Membrane Targeting by C1 and C2 Domains*

Many peripheral proteins involved in cell signaling translocate to different cell membranes in response to specific cell stimuli. Because cellular functions and regulation of these proteins depend on their specific subcellular localization (1), understanding the mechanisms of membrane targeting is of great importance. The membrane targeting of diverse peripheral proteins is mediated by a limited number of membrane-targeting domains, including pro- tein kinase C (PKC) 1 conserved 1 (C1), PKC conserved 2 (C2), and pleckstrin homology domains. Recent structural and functional studies of individual membrane targeting domains as well as the peripheral proteins harboring these domains have provided new insights into the molecular mechanisms underlying the specific subcellular targeting and activation of peripheral proteins. This review summarizes the recent progress in our understanding of the mechanisms of C1 and C2 domain-mediated membrane targeting, with an emphasis on the correlation between the membrane binding properties of the C1 and C2 domains and the peripheral pro- teins containing these domains and their subcellular targeting behaviors. There are several excellent reviews (2–6) that contain more exhaustive surveys on the membrane targeting domains.


Membrane Binding of C1 Domains
Structural (12) and mutation (14) studies of PKC␦-C1B have defined the phorbol ester/DAG binding pocket. The polar binding pocket is located at the tip of the molecule and is surrounded by aliphatic and aromatic residues, which are adjoined by a ring of cationic residues in the middle part of the molecule (Fig. 2). A NMR study of PKC␥-C1B (15) and a monolayer penetration study of PKC␣ (16) showed that the hydrophobic residues of the C1 domain penetrate the membrane for DAG/phorbol ester binding. The phorbol ester binding seals the polar surface in the binding pocket and thereby generates a contiguous hydrophobic surface (12), which in turn greatly enhances the stability of the C1-membrane complex (17,18). Further mutational studies of PKC␣ showed that clustered cationic residues in the C1A domain are involved in nonspecific electrostatic interactions with anionic phospholipids (19). In agreement with this finding, the isolated C1 domain repeat (i.e. C1A ϩ C1B) of PKC␣ exhibits little head group specificity among anionic phospholipids while discriminating against PC (16). Together these studies on PKC C1 domains have led to a model for C1 domain-membrane interactions illustrated in Fig. 2. In this model, cationic residues accelerate the initial adsorption of the C1 domain to the anionic membrane surfaces and properly position the C1 domain at the membrane surface. Then, the hydrophobic tip of the domain penetrates the membrane to bind DAG that is partially buried in the membrane because of its hydrophobic nature. Because C1 domains with the exposed hydrophobic patch are subject to protein aggregation in solution, a tightly controlled triggering mechanism would be necessary for the C1 domain-mediated membrane targeting. In the case of conventional PKCs, it has been shown that C1 domains are buried at the resting state and become accessible to DAG or phorbol esters only after Ca 2ϩ -dependent membrane binding (19,20). The C1 domains of PKD/PKC (21,22), Ras-GRP (23), and ␤ 2 -chimaerin (24) have also been shown to drive the cellular membrane targeting of the corresponding peripheral proteins in response to DAG and phorbol esters. Because no detailed analysis of their interactions with membranes has been reported, it is not known whether the mechanisms of C1 domain-mediated membrane binding of these peripheral proteins are similar to that of PKCs.

Differential Roles of Multiple C1 Domains in PKC
Conventional and novel PKCs contain two copies of C1 domains. Earlier studies on conventional PKC reported the one-to-one stoichiometry of PKC-DAG (25) and PKC-phorbol ester binding (26,27) domains (28). The only exception was PKC␥, the C1A and C1B domains of which show comparably high affinities. For PKC␦, good correlation was observed between the intrinsic phorbol ester affinities of C1A and C1B domains and their relative importance in phorbol ester-induced activation (29), supporting the notion that the disparate roles of C1 domains mainly derive from their different intrinsic affinities for phorbol esters. More recently, however, it was shown that C1A and C1B domains of PKC␣ play an equivalent role in cellular membrane translocation in response to phorbol 12-myristate 13-acetate (PMA) despite their distinct phorbol ester affinities (30). To date, correlation between the intrinsic DAG affinities of C1 domains and their relative importance in DAG-dependent PKC activation has not been documented. A series of spectroscopic vesicle binding studies of PKC␣ indicated that PKC␣ contains two phorbol ester binding sites with high and low affinities and that DAG and phorbol esters bind to the two sites with opposite affinity (31,32). The notion that DAG and phorbol esters might have different C1A versus C1B selectivity was further supported by the finding that the C1A domain is exclusively involved in the DAG-induced vesicle binding and activation of PKC␣ (16). Clearly, more studies are needed to fully understand these complex interactions of the C1 domains of PKC with their ligands.

