose gradient centrifugation of cell lysate was analyzed on SDS-PAGE followed by immunoblotting with anti-MacMARCKS and anti-dynamitin antibodies. Mac-MARCKS, which usually migrates on SDS-PAGE as a double band, was found to cosediment with the dynamitin at 20 S (Fig.  5C) in the sucrose gradient; this is a signature density of dynactin complex during the purification of dynactin complex from the brain tissue. The presence or absence of MacMARCKS in the 20 S complex seemed to have no effect on its integrity. In COS-7 cells, a cell line that does not express MacMARCKS (38), dynactin complex sediments at 20 S without MacMARCKS being detected (Fig. 5C). Because the 20 S complex represents intact dynactin complex, this result implies that MacMARCKSassociated dynamitin is bound with the dynactin complex. This result is also supported by the presence of p150 Glued during MacMARCKS purification (Fig. 5B). However, with or without MacMARCKS, the integrity of 20 S did not change; the 20 S dynactin complex from COS-7 cells appears intact, whereas these cells do not express MacMARCKS. We also tested whether the in vivo association of dynamitin-2 and MacMARCKS is regulated by PKC-mediated phosphorylation. In the cells treated with PMA, in which MacMARCKS was heavily phosphorylated by PKC (38), we found that dynamitin-2 was no longer coimmunoprecipitated with Mac-MARCKS (Fig. 5A). The phosphorylation-mediated regulation is also confirmed by the in vitro binding assay. We found that phosphorylation of MacMARCKS by purified PKC abolished its binding with dynamitin-2 (Fig. 5D). This observation, along with other observations (62)(63)(64)(65), provides evidence of the involvement of phosphorylation in microtubule-related functions.
As further support for the in vivo interaction of these two proteins, we also observed partial colocalization of these proteins in 293 epithelial cells and 774 macrophage cells. Immunofluorescence confocal microscopy was used to observe the subcellular localization of dynamitin and MacMARCKS in a thin section of cell, a way that eliminated the false impression of protein concentration because of thick membrane folding at the cell-cell borders. The subcellular localization of dynamitin has been described in HeLa, PtK1, COS-7, and Rat2 cells as fine punctate staining throughout the cytoplasm and concentrated in the centrosomal area (32). We observed a similar pattern in 293 epithelial cells. In addition, we also observed that dynamitin concentrated in the cell-cell border where it colocalized with MacMARCKS (Fig. 6, 293).
To investigate the in vivo function of the dynamitin-Mac-MARCKS interaction in macrophages, we examined the subcellular localization of these two proteins in macrophages. The centrosomal staining of dynamitin was not visible in the J774 macrophages. In addition to the fine punctate staining, dense patches of dynamitin were also observed in the membrane edges of these cells where it colocalized with MacMARCKS protein (Fig. 6, J774). MacMARCKS has been shown to localize both on the plasma membrane and on the vesicle membranes (47,52). The localization pattern of dynamitin in the plasma region is subject to PKC regulation; adding PMA abolished dynamitin from the membrane edge in macrophages, whereas some of MacMARCKS protein remained at the edge (Fig. 6, J774 ϩ PMA). PMA treatment also dispersed dynamitin from the cell-cell borders in 293 cells (Fig. 6, 293 ϩ PMA). The fact that dynamitin dislocated from the membrane edge correlates with the finding that PKC regulates the MacMARCKS-dynamitin interaction. This observation further supports that dynamitin-MacMARCKS may interact in vivo.
Dynamitin-MacMARCKS Interaction Is Involved in Cell Spreading-To explore the potential function of the dynamitin-MacMARCKS interaction in vivo, MacMARCKS-binding peptide of dynamitin-2 (MB) was introduced into the cells to competitively disrupt this interaction. The peptide was shown to inhibit MacMARCKS-dynamitin interaction (Fig. 4C). Although our current report focuses on the effect of this peptide on cell spreading, its effect on mitosis and lysosomal distribution was also observed to be similar to that reported by Echeverri et al. (32), which will be studied further.
