Mammalian Nonmuscle Myosin II Binds to Anionic Phospholipids with Concomitant Dissociation of the Regulatory Light Chain*

Mammalian cells express three Class II nonmuscle myosins (NM): NM2A, NM2B, and NM2C. The three NM2s have well established essential roles in cell motility, adhesion, and cytokinesis and less well defined roles in vesicle transport and other processes that would require association of NM2s with cell membranes. Previous evidence for the mechanism of NM2-membrane association includes direct interaction of NM2s with membrane lipids and indirect interaction by association of NM2s with membrane-bound F-actin or peripheral membrane proteins. Direct binding of NM2s to phosphatidylserine-liposomes, but not to phosphatidylcholine-liposomes, has been reported, but the molecular basis of the interaction between NM2s and acidic phospholipids has not been previously investigated. We now show that filamentous, full-length NM2A, NM2B, and NM2C and monomeric, non-filamentous heavy meromyosin bind to liposomes containing one or more acidic phospholipids (phosphatidylserine, phosphatidylinositol 4,5-diphosphate, and phosphatidylinositol 3,4,5-triphosphate) but do not bind to 100% phosphatidylcholine-liposomes. Binding of NM2s to acidic liposomes occurs predominantly through interaction of the liposomes with the regulatory light chain (RLC) binding site in the myosin heavy chain with concomitant dissociation of the RLC. Phosphorylation of myosin-bound RLC by myosin light chain kinase substantially inhibits binding to liposomes of both filamentous NM2 and non-filamentous heavy meromyosin; the addition of excess unbound RLC, but not excess unbound essential light chain, competes with liposome binding. Consistent with the in vitro data, we show that endogenous and expressed NM2A associates with the plasma membrane of HeLa cells and fibrosarcoma cells independently of F-actin.

Mammalian nonmuscle cells express three Class II myosins, NM2A, 3 NM2B, and NM2C, which, like all Class II myosins (1), consist of six polypeptides: two identical heavy chains, a pair of essential light chains, and a pair of regulatory light chains (see Fig. 1). Each heavy chain has an N-terminal globular motor domain with an ATPase site and an F-actin-binding site followed by a lever arm that binds both an essential light chain and a regulatory light chain. The HC terminates in a long ␣-helical tail with a short nonhelical tailpiece (1,2). The helical tails of the two HCs form a stable coiled-coil (2,3) creating the six-polypeptide molecule, hereafter referred to as the monomer. Monomers assemble via their helical tails forming short (ϳ300 nm) antiparallel, bipolar filaments consisting of a mean value of 29, 30, and 14 molecules for NM2A, NM2B, and NM2C, respectively (3). The coiled-coil helical tail of NM2 can be cleaved ( Fig.  1), producing heavy meromyosin (HMM), a non-polymerizable monomer with actin-activated ATPase activity, and light meromyosin (LMM), which can polymerize but differently than full-length NM2.
Interaction of actin filaments with the motor domains of myosin filaments and HMM activates myosin's ATPase activity. The energy produced by ATP hydrolysis is converted into mechanical force by coupled lever movement of the light chain binding regions of the HCs, which allows the motor domains both to translocate actin filaments and move along actin filaments. As described originally for smooth muscle myosin (4 -7), actin-activated ATPase activity of NM2s is stimulated by phosphorylation of Thr-18 and/or Ser-19 of the RLC (8 -11).
NM2s have well documented essential roles in migration, adhesion, and cytokinesis of nonmuscle cells (12,13). The functions of NM2s in vesicle transport are less well established, although there is evidence that NM2s are the motors for vesicle transport along actin filaments in clam oocytes (14) and for the involvement of NM2s in the formation and transport of Golgi vesicles (15)(16)(17)(18)(19)(20), squid axoplasmic organelles (21), brain neuronal membranes (22,23), and epithelial membranes (24,25). However, it is not known whether the associations of NM2s and membranes are mediated by the interaction of NM2s with membrane-associated actin and/or peripheral membrane proteins or if NM2s bind directly to the lipid bilayer. For example, there are reports that binding of NM2 to membranes of mouse forebrains requires interaction of the motor domain of NM2 with F-actin (23), that the tail (and not the motor domain) of single NM2 molecules may bind to the peripheral membrane proteins of Golgi stacks (19) and lytic granules from lymphocytes (26), and, to the contrary, that the association of NM2 with plasma membranes is independent of F-actin and other proteins that can be extracted by 0.1 M Na 2 CO 3 (27) or 0.6 M KI (22).
