MARCKS Is a Natively Unfolded Protein with an Inaccessible Actin-binding Site

Myristoylated alanine-rich C kinase substrate (MARCKS) is an unfolded protein that contains well characterized actin-binding sites within the phosphorylation site domain (PSD), yet paradoxically, we now find that intact MARCKS does not bind to actin. Intact MARCKS also does not bind as well to calmodulin as does the PSD alone. Myristoylation at the N terminus alters how calmodulin binds to MARCKS, implying that, despite its unfolded state, the distant N terminus influences binding events at the PSD. We show that the free PSD binds with site specificity to MARCKS, suggesting that long-range intramolecular interactions within MARCKS are also possible. Because of the unusual primary sequence of MARCKS with an overall isoelectric point of 4.2 yet a very basic PSD (overall charge of +13), we speculated that ionic interactions between oppositely charged domains of MARCKS were responsible for long-range interactions within MARCKS that sterically influence binding events at the PSD and that explain the observed differences between properties of the PSD and MARCKS. Consistent with this hypothesis, chemical modifications of MARCKS that neutralize negatively charged residues outside of the PSD allow the PSD to bind to actin and increase the affinity of MARCKS for calmodulin. Similarly, both myristoylation of MARCKS and cleavage of MARCKS by calpain are shown to increase the availability of the PSD so as to activate its actin-binding activity. Because abundant evidence supports the conclusion that MARCKS is an important protein in regulating actin dynamics, our data imply that post-translational modifications of MARCKS are necessary and sufficient to regulate actin-binding activity.

molecule (1-3) with a centrally located active site known as the phosphorylation site domain (PSD). Consistent with the paradigm for natively unfolded proteins, MARCKS is thought to interact with several ligands so as to integrate information from various signal transduction pathways to produce an output signal that regulates cell motile and contractile function. Numerous studies of the MARCKS protein have utilized a peptide with a sequence that corresponds to the PSD peptide as a substitute for studying interactions between the intact protein and its multiple ligands (3,4). Although this approach intuitively appears to be logical, given the unfolded state of the native protein, the substitution of the PSD peptide for the intact protein has never been rigorously justified. In fact, there are several reported experiments that imply that the PSD peptide behaves differently from intact MARCKS. The nonphosphorylated PSD peptide is known to have extended structure, to nucleate polymerization, and to cross-link F-actin filaments (5-7), presumably because of two binding sites with a site-specific K d of ϳ0.5 M for F-actin (8). Although the PSD of MARCKS and its homolog MARCKS-related protein have both been shown to bind to actin with similar affinity, intact recombinant MARCKS-related protein, with or without myristoylation, exhibits a lower affinity for actin (much greater than 1 M) and does not cross-link F-actin or induce G-actin polymerization (9). Although full-length MARCKS has been shown to bind and bundle F-actin (4,10), the published data are only semiquantitative. Comparison of the available data also suggests that intact MARCKS binds to vesicles containing acidic phospholipids with 10 4 -fold lower affinity than does the PSD peptide alone (11).
Calcium-dependent interactions between the PSD peptide and the Ca 2ϩ -binding protein calmodulin have been extensively characterized, and a crystallographic structure is available that reveals that the phenylalanine residues of the PSD are buried in a hydrophobic tunnel of calmodulin and that the highly charged termini of the peptide interact with patches of opposite charge on the surface of calmodulin (12). Once again, the PSD peptide is said to interact with higher affinity than the intact protein (K d ϭ 3.8 versus 12.7 nM, respectively, in 0.1 M KCl) with this ligand (13). Recently, it has been speculated that myristoylation of MARCKS adds a second, low affinity, calmodulin-binding site to MARCKS without evidence of cooperativity (13). Importantly, the addition of a second non-cooperative binding site cannot explain the long-known result that myristoylated MARCKS binds to calmodulin with higher affinity than does non-myristoylated MARCKS (14) unless both binding sites can interact with a single calmodulin molecule simultaneously or myristoylation itself changes the accessibil-ity of the PSD to bind to calmodulin. Independent binding would change the stoichiometry of the interaction to two calmodulin molecules/MARCKS, but could not significantly increase the apparent affinity unless the calmodulin ligand was multivalent (oligomerized or attached to a bead). A recent crystallographic structure of a myristoylated peptide bound to calmodulin (15), with the myristoyl group in the same hydrophobic tunnel and interacting with many of the same residues as the phenylalanines of the PSD peptide, suggests that both the N-terminal myristoyl-and PSD-binding sites could not simultaneously interact with calmodulin without significant steric effects. However, the hydrophobic tunnel through calmodulin has been shown to be quite flexible, and there are no experimental data to rule out the possibility that calmodulin could adjust to accommodate both putative binding regions. Of note, myristoylation of MARCKS is likely a dynamically regulated post-translational event (16). Non-myristoylated MARCKS has been isolated from bovine brain (14), and a demyristoylase activity has been characterized (13), thus making the acronym somewhat of a misnomer. For the purposes of this study, except in instances in which there may be some confusion, we refer to MARCKS as the non-myristoylated protein, for which, because of its natively unfolded structure, there is reason to believe that the native protein is equivalent to the recombinant protein.
Many in vivo studies have implicated MARCKS in an actinregulating function (18 -22), but the evidence is indirect and could in all cases be explained by invoking schemes in which MARCKS alters events that in turn regulate cytoskeletal dynamics. However, in several specific examples, MARCKS colocalizes with F-actin, and dissociation from actin is temporally associated with alterations in actin dynamics (22,23), or MARCKS localization is altered after treatment that disrupts actin filaments (18,22). Such data are most readily interpreted as direct effects of MARCKS on actin. One target of MARCKS, phosphatidylinositol 4,5-bisphosphate, has been implicated in controlling the actin cytoskeleton by binding to many other actin-regulating proteins such as neural Wiskott-Aldrich syndrome protein (24), suggesting one possible indirect mechanism of actin control by MARCKS. MARCKS may sequester phosphatidylinositol 4,5-bisphosphate in the plasma membrane by reversible PSD binding (25). However, it should be noted that published estimates of the intracellular MARCKS concentration (12 M) (26) are based on the yield of a bovine forebrain preparation, but the original data are probably more consistent with an intracellular concentration of 1.2 M (27). Thus, although there may not be sufficient MARCKS to globally regulate phosphatidylinositol 4,5-bisphosphate metabolism, local pools of phosphatidylinositol 4,5-bisphosphate could potentially be regulated by MARCKS.
