Overexpression of the Myristoylated Alanine-rich C-kinase Substrate Inhibits Cell Adhesion to Extracellular Matrix Components*

Mice lacking the myristoylated alanine-rich C-kinase substrate, or MARCKS protein, exhibit abnormalities consistent with a defect in the ability of neurons to migrate appropriately during forebrain development. To investigate the possibility that this phenotype could be due to disruption of normal cellular adhesion to extracellular matrix, an assay was developed in which 293 cells co-expressing MARCKS and green fluorescent protein were tested for their adhesion competence on various substrates. Fluorescence-activated cell sorting of adherent and non-adherent green fluorescent protein-expressing cells demonstrated that wild-type MARCKS inhibited adhesion of cells to fibronectin, whereas a non-myristoylated mutant did not inhibit adhesion of cells to a variety of substrates. The fibronectin competitive inhibitor RGD peptide inhibited adhesion of cells expressing all MARCKS variants equally. Cytochalasin D inhibited the adhesion of cells expressing non-myris-toylated MARCKS, but did not further decrease the adhesion of cells expressing adhesion-inhibitory proteins. Confocal microscopy demonstrated the presence of inhibitory, myristoylated MARCKS at the plasma membrane, suggesting that localization at this region might be important for MARCKS to inhibit cellular adhesion. These data suggest a possible myristoylation-dependent function of MARCKS to inhibit cellular adhesion to extracellular matrix proteins, indicating a potential mechanism for the cell migration defects seen in the MARCKS-deficient mice. Cells 37 5% CO 95% incubator. Plasmids Constructs— The non-myristoylatable and PSD mutants of the bovine MARCKS cDNA have been described previously (11). They are designated here as follows: wild-type (WT); alanine, glycine, asparagine, or aspartic acid replacement of the four phospho- rylatable serines within the PSD (AS, GS, NS, or DS, respectively); alanine replacement of the amino-terminal glycine preventing myris- toylation (A2G2); and the non-myristoylated (A2G2) and “pseudo-phos-phorylated” (DS) double mutant (DBL). A Hin dII/ Eco RI insert from each of the parent mutant constructs was purified, and the ends were made flush with the Klenow fragment of DNA polymerase. Blunt-ended inserts were ligated into the pCMV4 vector (31), which had been digested with Sma I and treated with calf intestinal alkaline phosphatase (Life Technologies, Inc.). JM109 cells were transformed, ampicillin-resistant colonies were picked, and plas- mid DNA was purified. Clones containing inserts were tested for MARCKS expression by transient transfection into 293 cells followed by radiolabeling with L -[ 35 S]cysteine and immunoprecipitation with MARCKS-specific antibodies (32). Expressing clones were re-trans- formed, and DNA was prepared by large scale plasmid preparations followed by double-band purification on cesium chloride density gradi- ents

The myristoylated alanine-rich C-kinase substrate or MARCKS 1 protein was originally identified because of its prominence as a cellular substrate for protein kinase C (PKC) (reviewed in Refs. 1 and 2). Although cloned more than a decade ago (3)(4)(5), a functional role for MARCKS remains elusive. Gene targeting experiments have demonstrated that MARCKS is essential for central nervous system development and postnatal life in the mouse (6). Defects seen in the brains of developing MARCKS-deficient embryos included frequent abnormal neurulation and failure of fusion of forebrain hemispheres. There was also universal neuronal and retinal ectopia, resulting from the inappropriate migration of neurons, as well as deficiencies in laminin and chondroitin sulfate proteoglycans (7), two components of the extracellular matrix (ECM) involved in migration of developing neurons (8,9). These experiments suggested that MARCKS can regulate the ability of neurons to migrate normally; however, it was unclear whether this abnormality was at the level of cell:cell interactions, cell: ECM interactions, or even in the regulation of proteases involved in the remodeling of the ECM.
MARCKS and its related protein, the MARCKS-like protein (MLP; also known as F52, MRP, or MacMARCKS) constitute a small family of heat-stable, acidic proteins, containing three regions of near-identity (2). An eight-residue domain in the amino-terminal portion of the protein surrounds the single intron splice site and is identical to a region of unknown function within the cytoplasmic domain of the mannose 6-phosphate/insulin-like growth factor II receptor. The other two regions include an amino-terminal consensus sequence, which directs the co-translational addition of the 14-carbon myristoyl moiety, and a highly basic region of 25 amino acids containing the PKC phosphorylatable serines, known as the phosphorylation site domain (PSD). These two regions are responsible, respectively, for the hydrophobic and electrostatic interactions of MARCKS with negatively charged lipids in cellular membranes (10 -17). PKC-mediated phosphorylation of the PSD decreases MARCKS affinity for the plasma membrane, calmodulin, and actin, and prevents cathepsin B cleavage at the PSD (14, 18 -30). However, a direct interaction between MARCKS and these proteins in intact cells has not been demonstrated convincingly.
Based on the morphological defects seen in the MARCKSdeficient mice, we speculated that MARCKS could be directly involved in cell:ECM and/or cell:cell interactions. To begin to analyze a role for MARCKS in the regulation of cell:ECM interactions, we have designed an assay based on the overexpression of MARCKS in human embryonic kidney 293 cells. By co-transfecting green fluorescent protein (GFP) and using fluorescence-activated cell sorting (FACS), we were able to demonstrate that overexpression of MARCKS reduces the ability of these cells to adhere to various extracellular matrix components. In addition, we demonstrate that myristoylation of the protein and plasma membrane localization are both necessary, but neither is sufficient, for this inhibitory effect on cellular adhesion to occur.