NM2A and two other proteins from lysates of human breast cancer cells have been shown to bind to PIP 3 -coated beads in vitro, and NM2A has been shown to co-localize with PIP 3 in the plasma membrane of breast cancer cells in situ (28). However, direct interaction between NM2A and PIP 3 was not shown in those experiments. The possibility of direct binding of NM2 to membrane anionic lipids is suggested, however, by the observation (22) that purified bovine brain NM2, but neither skeletal nor cardiac muscle myosin II, binds to 100% phosphatidylserine liposomes but not to 100% phosphatidylcholine liposomes. The molecular basis of the binding of NM2 to the PS-liposomes was not explored in those experiments.
In this paper we report that recombinant NM2A, NM2B, and NM2C bind to negatively charged liposomes containing either PS, PIP 2 , or PIP 3 but not to liposomes composed only of PC, that liposome binding depends substantially, but perhaps not exclusively, the RLC-binding site of the HC with concomitant dissociation of the bound RLC, and that NM2A localizes at the plasma membrane of HeLa cells and HT1080 fibrosarcoma cells even when F-actin is depolymerized by latrunculin A.

Full-length NM2 Filaments Bind to Phosphatidylserine
Liposomes-Filaments of full-length NM2A, NM2B, and NM2C were incubated with and without liposomes consisting of 100% PS or 100% PC. The mixtures were centrifuged at 17,000 ϫ g for 20 min (see "Experimental Procedures"), and the amounts of NM2 HC in the supernatants and pellets were determined by SDS-PAGE. None of the three NM2s pelleted when incubated with 100% PC-liposomes; the small amount of HC in the pellet was the same as in the absence of liposomes ( Fig. 2A). However, all three myosins pelleted when incubated with 100% PS-liposomes ( Fig. 2A) with a maximum of ϳ80% of the 100 nM NM2s binding to 60 -125 M PS-liposomes (Fig.  2B).
Notably, binding of the NM2s to PS-liposomes was accompanied by substantial displacement of the RLC from the myosin HC ( Fig. 2A). The ratios of RLC:HC for NM2s bound to PSliposomes in the centrifugal pellet were 0.4 -0.5:1 (Fig. 2C), and the RLC:HC ratios for the unbound proteins in the centrifugal supernatant were 4 -5.5:1 (Fig. 2C), demonstrating the presence of excess free RLC in the supernatant. The ELC:HC ratios remained 1:1 in both the supernatant and pellet. These results suggested that full-length filamentous NM2s bind to PS-liposomes via the RLC-binding site on the HC with the liposomes displacing the RLC but not the ELC. Electron micrographs of mixtures of NM2s and PS-liposomes before centrifugation (  (3), not all of the myosin molecules in the filaments that pelleted with the liposomes were necessarily individually bound to the liposomes. More quantitative data were obtained from the binding to liposomes of monomeric HMM, which does not form filaments (Fig. 4).
As observed for filaments of full-length myosin (Fig. 2), SDS-PAGE analysis showed no detectable binding of NM2A-HMM to 100% PC-liposomes but substantial NM2A-HMM bound to 100% PS-liposomes (Fig. 4A). The decrease in the RLC:HC ratio to 0.29:1 in the liposome-bound HMM (Fig. 4B, HMMϩPS) indicated that both RLCs were displaced from ϳ40% of the HMMs. RLC was not displaced when NM2A was bound to, and sedimented with, F-actin ( Fig. 4C) showing that displacement of RLC was specific for liposome binding.