If the PSD of MARCKS is incompletely accessible to its many ligands in the intact protein, then two questions emerge that are addressed in this study. 1) How can a natively unfolded protein maintain the PSD in a buried unavailable position? 2) Do mechanisms exist to alter the availability of the PSD? Here, we test a novel structural hypothesis related to the unusual charge distribution in the primary sequence of MARCKS and attribute physical significance to post-translational modifications that are now shown to regulate the actinbinding functions of MARCKS.
Preparation of MARCKS and Myristoylated MARCKS-Murine fulllength MARCKS DNA (GenBank TM /EBI accession number M60474) was inserted into the pMW172 vector (30) and transformed into Escherichia coli BL21(DE3) competent cells. Cells (300 ml) were grown overnight in LB medium and used to inoculate 2 liters of culture. After 3 h, protein expression was further induced by the addition of isopropyl ␤-D-thiogalactopyranoside (100 M) for 1 h. Cultures were spun down and frozen at Ϫ80°C. Frozen cultures were resuspended in buffer consisting of 10 mM Tris-HCl, 100 M EGTA, 5 mM ␤-mercaptoethanol, 2 mM EDTA, 100 M phenylmethylsulfonyl fluoride, and 0.6 mM diisopropyl fluorophosphate (pH 7.9). Resuspended cells were sonicated, heated to 85°C for 10 min, and centrifuged at 38,000 rpm for 1 h. The supernatant was loaded onto a DEAE column, and fractions were collected with a 100 -400 mM KCl gradient in 10 mM Tris-HCl (pH 7.9). Fractions containing MARCKS (as shown by SDS-PAGE and Western blotting using goat anti-MARCKS polyclonal antibody raised against a C-terminal synthetic peptide (Serotec Inc., Raleigh, NC)) were combined and concentrated on a hydroxylapatite column and further purified by gel filtration on Sephacryl 300 HR. Concentration was determined by UV absorption at 258 nm (⑀ M ϭ 1100 at 258 nm) or by amino acid analysis. MARCKS typically eluted as a monomer at 40 -70 M and was stored at Ϫ80°C in the column buffer, which contained 5 mM Tris-HCl, 5.0 mM ␤-mercaptoethanol, and 50 mM KCl (pH 7.9) (MARCKS buffer).
For myristoylated MARCKS, E. coli strain BL21 was transformed with both the pBB131NMT plasmid (a gift from Dr. J. I. Gordon, Washington University), which contains the gene for yeast N-myristoyltransferase (31), and the pMW172-MARCKS plasmid described above and selected in the presence of 50 g/ml kanamycin and ampicillin. A frozen stock of transformed colonies was used to inoculate 200 ml of LB medium containing 50 g/ml kanamycin and ampicillin. The overnight culture (2 liters) was grown to log phase; 400 mM isopropyl ␤-D-thiogalactopyranoside was added to induce protein expression; and the culture was grown for an additional 3 h. Myristoylated MARCKS was purified according to the protocol for MARCKS described above. Myristoylated MARCKS runs at ϳ83 kDa on SDS-polyacrylamide gel (a slightly higher apparent molecular mass than that of MARCKS). It was characterized by mass spectroscopy with a peak at 29,874 Da compared with 29,664 Da for non-myristoylated MARCKS, and these correspond to the predicted masses based on the sequence.
To test for the propensity of MARCKS to aggregate, gel-filtered monomeric MARCKS was concentrated to 300 M in an Microcon filtration device (Millipore Corp., Billerica, MA) and then diluted to 100, 18, or 3 M in MARCKS buffer. Samples were centrifuged at 150,000 ϫ g for 15 min through a 20% sucrose cushion either 1 or 12 h after dilution. Pellets and supernatants were analyzed by SDS-PAGE with loading volumes inversely proportional to the protein concentration.
Assays of Actin-binding Function-Binding of MARCKS, covalently modified MARCKS, myristoylated MARCKS, or calpain-digested MARCKS to F-actin was detected by a high speed centrifugation assay. Mg 2ϩ -F-actin was prepared by converting Ca 2ϩ -G-actin to Mg 2ϩ -Gactin by the addition of 0.125 mM EGTA and 0.05 mM MgCl 2 for 10 min at room temperature and then polymerizing by the addition of MgCl 2 to a 2.0 mM final concentration. MARCKS (0 -10 M) was added to Mg 2ϩ -F-actin (0 -40 M) in a total volume of 100 l and equilibrated for varying times (20 min to 24 h). F-actin was pelleted at 140,000 ϫ g in a tabletop ultracentrifuge for 1 h. Supernatants (60 l) were removed, and pellets were washed gently three times to remove trapped unbound protein. Supernatants and pellets were then analyzed by SDS-PAGE or by fluorescence spectrometry to determine bound (pellet) or free (supernatant) MARCKS. For SDS-PAGE of covalently cross-linked MARCKS, 12% polyacrylamide gels were stained with SYPRO Ruby protein stain (Molecular Probes, Inc., Eugene, OR) after the samples were concentrated in a filtration device. Actin filament cross-linking or bundling was assessed by a low speed pelleting assay. Filament aggregates of either ordered bundles or isotropic networks of cross-linked filaments sediment at low centrifugal forces. Proteins or peptides were added to Mg 2ϩ -F-actin (7 M final concentration) to a final volume of 80 l. After a 10-min incubation, samples were centrifuged at 8000 ϫ g for 20 min to pellet actin filament aggregates and any associated proteins. Supernatants (30 l) were removed and loaded onto 10% SDS-polyacrylamide gel. The presence of bundling or cross-linking is indicated by the depletion of actin from the supernatant. The effects of MARCKS on the time course of actin filament polymerization were measured by the fluorescence change associated with the polymerization of pyrenyl-actin (29). Ca 2ϩ -G-actin (3 M, 4% pyrenyl-actin) was converted to Mg 2ϩ -actin as described above, and polymerization was initiated by adjustment to 50 mM KCl and 2 mM MgCl 2 . (Experiments without KCl are specifically indicated below.) Seeded polymerization assays employed cross-linked oligomeric F-actin seeds (32) and 4% pyrenyl-labeled Mg 2ϩ -actin monomer (0.5 M). The assay was shown to be linear in response to seed concentration and actin monomer concentration with variation from these conditions. Polymerization in the presence or absence of MARCKS was measured using pyrene fluorescence, and the initial polymerization rates were determined using time course data that could be fit with a line without systematic deviation. The time course of actin filament depolymerization was assayed by dilution of 10% pyrenyl-labeled F-actin (10 M) polymerized with 2 mM MgCl 2 to 0.1 M in the same buffer containing 0 or 2.0 M MARCKS.