MATERIALS AND METHODS
Cells and Cell Maintenance-Human embryonic kidney 293 cells, obtained from the American Type Culture Collection (Manassas, VA), were grown in Eagle's minimum essential medium (Life Technologies, Inc.) containing 10% (v/v) heat-inactivated fetal calf serum, 2 mM glutamine, 100 units/ml penicillin, and 100 units/ml streptomycin. Cells were maintained at 37°C and 5% CO 2 , 95% air in a water-jacketed incubator.
Plasmids and Constructs-The non-myristoylatable and PSD mutants of the bovine MARCKS cDNA have been described previously (11). They are designated here as follows: wild-type (WT); alanine, glycine, asparagine, or aspartic acid replacement of the four phosphorylatable serines within the PSD (AS, GS, NS, or DS, respectively); alanine replacement of the amino-terminal glycine preventing myristoylation (A2G2); and the non-myristoylated (A2G2) and "pseudo-phosphorylated" (DS) double mutant (DBL).
A HindII/EcoRI insert from each of the parent mutant constructs was purified, and the ends were made flush with the Klenow fragment of DNA polymerase. Blunt-ended inserts were ligated into the pCMV4 vector (31), which had been digested with SmaI and treated with calf intestinal alkaline phosphatase (Life Technologies, Inc.). JM109 cells were transformed, ampicillin-resistant colonies were picked, and plasmid DNA was purified. Clones containing inserts were tested for MARCKS expression by transient transfection into 293 cells followed by radiolabeling with L-[ 35 S]cysteine and immunoprecipitation with MARCKS-specific antibodies (32). Expressing clones were re-transformed, and DNA was prepared by large scale plasmid preparations followed by double-band purification on cesium chloride density gradients (33).
MARCKS-GFP and MARCKS-RFP fusion constructs were made by digesting the pCMV4 plasmids containing various MARCKS inserts (above) with BsgI and digesting the 3Ј overhangs of the inserts with T4 DNA polymerase and Klenow fragment, followed by digestion with HindIII and purification of the MARCKS inserts. The inserts were subcloned into HindIII/SmaI-digested and phosphatase-treated pEGFP-N1 or pDsRed1-N1 (CLONTECH Laboratories, Palo Alto, CA). Transformed JM109 cells were selected with kanamycin (30 g/ml, Sigma), and clones containing inserts were tested for expression by transient transfection into 293 cells followed by immunoblotting of lysates with GFP-, RFP-, and MARCKS-specific antibodies (see below).
The GFP fusion peptide containing the myristoylated amino-terminal 58 amino acids of bovine MARCKS, Myr58, was made as follows; pCMV4AS was digested with SstI and PstI, and a 220-base pair fragment was purified and ligated into the pEGFP-N1 vector that had been digested with the same enzymes. Clones were analyzed for expression by immunoblotting as described above.
All DNA modifying enzymes were from Life Technologies, Inc. Cell Transfections-293 cells were transfected with DNA by the calcium phosphate precipitation method (34). Cells were plated at a density of 1-1.5 ϫ 10 6 cells/100-mm (58 cm 2 ) tissue culture plate (Costar, Cambridge, MA) or an equivalent cell density on different size plates. The specific amounts of DNA transfected are indicated in the figure legends. Following 4 h of exposure to DNA, cells were rinsed once in 0.9% NaCl, refed with growth medium, and grown for an additional 48 h before harvesting.
Adhesion Assays-Multiwell tissue culture plates (Costar) were coated with 5 g/cm 2 poly-D-lysine (PDL), 2 g/cm 2 fibronectin or collagen, or 4 g/cm 2 laminin. All were obtained from Collaborative Biomedical Products (Bedford, MA). All matrix components were diluted in phosphate-buffered saline (PBS) except collagen, which was diluted in 0.05 N HCl. Plates were coated for 1-2 h at room temperature with gentle shaking. Plates were rinsed three times with deionized H 2 O, followed by blocking of nonspecific sites with 1% (w/v) heat-denatured bovine serum albumin (low fatty acid content, Sigma) in PBS (35). Blocking was either overnight at 4°C or at room temperature for 1-2 h. Plates were rinsed five times with H 2 O and air-dried under a laminar flow hood.
To test the adhesion of 293 cells, transfected cells were collected by pipetting a stream of growth medium at the monolayer. Cells were pelleted by centrifugation at 600 ϫ g at room temperature for 5 min, followed by rinsing with serum-free medium and re-pelleting. Cell pellets were gently resuspended in 3 ml of serum-free medium by pipetting to ensure that no visible clumps of cells were present. Onehalf ml was plated on each of 3 pre-coated wells of a 24-multiwell tissue culture cluster plate (Costar). The remaining cell suspension was analyzed for MARCKS expression (see below). Following adhesion for 10 min at 37°C, the medium containing non-adherent cells was collected at room temperature by gently pipetting the medium three times over the adherent cells in order to release all non-adherent cells without disturbing the adherent cells. The adherent cells were gently rinsed once with saline (0.9% NaCl) at room temperature and released from the plates by treatment with 0.25% trypsin (w/v) in calcium-and magnesium-free Hanks' balanced salt solution (Life Technologies, Inc.) containing 1 mM EDTA at 37°C for 10 min, or until cells were released from the dish. Final suspensions of adherent and non-adherent cells prepared for flow cytometry contained 3% (v/v) FCS, 0.07% (w/v) trypsin, and 1% (v/v) formaldehyde.