Binding of HMM to PS-liposomes increased with both the concentration of liposomes and the percentage of PS in the liposomes (Fig. 5A). As with filaments of full-length NM2A (Fig. 2B), maximum binding of NM2A-HMM occurred at about 120 M concentrations of PS-liposomes (Fig. 5A). In contrast to HMM, NM2A-LMM ( Fig. 1) bound weakly to 100% PS-liposomes, a maximum of ϳ20% for LMM compared with ϳ80% for HMM, as analyzed by the sedimentation assay (Fig. 5B). No, or very little, binding of LMM was observed by electron microscopy (Fig. 5C).
The affinities of NM2A-HMM for PS-, PIP 2 -, and PIP 3 -liposomes were proportional to the net negative charges of the phospholipids, assuming that PIP 2 has four times and PIP 3 six times the negative charge of PS (Fig. 6, A and B). Similarly, NM2A bound to liposomes with the lipid composition of plasma membranes (29,30) in proportion to their net negative charge (Fig. 7A) and with displacement of the RLC (Fig. 7, B and C).
Deletion, Addition, and Phosphorylation of RLC Affect Binding of NM2A-HMM to Liposomes-We found that binding of NM2A-HMM to PS-liposomes was enhanced ϳ20% when HMM was expressed without the RLC (Fig. 8, A and C). The addition of excess free RLC inhibited binding of HMM to lipo-FIGURE 1. Schematic representation of nonmuscle myosin II. NM2 consists of two heavy chains and two pairs of light chains. The two heavy chains have an N-terminal globular head, a helical neck followed by a coiled-coil helical tail, and a C-terminal non-helical tailpiece. The ELCs and the RLCs bind to the neck domains of the heavy chains. NM2 can be proteolytically cleaved into HMM, the two heads and ϳ 1 ⁄3 of the tail domain, and LMM, the C-terminal ϳ 2 ⁄3 of the tail domain.
somes ϳ45%, but the addition of excess free ELC had no effect on HMM-binding to PS-liposomes (Fig. 8, B and C). As neither free RLC nor free ELC bound to PS-liposomes (Fig. 8D), these results are consistent with liposomes competing with RLC for binding to the RLC-binding site on the HC. Phosphorylation of the RLC bound to HMM and to full-length NM2A (Fig. 9A) inhibited binding of both HMM and NM2A to PS-liposomes ( Fig. 9B), indicative of pRLC having a higher affinity than RLC for the RLC-binding site, which is consistent with the basic charge of the RLC-binding site (see "Discussion"). Deletion of the RLC-binding site (IQ2 motif) of NM2A-HMM (HMM⌬IQ2) reduced by ϳ60% (Fig. 9, C and D) but did not eliminate binding of HMM to PS-liposomes (see "Discussion").
NM2A Binds to the Plasma Membrane in Situ-Immunofluorescence microscopy of formaldehyde-fixed cells showed that endogenous NM2A localizes with F-actin in stress fibers and at the plasma membrane of HeLa cells (Fig. 10A). Plasma membrane association of NM2A was independent of actin, however, as the myosin remained associated with the plasma membrane, including on blebs, when actin filaments were depolymerized by latrunculin A and actin was aggregated in the cytoplasm (Fig. 10B). Moreover, expressed NM2A-Tomato also associated with both the plasma membrane and cytoplasmic stress fibers when expressed in live HeLa cells (Fig.  10C, 0 s) and concentrated at the plasma membrane when the actin was depolymerized by latrunculin A (Fig. 10C, 1200  s). Furthermore, as shown in sequential images of live cells (Fig. 11), expressed NM2A localized at the plasma membrane of HT1080 cells, especially after the cells were exposed to latrunculin A for 1000 s. PIP 2 was also present in the plasma membrane of these cells (Fig. 11), in addition to its expected occurrence (31) in endosomal vesicles.

