|Home | About | Journals | Submit | Contact Us | Français|
The C2 domain is a ubiquitous, conserved protein signaling motif widely found in eukaryotic signaling proteins. Although considerable functional diversity exists, most C2 domains are activated by Ca2+ binding and then dock to a specific cellular membrane. The C2 domains of protein kinase Cα (PKCα) and cytosolic phospholipase A2α (cPLA2α), for example, are known to dock to different membrane surfaces during an intracellular Ca2+ signal. Ca2+ activation targets the PKCα C2 domain to the plasma membrane and the cPLA2α C2 domain to the internal membranes, with no detectable spatial overlap. It is crucial to determine how such targeting specificity is achieved at physiological bulk Ca2+ concentrations that during a typical signaling event rarely exceed 1 μM. For the isolated PKCα C2 domain in the presence of physiological Ca2+ levels, the target lipids phosphatidylserine (PS) and phosphatidylinositol-4,5-bisphosphate (PIP2) are together sufficient to recruit the PKCα C2 domain to a lipid mixture mimicking the plasma membrane inner leaflet. For the cPLA2α C2 domain, the target lipid phosphatidylcholine (PC) appears to be sufficient to drive membrane targeting to an internal membrane mimic at physiological Ca2+ levels, although the results do not rule out a second, unknown target molecule. Stopped-flow kinetic studies provide additional information about the fundamental molecular events that occur during Ca2+-activated membrane docking. In principle, C2 domain-directed intracellular targeting, which requires coincidence detection of multiple signals (Ca2+ and one or more target lipids), can exhibit two different mechanisms: messenger-activated target affinity (MATA) and target-activated messenger affinity (TAMA). The C2 domains studied here both utilize the TAMA mechanism, in which the C2 domain Ca2+ affinity is too low to be activated by physiological Ca2+ signals in most regions of the cell. Only when the C2 domain nears its target membrane, which provides a high local concentration of target lipid, is the effective Ca2+ affinity increased by the coupled binding equilibrium to a level that enables substantial Ca2+ activation and target docking. Overall, the findings emphasize the importance of using physiological ligand concentrations in targeting studies because super-physiological concentrations can drive docking interactions even when an important targeting molecule is missing.
Many signaling pathways are regulated by signaling lipids, membrane proteins, or membrane-bound complexes associated with the plasma or internal cell membranes. Such membrane-associated signaling components control essential processes, such as cellular movement, growth, gene regulation, metabolism, hormone release, and inflammation. One of the most common regulatory elements in membrane-associated signaling pathways is the C2 domain, a ubiquitous, conserved signaling motif recognized in over 200 mammalian proteins (1). Structurally, the C2 domain motif comprises eight antiparallel β-strands assembled in a β-sandwich architecture (2-5). Functionally, although diversity exists (6), the C2 motif typically serves as a reversible membrane-targeting element activated by the binding of multiple Ca2+ ions (2-5, 7, 8). During a cytoplasmic Ca2+ signal, freely diffusing C2 proteins are activated by Ca2+ and then dock to specific cellular membranes. The resulting association brings the signaling domains of these proteins to the appropriate target membrane surface, thereby greatly facilitating interactions with membrane-bound substrates or effectors during the duration of the Ca2+ signal. Clearly, the ability of a C2 domain to preferentially dock to a specific membrane plays a central role in its function as a targeting element; thus, it is important to understand the molecular mechanisms underlying target-membrane recognition and docking.
Two representative C2 domains, those of protein kinase Cα and cytosolic phospholipase A2α, have been shown to be useful in comparative studies of membrane-targeting mechanisms (7, 9-14). These two well-characterized domains target to mutually exclusive intracellular membrane targets, and their structures (15-19) and functions (20-22) are well-studied. More generally, these two domains are representative of broader classes of C2 domains that dock to plasma or internal membranes, respectively, where they carry out essential regulatory functions. The goal of the present study is to elucidate the molecular mechanisms by which these two C2 domains are recruited specifically to their target membranes, thereby bringing the other domains of their parent proteins into the vicinity of membrane-associated protein or lipid targets.