Using a FacSort flow cytometer and CellQuest software (BD Biosciences, San Jose, CA), cells were sorted for size ( Fig. 1, R1), and then for fluorescence due to GFP (Fig. 1, R2). Five to ten thousand cells were sorted for each sample, and the percentage of GFP-expressing cells within each population was analyzed. The percentage of GFP-expressing cells in the adherent population was divided by the percentage of GFP-expressing cells in the non-adherent population; this figure was used as an index of overall cell adhesion. The efficiency of transfection among different experiments ranged from 10% to 30%. However, differences in transfection efficiency did not alter experimental outcome. Each point represents the mean Ϯ S.D. of three individual values. Each experiment was repeated at least three times with similar results. Both trypan blue exclusion and propidium iodide staining were used to rule out possible artifactual effects due to cell death (data not shown).
In the case of RGD peptide treatment of cells, transfected 293 cells were prepared for FACS analysis essentially as described above, except for the following modifications: Transfected 293 cells were resuspended in serum-free medium containing GRGDSP or control GRADSP peptides (Life Technologies, Inc.) at a final concentration of 1 mM. Cells were then allowed to adhere to fibronectin-coated plates for 15 min. The cells from a single 100-mm plate were plated on a fibronectincoated 9.6-cm 2 dish. The final cell suspensions of both adherent and nonadherent cells subjected to FACS analysis contained 1% (v/v) formaldehyde, and the adherent cell samples also contained 0.17% (w/v) trypsin.
In the case of cytochalasin D treatment of cells, transfected cells were incubated with 1 M cytochalasin D prepared in Me 2 SO, or an equivalent amount of Me 2 SO, for 30 min at 37°C. Cells were then analyzed for were transfected by the calcium phosphate precipitation method with 2 g of MARCKS-GFP and 8 g of control vector DNA. Forty-eight h following transfection, cells were collected and prepared for FACS analysis, as described under "Materials and Methods." Sorting of cells proceeds until 5000 cells, each of whose size is consistent with the previously determined forward and side scatter corresponding to 293 cells, is collected as indicated in R1 of the upper panel. In the lower panel, cells exhibiting forward scatter corresponding to the size of 293 cells were then sorted based on their fluorescence due to GFP expression following laser excitation at 488 nm and emission at 530 nm. Cells demonstrating fluorescence above the autofluorescence background are indicated in the R2 quadrant. Green and blue dots correspond to GFP-expressing and non-expressing cells, respectively. In this particular experiment, R2 contains 1,619 dots corresponding to GFP-expressing cells, representing 33% of the total cell population. adhesion as described above, including the modifications used for the RGD treatment.
Immunoblotting-Cell suspensions, 1.5 ml of the suspensions used for the adhesion assays (see above), were processed for immunoblotting by rinsing with PBS and lysing in immunoprecipitation buffer containing protease inhibitors (32). Solubilized pellets were snap-frozen in liquid nitrogen, followed by thawing and centrifugation at 10,000 ϫ g for 10 min at 4°C. Clarified supernatants were treated with sodium dodecyl sulfate (SDS) sample buffer (32), and proteins were separated by SDS-polyacrylamide gel electrophoresis on 10% polyacrylamide gels, followed by electrophoretic transfer to nitrocellulose filters. Filters were blocked with 5% (w/v) dry milk in Tris-buffered saline containing 3% (v/v) Tween 20, followed by incubation for 1 h at room temperature with the appropriate primary antibody in the same buffer. A commercial rabbit anti-GFP or anti-RFP antibody was used according to the manufacturer's instructions (CLONTECH Laboratories). An IgG fraction from a polyclonal MARCKS-specific antibody (36) was used at 1:100. Goat anti-rabbit horseradish peroxidase-conjugated second antibodies (Bio-Rad) were used at 1:5000 in the same buffer, and reactive proteins were detected with enhanced chemiluminescence (Amersham Pharmacia Biotech). When necessary, films were scanned by scanning densitometry followed by analysis using the public domain NIH Image program.
Confocal Microscopy-293 cells were plated and transfected on twowell Lab-Tek chamber slides (Nalge Nunc International, Naperville, IL). In some cases the slides were pre-coated with fibronectin to ensure that the cells remained attached to the surface throughout the fixation protocol. Preliminary experiments demonstrated similar morphology of 293 cells on uncoated or treated surfaces (data not shown). To prepare cells for the detection of GFP or RFP fluorescence, the medium was removed 48 h following transfection, the monolayers were rinsed briefly in PBS, and the cells were fixed for 10 min at room temperature in freshly prepared 4% (w/v) paraformaldehyde in PBS (37). Residual fix solution was removed, and the cells were rinsed briefly in PBS. The upper chamber portions of the slides were removed, and the fixed cells were treated with Prolong Anti-fade (Molecular Probes, Eugene, OR) and mounted under a coverslip. Images were collected using a Zeiss Inverted Confocal Laser Scanning Microscope, LSM-510, (Zeiss, Thornwood, NY); an excitation wavelength of 488 nm with an LP 505 emission filter, or wavelength of 543 nm with an LP 560 emission filter, was used to detect GFP or RFP, respectively. Images were analyzed using the LSM-510 Image software.