Subcellular Targeting by C1 Domains
The subcellular targeting of isolated C1 domains and PKC holoenzymes in response to DAG and phorbol esters has been measured in various mammalian cells transfected with green fluorescent protein (GFP)-tagged proteins (20,(33)(34)(35)(36)(37)(38)(39). PKC␥ translocated to plasma membrane in response to PMA in COS-7 cells (33). In RBL cells, PKC␥ and its C1A, C1B, and C1A-C1B domains all showed the translocation to plasma membrane in response to PMA or DAG (35). As is the case with PKC␥, PKC␤ II (40,41), PKC␦ (34), PKC␣ (39), and PKC␣-C1A-C1B (38) were shown to translocate to plasma membrane in response to PMA or DAG. In the case of PKC holoenzymes, Ca 2ϩ enhanced the rate of translocation of proteins but did not affect their localization. Later studies showed that whereas PMA and other hydrophobic phorbol esters induced the initial translocation of PKC␦ to the plasma membrane, less hydrophobic DAG and PDB caused PKC␦ to translocate to the perinuclear region (36,37). Our cell study 2 of GFP-PKC␣ with a fluorescent phorbol ester analog, sapintoxin D, supported the notion that the differential subcellular localization of C1 ligands is a main determinant of the differential subcellular targeting of PKCs. This in turn suggests that the in vitro membrane binding and the subcellular targeting of C1 domain are driven by the same forces. Binding to neither adapter proteins nor cytoskeleton appears to contribute significantly to the subcellular targeting of PKC-C1 domains because the putative adapter binding site is not located in the C1 domains (42) and the cytoskeleton inhibitors show little effect on C1 domain translocation (35,39).

Structure, Function, and Occurrence of C2 Domains
The C2 domain (ϳ130 residues) was first discovered as the Ca 2ϩ -binding site in conventional PKCs (15). A great number of proteins containing the C2 domain have been identified since, and most of them are involved in signal transduction (e.g., PKC, cyto-solic PLA 2 (cPLA 2 ), phospholipases C, plant phospholipase D, and phosphatidylinositol 3-kinase) or membrane trafficking (e.g., synaptotagmins, rabphilin-3A, and Unc-13) (2, 3). Structural analyses of multiple C2 domains have indicated that all C2 domains share a common fold of eight-stranded antiparallel ␤-sandwich connected by variable loops, with the Ca 2ϩ -binding sites located at one side of the domain (Fig. 2) (43)(44)(45)(46)(47). When compared with C1 domains, C2 domains show a larger degree of variation in amino acid sequence, particularly in the loop regions (2,3). Consistent with this finding, C2 domains show greater functional diversities. Most Ca 2ϩ -dependent membrane-binding C2 domains prefer anionic membranes to zwitterionic ones; however, cPLA 2 -C2 strongly favors PC membranes (48). Among anionic lipid-selective C2 domains, PKC␣-C2 (16) and PLC␦ 1 -C2 (49) exhibit PS selectivity. Furthermore, there are many non-Ca 2ϩ -binding C2 domains; some of them, such as PTEN-C2 (50), still bind the membrane, and others might be involved in protein-protein interactions (2,3). This review will mainly deal with the C2 domains that bind phospholipids in a Ca 2ϩ -dependent manner.