Protein kinase C isoform α (PKCα1) is a ubiquitous signaling protein and a member of the conventional protein kinase C subfamily of serine/threonine kinases (5, 22). The PKCα enzyme regulates a wide array of important pathways ranging from cellular taxis to growth and transformation. The C2 domain of PKCα is an independent folding domain that binds two Ca2+ ions (19) and drives docking to the plasma membrane, where it recognizes both phosphatidylserine (PS) and phosphatidylinositol-4,5-bisphosphate (PIP2) as lipid targets (12, 23-28). These target lipids bind to two C2 domain sites: PS associates with the two Ca2+ ions and nearby amino acids in the Ca2+ binding site formed by three inter-strand loops (19), whereas PIP2 binds at a distinct site dominated by side chains on the β3–β4 hairpin (24-28). When docked to the membrane, the C2 domain contacts the headgroup region of the bilayer in a geometry that enables both of these sites to simultaneously contact their target lipid headgroups (29, 30). The resulting Ca2+-triggered recruitment tethers the C2 domain to the membrane, while another domain, C1, searches for rare, membrane-embedded diacylglycerol molecules that further stabilize the membrane-bound state (31, 32). It is this active, membrane-bound form of PKCα that phosphorylates an array of plasma membrane proteins. Potential targets of PKCα and other conventional PKC isozymes include MARCKS, Raf, coronin, gravin, Ca2+ channels, GTPase regulatory proteins, cytoskeletal proteins, and caveolar proteins (33-39). As observed for other conventional PKC isozymes (23, 32, 40-42), the C2 domain of PKCα is the primary driving force underlying Ca2+-regulated membrane docking such that the isolated C2 domain exhibits the same plasma membrane distribution during an intracellular Ca2+ signal as that of the full length protein (13). Thus, the isolated C2 domain is fully functional in targeting, and its mechanism of membrane specificity can be studied independent of the other protein domains.
Cytosolic phospholipase A2 isoform α (cPLA2α) is a ubiquitous signaling protein that hydrolyzes specific phospholipids to release arachidonic acid, a biosynthetic precursor of prostaglandins and leukotrienes that serve as inflammatory agents and chemoattractants (43, 44). The C2 domain of cPLA2α folds independently and is coupled to the catalytic domain by a long, flexible linker (18). During a cytoplasmic Ca2+ signal, the C2 domain binds two Ca2+ ions (45) and docks to intracellular membranes, primarily nuclear, Golgi, and endoplasmic reticulum membranes, where lipids containing arachidonate in the sn-2 position are found in highest abundance (12, 13, 46). The membrane-docked C2 domain is oriented with its three Ca2+-binding loops penetrating into the membrane, where they contact the target lipid phosphatidylcholine (PC) in the headgroup layer and also penetrate more deeply into the hydrocarbon core (47-50). The docking of the C2 domain to the membrane tethers the catalytic domain in the vicinity of the membrane surface, greatly facilitating its search for substrate lipid molecules. When the C2 and catalytic domains are separated, they retain their distinct targeting and enzymatic functions, respectively (20). The isolated C2 domain exhibits the same intracellular membrane distribution during a cytoplasmic Ca2+ signal as that of the full length protein (51). Thus, like the isolated PKCα C2 domain, the isolated cPLA2α C2 domain is fully functional as a targeting motif, and its mechanism of membrane specificity can be studied in the absence of the catalytic domain.
Previous studies have identified target lipids proposed to dominate membrane recognition during the Ca2+-triggered membrane-docking reactions of C2 domains. For the PKCα C2 domain, the primary target lipids appear to be PS and PIP2 (12, 23, 27), whereas for the cPLA2α C2 domain, the primary target lipid is PC (10, 12, 52). However, no in vitro study has yet successfully reproduced C2 domain targeting under physiological conditions. During a typical cytoplasmic Ca2+ signal, the bulk Ca2+ concentration rises from approximately 0.1 μM up to a peak level of 0.5–0.9 μM, thus approaching but rarely exceeding 1 μM (53). In the cytoplasmic compartment, the effective concentration of lipids on the plasma membrane inner leaflet is approximately 400–800 μM, whereas the effective concentration of lipids on the cytoplasmic leaflet of the nuclear–Golgi–ER membrane system is much higher (54). Current models propose that C2 domain docking and membrane specificity are driven purely by interactions of the C2 domain with the structural and electrostatic features of lipid bilayers, including specific lipid headgroups, without contributions from protein–protein interactions (10, 12, 25, 27). If this hypothesis is true, then it should be possible to find lipid mixtures that enable C2 domain docking to specific target membranes at micromolar Ca2+ concentrations. It is crucial to carry out such in vitro studies at physiological concentrations of Ca2+ because previous studies have demonstrated that super-physiological concentrations of Ca2+ can drive docking to membranes even when they are missing an important targeting element (10, 12, 27).