In the case of cytochalasin D treatment and rhodamine-phalloidin staining of actin filaments prior to confocal microscopy, transfected cells were treated with Me 2 SO or cytochalasin D as described above. Cells were fixed and stained with rhodamine-phalloidin according to the manufacturer's instructions (Molecular Probes, Eugene, OR) and detected as described above; an excitation wavelength of 488 nm with an LP 505 emission filter, or wavelength of 543 nm with an LP 560 emission filter, was used to detect GFP or rhodamine-phalloidin, respectively

Effect of MARCKS Expression on 293 Cell Adhesion to Fi-
bronectin-In order to analyze the ability of MARCKS to regulate cellular adhesion on a single cell basis, we used GFP as an indicator of transfection that would also allow cells to be sorted by FACS. Two approaches were used. In the first approach, cells were co-transfected with separate GFP-and MARCKS-expressing plasmids, using the assumption that all GFP-expressing cells should also have been transfected with the MARCKS cDNA. In the second approach, single plasmids expressing MARCKS-GFP fusion proteins were transfected. These independent methods were used to exclude the possibility that results obtained with the fusion proteins were anomalous.
The ability of 293 cells to adhere to various protein matrices was first tested (data not shown). The most consistent levels of adhesion between experiments were obtained following adhesion of the cells to fibronectin; this matrix protein was therefore used in subsequent experiments.
In the standard assay, 48 h following transfection, cells were allowed to adhere to fibronectin for 10 min at 37°C, followed by collection for FACS analysis as described under "Materials and Methods." To compensate for the fact that variable percentages of cells expressed the transfected proteins, the percentage of GFP-expressing cells in the adherent population divided by the percentage of GFP-expressing cells in the non-adherent population was used as the index of cellular adhesion.
When the cells were co-transfected with both the GFP expression plasmid and the CMV vector alone, the ratio of adherent/nonadherent cells was 1.3 Ϯ 0.07 ( Fig. 2; mean Ϯ S.D. of three values), whereas cells co-transfected with the pCMV4-MARCKS plasmid and the GFP plasmid exhibited a ratio of 0.6 Ϯ 0.03 adherent/nonadherent cells (p Ͻ 10 Ϫ4 ). In a parallel experiment, in which GFP was expressed alone or as a MARCKS-GFP fusion protein, the ratios of adherent/nonadherent cells were 0.92 Ϯ 0.04 and 0.47 Ϯ 0.07, respectively (p Ͻ 10 Ϫ4 ) (Fig. 2). Using scanning densitometry of immunoblots probed with MARCKS-specific antibodies (Fig. 2, inset; data not shown), the level of ectopic MARCKS-GFP expression was determined to be ϳ60 -100-fold greater than endogenous MARCKS. When adjusted for the percentage of cells expressing the GFP fusion proteins as ascertained through FACS analysis, 15-30%, it was determined that MARCKS-GFP expression ranged from 200-to 700-fold over endogenous MARCKS. Using the same approach, it was determined that cells expressing MARCKS as a non-fusion protein exhibited much lower levels, 10 -50-fold above endogenous. These data demonstrated that increased wild-type MARCKS expression, when co-expressed with GFP either as separate proteins or as a fusion protein, resulted in the decreased ability of 293 cells to adhere to a fibronectin-coated surface.
Myristoylation of MARCKS Is Necessary but Not Sufficient for Inhibition of Adhesion-MARCKS is thought to associate with the cytoplasmic face of the plasma membrane through hydrophobic and electrostatic interactions of its myristoylated In the latter case, the test DNA was mixed with the appropriate amount of CMV vector DNA to adjust the total amount of transfected DNA to 10 g. In both cases, transfected cells were plated on dishes previously coated with fibronectin, allowed to adhere for 10 min, and then collected and analyzed for cell adhesion by FACS as described under "Materials and Methods." Each bar represents the mean Ϯ S.D. of three values of adherent/nonadherent cells from representative experiments; in each experiment, 5000 cells were counted from both the adherent and non-adherent fractions. The means for MARCKS-expressing cells were significantly lower than control in both cases (p Ͻ 10 Ϫ4 ) using the Student's t test. The inset in the right panel shows clarified lysates from cells transfected with GFP alone (Ϫ) or MARCKS-GFP (WT) that were analyzed by immunoblotting with a MARCKS-specific antibody as described under "Materials and Methods." The positions of protein molecular weight standards are indicated. Endogenous human MARCKS and the GFP fusion protein migrate at ϳ80 and ϳ120 kDa, respectively. The MARCKS-GFP fusion protein (arrow) was expressed at levels considerably greater than those of endogenous MARCKS, which was only visible at longer chemiluminescence exposures. amino terminus and its PSD, respectively. Mutants in these two domains that were used to determine the requirement of both of these regions for membrane association (11,12) were used to evaluate whether either or both domains were responsible for the observed decrease in 293 cell adhesion to fibronectin. Expression of the wild-type MARCKS-GFP fusion protein, and the various non-phosphorylatable, constitutively membrane-associated