Calmodulin Binding Assays-Binding of the Rh-PSD peptide or MARCKS to CaM-Sepharose beads was determined in a pull-down assay as described previously (33). Varying amounts of CaM-Sepharose suspended beads (total CaM concentration of 0 -2 M) were pipetted into 0.5-ml Eppendorf tubes containing fixed amounts of Rh-PSD (0.6 or 0.06 M) in 50 mM KCl, 0.1 mM CaCl 2 , and 10 mM Tris-HCl (pH 7.9) to a final volume of 400 l. Tubes were centrifuged at 7840 ϫ g for 1 min to pellet the CaM⅐Rh-PSD bead complex, and the amount of free Rh-PSD was quantified by fluorescence spectroscopy using excitation at 522 nm and emission at 575 nm. The concentration of available CaM on the Sepharose beads was calibrated based on data obtained at 0.6 M Rh-PSD instead of using the manufacturer's estimate, which was ϳ20% higher than the value we obtained (see "Results" for details). For fulllength recombinant MARCKS, the amount of MARCKS in the supernatant was quantified by Coomassie Blue staining after SDS-PAGE. Because the fluorescence of the Rh-PSD peptide increased by a factor of 2.3 upon binding to CaM, the increment in fluorescence could also be used to measure the fraction of bound Rh-PSD at varying concentrations of CaM in solution. Also, the fluorescence anisotropy of the Rh-PSD peptide changed upon binding to CaM, providing another quantitative assay for binding, which is discussed more completely below. Finally, wheat germ MIANS-CaM, labeled specifically with a fluorophore at its single cysteine residue at position 27, has been used successfully to generate a binding isotherm for calmodulin-binding proteins based on a large increment in fluorescence intensity (34). Saturation of MIANS-CaM by recombinant MARCKS, but not by the PSD peptide, yielded a significant increase in fluorescence intensity with excitation at 322 nm and emission at 438 nm.
Fluorescence Anisotropy-Data were collected on a Photon Technology International spectrofluorometer. The Rh-PSD was excited with vertically polarized light at 546 nm. The horizontal (I h ) and vertical (I v ) components of the emitted light were detected distal to a long-pass filter with a nominal cutoff of 570 nm (catalog no. 03 FCG 089; Melles Griot, Rochester, NY) for ϳ20 s for each component. The total intensity of the Rh-PSD peptide fluorescence (I v ϩ 2GI h ) was found to change proportionally upon MARCKS or CaM binding. The ratio (R) of the fluorescence of the bound species divided by that of the free species was calculated. The fluorescence anisotropy (r) was calculated using r ϭ The G factor was determined for each experiment with the peptide in solution excited with horizontally polarized light and averaged over ϳ10 measurements. The experiments were performed using 0.3-ml samples in glass cuvettes. For competition assays, the direct binding assay was used to estimate an amount of MARCKS that bound approximately two-thirds of the Rh-PSD peptide, and the anisotropy was measured at these concentrations of MARCKS and the Rh-PSD with varying concentrations of the competing unlabeled PSD.
The fitting procedure for the fluorescence anisotropy data included correction for the fact that the total fluorescence intensity changed upon binding of the PSD to CaM (see Fig. 2, B and C) or MARCKS (see Fig.  4). Because bound MARCKS or CaM changed the fluorescence intensity of the Rh-PSD, the observed anisotropy (r) was a nonlinear function of the fraction of PSD that bound to MARCKS or CaM: , where r f is the anisotropy of the free PSD, r b is the anisotropy of the PSD bound to MARCKS or CaM, b is the fraction of PSD bound to MARCKS or CaM, and the ratio R is defined above.
PSD-MARCKS Interactions-MARCKS (5 M) was covalently crosslinked to a rhodamine-labeled N-terminal PSD peptide with sequence KKKKKRFSFKK (25 M) using 50 mM 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), 16.5 mM N-hydroxysulfosuccinimide, and 25 mM sodium phosphate dibasic to give a final EDC concentration of 14 mM. Samples were incubated at room temperature for 30 min, and cross-linked proteins were separated by SDS-PAGE. This rhodaminelabeled peptide gave results indistinguishable from those of the Rh-PSD in the anisotropy assay measuring competition with the unlabeled PSD peptide for binding to MARCKS (data not shown). Binding of the rhodamine-labeled N-terminal PSD peptide to MARCKS was also assessed in a gel overlay (far-Western blot). After overnight transfer to polyvinylidene difluoride membrane of purified recombinant murine MARCKS or a whole cell extract prepared by the addition of SDS gel buffer to washed cultured macrophages (RAW 264.7 murine macrophage cell line) and blocking with 1% bovine serum albumin and 100 mM L-lysine, membranes were probed with peptide (5-10 M) in MARCKS buffer for 1 h. Excess probe was removed by washing in PBS, and the membrane was visualized and imaged by indirect fluorescence.
Preparation of EDC-modified MARCKS-As described previously for modification of the acidic residues of caldesmon (35), MARCKS (28 M) in 100 mM MES and 400 mM ethanolamine (pH 5.5) was incubated for 10 min at room temperature. Freshly prepared EDC (200 mM) was added to give varying final EDC concentrations (0, 4, 12, and 20 mM). Reactions were incubated at room temperature for 1 h, stopped by the addition of ␤-mercaptoethanol (100 mM), and dialyzed into 5 mM Tris-HCl and 5 mM ␤-mercaptoethanol (pH 7.9). Aliquots of treated MARCKS from each reaction mixture were subjected to 12% SDS-PAGE to confirm modification by EDC as detected by an expected shift in electrophoretic mobility, as unmodified MARCKS runs anomalously as an ϳ80-kDa protein because of its unusual pI, and modification of the acidic residues shifts it toward its expected position at 30 kDa (35). Modification of the acidic (but not basic) residues was confirmed by amino acid analysis.
Calpain Digestion of MARCKS-MARCKS was digested with calpain I (75 nM) in 50 mM Tris-HCl, 100 mM NaCl, and 2 mM CaCl 2 (pH 7.9) for varying lengths of time. The reaction was stopped by addition of SDS-PAGE sample buffer. For N-terminal sequencing of calpain-digested MARCKS fragments, the digested fragments were first separated by SDS-PAGE and then transferred by electroblotting in 20% methanol and 10 mM MES (pH 6.0) to polyvinylidene difluoride membrane and stained with 0.02% Coomassie Blue R-250 in 40% methanol and 5% acetic acid for 30 s, followed by destaining (40% methanol and 5% acetic acid) and air-drying. Amino acid sequence was obtained by Edman degradation on a Procise instrument (Model 494 HT, Applied Biosystems).