The present study compares the intracellular Ca2+-activated targeting of the isolated PKCα- and cPLA2α-C2 domains in a macrophage model system. Subsequently, synthetic lipid mixtures are used in vitro to determine the lipid components and concentrations needed to drive membrane association at micromolar Ca2+ concentrations. Both the equilibrium and kinetic features of these interactions are investigated. The results further define the lipid bilayer features required for efficient Ca2+-driven targeting of these two C2 domains to specific membrane surfaces under physiological conditions and further illuminate the molecular events that occur during these targeting reactions.
All lipids were synthetic unless otherwise indicated. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (phosphatidylcholine, POPC, PC) and 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine (phosphatidylcholine, PAPC); 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (phosphatidylethanolamine, PE); phosphatidylinositol (PI) natural from bovine liver; 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoserine (phosphatidylserine, PS); sphingomyelin (SM) natural from brain; and dipalmitoyl-d-myo-phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2) natural from brain were all from Avanti Polar Lipids. Cholesterol (CH) was from Sigma. N-[5-(Dimethylamino)naphthalene-1-sulfonyl]-1,2-dihexadecanoyl-sn-glycero-3-phosphoethanolamine (dansyl-PE, dPE) was from Molecular Probes.
A plasmid encoding the murine cPLA2α C2 domain fused to the C-terminus of mRFP1 (RFP-cPLA2αC2) was constructed by amplifying IMAGE clone BC003816 using primers that generated XhoI and XmaI sites on the polymerase chain reaction (PCR) product and ligating the digested product into complementary sites in pmRFP1-(C3), a plasmid constructed by substituting mRFP1 for EGFP in pEGFP(C3) (BD Biosciences Clontech). The resulting construct, RFP-cPLA2αC2, encodes residues 17 to 148 of the murine cPLA2α protein and includes the entire C2 domain. The plasmid encoding the human PKCα C2 domain fused to YFP (YFP-PKCαC2) was previously described (13). For in vitro lipid-binding studies, the PKCαC2 and cPLA2αC2 domains were subcloned by PCR into the EagI/EcoR1 site of a glutathione S-transferase (GST)-fusion vector as previously described (27).
RAW264.7 cells obtained from American Type Culture Collection (Manassas, VA) were plated on 35 mm glass-bottomed dishes (MatTek, Ashland, MA) at a density of 1 × 104 cells/cm2 and cultured in DMEM containing 10% heat-inactivated FBS, 100 U/mL penicillin, 100 μg/mL streptomycin, 0.292 mg/mL glutamine, and 20 mM HEPES in 5% CO2 at 37 °C. The cells were transfected with 1–2.5 μg each of the relevant plasmids using Lipo-fectamine 2000 (Invitrogen) in OptiMEM (Invitrogen), following the manufacturer's protocol.
Cotransfected RAW264.7 cells were rinsed with and incubated in HBSS additionally buffered with 25 mM HEPES at pH 7.4 (HHBSS) containing 0.01% endotoxin-free BSA. Images were acquired using a Nikon inverted microscope equipped with a 60 × 1.4 N.A. oil immersion objective, a CFP/YFP/RFP dichroic mirror, corresponding single band excitation and emission filters (Chroma Technology), and a CoolSNAP ES camera (Photometrics, Tucson, AZ). Excitation light was provided by a mercury lamp. Cells were stimulated with 5 μM ionomycin between the acquisition of the first and second YFP/RFP image sets. Each image set began with 300 ms YFP and 300 ms RFP acquisitions, followed by a closed shutter period, yielding a total interval of 3 s between the starting points of subsequent image sets. Final images were produced using Adobe Photoshop (Adobe) and ImageJ (NIH, http://rsb.info.nih.gov/ij/).