mutants, AS-, GS-, and NS-GFP fusion proteins, exhibited essentially identical adherent/nonadherent cell ratios of 0.5 Ϯ 0.07, 0.5 Ϯ 0.03, 0.5 Ϯ 0.06, and 0.5 Ϯ 0.01, respectively (p Ͻ 0.002 when compared with the mean of the values from the control GFP-expressing cells using the Bonferroni correction for multiple means; Ref. 38) (Fig. 3A). The DS (pseudo-phosphorylated)-GFP fusion protein resulted in an adherent/nonadherent cell ratio similar to that of the constitutively membrane-bound mutants of 0.5 Ϯ 0.001 (p Ͻ 0.002 when compared with the control GFP-expressing cells, using the Bonferroni correction for multiple means). However, the A2G2 (non-myristoylated)-and DBL (non-myristoylated, pseudo-phosphorylated)-GFP fusion proteins resulted in adherent/ nonadherent cell ratios of 0.8 Ϯ 0.05 and 0.97 Ϯ 0.01, respectively (Fig. 3A). Both values were not significantly different from those obtained with the control GFP-expressing cells, which exhibited a ratio of 0.92 Ϯ 0.04 adherent/nonadherent cells (Fig. 3A). Therefore, whereas myristoylation of the protein appeared to be necessary for inhibition of adhesion under these conditions, the pseudo-phosphorylated version of the PSD, in a normally myristoylated protein, did not interfere with the MARCKS inhibition of 293 cell adhesion to fibronectin. Consistent with the observation that the pseudo-phosphorylated mutant inhibited adhesion to the same degree as the wild-type protein and the non-phosphorylatable, myristoylated mutants, was the observation that phorbol 12-myristate 13-acetate treatment (1.6 M in 0.01% Me 2 SO for 10 min before plating) of 293 cells expressing wild-type or mutant MARCKS did not alter the MARCKS effects on adhesion (data not shown), supporting our finding that MARCKS inhibition of 293 cells is likely to be PKC-independent.
The data suggesting that myristoylation was necessary for MARCKS to inhibit adhesion to fibronectin suggested the possibility that a similar inhibition of adhesion could occur with the overexpression of any myristoylated peptide. To test this, a GFP fusion peptide was tested that consisted of the aminoterminal 58 amino acids of MARCKS, including the myristoylation site. The effect of this peptide on adhesion was compared with the effects of the NS-, DS-, and DBL mutant MARCKS-GFP fusion proteins. Control GFP transfections resulted in a ratio of 0.88 Ϯ 0.03 adherent/nonadherent cells (Fig. 3B). The NS-and DS-GFP mutant protein-expressing cells exhibited adherent/nonadherent cell ratios of 0.60 Ϯ 0.06 and 0.64 Ϯ 0.04, respectively, values that were significantly different from those obtained in the control GFP-expressing cells (p Ͻ 0.004) (Fig. 3B). The DBL-GFP-expressing cells exhibited an adherent/nonadherent cell ratio of 0.91 Ϯ 0.03, which was not significantly different from that obtained with the GFP-expressing cells (Fig. 3B). Expression of the myristoylated 58-residue amino-terminal MARCKS peptide (Myr58) demonstrated an adherent/nonadherent cell ratio of 0.78 Ϯ 0.004 (Fig. 3B). Although this value was slightly less than that obtained for the GFP-expressing cells (p Ͻ 0.01), in no experiment did this mutant ever inhibit adhesion to the same degree as the fulllength myristoylated protein, and in certain instances, the adherent/nonadherent ratio seen with this mutant was the same as that seen for GFP or GFP-DBL expressing cells. These data demonstrated that expression of at least one myristoylated peptide alone is not sufficient to inhibit adhesion of 293 cells to fibronectin in this assay, at least not to the same extent as the full-length MARCKS protein, and that other regions of MARCKS are required, in addition to the myristoyl moiety, to promote inhibited adhesion.
Approximately equal expression of all full-length mutants was demonstrated by immunoblotting, indicating that the observed differences in adhesion were not due to differences in protein expression (Fig. 3C). The higher levels of expression seen with GFP alone and the Myr58-GFP peptide were most likely due to greater stability of these smaller proteins in the cells.
MARCKS Inhibits Adhesion of 293 Cells to Various Matrices-To determine whether MARCKS inhibition of 293 cell adhesion was specific to fibronectin, transfected cells were analyzed for their ability to adhere to various matrices. Cells expressing wild-type MARCKS, as well as all of the myristoylated full-length PSD mutant proteins, exhibited an average inhibition of adhesion to PDL (38%, p Ͻ 5 ϫ 10 Ϫ2 ), fibronectin (49%, p Ͻ 2 ϫ 10 Ϫ3 ), collagen (54%, p Ͻ 8 ϫ 10 Ϫ5 ), and laminin (47%, p Ͻ 2 ϫ 10 Ϫ3 ), respectively, when compared with control cells transfected with vector alone (Fig. 4A). As with fibronectin, the non-myristoylated DBL mutant did not affect adhesion to PDL, and only slightly inhibited adhesion to collagen and laminin (9% and 14% inhibition, respectively; Fig. 4A). Expression of non-myristoylated but otherwise wild-type MARCKS did not inhibit adhesion to PDL; however, expression of this mutant protein inhibited adhesion to collagen and laminin by 29% (p Ͻ 2 ϫ 10 Ϫ4 ) and 24% (p Ͻ 2 ϫ 10 Ϫ3 ), respectively. These experiments were repeated several times with similar results. We also found that the myristoylated NS and DS mutants inhibited adhesion to plastic, by ϳ50%, whereas the non-myristoylated A2G2 and DBL mutants did not (data not shown).