Ca 2؉ -dependent Membrane Binding of C2 Domains
The Ca 2ϩ -binding sites of C2 domains are composed of three variable loops that contain ligands for multiple Ca 2ϩ ions. Structural (43)(44)(45)(46)(47) and binding analyses (51)(52)(53) have determined the calcium binding stoichiometry, geometry, and affinity for several C2 domains. Two major roles of Ca 2ϩ ions in the membrane targeting of the C2 domain have been experimentally demonstrated. The first role of Ca 2ϩ ions is to provide a bridge between the C2 domain and anionic phospholipids. This Ca 2ϩ bridge model is supported by an x-ray structure of the PKC␣-C2-(Ca 2ϩ ) 2 -PS complex (46), showing that a short-chain PS molecule is specifically coordinated to a Ca 2ϩ ion and other residues in the Ca 2ϩ -binding loops. This structure also accounts for the PS selectivity of PKC␣-C2 (16). The second role of Ca 2ϩ ions is to induce intra-or interdomain conformational changes, which in turn trigger membrane-protein interactions. Despite earlier negative reports (43), accumulating evidence has supported the occurrence of Ca 2ϩ -induced conformational changes in the C2 domain (44,(52)(53)(54)(55). Recent structurefunction studies of the C2 domains of PKC␣ (18) and cPLA 2 (56), both of which bind two Ca 2ϩ ions, showed that two Ca 2ϩ ions play distinct roles, with one primarily involved in inducing the conformational changes and the other in Ca 2ϩ bridging. The main difference is that Ca 2ϩ -induced conformational changes are critical for the membrane binding of cPLA 2 -C2, whereas Ca 2ϩ bridging is a relatively more important role of PKC␣-C2.
Non-Ca 2ϩ -coordinating protein residues in the Ca 2ϩ -binding loops also play important roles in the membrane binding and phospholipid selectivity of the C2 domain. A large degree of structural variations have been found in the Ca 2ϩ -binding loops of C2 domains in terms of both primary and tertiary structures (43)(44)(45)(46)(47). Mutational (56,57) and labeling (58,59) studies have identified the residues in the Ca 2ϩ -binding loops that play a key role in membrane binding. In general, cationic residues on the surface of Ca 2ϩbinding loops are important for anionic lipid-selective C2 domains (57,60), whereas aliphatic and aromatic residues are essential for PC-selective cPLA 2 -C2 (56,58,59). A predominant cationic cluster in the ␤-sandwich region has been implicated in inositol polyphosphate binding of synaptotagmin II (61) but does not significantly affect the membrane binding of conventional PKCs (62) and cPLA 2 (56). Together, these in vitro studies have indicated that anionic lipid-selective C2 domains and cPLA 2 -C2 have distinct membrane binding modes. As shown in Fig. 2, PKC␣-C2 binds to the membrane in an orientation that optimizes its electrostatic interactions with the anionic membrane (46,57,60), whereas cPLA 2 -C2 binds to the membrane in an orientation that optimizes the membrane penetration of its hydrophobic residues (56,58,59,63). Interestingly, membrane binding properties of cPLA 2 and its C2 domain are similar (56,64), whereas those of PKC␣ and its C2 domain have noticeable differences (16,18). This discrepancy implies that the relative contribution of a C2 domain to the membrane binding of a peripheral protein depends on the structural context of the protein, particularly on the presence of other membrane targeting domains in the same molecule. 2 J. Rafter and W. Cho, manuscript in preparation.

FIG. 1. Different types of membrane-protein interactions.
Interactions of peripheral proteins with bulk lipids include coulombic interactions between cationic residues and anionic lipids (red) (a), interactions between aromatic residues (green) and zwitterionic lipid (blue) (b), and hydrophobic interactions involving aliphatic residues (and Phe) that are, in general, exposed by conformational changes of protein (c). d, peripheral proteins can also interact with a lipid second messenger (magenta) such as DAG.

Subcellular Targeting by C2 Domains
The subcellular targeting of GFP-tagged C2 domains and C2containing peripheral proteins, most notably conventional PKCs and cPLA 2 , has been studied in different mammalian cells. In general, the subcellular localization behaviors of C2 domains are consistent with their in vitro membrane binding properties. For instance, C2 domains of conventional PKCs that prefer anionic phospholipids rapidly translocate to plasma membrane (20) and PC-selective cPLA 2 -C2 to the perinuclear region in response to Ca 2ϩ import (65,66). Also this subcellular localization pattern of isolated C2 domains correlates with that of peripheral proteins harboring the C2 domains: i.e. conventional PKCs translocate to the plasma membrane (33)(34)(35)39) whereas cPLA 2 moves to the perinuclear region (67). Furthermore, chimera proteins of PKC␣ and cPLA 2 containing the C2 domains of cPLA 2 and PKC␣, respectively, are localized to the perinuclear region and plasma membrane, respectively. 3 Thus, it is expected that anionic lipid-selective C2 domains (and C2-bearing proteins) will translocate to plasma membrane and that PC-selective C2 domains (and C2-bearing proteins) will translocate to the perinuclear region. When conventional PKCs are activated by both C1 ligand and Ca 2ϩ ionophore, however, their subcellular localization is governed primarily by the subcellular location of C1 ligands although the C2 domain also contributes to the kinetics and energetics of translocation (20). Our recent study 3 of PKC␣-C2, cPLA 2 -C2, and their mutants demonstrated quantitative correlation between the subcellular targeting and in vitro membrane binding: i.e. PKC␣-C2 rapidly but transiently translocates to plasma membrane, but cPLA 2 -C2 slowly but irreversibly translocates to the perinuclear region in response to Ca 2ϩ import. Also subcellular translocation kinetics of mutants exhibited good correlation with their altered in vitro membrane binding kinetics. Thus, it appears that the subcellular targeting of these C2 domains is primarily driven by forces that govern their membrane binding.