Full-length Recombinant MARCKS Fails to Interact with
Actin-MARCKS did not bind to F-actin, as shown by SDS-PAGE after pelleting F-actin by ultracentrifugation (Fig. 1A). Neither a dose-dependent increase in MARCKS in the pellets nor a dose-dependent decrease in MARCKS in the supernatants was observed. There were no reproducible differences in the amount of actin in the supernatants as a function of MARCKS concentration, i.e. MARCKS had no detectable effect on the critical concentration of actin. Despite a different interpretation, these results are qualitatively similar to those reported by Hartwig et al. (5), in whose study pelleting assays were assessed by Western blotting, and the molar ratio of bound MARCKS to actin was ϳ0.01-0.02 at saturation by MARCKS. Also, Wohnsland et al. (7) reported that only 1-5% of the total amount of MARCKS-related protein was associated with F-actin upon pelleting at 100,000 ϫ g. The results reported here by us and elsewhere by others are consistent either with very low affinity (K d Ͼ 100 M) or with a very low stoichiometry of MARCKS to actin at saturation by MARCKS. Although several F-actin-binding proteins do exhibit low stoichiometry to actin, known examples such as capping proteins or proteins that use repetitive actin-binding sequences to associate with multiple subunits as a linear lattice have much different actin-binding properties than the PSD of MARCKS.
Alternative theoretical explanations can be imagined to explain the low stoichiometry of binding. A MARCKS-binding site could be buried within F-actin so that only a few sites can be simultaneously accessed, or a MARCKS-binding site could be created by two adjacent actin filaments in a filament bundle aligned so that only a few of the F-actin protomers have the correct orientation for binding. However, these explanations are not readily made consistent with binding data for the PSD peptide alone showing a 1:1 stoichiometry with F-actin subunits at saturation by the PSD (8).
MARCKS does not have actin-regulating functions previously associated with the PSD. We previously identified two potential actin-binding sites within the PSD that could explain how MARCKS can induce aggregation of actin filaments (8). In the presence of the PSD peptide, Mg 2ϩ -actin filaments pelleted at low centrifugal forces (8000 ϫ g), but the filaments remained in the supernatant despite a large excess of intact MARCKS (Fig. 1B). The results in Fig. 1B, in which the concentration of MARCKS exceeded that of actin, could possibly be explained by saturation of the actin-binding sites for MARCKS, leaving no free binding sites for MARCKS to exert a cross-linking effect. However, lower concentrations of MARCKS were also unable to produce any detectable filament bundling or cross-linking (data not shown). In the presence of 4 M MARCKS, the time course of actin filament polymerization for Mg 2ϩ -actin in 50 mM KCl and 2 mM MgCl 2 was similar to the control (Fig. 1C). The PSD peptide showed an ϳ4-fold increase in the rate of actin filament polymerization over the first 100 s at the same concentration (4 M) as intact recombinant MARCKS. The same experiment performed in the absence of KCl showed a much greater discrepancy between MARCKS and the PSD peptide, again with no effect attributable to MARCKS, but an ϳ30-fold increase in the rate of polymerization by the PSD peptide (Fig. 1C, inset). Because the effects of KCl on this particular assay in the absence of the PSD were undetectable, this result is consistent with prior reports of a significant difference in activity of the PSD in the presence and absence of KCl (6).
Binding of the PSD to CaM-Three different assays were used to investigate binding of the PSD to CaM. First, the Rh-PSD was bound to CaM-Sepharose beads, and the amount bound was determined from the observed decrease in supernatant fluorescence after pelleting the Rh-PSD bound to CaM-Sepharose beads (Fig. 2A). The experiment was done at two different concentrations of the Rh-PSD peptide (0.6 and 0.06 M), one of which was substantially higher than the K d , so as to provide a more precise estimate of the concentration of CaM oriented on the bead appropriately for binding to PSD. The x intercept of the essentially linear data for 0.6 M Rh-PSD and varying amounts of CaM-Sepharose beads in Fig. 2A indicates a theoretical point at which the concentrations of the Rh-PSD and CaM are equal, assuming a stoichiometry of 1:1. The amount of CaM-Sepharose beads required to bind 0.6 M Rh-PSD was then assumed to provide an effective concentration of 0.6 M CaM, and this result was used to calibrate all data using CaM-Sepharose beads. Binding isotherms for both sets of data are consistent with a K d of 10 nM for binding of the Rh-PSD to CaM. Second, CaM (in solution, not bead-bound) was equilibrated with the Rh-PSD peptide, and a binding curve was obtained based on the change in fluorescence that was observed upon binding of the Rh-PSD to CaM (Fig. 2B). At saturation by CaM, the maximum increase in fluorescence intensity was 2.3-fold. Assuming that the change in fluorescence intensity varied linearly with the fraction of bound Rh-PSD, then the data for two different concentrations of the Rh-PSD can be globally fit to a K d of 9 nM. As long as the stoichiometry of the interaction between the PSD and CaM is 1:1 and the PSD interacts in a site-specific manner with CaM, then this assumption is very likely correct. Published crystallographic data for the complex of the PSD with CaM provide direct support for both of these claims (12). When the unlabeled PSD peptide was added to a complex of the Rh-PSD peptide and CaM, the PSD competed directly with the Rh-PSD for binding to CaM; so at high concentrations of the PSD, the fluorescence intensity of the Rh-PSD mixed with CaM was about the same as that of the Rh-PSD alone (Fig. 2B, inset). The theoretical fit to these competitive binding data shows that the rhodamine label on the PSD does not significantly influence binding of the PSD to CaM.
The third assay was steady-state fluorescence anisotropy of the Rh-PSD, which varied from 0.06 in the free state to 0.18 in the bound state (bound to CaM). Anisotropy data were collected from the same samples used in Fig. 2B for steady-state fluorescence measurements. Because of the increment in fluorescence intensity upon binding CaM, equivalent molar quantities of CaM⅐Rh-PSD contributed 2.3 times as much to the measured value of fluorescence anisotropy as did the free Rh-PSD, necessitating correction of the data by this weighting factor (Fig. 2C). As observed with the fluorescence intensity measurements, the unlabeled PSD competed with the Rh-PSD; so at high concentrations of the unlabeled PSD, the anisotropy of a solution of the Rh-PSD plus CaM was nearly identical to that of the Rh-PSD alone (Fig. 2C, inset). With the assumption that the change in fluorescence intensity varied linearly with the fraction of bound Rh-PSD, a global fit of all the fluorescence and anisotropy data for direct and competitive binding yields K d ϭ 9 Ϯ 4 nM for the Rh-PSD and 6 Ϯ 2 nM for the unlabeled PSD (Fig. 2, A-C). These results show that the Rh-PSD and unlabeled PSD bind to the same site on CaM with similar affinity.