Inhibition of adhesion to various protein and non-protein substrates suggested that MARCKS-mediated effects were independent of specific integrin receptors. To determine whether MARCKS-inhibition of cellular adhesion to fibronectin was independent of integrin receptors, 293 cells expressing GFP, NS-, or DBL-GFP were treated with an arginine-glycine-aspartate (RGD) peptide, which inhibits the binding of normal rat kidney cells to fibronectin (39). Treatment of 293 cells with 1 mM RGD peptide during plating on fibronectin resulted in 32%, 36%, and 44% (p Ͻ 0.002) decreases in adhesion of GFP-, NS-GFP-, and DBL-GFP-expressing cells, respectively, when compared with treatment with the arginine-alanine-aspartate (RAD) control peptide. Inhibition of 293 adhesion to fibronectin by the competitive peptide regardless of MARCKS expression is consistent with the data presented in Fig. 4A, which suggest that MARCKS-mediated inhibition of adhesion is independent of specific integrin receptors.
Subcellular Localization of Myristoylated and Non-myristoy-lated MARCKS in 293 Cells-Confocal laser microscopy was used to visualize the subcellular localization of the MARCKS-GFP or MARCKS-RFP fusion proteins (Figs. 5 and 6). Approximately 10 -30% of the cells present in a representative field expressed GFP and the MARCKS-GFP fusion proteins (data not shown). As expected, GFP alone was expressed at approximately equal levels in both the nucleus and cytosol (data not shown). RFP alone, or MARCKS-RFP fusion proteins, exhibited the same localization and levels of expression as their GFP counterparts (data not shown; Fig. 6). The confocal images shown in Fig. 5 compare the different subcellular localizations exhibited by the various MARCKS-GFP fusion proteins. The fully myristoylated NS-GFP protein was localized predominantly at the plasma membrane, as well as in long cellular processes that extended from the cells (Fig.  5A). The wild-type protein exhibited both membrane and cytosolic localization (data not shown), probably because a substantial proportion of wild-type MARCKS would be expected to be phosphorylated and localized to the cytosol (11,12). Both of the non-myristoylated proteins, DBL-and A2G2-GFP, were dispersed throughout the cytosol, and, when highly overexpressed, also appeared in the nucleus (Fig. 5, C and D). The PSD of MARCKS contains sequences identical to known nuclear localization sequences (40,41), and it is possible that lack of the myristate moiety may allow translocation of these mutants to the nucleus. In addition, the A2G2 mutant could also be seen in cellular processes (Fig. 5C) as seen with the wildtype and NS proteins. The DS-and the Myr58-GFP proteins exhibited predominantly non-nuclear, cytosolic staining (Fig. 5,  B and E).
The localization of the various MARCKS-GFP mutants did not correlate perfectly with the ability of MARCKS to inhibit  (Fig. 3), yet localized differently in the cells (Fig. 5, A and B). Furthermore, the Myr58 peptide and the DS mutant localized to the same regions of the cell (Fig. 5,  B and E), but only the DS mutant inhibited cellular adhesion (Fig. 3). In order to determine the regions of apparent exclusion and/or overlap among the various mutant proteins, cells were co-transfected with combinations of MARCKS-GFP and -RFP mutant plasmids (Fig. 6).
The confocal images of 293 cells co-transfected with both NS-RFP and DS-GFP are shown in Fig. 6 (A-C). The NS-RFP protein is predominantly localized to the plasma membrane (red; Fig. 6A), and the DS-GFP protein is cytosolic (green; Fig.  6B). The minimal region of resulting overlap near the plasma membrane is shown in Fig. 6C (yellow).
Co-transfection of the NS-RFP and A2G2-GFP mutants (Fig.  6, D-F), resulted in distinct cytosolic and membrane localizations of A2G2-GFP (green; Fig. 6E) and NS-RFP (red; Fig. 6D), respectively. Again, there was a region of overlap between these two mutants at the plasma membrane (yellow; Fig. 6F).
Finally, the co-localization of the DS-RFP protein and the Myr58-GFP peptide were examined (Fig. 6, G-I). The DS-RFP protein (red, Fig. 6G) and the Myr58-GFP peptide (green; Fig.  6H) were both distributed throughout the cytosol. Fig. 6I shows that these two mutants were expressed in the same regions of the cell, as indicated by the extensive yellow fluorescence and essentially undetectable green or red fluorescence.
These images suggest that the presence of MARCKS at the cytoplasmic face of the plasma membrane may be necessary for MARCKS to inhibit cellular adhesion. However, localization to this region appears not to be sufficient to inhibit adhesion.