The C2 domains of PKCα and cPLA2α were expressed as glutathione S-transferase (GST)-fusion proteins and isolated on a glutathione affinity column prior to cleavage with thrombin and elution of the free C2 domain. The free cPLA2-C2 domain was further purified by Ca2+-dependent binding to PC-phenyl sepharose resin, followed by elution with ethylenediaminetetraacetic acid (EDTA) as described (45). The mass of the PKCα-C2 and cPLA2α-C2 domains were confirmed by matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy (MALDI-TOF). Protein purity was determined by SDS–PAGE (55), and protein concentration was determined by both absorbance at 280 nm using the calculated extinction coefficient and by the tyrosinate difference spectral method (56).
Lipids were dissolved in chloroform/methanol/water (1/2/0.8) to give the desired lipid ratios, dried under vacuum at 45 °C until all solvents were removed, and then hydrated with buffer A (25 mM N-(2-hydroxyethyl)piperazine-N′-2-ethanesulfonic acid (HEPES) at pH 7.4 with KOH, 140 mM KCl, 15 mM NaCl, and 1 mM MgCl2) by rapid vortexing. Small unilamellar phospholipid vesicles were generated by sonication of the hydrated lipids to clarity with a Misonix XL2020 probe sonicator. The vesicle stock solutions used in the equilibrium calcium titrations and kinetic experiments were prepared with a total lipid concentration of 3 mM with the following mole percentages for simple membranes: PE/PC/PS/dPE (65/10/20/5) and PE/PC/PS/dPE (40/50/5/5); for physiological plasma membrane variations: PM [6% PIP2], CH/PE/PS/PC/PIP2/dPE/PI/SM (25/23.5/21/10.5/6/5/4.5/4.5); PM [2% PIP2], PE/CH/PS/PC/dPE/PI/SM/PIP2 (27.5/25/21/10.5/5/4.5/4.5/2); PM [6% PIP2] (−) PS, PE/CH/PC/PIP2/dPE/PI/SM (44.5/25/10.5/6/5/4.5/4.5); PM [2% PIP2](−) PS, PE/CH/PC/dPE/PI/SM/PIP2 (48.5/25/10.5/5/4.5/4.5/2); PM [0% PIP2, 5%CH], PE/PS/PC/CH/dPE/PI/SM (49.5/21/10.5/5/5/4.5/4.5); PM [0% PIP2], PE/CH/PS/PC/dPE/PI/SM (29.5/25/21/10.5/5/4.5/4.5); PM [0% PIP2, 38.5% PC], PC/CH/PS/dPE/PI/SM/PE (38.5/25/21/5/4.5/4.5/1.5); and PM [0% PIP2] (−) PS, PE/CH/PC/dPE/PI/SM (50.5/25/10.5/5/4.5/4.5); and for physiological internal membrane variations: IM, PC/PE/CH/dPE/PS/PI/SM (49.5/27/5/5/4.5/4.5/4.5); IM [12%PS], PC/PS/PE/CH/dPE/PI/SM (49.5/21/10.5/5/5/4.5/4.5); IM [25%CH], PC/CH/PE/dPE/PS/PI/SM (49.5/25/7/5/4.5/4.5/4.5); IM [13.5%PC], PE/PC/CH/dPE/PS/PI/SM (63/13.5/5/5/4.5/4.5/4.5). Table 1 summarizes these lipid mixtures. Additional vesicle stock solutions for use in the equilibrium calcium titrations that required final total lipid concentrations greater than 200 μM were prepared at a total lipid concentration of 30 mM with the following mole percentages: PE/CH/PS/PC/dPE/PI/SM/PIP2 (28.5/25/21/10.5/5/4.5/4.5/1) and PC/PE/CH/dPE/PS/PI/SM (49/27/5/5/4.5/4.5/4.5). Following sonication, the insoluble material was removed from all lipid mixtures by centrifugation at 17,970g for 5 min.
Steady-state fluorescence experiments were carried out on a Photon Technology International QM-2000-6SE fluorescence spectrometer at 25 °C in buffer A. The excitation and emission slit widths were 1 and 8 nm, respectively, for all measurements. All buffers were made with Chelex-treated Ca2+-free water. Protein and lipid solutions were incubated with Chelex resin to remove residual Ca2+ before use. Quartz cuvettes and stir bars were decalcified by soaking in 100 mM EDTA and extensive rinsing with Ca2+-free water prior to use (10, 23, 45).