MARCKS Binds with Lower Affinity than the PSD Peptide to CaM, and Myristoylation of MARCKS Alters the Affinity for
CaM-Based on the CaM-Sepharose binding assay, a K d of 1.3 M was calculated for MARCKS binding to CaM in 0.1 mM CaCl 2 , compared with a K d of 10 nM for the PSD under the same conditions ( Fig. 2A). Schleiff et al. (36) have shown that maximum binding of MARCKS-related protein to CaM occurs in 0.1 mM CaCl 2 and that the affinity of the complex is decreased by 6-fold as the concentration of Ca 2ϩ is increased to 1 mM. Using the increment in fluorescence of MIANS-CaM to determine binding of MARCKS to CaM, we found a qualitatively similar dependence on calcium for recombinant MARCKS. Our data are consistent with a K d of 2.5 M for binding of MARCKS to MIANS-CaM in 0.6 mM CaCl 2 , but the measured K d dropped to 0.1 M in 0.1 mM CaCl 2 (Fig. 3). Titration of MIANS-CaM with CaCl 2 alone had an insignificant effect on fluorescence intensity, thus ruling out the possibility that the change in fluorescence upon the addition of MARCKS was caused indirectly by binding of MARCKS to Ca 2ϩ with a secondary decrease in free Ca 2ϩ . Although the assays are different, the measured values for K d are all significantly higher than those obtained for the PSD peptide at either 0.1 or 0.6 mM CaCl 2 . We believe that the CaM-Sepharose bead assay gave a lower affinity for MARCKS than the MIANS-CaM assay because the bead pull-down assay is performed under non-equilibrium conditions, and in this low affinity interaction, the MARCKS protein detached from the beads before the supernatant was recovered for electrophoresis, thus giving a falsely elevated estimate of K d . Qualitatively, upon binding CaM in B. At saturation with CaM, this increment is a factor of 2.3, so, for example, when the anisotropy change is 2.3/3.3 or 70% of the maximum anisotropy change, then one-half of the Rh-PSD is free and one-half is bound. A global fit to the fluorescence and anisotropy data with the same set of parameters yields K d ϭ 6 Ϯ 2 nM for binding of the PSD peptide and 9 Ϯ 4 nM for binding of the Rh-PSD peptide to CaM (solid lines). cps, counts/s. however, this assay uniquely provides a straightforward comparison of binding of MARCKS relative to the PSD peptide. Unfortunately, the other available assays could not be performed on both MARCKS and the PSD peptide because only the PSD peptide was covalently modified with rhodamine, and interactions between the PSD and MIANS-CaM did not alter the fluorescence intensity of MIANS-CaM. These results independently confirm that the interaction of the PSD with CaM is attenuated by the presence of the N and C termini despite the natively unfolded structure of full-length MARCKS (13).
Consistent with earlier observations that myristoylated MARCKS binds better than non-myristoylated MARCKS to a calmodulin affinity column (14) and subsequent quantification of this effect (13), myristoylated MARCKS exhibited higher affinity for MIANS-CaM by a factor of 3 in 0.6 mM CaCl 2 and by a factor of Ͼ10 in 0.1 mM CaCl 2 (Fig. 3). A simple explanation for the attenuation of binding by full-length MARCKS relative to the PSD peptide and augmentation by myristoylation is that the PSD is less accessible in full-length MARCKS due to intramolecular interactions within MARCKS that are modulated by myristoylation. Moreover, the unusual inverse relationship between CaM affinity and increasing calcium concentration over higher ranges of calcium could have a similar explanation. Others have surmised that calcium likely interacts with an acidic portion of MARCKS-related protein (36), and if calcium interacts similarly with an acidic portion of MARCKS (the N or C terminus), then calcium would not be expected to influence binding between the PSD and CaM unless the acidic termini control binding events occurring at the PSD.
Investigating Intramolecular Interactions within MARCKS: Binding of Externally Added PSD to MARCKS-Based on the hypothesis that the positively charged PSD of MARCKS associates intramolecularly with one or more sites on the negatively charged ends of MARCKS, we tested for evidence that a synthetic peptide corresponding to the PSD may associate intermolecularly with MARCKS. Such binding could be site-specific and/or specific to the PSD sequence. Also, depending on the flexibility of MARCKS and the position of binding site(s) on the primary sequence of MARCKS, any intramolecular binding between the PSD and a site on an oppositely charged terminus would be expected to have complicated equilibrium kinetics. The consequences of localization of a binding site for the PSD on a single MARCKS molecule (creating a high, effective, local concentration of ligands) would be to some degree mitigated by any structural restraints imposed by physical position and intrinsic flexibility. The resulting equilibrium association reaction for intramolecular binding therefore might be either lower or higher affinity than experimentally observed for intermolecular PSD peptide-MARCKS interactions. Moreover, the intermolecular binding events measured experimentally using a synthetic PSD peptide would occur in competition with intramolecular binding, so the intermolecular equilibrium kinetics may also be expected to be complicated. Using the Rh-PSD peptide, we observed saturable binding by MARCKS based on the changes in fluorescence amplitude and anisotropy that were consistent with association of the Rh-PSD and MARCKS (Fig. 4A). The unlabeled PSD was able to displace the Rh-PSD from MARCKS as determined by a return of the amplitude of fluorescence intensity and the anisotropy to levels measured for the free Rh-PSD peptide (Fig. 4A, inset). Qualitatively, the decline in anisotropy as a function of the unlabeled PSD is steep, suggesting that the unlabeled PSD has high affinity for MARCKS.