MARCKS-expressing 293 Cells Are Refractory to Cytochalasin D-mediated Inhibition of 293 Adhesion to Fibronectin-To begin to understand the mechanism by which MARCKS decreased the adhesion of 293 cells to various substrates, we employed the pharmacological agent, cytochalasin D, to destabilize actin filaments. Cytochalasin D treatment (1 M for 30 min at 37°C) of GFP-or DBL-GFP-expressing 293 cells resulted in a 40% (p Ͻ 0.002) decrease in adhesion to fibronectin when compared with Me 2 SO control-treated cells (Fig. 7). In contrast, cytochalasin D treatment of 293 cells expressing either the NS-or GS-GFP myristoylated, full-length mutants did not further inhibit binding over that already occurring in response to MARCKS expression (Fig. 7).
To confirm the destabilization of actin filaments in 293 cells in response to cytochalasin D treatment, and the subcellular localizations of the MARCKS-GFP fusion proteins in response to cytochalasin D, cells transfected with GFP (Fig. 8, A and B), NS-GFP (Fig. 8, C and D), or DBL-GFP (Fig. 8, E and F) cDNA constructs, followed by treatment with Me 2 SO as a control (Fig.  8, A, C, and E), or cytochalasin D (Fig. 8, B, D, and F), for 30 min at 37°C, were fixed and incubated with rhodamine-conjugated phalloidin used to specifically stain actin. The subcellular localization of GFP expression (green) in cells expressing control GFP, NS-GFP, and DBL-GFP, following Me 2 SO treatment, was similar to that seen in Figs. 5 and 6. GFP was expressed throughout the cell (Fig. 8A), NS-GFP was expressed predominantly at the plasma membrane and in cellular extensions (Fig. 8C), and DBL-GFP was expressed throughout the cytoplasm and, when overexpressed, also in the nuclei (Fig.   8E). Rhodamine-phalloidin fluorescence (red) showed actin filaments and cortical actin concentrated at the edges of the cells. This was detected in 293 cells regardless of the MARCKS construct that was transfected (Fig. 8, A, C, and E). In addition, when GFP alone or as a MARCKS fusion protein was overexpressed in these cells, at least some degree of co-localization with actin was noted (yellow). Following treatment with cytochalasin D, the actin filaments were fragmented, as indicated by the loss of long filaments and the appearance of small rhodamine-phalloidin-staining fragments (Fig. 8, B, D, and F). The effect of cytochalasin D on the localization of GFP, NS-GFP, and DBL-GFP fusion proteins appeared to be negligible. GFP and DBL-GFP were detected throughout the cell (Fig. 8, B and F), and NS-GFP was still confined to the perimeter of the cell (Fig. 8D). In addition, NS-GFP was detected in long cellular processes (Fig. 8D). These images demonstrate that the cytochalasin D treatment used in these studies was effective at causing the fragmentation of actin filaments, but that this fragmentation did not appear to significantly alter the cellular localization of either the membrane-associated, myristoylated NS-GFP or the non-membrane-associated, DBL-GFP MARCKS mutant proteins.

DISCUSSION
The major finding of this study is that overexpression of myristoylated, full-length MARCKS in 293 cells resulted in decreased adhesion of cells to a variety of matrix protein-coated surfaces. This inhibition appeared to require myristoylation of the protein as well as association of MARCKS with the plasma membrane, although neither of these properties was sufficient to inhibit adhesion. These data suggest that MARCKS may influence processes requiring cellular adhesion, such as cell: matrix interactions or cellular migration, by decreasing cell adhesion to extracellular substrates.
Although the cellular functions of MARCKS remain unknown, the phenotype exhibited by MARCKS-deficient mice (6,7,42) is consistent with an inability of developing forebrain cortical neurons to migrate to their appropriate final destinations. In this phenotype, radially migrating cortical neurons often migrate through the pial-glial interface and into the subarachnoid space, a phenomenon known as neuronal leptomeningeal ectopia. Among the several possible mechanisms for this effect are: 1) the migrating cells may not interact appropriately with the surrounding ECM, which in turn could affect the ability of cells to migrate to and/or stop migrating at their appropriate final destinations; 2) the cells may adhere to each other (both neuron-neuron and neuron-radial glia interactions) abnormally, which could also affect their final destinations; 3) the cells may respond abnormally to extracellular signals regulating their migration, or these signals themselves may be abnormal; and 4) the synthesis or breakdown of ECM components may be abnormal in the MARCKS-deficient forebrain cortex.
The data presented here suggest that overexpression of normally myristoylated MARCKS results in decreased cellular adhesion to ECM proteins; conversely, it is possible that MARCKS deficiency could result in enhanced adherence to the matrix and abnormal migration. Most cell:matrix interactions are thought to occur through the interactions of integrins with their matrix ligands (43,44). It has been shown previously that 293 cells contain integrin receptors, formed by the pairing of a ␤ 1 subunit with various ␣ subunits (45), for the ligands tested in this study. Overexpression of MARCKS inhibited cell adhesion to all of the matrix proteins tested, suggesting that this inhibition was not specific for a single receptor. In addition, myristoylated MARCKS also inhibited 293 cell adhesion to both PDL and plastic, substrates to which cell binding is not thought to be integrin-dependent (46,47). Consistent with these observations, the fibronectin-specific competing peptide, RGD, inhibited 293 cell adhesion to approximately the same extent regardless of the MARCKS mutant expressed, supporting the likelihood that MARCKS inhibition of 293 cell adhesion is largely independent of direct integrin receptor modulation.