Discussion
As outlined in the Introduction, there is considerable evidence that NM2s, in addition to their association with the actin cytoskeleton, are bound to diverse membranes in many cell types. Also, full-length NM2s (22,32) and the C-terminal half of LMM (32,33) have been shown to bind to acidic liposomes, but there is no evidence for the molecular basis of these interactions. The data in this paper show that both filamentous fulllength recombinant NM2A, NM2B, and NM2C and monomeric recombinant HMM bind to liposomes containing acidic phospholipids (PS, PIP 2 , or PIP 3 ) principally, if not exclusively, through the regulatory light chain binding site of the heavy chain with concomitant displacement of the RLC. Also, endogenous and expressed NM2A remained associated with the plasma membrane of live HeLa and HT1080 fibrosarcoma cells in which actin filaments were removed by latrunculin A. The relatively weak binding to PS-liposomes of recombinant HMM minus the RLC-binding site and of recombinant LMM may result from structural differences between these molecules and native myosin. For example, LMM associates into paracrystals with a very different structure than bipolar myosin filaments.
The RLC-binding sites (IQ2 domains) of NMIIA (AYL-KLRNWQWWRLFTKV), NMIIB (AYLKLRHWQWWRV-FTKV), and NMIIC (AYLKLRNWQWWRLFIKV) have a high percentage of basic amino acids (ϳ24%) and hydrophobic amino acids (ϳ55%). Similar basic/hydrophobic sites in the non-helical tails of Acanthamoeba and Dictyostelium class I myosins have been shown to be the sites by which these myosins bind to acidic phospholipid vesicles in vitro and to PIP 2 / PIP 3 -enriched regions of the plasma membrane in vivo (34,35). Also, Ca 2ϩ -stimulated binding of chicken brush border myosin I to phosphatidylserine vesicles occurs with concomitant dissociation of calmodulin from sites analogous to the light chain binding sites of NM2s (36), and Ca 2ϩ -enhanced binding of mammalian myosin IC to acidic phospholipids (37) occurs at a Ca 2ϩ concentration at which calmodulin dissociates from the HC (38). These in vitro observations are directly relevant to in vivo studies showing that myosin IC interacts with hair cell receptors via its calmodulin binding domains (39) and that PIP 2 is required for hair cell activities that involve myosin IC (40).
The present paper is the first in which observations similar to those summarized above for class I myosins have been made for class II myosins. Breckenridge et al. (41) did observe, however, that contrary to endogenous NM2A, expressed NM2A with the  RLC-binding site deleted did not relocalize from the cytoplasm to the cell periphery during spreading of cultured human cervical cancer cells. This observation was interpreted as demonstrating that HC-bound RLC is required for NM2A to form a folded monomer that could diffuse through the cytoplasm to the periphery. However, it is possible that the RLC-binding site was required in those experiments, not for binding of RLC to NM2A but for binding of NM2A to the plasma membrane.
Although an NM2 monomer bound to membranes by its RLC-binding site would probably not be accessible to F-actin, a membrane-bound monomer might initiate polymerization of myosin filaments, or myosin filaments could bind directly to membranes as they do to liposomes. Membrane-bound filamentous myosin would interact with F-actin, and if the myosin filaments contained RLC-phosphorylated subunits (which is likely because phosphorylated RLC is more resistant than unphosphorylated RLC to displacement by lipids) the membrane-bound myosin filaments would have actin-activated ATPase activity. Whether ATP hydrolysis by membrane-bound actomyosin could generate a useful powerstroke requires further investigation.

Experimental Procedures
Cloning of DNAs-Full-length cDNAs of the HCs of NM2A (Homo sapiens myosin heavy chain 9), NM2B (H. sapiens myo-  sin heavy chain 10 transcript variant 2), and mouse NM2C (Mus musculus myosin heavy chain 14 transcript variant 1) were cloned onto pFastBac1, the Bac-to-Bac plasmid for expression in SF-9 cells (Invitrogen). A FLAG tag (DYKDDDDK) was added to the N termini of the HCs for purification of the myosins by affinity chromatography. The cDNAs for the HCs of NM2A-HMM (terminating at Leu-1355), NM2A-LMM (beginning at Met-1565), and deletion of the RLC-binding site (Gln-804 -Val-834) from HMM (HMM⌬IQ2) were prepared according to standard molecular biology protocols. For expression in SF-9 cells, the cDNAs of human NM2 RLC and ELC were synthesized and cloned onto pFastBac1. For their expression in Escherichia coli, both light chain cDNAs with a FLAGtag sequence at the N terminus were cloned onto pETDuet-1. All sequences obtained through PCR were confirmed by DNAsequencing by GENEWIZ (South Plainfield, NJ).