Because of the aforementioned potential complications, the quantitative analysis of these data is quite speculative. In our analysis, we attempted to globally fit the data for both samples in which variable amounts of MARCKS were added to the Rh-PSD peptide and samples in which variable amounts of unlabeled PSD peptide were added to a fixed mixture of MARCKS and the Rh-PSD. The straightforward assumption that the intramolecular PSD of MARCKS competes equivalently with unlabeled and labeled PSD peptides for a single binding site on MARCKS was employed. Based on this assumption, the results are consistent with the K d for binding of the PSD (labeled or unlabeled) to the binding site on MARCKS of Յ40 nM, and the apparent (effective) concentration of the intrinsic PSD was estimated as 0.3-0.8 M. The fitting algorithm for the anisotropy data included a correction for the larger fluorescence contribution of the free Rh-PSD relative to the bound Rh-PSD. The effective local concentration of the intramolecular PSD depends on the distance between the PSD and its binding site on MARCKS and on steric factors that reflect limitations in relative orientations and conformations of the MARCKS protein. Based only on geometric considerations and no steric constraints, the theoretical maximum effective concentration is ϳ50 M (this assumes a 20-nm radius). Although the data fit excellently based on the assumption of site-specific binding, we could not rule out the possibility that combinations of complicating factors, including non-site-specific binding, could explain our results equally well.
Experiments using a short oligomer of polylysine, an ϳ24mer with low polydispersity (32), to displace the rhodaminelabeled PSD peptide (44 nM) from MARCKS (0.25 M) also are strongly suggestive of site-specific binding, but also point to a lack of sequence specificity (Fig. 4A, inset). Qualitatively, it is clear from the steep, nearly stoichiometric dependence of the competitive binding isotherm on polylysine that polylysine interacts with a single site or, at most, a very few sites on MARCKS, otherwise polylysine would have to saturate its several sites and require high stoichiometry relative to MARCKS to be an effective competitor. The relative high affinity of polylysine (Յ14 nM) is likely related to the polyvalency of this ligand, as it likely can bind to its target in any of several registers. The lack of sequence specificity is presumably compensated for in vivo by the high effective local concentration of the intramolecular PSD.
MARCKS (5 M) could be successfully cross-linked to a synthetic rhodamine-labeled PSD peptide (25 M) using the zerolength cross-linking reagent EDC (Fig. 4B), indicative of close proximity of the peptide and MARCKS in solution. Also, a small fraction of EDC-modified MARCKS exhibited decreased electrophoretic mobility (Fig. 4B, lanes 2 and 3), perhaps because of an intermolecular cross-link consistent with limited self-association, as discussed below in relation to Fig. 4D. The same rhodamine-labeled peptide bound to a band on polyvinylidene difluoride membrane (after SDS-PAGE and transfer) that was at the same position as MARCKS (Fig. 4C). Not only did the peptide bind to purified recombinant protein, but it also demonstrated apparent selectivity for MARCKS, as it recognized a band at the expected position for MARCKS in whole cell extracts from RAW 264.7 macrophages. The position of the murine MARCKS band for both the recombinant protein and whole cell extract after transfer was identified by traditional Western blotting. The faint doublet seen on one Western blot is consistent with expectations for incompletely myristoylated MARCKS (14), but may also reflect other post-translational modifications. In this assay, the PSD appeared to recognize only the band that would correspond to post-translationally modified MARCKS, but because the Western blots are nonquantitative (and in this instance, demonstrate some variation even when using the same samples), this may be due either to the relative absence of unmodified MARCKS or to selectivity for modified MARCKS.
If the free PSD can bind to MARCKS, then perhaps MARCKS can also self-associate into homo-oligomers. Indeed, we found evidence of reversible MARCKS aggregation after concentration to 300 M. High speed pelleting of concentrated samples occurred with dilution from 300 to 100 M for 1 h, and possibly a small amount of material was depleted from the supernatant after dilution to 18 M, but none was depleted from the 3 M sample (Fig. 4D). After 12 h, only the 100 M sample showed any evidence of pelleting (data not shown). These results are consistent with competition between intermolecular and intramolecular binding sites for the PSD, with aggregation due to intermolecular binding only when the concentration of MARCKS molecules is much higher than the effective local concentration of the intramolecular binding site.

Effect of EDC Neutralization of Negative Charges on Interactions of MARCKS with Actin and Calmodulin-Treatment
with EDC in the presence of excess amine (400 mM ethanolamine) at low pH is expected to modify acidic (but not basic) residues and not to induce intra/intermolecular cross-links (35). We used a conservative treatment of EDC at various concentrations for only 1 h. Amino acid analysis showed that, at the highest concentration of EDC (20 mM), only an average of 6 of the 59 acidic residues were modified, whereas averages of 5 and 2 residues were modified at 12 and 4 mM EDC, respectively. Prior reports suggest that the anomalously low electrophoretic mobility of proteins with very low pI is due to a failure to become uniformly charged in SDS and that neutralization of these charges with EDC will cause such a protein to migrate as expected relative to mass (35). Although SDS-PAGE showed that the electrophoretic mobility of MARCKS generally increased with neutralization ( Fig. 5A) (experimental repeats not shown), the bandwidth also increased, presumably because of heterogeneity in the covalent modifications (Fig. 5A). High speed pelleting of F-actin (24 M) and EDC-treated MARCKS (16 M) resulted in depletion of MARCKS from the supernatant, in contrast to the results for untreated MARCKS (compare Figs. 1A and 5A). Thus, EDC modification of regions of MARCKS outside of the PSD convert MARCKS to an F-actinbinding protein.
The affinity of MARCKS for MIANS-CaM was similarly increased by EDC treatment (Fig. 5B). The data for untreated MARCKS are superimposable in Figs. 3 and 5C, but are shown on different scales. Treatment of MARCKS with EDC caused an increase in the fluorescence increment of MIANS-CaM that was associated with saturation by MARCKS, implying a greater influence of bound EDC-treated MARCKS on the fluorophore environment and a change in the shape of the binding isotherm consistent with augmented affinity. Untreated or mock-treated MARCKS gave K d ϭ 2.5 M, whereas 4, 12, and 20 mM EDC-treated MARCKS gave K d ϭ 1.01, 0.65, and 0.37 M, respectively. These data imply that incremental neutralization of negative charges on the C and N termini of MARCKS allows binding of the PSD to actin or CaM.