However, the observation that inhibition of 293 cell adhesion by the overexpression of MARCKS appears to be independent of integrin-mediated binding does not rule out the possibility that MARCKS may be interacting with other proteins that bridge integrins and other membrane receptors to the actin cytoskeleton. Several studies have attempted to functionally link MARCKS to various cytoskeletal proteins including vinculin, talin, paxillin, and tetraspanin (19,44,48,49). In addition, many reports have attempted to link MARCKS to cytoskeletal events through its interaction with actin (28,29). Although the data are convincing that the MARCKS PSD can bind actin in vitro, this characteristic is not restricted to the MARCKS protein, and proteins containing domains similar to the MARCKS PSD, such as adducin, can also bind actin filaments (50).
The fungal cytochalasins are thought to disrupt actin filaments by capping the faster growing barbed ends of filaments and cleaving the microfilaments (51). Various studies have demonstrated cytochalasin effects on cellular adhesion and/or cell spreading (52)(53)(54)(55)(56)(57)(58). In addition, several groups have demonstrated an intact focal adhesion signaling pathway in 293 cells and its sensitivity to cytochalasin D (59 -61).
In the studies presented here, expression of either membrane-associated or cytosolic MARCKS did not appear to alter actin filament organization when compared with control transfected 293 cells. Furthermore, the observed collapse of actin filaments in response to cytochalasin D was similar regardless of MARCKS expression. It is intriguing, however, that cells expressing full-length myristoylated MARCKS were refractory to further inhibition of adhesion by cytochalasin D. MARCKS and cytochalasin D inhibition of adhesion were not additive, suggesting that they inhibit adhesion by affecting the same pathway; however, the data do not indicate the position of MARCKS in the actin filament destabilization pathway.
In a recent study, involving the MARCKS relative MLP, Zhou and Li (62) demonstrated that an effect of MLP to increase the diffusion rate of ␤ 2 integrin receptors in macrophages could be mimicked by cytochalasin D. Both MLP expression and cytochalasin D treatment caused increased cellular adhesion. Their studies differed from ours in that macrophages were used instead of 293 cells, they used MLP instead of MARCKS, and MLP required PKC-mediated phosphorylation to increase ␤ 2 integrin lateral diffusion (62), whereas our data demonstrated that MARCKS inhibition of 293 cell adhesion was likely to be a PKC-independent event. Nonetheless, the implications of both sets of data are that both MARCKS and MLP can regulate cytoskeletal events, perhaps resulting in alterations of cellular adhesion to the substratum.
The apparent requirement that MARCKS be myristoylated to inhibit 293 cell adhesion suggested that overexpression of any myristoylated protein might interfere with the cell's ability to adhere to surfaces. However, expression of a myristoylated, 58-amino acid MARCKS amino-terminal peptide at levels even higher than the overexpressed MARCKS did not inhibit 293 cell adhesion. We used this peptide as a control rather than other myristoyl proteins, such as Src, since many of these proteins can affect cellular adhesion (63). The requirement for the myristoyl moiety suggests that MARCKS needs to associate with membranes through its hydrophobic interactions in order to exert its inhibitory effects on cellular adhesion, although a myristoyl-dependent interaction with other proteins cannot be ruled out.
Myat et al. (64) showed that a mutant MARCKS protein, in which the amino-terminal myristoyl moiety of MARCKS was replaced by two palmitoyl groups, bound with higher than normal affinity to the plasma membrane. Expression of this palmitoylated mutant MARCKS in mouse fibroblasts resulted in decreased cellular adhesion to fibronectin-coated surfaces when compared with cells expressing the wild-type protein.
Their experiments did not address whether or not the overexpression of wild-type MARCKS caused a difference in cellular adhesion when compared with cells expressing endogenous MARCKS. However, their data and ours support the theory that hydrophobic interactions of fatty acyl-modified MARCKS with either the lipid bilayer, other lipid-modified proteins, or hydrophobic domain-containing proteins can inhibit cellular adhesion to surfaces.
To test whether the inhibition of cell-matrix adhesion by MARCKS might require its localization to the cytoplasmic face of the plasma membrane, we performed confocal microscopy of the various GFP-and RFP-MARCKS fusion proteins. As expected, the NS mutant was found almost entirely along the plasma membrane. However, the DS mutant, which inhibited cellular adhesion of 293 cells as well as the NS mutant, was predominantly cytosolic in this cell system. Co-transfection of NS-RFP and DS-GFP resulted in a zone of co-localization of the two proteins at the cytoplasmic face of the plasma membrane, suggesting the possibility that the presence of MARCKS in this region may be sufficient to inhibit cellular adhesion. However, the A2G2 protein also partially co-localized with the NS protein at the cell membrane and in cellular extensions, but was not able to inhibit cellular adhesion. These data suggest that membrane localization is required for binding inhibition, leaving open the possibility that MARCKS exerts its effects on cellular adhesion through interactions with other protein(s) in close proximity to the plasma membrane, rather than through hydrophobic interactions with the lipid bilayer alone.
A caveat of the work presented here is that it involved overexpression of MARCKS in 293 cells, often used for expression studies because of their ease of transfection and their robust expression of proteins following cDNA transfection. The reproducible inhibition of adhesion caused by full-length myristoylated MARCKS in these cells may have been detectable only because of the high levels of MARCKS expression achieved. However, this system allows for the convenient overexpression of normal, mutated, and labeled MARCKS, which can then be used as an affinity probe for potential interacting proteins or cellular structures.