The quantitation of the Ca2+-dependent increase in protein-to-membrane FRET for C2 domain binding to membranes was carried out according to previously developed methods (10, 23, 45). Briefly, Ca2+-free C2 domain (0.75 μM) and Ca2+-free sonicated lipids (200 μM total = 100 μM accessible, except where indicated otherwise) in buffer A were mixed with a small volume of a concentrated Ca2+ stock solution in buffer A, and the protein-to-membrane FRET was quantitated from dPE emission (ex-citation and emission wavelengths were λex = 284 nm and λem = 522 nm, respectively). In a separate sample, identical Ca2+ was added to a Ca2+-free lipid solution in buffer A lacking protein to control for any changes in background emission arising from light scattering associated with Ca2+-mediated liposome aggregation and photobleaching of the dPE. Following the correction for dilution and subtraction of the background emission, the Ca2+ dependence of the fluorescence increase (ΔF) was plotted as a function of free Ca2+ ([Ca2+]), and the best fit was obtained using the following Hill equation (eq 1):
where ΔFmax represents the calculated maximal fluorescence change (normalized to unity, except where noted otherwise, to simplify graphical presentations), H represents the Hill coefficient, and [Ca2+]1/2 represents the free Ca2+ concentration that induces a half-maximal fluorescence change. In most cases, the free Ca2+ concentration was estimated to be the same as the concentration of Ca2+ added to the decalcified reaction solution over the course of the titration. This approximation yielded accurate [Ca2+]1/2 values when [Ca2+]1/2 values sufficiently exceeded the concentration of the C2 domain present in the decalcified starting reaction (0.75 μM). For [Ca2+]1/2 values below 4 μM, the free Ca2+ concentration was calculated by correcting the total Ca2+ concentration for Ca2+ bound to the C2 domain and for trace background amounts of Ca2+ present after decalcification.
All kinetic experiments were done on an Applied Photophysics SX.17 stopped-flow fluorescence instrument at 25 °C in buffer A as previously described (10, 23, 45) with the following modifications. The deadtime of the instrument was 0.9 ± 0.1 ms; thus, all data points prior to 1 ms were eliminated prior to quantitative analysis. To measure the protein-to-membrane FRET in this instrument, the excitation wavelength and slit-width settings on the excitation monochromator were 284 and 6 nm, respectively, whereas a 475 nm long-pass filter was used to select the detected wavelengths of emitted light.
To determine the observed rate constant for membrane association (kobs), C2 domains (1 μM, all concentrations prior to mixing) and Ca2+ (5, 50, or 1000 μM) in buffer A were mixed by stopped-flow with vesicles (400 μM total lipid) in the same buffer and Ca2+ concentration. The resulting time course yielded an increasing protein-to-membrane FRET with time and was subjected to nonlinear least-squares best-fit analysis using the following single-exponential function (eq 2).
To simplify graphical presentations, the best-fit offset C was subtracted from all data points, and the best-fit ΔFmax value was normalized to unity.
To determine the rate constant for the dissociation (koff) of C2 domains from the membranes, the experiment began with the preformed ternary complex of the C2 domain (1 μM), vesicles (400 μM total lipid), and Ca2+ (5, 10, or 1000 μM) in buffer A. At time zero, the ternary complex was rapidly mixed with an equal volume of EDTA (20 mM) in the same buffer. The resulting approach to equilibrium was monitored as a decrease in the protein-to-membrane FRET as the C2 domain dissociated from the membrane. The time course was subjected to nonlinear least-squares best-fit analysis using a single- or double-exponential function eq 3 or 4, respectively.
To simplify graphical presentations, the best-fit offset C was subtracted from all data points, and the best-fit ΔFmax (or, for the latter equation, ΔFmax1 + ΔFmax2) value was normalized to unity.
The present study investigates the mechanism of Ca2+-activated targeting of C2 domains to specific intracellular membranes, using the C2 domains of the important signaling enzymes PKCα and cPLA2α as representative examples. Previous studies carried out in various cell types have found that cytoplasmic Ca2+ signals drive the PKCα C2 domain primarily to the plasma membrane, whereas the cPLA2α C2 domain primarily targets to internal membranes (12, 13, 27, 46, 57). Previous in vitro studies have established important elements of this specificity: PKCα C2 prefers membranes containing phosphatidylserine (PS) (12, 23) and phosphatidylinositol-4,5-bisphosphate (PIP2) (27), whereas cPLA2α C2 prefers membranes rich in phosphatidylcholine (PC) (10, 12, 52). But previous in vitro studies have not yet rigorously explained the highly efficient and specific membrane docking of these C2 domains at the bulk concentration of Ca2+ achieved in the cytoplasm during the peak of a typical Ca2+ signaling event (approximately 500 to 900 nM, hereafter referred to as 1 μM) (53).