Expression and Purification of Proteins-Full-length NM2A, NM2B, NM2C, and HMM heavy chain viruses were individually co-expressed with RLC and ELC viruses in SF-9 cells. The recombinant wild-type and mutant NM2s and recombinant HMMs were purified from cell lysates by affinity chromatogra-  phy on anti-FLAG-resin (Sigma) and eluted with FLAG peptide, as described (42). LMM, ELC, and RLC with N-terminal FLAG tags were individually expressed in E. coli. Bacteria were lysed by the same procedure as for SF-9 cells (42). Purified proteins were dialyzed against buffer (20 mM Tris, pH 7.0, 600 mM NaCl) to remove the FLAG peptides used to elute the proteins from the affinity columns, and the proteins were stored in liquid nitrogen until use.
Preparation of Unilamellar Phospholipid Liposomes-All phospholipids were from Avanti Polar Lipids: dioleoyl-PC (catalog #850375C), dioleoyl-PS (catalog #840035C), dioleoyl-PE (catalog #850725C), dioleoyl-PI (catalog #850149P0), TopFluor PC (catalog #810281C), brain PIP 2 (catalog #840046x), brain PIP 3 (catalog #850156x), and cholesterol (catalog #700000P). Phospholipid liposomes were prepared as described (32) with modifications. Phospholipids in chloroform and/or methanol were mixed in appropriate proportions and dried sequentially for 5 min each under gentle and strong streams of N 2 , 30 min under vacuum, and 30 min under N 2 . Dried lipids were resus-pended in 10 mM imidazole, pH 7.0, and 180 mM sucrose to a final concentration of total lipid of 5 mM. The mixture was vortexed for 1 min and sonicated in a bath sonicator 5 times for 5 min each. Vesicles were frozen and thawed in liquid N 2 five times and then manually passed 30 times through an Avanti Mini-Extruder fitted with two 100-nm membranes. The final liposomes contained either 100% PC, 100% PS, or PC with varied percentages of PS, PE, PI, PIP 2 , PIP 3 , and cholesterol. The pH of the final liposome preparations varied between 6.9 and 7.2 with no correlation with their lipid compositions. All liposomes contained 0.25% fluorescent TopFluor PC, and liposome concentrations were determined by fluorescence measurements in a LS55 luminescence spectrometer with an excitation wavelength of 495 nm, an emission wavelength of 503 nm, and a slit width of 4 nm.
Phosphorylation of Myosins-Phosphorylation was done as described (3) in a reaction mixture containing 10 mM MOPS, pH 7.0, 150 mM NaCl, 2 mM MgCl 2 , 1 mM ATP, 0.2 mM CaCl 2 , 1 mM DTT, 100 nM calmodulin, and ϳ10 nM myosin light chain kinase for 1 h at room temperature. Binding Assays-Myosins (100 -200 nM) and different concentrations of liposomes (0 -125 M) were mixed together in 20 mM imidazole, pH 7.0, 150 mM NaCl, and 1 mM EGTA containing 0.5 mg/ml BSA to block nonspecific binding of myosins to liposomes. The mixture, 100 l, was incubated at room temperature for 1 h. To separate liposome-bound myosin from unbound myosin, mixtures of liposomes and filaments of fulllength myosins or LMM were centrifuged in a Beckman TL-100 for 20 min at 13,000 ϫ g. Under these conditions, liposomes with either bound myosin or bound LMM pellet but neither myosin filaments, LMM, nor liposomes alone, did. To separate liposome-bound HMM from unbound non-filamentous HMM, mixtures were centrifuged for 30 min at 189,000 ϫ g, conditions under which liposomes and liposome bound the HMM pellet but not unbound HMM. For NM2A binding to F-actin, 200 nM NM2A was incubated with 1 M muscle F-actin for 1 h at room temperature, and actin-bound myosin separated from unbound myosin by centrifugation for 20 min at 13,000 ϫ g. After centrifugation, 100-l aliquots of the supernatants were added to 15 l of 2ϫ SDS sample buffer, and the pellets were suspended in 115 l of the SDS buffer with the same composition. After SDS-PAGE, the bound (P) and unbound (S) proteins were quantified by Odyssey infrared imaging (LI-COR Biosciences, Lincoln, NE) of the protein bands.