We studied the short-term effects (in the minute range) of the disruption of this interaction. Instead of transfecting the cDNA construct, which usually takes 10 -20 h to accumulate enough mutant protein, GST-MB fusion peptide was introduced into J774 macrophage cells by microinjection. As a control, we also injected GST alone and a GST fusion peptide of the C-terminal 48 amino acids of dynamitin-2 (amino acids 288 -335, GST-C), which is not involved in dynamitin-MacMARCKS interaction. Although the uninjected and control injected cells mostly remained spherical, GST-MB injected cells became flattened within a time range between 5 and 20 min (Fig. 7, GST-MB). In five separate experiments, we counted 219 cells in which 96% (210) of cells spread after injection. However, only 21% (105 of 500 cells of five experiments) of uninjected cells and 20% (61 of 300 cells of three experiments) of GST injected cells spread spontaneously. It is important to note that the C-terminal domain peptide of dynamitin did not show any effects (Fig.  7, GST-C), and only 19% (47 of 250 cells of three experiments) spread spontaneously. To be sure that the effect of MB was not a result of unknown contamination in the sample, we also passed the MB solution through the anti-dynamitin antibody column. The MB-depleted solution was also injected in the cells, and no effect was observed (data not shown). This result strongly suggests that dynamitin-MacMARCKS interaction is involved in regulating cell spreading. This phenotype of dynamitin-MacMARCKS interaction agrees with the observation that MacMARCKS is involved in cell spreading (47).
In addition, we injected GST-MB into J774 macrophages expressing the effector domain deletion mutant of Mac-MARCKS (47). Expression of this dominant negative mutant in J774 macrophage cells blocks phorbol ester-stimulated or immune complex-stimulated cell spreading. Using these mutant cells allows us to determine whether MacMARCKS is required for the effect of GST-MB. If dynamitin functions independently from MacMARCKS or if it functions down stream of Mac-MARCKS, then mutation of MacMARCKS should not change the effect of injecting GST-MB. One would expect cell spreading after injection of this peptide. If these cells did not spread, it would suggest that the role of dynamitin in cell spreading depends on a functional MacMARCKS or dynamitin functions upstream from MacMARCKS. The experiment showed that injection of GST-MB into these MacMARCKS mutant cells did not cause cell spreading as it did in normal J774 cells (Fig. 7). Therefore, this experiment suggests that MacMARCKS is required for this MB-induced effect, which could be either due to the fact that MacMARCKS is downstream of dynamitin or due to the fact that MacMARCKS and dynamitin collaborate in one function.

DISCUSSION
The data presented here show that dynamitin-2 binds to the PKC substrate MacMARCKS under both in vitro and in vivo conditions and that this binding is regulated by PKC-mediated phosphorylation of MacMARCKS. In addition to the in vitro binding assay using purified fusion proteins, the in vivo interaction between dynamitin and MacMARCKS was demonstrated in three different ways including coimmunoabsorption and copurification in both directions. The N-terminal domain of dynamitin, very well conserved among different species (34), is required in this interaction. The effector domain of Mac-MARCKS, which contains the phosphorylation sites (serines 93 and 104), is also required for the interaction.
The PKC-mediated phosphorylation of MacMARCKS regulates its binding to dynamitin both in vitro and in vivo. When MacMARCKS was phosphorylated in vitro with purified PKC, it no longer binds to the dynamitin. In PMA-treated cells, MacMARCKS was heavily phosphorylated by PKC (38). In this case, dynamitin was no longer coimmunoprecipitated with MacMARCKS. Therefore, we concluded that phosphorylation of MacMARCKS by PKC is an important regulation of dynamitin-MacMARCKS interaction.
PKC is known to phosphorylate serines 93 and 104 of Mac-MARCKS (38), consistent with our finding that the effector domain mediates the interaction with dynamitin. As expected, the substitution of aspartic acid at these two sites to mimic phosphorylation resulted in decreased binding to dynamitin; however, substitution of alanine also resulted in decreased binding, suggesting that serines 93 and 104 may represent critical residues for the interaction between MacMARCKS and dynamitin. Currently, the affinity and stoichiometery of the MacMARCKS-dynamitin interaction are unknown, but future research in this area may shed light on the involvement of the serine residues.
Experiments in vivo further suggest that the MacMARCKSdynamitin interaction may be involved in cell spreading. Microinjection of the N-terminal MacMARCKS-binding peptide of dynamitin (MB) caused macrophages to spread, whereas injection of the C-terminal peptide of dynamitin showed no such effect. In addition, a functional MacMARCKS is required for this effect of MB, because injection of this peptide into cells expressing MacMARCKS mutant showed no effect. These data agree with the reported function of MacMARCKS in regulating cell spreading (47) and suggest one of the possible functions of MacMARCKS-dynamitin interaction.
It is not surprising that dynamitin, a protein involved in microtubule function, participates in cell spreading. The involvement of the microtubule cytoskeleton in cell adhesion and spreading has been well documented, and such an involvement is more prominent in macrophages than in fibroblasts. Recent reports suggest that microtubules do not merely provide a structural reinforcement during cell spreading but also are involved in regulating focal adhesion (12,13). Recently, the p62 subunit of dynactin was shown to be involved in cell adhesion (66). Because dynamitin is an important regulator of the dynactin and microtubule function, it is feasible for dynamitin to be involved in cell spreading. Our data show that injecting the MacMARCKS-binding peptide of dynamitin disrupts Mac-MARCKS-dynamitin interaction and promotes macrophage spreading. These observations suggest that the binding of dynamitin with MacMARCKS inhibits cell spreading. This suggestion is also reasonable because MacMARCKS is involved in PMA-stimulated macrophage spreading (47) through its regulation of the activation of ␤ 2 integrin (49,50). Our recent study shows that MacMARCKS acts on the cytoskeletal constraint on the mobility of integrin molecules (51).