To directly compare the membrane-targeting specificities of the PKCα and cPLA2α C2 domains, the present study begins by coexpressing the two domains in RAW264.7 cells, a macrophage cell line. Subsequent in vitro studies are designed to analyze the mechanism of specific membrane targeting using synthetic membranes composed of carefully controlled lipid mixtures. To determine the lipid mixtures and concentrations needed to achieve physiological membrane targeting, the present study employs model membranes composed of both simple and complex lipid mixtures, the latter designed to closely approximate the cytosolic leaflets of plasma and intracellular membranes. A protein-to-membrane FRET assay is used to elucidate the equilibrium and kinetic parameters for C2 domain docking to these model membranes in vitro. The approach provides new insights into the mechanisms of target-membrane recognition by the PKCα and cPLA2α C2 domains and into the molecular events that occur during Ca2+-activated membrane docking.
The PKCα and cPLA2α C2 domains were fused to yellow fluorescent protein (YFP) and red fluorescent protein (RFP), respectively. The resulting fusion proteins were simultaneously introduced into the mouse macrophage cell line, RAW264.7, and the fluorescent cells were treated with the Ca2+ ionophore ionomycin to elicit a homogeneous cytoplasmic Ca2+ increase of magnitude approximating the micromolar peak of a physiological Ca2+ signal ((58) and Evans, unpublished results), as illustrated in Figure 1. In untreated cells, both fusion proteins exhibited a uniform distribution in the cytoplasm and nucleoplasm (Figure 1A and C). Within seconds of the cytoplasmic Ca2+ increase, YFP-PKCαC2 translocated to the plasma membrane and RFP-cPLA2αC2 to the internal membranes (Figure 1B and D). Notably, the merged image (Figure 1E, PKCαC2 in green and cPLA2αC2 in red) highlights the mutually exclusive targeting of the two C2 domains, consistent with previous findings that the PKCα C2 domain exclusively targets the plasma membrane, whereas the cPLA2α C2 domain exclusively targets the internal membranes (12, 13, 27, 46, 57).
Similar targeting specificities have been observed for these two C2 domains in other cell types (12, 13, 27, 46, 57). The most variability is observed for the cPLA2α C2 domain, which in RAW cells primarily targets to the nuclear membrane, whereas a small but detectable fraction of the C2 domain population targets to the Golgi and endoplasmic reticulum (ER) membranes. The same type of targeting pattern, primarily localized to the nuclear membrane, has been previously observed in primary leukocytes and in HEK293 cells, both of which possess a relatively small density of Golgi membranes (12, 46). By contrast, in CHO and MDCK cells, both of which possess extensive Golgi systems, the cPLA2α C2 domain primarily targets to Golgi rather than to other internal membranes (13, 27, 59). Thus, the distribution of the cPLA2α C2 domain between the nuclear, Golgi, and ER membranes is closely tied to the relative densities of these membrane systems in the cell interior. In cell types with poorly developed Golgi systems, most of the targeting is to the nuclear membrane, whereas in cell types with extensive Golgi systems, the majority of targeting is to the Golgi membrane.
Recombinant PKCα-C2 and cPLA2α-C2 domains were cloned, expressed, and purified (see Materials and Methods) for use in model membrane-binding studies designed to investigate the mechanisms of specific membrane targeting at physiological Ca2+ concentrations. The purity of the resulting domains was determined by SDS–PAGE and was found to exceed 90%. The masses of PKCα-C2 and cPLA2α-C2 were found to be 16,288 and 16,267 Da, respectively, by MALDI-TOF mass spectroscopy. Both experimental masses are within the error of the predicted masses (16,283 for PKCα-C2 and 16,267 for cPLA2α-C2).