Membrane Targeting by C2-like Domains
Some Ca 2ϩ -dependent membrane-binding proteins, including the amino-terminal ␤-barrel domain of 5-lipoxygenase (68) and the domain III of calpains (69,70), have C2 domain-like structures. In particular, the C2-like domain of 5-lipoxygenase has been shown to have specific Ca 2ϩ ligands (68) and to be responsible for the Ca 2ϩdependent nuclear translocation of 5-lipoxygenase (71). The C2like domain of 5-lipoxygenase has been shown to have PC selectivity because of the presence of aromatic residues in the putative Ca 2ϩ -binding loops. 4

Interactions of C1 and C2 Domains
The majority of peripheral proteins have two (or more) membrane targeting domains although the presence of extra domains is not absolutely required for their membrane translocation. This suggests that their membrane targeting and activation might involve a synergistic and/or regulatory interplay of the membrane targeting domains. A synergistic action of C1 and C2 domains in the prolonged membrane localization of PKC was demonstrated for PKC␥ (20) and PKC␤ II (41). The regulatory interactions between C1 and C2 domains have also been suggested for the targeting and regulation of different PKC isoforms. In the case of PKC␣, a single anionic residue in the C1A domain is implicated in the tethering of C1A to the other part of the molecule, most likely the C2 domain (19). The putative tethering keeps the protein in an inactive conformation at the resting state and is disrupted specifically by the Ca 2ϩ -dependent binding of the C2 domain to PS in the membrane, which in turn leads to the membrane penetration and DAG binding of the C1A domain and PKC activation. Similarly, it was suggested that the C2-like domain of novel PKC interacts with the C1 domain to regulate the enzyme activity (72). Further studies will reveal more examples of interdomain interactions in the membrane targeting of peripheral domains.

Roles of Bulk Lipids and Lipid Second Messengers
In both protein-protein and membrane-protein interactions, the initial formation of nonspecific collisional complexes, driven by diffusion and electrostatic forces, is followed by the formation of tightly bound complexes, which are stabilized by specific interactions (8,73). The former interactions mainly enhance the association rate whereas the latter interactions primarily decrease the dissociation rate. For C1 domains, the initial binding is driven by nonspecific electrostatic interactions between cationic residues with bulk anionic lipids, but the tight complex formation is achieved by both hydrophobic interactions between hydrophobic C1 residues and the membrane core and hydrogen bonds between polar C1 residues and DAG (Fig. 2). Accordingly, the subcellular localization of the C1 domain is determined primarily by the location of DAG and phorbol ester, although the bulk lipid composition of the targeting membrane can affect the kinetics of translocation (20). For anionic lipid-selective C2 domains, the initial binding is driven by electrostatic interactions via C2-bound Ca 2ϩ , cationic residues, or a combination of both. However, the paucity of multiple specific interactions required for tight complex formation renders the binding relatively weak and transient (20). On the other hand, the membrane binding of PC-selective cPLA 2 -C2 is slowly driven by interactions between aromatic residues and PC molecules (8), but the binding is tight due to membrane penetration and the resulting hydrophobic interactions. In accordance with these membrane 3 R. Stahelin, J. Rafter, and W. Cho, manuscript in preparation. 4 S. Kulkarni and W. Cho, manuscript in preparation. Minireview: C1 and C2 Domains 32409 binding properties, cPLA 2 -C2 alone can achieve the membrane localization of cPLA 2 in the cell, whereas both C1 and C2 domains are required for the prolonged membrane localization of conventional PKC (20,41). Unlike the case of the C1 domain, the bulk lipid composition of the membrane is an important determinant of Ca 2ϩ -dependent localization of the C2 domains. For the reliable prediction of subcellular localization of C2 domains based on their lipid selectivity, the lipid compositions of various cellular membranes of different cells need to be fully characterized.

Conclusions and Future Direction
Results summarized in this review show that much of the spatiotemporal dynamics of the peripheral proteins harboring membranebinding C1 and C2 domains can be accounted for by the physicochemical principles that govern their in vitro membrane binding properties. The subcellular targeting of peripheral proteins containing a single targeting domain (e.g. cPLA 2 ) usually reflects the membrane binding properties of the domain, whereas that of peripheral proteins with multiple C1 and C2 domains is determined by the delicate interplay of the domains and the synergistic actions of their agonists (e.g. Ca 2ϩ and DAG). The principles learned from these studies should help in understanding the subcellular targeting of peripheral proteins containing other membrane targeting domains. Because C1 and C2 domains can also interact with other proteins (42), the subcellular targeting of some C1-and C2-containing proteins might involve both membrane-protein and protein-protein interactions. Moreover, protein phosphorylation might play an important regulatory role in the subcellular targeting of peripheral proteins (40,41). Further systematic studies are required to comprehensively address these important issues. Finally, cellular translocation studies of peripheral proteins have been focused mainly on membrane translocation so far. To fully understand spatiotemporal dynamics and regulation of signaling peripheral proteins, however, it is necessary to simultaneously monitor the spatiotemporal dynamics and activities of peripheral proteins as well as the spatiotemporal dynamics of their stimuli in living cells. Recent technological advances in light microscopy and cell biology should greatly facilitate this challenging endeavor.