Not only is it feasible that dynamitin may be involved in macrophage spreading, it is also feasible that the role of dynamitin in cell spreading occurs through its interaction with MacMARCKS, whose role in macrophage spreading is known through its regulation of the ␤ 2 integrin (47, 49 -51). An important question regarding the dynamitin/dynactin complex is how this complex connects dynein motor protein to cargo membrane (15,27). Although a bridge between spectrum and Arp is proposed, a regulatory mechanism is still needed. Therefore, it is reasonable to propose that the dynamitin-MacMARCKS interaction may be part of the proposed bridge and may provide a regulation to this complex through PKC-mediated phosphorylation of MacMARCKS. Many microtubule-related functions are regulated by phosphorylation. For example, dyneindependent vesicle transportation is regulated by phosphorylation (63,64). Phosphorylation also regulates kinesin activity (62). MacMARCKS is a PKC substrate, and it not only is required for integrin activation at cell surface but also associates with intracellular membrane compartments such as secretory vesicles of neuronal cells and phagosomes of macrophages (52,57). In addition, dynamitin-MacMARCKS interaction is regulated by PKC. Therefore, its interaction with dynamitin would be an ideal candidate to regulate the dynactin-membrane association.
The key issue is how dynamitin-MacMARCKS interaction may be involved in cell spreading. We believe that Mac-MARCKS may hold the key. MacMARCKS is required for PMA-stimulated cell spreading through its regulation of ␤ 2 integrin (47,49,50), an adhesion molecule essential for cell spreading. The first step of integrin activation is release of cytoskeletal constraint on the integrin molecules so that integrin diffuses faster on the cell membrane and has a better chance of binding a ligand (67). Our recent data suggest that MacMARCKS, after being phosphorylated by PKC, is responsible for breaking the cytoskeleton complex that restrains the integrin molecules (51). MacMARCKS might also participate in establishing focal adhesion, which is a later stage of integrin activation. It is possible that dynamitin-MacMARCKS interaction might be part of the cytoskeletal complex that restrains integrin molecules. When MacMARCKS is phosphorylated by PKC, it no longer binds to dynamitin, and concomitantly, the cytoskeletal constraint on integrin is released. It is important to note that PMA and GST-MB both caused the dissociation of dynamitin-2 from MacMARCKS and that both treatments stimulated cell spreading, which raises the possibility that PKC-regulated dissociation of MacMARCKS from dynamitin-2 may be an important step during cell spreading. This possibility remains to be tested.
Because dynamitin is involved in dynactin/dynein function, our data also suggest the potential involvement of dynactin/ dynein in cell spreading in addition to its known function in vesicle and spindle organization. This speculation is in agreement with the finding that the p62 subunit of dynactin is also involved in cell adhesion (66). Kinesin, another microtubuledependent motor protein, has already been shown to be involved in cell spreading and lysosomal distribution (10). It is interesting to note the opposite effects of blocking kinesin function and interfering with dynamitin-2 function. Although blocking kinesin inhibits cell spreading (10) and the formation of tubule lysosome (68), interfering with dynamitin-Mac-MARCKS interaction promotes cell spreading and formation of tubular lysosomes (data not shown).
As a working hypothesis, we propose that by interacting with MacMARCKS, which associates with plasma membranes, dynamitin may bridge the microtubule motor proteins to the plasma membranes. Therefore, dynamitin/microtubule may be a part of the unknown cytoskeletal complex that constrains integrin molecules at the resting stage to keep cells round. PKC is likely to regulate this bridge, because the interaction between dynamitin and MacMARCKS is regulated by PKC-mediated phosphorylation. As a consequence of MacMARCKS phosphorylation, dynamitin-MacMARCKS interaction is disrupted, freeing integrin to bind its ligand and mediate cell spreading.
If this hypothesis is true, how is it that depolymerization of microtubules causes macrophages to become round rather than to spread? A possible explanation is that integrin activation and cell spreading involve first breaking of the cytoskeleton and then reestablishing the cytoskeletal link later at focal adhesion points. Therefore, a total breakdown of microtubules by nocodazol causes cells to become round because the drug may further affect the integrity of focal adhesion point.