The affinities of PKCα-C2 and cPLA2α-C2 for membranes have been shown to be strongly dependent on membrane PS and PC content, respectively, under conditions of super-physiological Ca2+ concentrations (100–1000 μM) (10, 12, 23, 52). To examine whether simple lipid mixtures containing PS and PC can reproduce in vitro the micromolar Ca2+ sensitivity and orthogonal target membrane specificities observed for these C2 domains, the binding of C2 domains to synthetic sonicated, small unilamellar vesicles (SUVs) composed of simple PS and PC mixtures was examined. Protein-to-membrane FRET was employed to measure membrane docking as Ca2+ was titrated into a solution containing a given domain, a given type of SUV, and a physiological buffer ((45), see Materials and Methods).
For the PKCα-C2 domain, the Ca2+ titration profile in Figure 2A yielded a nonlinear least-squares best-fit [Ca2+]1/2 of 28 ± 2 μM for C2 domain docking to simple membranes with a PS content of 20 mol %, approximating the PS content of its target membrane, the inner leaflet of the plasma membrane (PE/PC/PS/dPE, 65/10/20/5 mol %) (12, 60, 61). By contrast, the [Ca2+]1/2 exceeded 1000 μM for C2 domain docking to membranes containing 5 mol % PS, corresponding to the PS content of internal membranes (PE/PC/PS/dPE, 40/50/5/5 mol %) (12, 60, 61).
For the cPLA2α-C2 domain, the Ca2+ titration in Figure 2B yielded a [Ca2+]1/2 of 7 ± 1 μM for C2 domain docking to membranes containing 50 mol % PC, resembling the PC content of its target internal membranes (PE/PC/PS/dPE, 40/50/5/5 mol %) (12, 60, 61). A [Ca2+]1/2 of 17 ± 2 μM was observed for C2 domain docking to membranes containing 10 mol % PC, as in the plasma membrane inner leaflet (PE/PC/PS/dPE, 65/10/20/5 mol %) (12, 60, 61).
These Ca2+ titrations for simple lipid mixtures confirm the previously noted PS and PC preferences of the PKCα-C2 and cPLA2α-C2 domains, respectively (10, 12, 23, 52). Such preferences make significant contributions to the intracellular targeting of the two domains because the highest PS and PC lipid mole percentages are found in the plasma and internal membranes, respectively (12, 60, 61). However, under the present experimental conditions, the Ca2+ titrations for simple PS/PC mixtures yield nonphysiological [Ca2+]1/2 values that are 28-fold and 7-fold too large, respectively, to explain the efficient docking of the PKCα- and cPLA2α-C2 domains to their target membranes during micromolar cytoplasmic Ca2+ signaling events. Recently, we observed that the addition of PIP2 to simple PS/PC membranes yielded a significantly lower [Ca2+]1/2 value; however, this value remained at least 3-fold too large to explain efficient docking during a micromolar cytoplasmic Ca2+ signal. In an effort to identify the protein–membrane interactions that bring the Ca2+ affinities of the PKCα- and cPLA2α-C2 domains into the physiological range, further experiments were conducted with synthetic membranes that more closely approximate the complexity of physiological membranes.
Physiological model membrane SUVs were generated with lipid compositions designed to mimic the cytosolic leaflet of either the plasma membrane or the internal cell membranes. These model membranes, detailed in Table 1, contained the predominant lipids of mammalian membranes exposed to the cytoplasm, including phosphatidylethanolamine (PE), phosphatidylcho-line (PC), phosphatidylserine (PS), phosphatidylinositol (PI), sphingomyelin (SM), cholesterol (CH), and phosphatidylinositol-4,5-bisphosphate (PIP2) as well as a small density of the FRET acceptor dansyl-phosphatidylethanolamine (dPE, 5 mol %). The plasma membrane mimic (PM) reflects the relatively high content of PS, cholesterol, and PIP2 and the low content of PC, found in the plasma membrane inner leaflet ((12, 60, 61), Table 1). Two versions of this PM mixture were utilized containing 21 mol % PS and either 2 mol % PIP2, corresponding to the bulk PIP2 concentration of the plasma membrane inner leaflet, or 6 mol % PIP2, representing the putative local PIP2 concentration in lipid rafts (62-65). The internal membrane mimic (IM) reflects the high content of PC, approaching 50 mol %, as well as the low content of PS, cholesterol, and PIP2 found in internal cell membranes, such as nuclear, Golgi, and ER membranes (12, 60, 61). Additional lipid mixtures based on variations of the PM and IM mixtures were also created to examine the roles of specific lipid components in membrane recognition by C2 domains. These variations lowered the mole percentage of PC, PS, CH, or PIP2 while correspondingly increasing the mole percent of PE, thereby maintaining constant densities of all other components (Table 1). Both previous work and the results from this study indicate that the PKCα-C2 and cPLA2α-C2 domains are relatively insensitive to membrane PE content (10, 12, 23, 52), making PE the best choice for a replacement component. Preliminary studies comparing the binding of C2 domains to SUVs versus large unilamellar vesicles (LUVs) of the same lipid composition revealed no detectable differences (data not shown), suggesting that the greater membrane curvature of SUVs does not significantly alter protein–membrane interactions in this system.