MARCKS treated with 20 mM EDC enhanced the rate of filament polymerization in 2.0 mM MgCl 2 , whereas MARCKS treated with 12 mM EDC had only a very small effect (Fig. 5C). These data suggest that the dose response to EDC treatment not only varies with the activity assayed, but also is not closely proportional to the extent of amino acid modification. This is not particularly surprising given that a prior report has shown that the effects of the PSD on actin vary non-monotonically with PSD concentration (37), so activation of MARCKS to Ͻ100% of the PSD activity may be expected to cause either no effect or activity opposite of that seen for the PSD alone, de- FIG. 4. Intramolecular interactions between the Rh-PSD and MARCKS. A, when MARCKS was added to the Rh-PSD (44 nM) in 50 mM KCl, 0.6 mM CaCl 2 , and 10 mM Tris-HCl (pH 7.9), the fluorescence of the Rh-PSD (not shown) decreased with increasing MARCKS concentrations, and at saturation, the fractional decrease was calculated as 0.53. Fluorescence anisotropy was measured in the same samples, with an observed increase in anisotropy consistent with a direct interaction between MARCKS and the smaller fluorescent ligand. Inset, the unlabeled PSD peptide was able to displace the Rh-PSD (44 nM) from MARCKS (1 M), consistent with site-specific binding and a 1:1 stoichiometry of MARCKS to PSD or Rh-PSD. When saturating amounts of unlabeled PSD peptide were included in the mixture of MARCKS and the Rh-PSD, the anisotropy of the Rh-PSD decreased to a level near that expected for the free Rh-PSD (compare inset with origin of the main panel), and the level of fluorescence increased, also to base-line levels (not shown). A simultaneous fit to all of the data shown using the assumption that the intramolecular PSD of MARCKS competed equivalently with the intermolecular labeled and unlabeled PSD peptides for a single binding site on MARCKS gives a K d in the range of 10 nM (solid lines), with good fits obtainable for any K d Յ 40 nM. Data consistent with site-specific binding were also obtained in the presence of increasing amounts of polylysine (OE). The apparent K d for polylysine is Յ14 nM. When one polylysine occupies that site, Rh-PSD is excluded. B, shown are the results from SDS-PAGE of untreated MARCKS (lanes 1 and 4), EDC-cross-linked MARCKS (lanes 2 and 5), and EDC-cross-linked MARCKS with the rhodamine-labeled N-terminal PSD peptide (lanes 3 and 6) visualized by Coomassie Blue staining (lanes 1-3) or by indirect UV fluorescence of the rhodamine label (lanes 4 -6). C, purified recombinant MARCKS (lanes 1, 4, 6, and 8) and two RAW 264.7 whole cell extracts (one in lanes 2, 5, 7, and 9 and the other in lanes 3 and 10) were stained with Coomassie Blue (lanes 1-3) or transferred to polyvinylidene difluoride membrane and Western-blotted for MARCKS (lanes 4 -7) or for a gel overlay using the rhodamine-labeled N-terminal PSD peptide as a probe (lanes 8 -10). The two Western blots are of identical samples transferred on different days and reveal that the major immunoreactive band in the cell extracts has slightly lower electrophoretic mobility compared with recombinant MARCKS. The gel overlay was imaged by indirect fluorescence, and the digital fluorescence image was electronically inverted. D, shown is the concentration dependence of pelleting for MARCKS after high speed centrifugation following dilution for 1 h from a 300 M stock.
pending on the precise extent of activation. Moreover, different modifications of the PSD have been shown to have different effects on specific actin-binding functions (8), so partial activation of MARCKS might not be expected to produce the full spectrum of PSD-related functions. Despite these arguments, the inhibition of actin depolymerization by EDC-treated MARCKS in 2.0 mM MgCl 2 has a very similar dose dependence compared with the effect on polymerization, with only the 20 mM sample having substantial activity (Fig. 5D). In this assay, 2.0 M MARCKS treated with 20 mM EDC had activity roughly equivalent to that of 0.2 M PSD alone. A prior study reported that the PSD slows depolymerization and attributed this effect to a lower critical concentration in the presence of the PSD (9). Alternatively, barbed end capping activity might be expected to produce the observed effect.
Modification of Full-length MARCKS by Myristoylation Enhances the Rate of Actin Filament Polymerization-A dose-dependent increase in the rate of actin filament polymerization was seen in the presence of myristoylated MARCKS in 2.0 mM MgCl 2 (Fig. 6). As discussed above, the PSD of MARCKS affects multiple parameters that alter the time course of actin polymerization, including effects on nucleation rates and filament capping and bundling. Measurements of the time course of polymerization alone do not delineate which of these variables has been modified to produce the overall variation relative to the control. Thus, it is possible that myristoylation selectively activates only some of the actin-binding properties of MARCKS. In this context, it should be noted that no evidence of bundling or significant binding of myristoylated MARCKS to F-actin was seen using the assays employed for MARCKS in Fig. 1 (data not shown). Also, there was no effect on elongation rates in a seeded polymerization assay (data not shown), and the implication is therefore that myristoylated MARCKS alters the time course of polymerization only by an effect on nucleation rates.
MARCKS-Actin Interactions May Be Regulated by Proteolysis-Inhibition of calpain has been reported to result in the accumulation of MARCKS in myogenic cells in culture (38) and rat hippocampal slices (39) and to be functionally related to myoblast migration and fusion (40,41). MARCKS was cleaved in vitro by calpain, resulting in several bands on SDS-polyacrylamide gel (Fig. 7A) by Western blotting using the goat anti-MARCKS antibody recognizing the C terminus was isolated for N-terminal sequencing. Consistent with the known epitope recognized by the antibody, the sequence SPKAEDGAA identifies this proteolytic fragment as a 17.7-kDa MARCKS C-terminal cleavage product. The N terminus of this fragment begins 17 residues before the PSD at Ser 134 , so the MARCKS cleavage site has the sequence SSTSϳSPKAE (where ϳ represents a scissile peptide bond). The primary structure of the substrate around the scissile bond has recently been confirmed as a recognition site for calpain cleavage and, interestingly, is identical to a calpain cleavage site previously identified in human protein kinase C␥ (42). Cleavage at this site leaves only 3 acidic residues N-terminal to the PSD. After digestion of MARCKS by calpain for time points up to 1 h, MARCKS increased the rate of actin polymerization, with the extent of the increase correlating with the duration of digestion (Fig. 7B). Compared with SDS-PAGE results for the same MARCKS samples used to stimulate polymerization, there was no obvious candidate that increased in abundance in a manner that correlated temporally with the effect on the rate of actin polymerization. Some proteolytic fragments may increase in concentration yet be unidentifiable because they may not stain well, so it is uncertain which, if any, of the bands on the gel identify MARCKS peptides that stimulate polymerization. In any case, these results suggest that cleavage of MARCKS by calpain may be a mechanism of activation of MARCKS that exposes the PSD by digestion and removal of the N terminus (presumably by cutting at Ser 134 ) and/or C terminus of MARCKS. DISCUSSION Initial experiments revealing the absence of actin-binding activity in recombinant MARCKS were unexpected. Yet we have documented that the DNA sequence of the MARCKS clone is correct, that the recombinant protein is recognized specifically by anti-MARCKS antibody, that it has the expected anomalous electrophoretic mobility on SDS-polyacrylamide gel, that it has the absence of significant secondary structure as determined by circular dichroism spectroscopy, that it yields distinctive correct amino acid analysis results, that it has the expected mass as determined by mass spectroscopy, and that it is not aggregated as assessed by analytical ultracentrifugation or gel filtration chromatography. All of these results support the conclusion that the recombinant protein is authentic MARCKS. Paradoxically, our results therefore imply that a natively unfolded protein has an inaccessible actin-binding site and a restricted CaM-binding site. Although our findings do not definitively provide an explanation, they clearly document the evidence confirming that such a paradox exists by demonstrating that the PSD is less available to bind to actin and CaM in the full-length protein than in a peptide that corresponds to the PSD alone. The observation that neutralization of charged residues outside of the PSD can alter PSD-CaM and PSD-actin interactions shows that the acidic termini influence binding events at the PSD and, not surprisingly, that these effects are charge-dependent. Direct binding of the PSD peptide to MARCKS shows that intramolecular interactions are likely to occur in MARCKS.