Preparation of Actin, Calmodulin, and Myosin Light Chain Kinase-Actin was purified from rabbit muscle acetone powder (PelFreez, Rogers, AZ) as described by Spudich and Watt (43). Rat calmodulin was expressed and purified from E. coli. FLAGtagged rabbit smooth muscle myosin light chain kinase (NP_001075775) was expressed using Baculovirus and purified from SF-9 cells.
Electrophoresis-SDS-PAGE was performed by standard procedures on 10% NUPAGE gels (Invitrogen). Urea-glycerol PAGE was performed by the method described by Ruppel et al. (44). Gels were stained by Instant Blue (Expedeon, San Diego, CA).
For immunolocalization, HeLa cells were seeded on coverslips for 24 h and then treated at 37°C with either 4 M latrunculin A, to depolymerize F-actin, or DMSO for 15 min before fixation with 2% paraformaldehyde in PBS at room temperature for 10 min. After fixation, cells were washed with PBS supplemented with 10% FBS for 1 h. Cells were then incubated with primary antibodies diluted in PBS/FBS containing 0.2% saponin for 1 h. Cells were washed 3 times in PBS and incubated with the appropriate secondary antibodies in PBS/FBS containing 0.2% saponin for 1 h. Cells were then washed 3 times in PBS and mounted on glass slides. Rabbit polyclonal NM2A-specific antibody (Sigma, M8064) was diluted 1:2000, and anti-PIP 2 mouse monoclonal IgG antibody (Abcam, Cambridge, MA), which also recognizes PIP 3 , was diluted 1:100. Fluorescently labeled secondary goat anti-rabbit and goat anti-mouse IgG antibodies   Live-cell imaging experiments were carried out 18 -24 h after transfection by transient electroporation of NM2A-mTomato and PLC␦1-PH-GFP in X-tremeGENE by Amaxa Nucleofection kit according to the manufacturer's instructions (Lonza, Allendale, NJ). Cells were treated with 4 M latrunculin A for 15 min to depolymerize F-actin, and images were acquired every 8 s for 30 min with an LSM-780 or LSM-880 laser scanning confocal microscope (Carl Zeiss) equipped with a PLAN apo63ϫ oil objective (NA ϭ 1.4). Pixel dimensions were 0.081 m ϫ 0.081 m (Fig. 10C), and 0.87 m ϫ 0.87 m (Fig. 11).
Negative Staining Electron Microscopy of Proteins Bound to Liposomes-Proteins and liposomes were diluted in buffer with final concentrations of 150 mM NaCl, 10 mM MOPS, pH 7.0, 0.1 mM EGTA, 1 mM MgCl 2 , and either 100 nM myosin or 125 M liposomes. In the case of mixtures, myosins and liposomes were first diluted separately to concentrations of 200 nM each. Equal volumes of protein and liposomes were then mixed and vortexed to avoid excessive aggregation, and 3-l samples were applied immediately to UV-treated carbon-coated copper grids and negatively stained with 1% uranyl acetate. Images were recorded on an AMT XR-60 CCD camera using a JEOL 1200EX II microscope operating at 80 kV. Catalase crystals were used as a size calibration standard.