In consideration of hypotheses that could explain these results, one idea was originally postulated as an explanation for the low affinity of MARCKS for negatively charged phospholipid vesicles (11). Perhaps the electrostatic potential of acidic residues of MARCKS that are distant from the PSD reduce the effective charge of the PSD. This explanation is more believable with regard to supposedly nonspecific interactions between anionic phospholipids and the PSD and less believable for site-specific interactions between CaM (12) or actin (8) and the PSD. Even in the instance of nonspecific binding, the contribution of distant acidic residues may be relatively small compared with the free energy contribution of the multiple electrostatic interactions occurring at the PSD-phospholipid junction. An analysis of the distance and salt dependence of coulombic interactions in proteins by Lee et al. (43), directly applicable to the analysis of MARCKS-protein interactions and likely applicable to electrostatic interactions between MARCKS and phospholipids, shows that, in 0.01 M KCl, ⌬G Յ 0.2 kcal/mol for charges separated by distances of Ն15 Å and that, in 0.1 M KCl, ⌬G Յ 0.1 kcal/mol for charges separated by distances of Ն12 Å. Thus, the 5-6 acidic residues that are likely to fall into the range limits of 12-24 Å are likely to have no more impact on binding to phospholipids than do the 5 Phe residues within the PSD. Each phenylalanine individually contributes ϳ0.2 kcal/ mol, and their complete replacement by Ala decreases binding by ϳ6-fold (44), rather than the 10 4 -fold observed when comparing MARCKS relative to the PSD peptide (11). In contrast, models of PSD-phospholipid structure show coulombic interactions occurring at a range of 3-5 Å (45,46), comparable with the length of a protein-protein salt bridge. Although the individual coulombic interactions that bind MARCKS to anionic phospholipids may contribute as little as ϳ0.6 kcal/mol, localization of the multiple charges in the vicinity of the ligand will have a major contribution to the overall free energy of interaction (47), and the free energy contribution of distant acidic residues would be expected to be insignificant in comparison.
An alternative possibility supported by our present results is that the acidic termini of MARCKS interact, either specifically or nonspecifically, with the PSD so as to shield the PSD from its ligands. The free energy difference between ligand binding to MARCKS and the PSD peptide is related to the free energy required to displace the acidic chains to a distance at which electrostatic effects become negligible. If the intramolecular interaction between MARCKS and its PSD were entirely nonsite-specific, then the observed differences between binding of myristoylated and non-myristoylated MARCKS to either CaM or actin would require still another explanation, as myristoylation has no effect on charge and therefore should not have an effect on non-site-specific ionic interactions between the PSD and the acidic termini. The hypothesis that site-specific intramolecular interactions between the MARCKS termini and the PSD regulate the availability of the PSD for its ligands is attractive from the perspective that the effects of myristoylation could be similarly explained: myristoylation may sterically inhibit the intramolecular interactions. The only data directly in support of this hypothesis are from Fig. 4, providing evidence of site-specific binding of the PSD peptide to MARCKS that includes the observation that the PSD competes directly with the Rh-PSD peptide to bind to MARCKS. The salt dependence of nearly every equilibrium reaction studied here does not help to distinguish between the site-specific and non-site-specific binding. Although it is true in general that short-range electrostatic interactions are less salt-dependent than long-range interactions (43), complicated interactions with multiple charges make it impossible to conclude that any particular reaction is too salt-dependent to be a short-range interaction or insufficiently salt-dependent to be a long-range interaction.
Given the inaccessibility of the PSD, what mechanisms are available for activation? Based on the available data, myristoylation and proteolysis are two post-translational modifications that are potential candidates. The calpains are ubiquitous, intracellularly located, Ca 2ϩ -dependent, neutral cysteine proteases. Because the calpains typically function through limited proteolysis that modifies rather than terminates substrate activity (48), calpains are implicated in several basic physiological functions related to the cytoskeleton such as activation of protein kinase C at the plasma membrane (49). The one specific cleavage site identified in this work would serve to isolate the PSD from the N terminus of MARCKS (Fig. 7), preventing intramolecular interactions that might block access to the PSD. Further degradation of MARCKS is clearly indicated by these data, and it is possible that additional proteolysis is necessary for the activation of actin-binding activity observed in Fig. 7. These data provide the first evidence that calpain cleavage of MARCKS alters molecular function, and in a broader context, the idea that calpain could regulate intramolecular interactions in MARCKS could be functionally relevant for any ligand that targets the PSD.
Myristoylation already has a functional role relative to MARCKS, contributing to membrane localization as a "myristoyl-electrostatic switch" (50). Independent data that myristoylation could also regulate accessibility of the PSD are limited to the effects on CaM binding reported here and by others (14) and the observation that myristoylated MARCKS-related protein incorporates significantly more phosphate than non-myristoylated MARCKS-related protein upon in vitro phosphorylation by the catalytic subunit of protein kinase C (51). Here, we have shown that myristoylation alters the actin-binding functions of MARCKS. This and the prior observations are all consistent with the unifying concept that myristoylation increases access to the PSD.
MARCKS is potentially the target of a myriad of additional post-translational modifications, including phosphorylation by multiple kinases (52,53), proteolysis by other enzymes (54), O-glycosylation (55), ADP-ribosylation (56), and evidence of other noncovalent ligands (17). Although we have shown that it is qualitatively possible to activate the MARCKS PSD with specific post-translational modifications, thereby providing these modifications with potential functional significance, other post-translational modifications of MARCKS may be operative in vivo, either alone or in combination. Furthermore, these modifications may selectively regulate specific PSD ligands and their function.