The protein-to-membrane FRET assay was used to measure membrane docking as Ca2+ was titrated into the system containing the PKCα-C2 domain and a given type of physiological model membrane SUVs, thereby revealing the effects of different lipid mixtures on the Ca2+ dependence of membrane docking. Figure 3A shows the resulting Ca2+ titration curves, and Table 2 summarizes the corresponding [Ca2+]1/2 values, determined by nonlinear least-squares best-fit using the Hill equation. Each Hill analysis utilized the actual free Ca2+ concentration calculated for the relevant titration by correcting for minimal background Ca2+ contamination (0.1 μM) and for Ca2+ bound to the C2 domain (see Materials and Methods).
The results of Figure 3A and Table 2 indicate that PM mixtures containing physiological levels of both PS and PIP2 yield high Ca2+ affinities that are adequate to explain the observed recruitment of the PKCα-C2 domain to the inner leaflet of the plasma membrane during cytoplasmic Ca2+ signals. The PM mixture containing 2 mol % PIP2, designated PM[2%PIP2], yielded a [Ca2+]1/2 value of 1.6 ± 0.1 μM. Similarly, the PM mixture containing 6 mol % PIP2, designated PM[6%PIP2], yielded a [Ca2+]1/2 value of 0.7 ± 0.1 μM (Figure 3A, Table 2). By contrast, the IM mixture yielded much weaker Ca2+-triggered membrane binding such that the measured [Ca2+]1/2 greatly exceeded 300 μM (Figure 3A, Table 2). These findings confirm that physiological lipid mixtures resembling the cytosolic leaflet of the plasma membrane but not that of inner membranes enable efficient recruitment of the PKCα-C2 domain at micromolar levels of Ca2+.
The results further indicate that efficient docking of the PKCα-C2 domain to model physiological membranes at micromolar Ca2+ levels requires both PS and PIP2 but is significantly less sensitive to PC and cholesterol. Thus, when the PS was removed from the PM mixture, yielding either the PM[2%PIP2](−)PS or the PM[6%PIP2](−)PS mixture, [Ca2+]1/2 increased 7-fold or 5-fold, respectively (Figure 3A, Table 2). Similarly, when PIP2 was removed from the PM mixture, yielding the PM(−)PIP2 mixture, [Ca2+]1/2 increased 20-fold or 50-fold, relative to the corresponding PM mixtures containing 2 or 6 mol % PIP2, respectively (Figure 3A, Table 2). Finally, when both PS and PIP2 were removed to yield the PM(−)PIP2(−)PS mixture, [Ca2+]1/2 increased to a value greatly exceeding 300 μM, similar to that observed for IM membranes (Figure 3A, Table 2). By contrast, the [Ca2+]1/2 value was found to be relatively independent of the PC or cholesterol content as long as lipid mixtures containing similar levels of PS and PIP2 were compared (Figure 3A, Table 2). Together, these results indicate that the exquisite plasma membrane targeting specificity of the PKCα-C2 domain triggered by a micromolar cytoplasmic Ca2+ signal is highly dependent on only the PS and PIP2 contents of the target membrane.
The presence of PIP2 in the membrane was found to have a significant effect on the Hill coefficient for the Ca2+-triggered docking of the PKCα C2 domain to membranes. Membranes lacking PIP2 exhibited Hill coefficients ranging from 1.2 to 1.7 (Table 2), overlapping the value previously measured for Ca2+ titrations of PKCα-C2 domain docking to simple membranes lacking PIP2 (H = 1.3 ± 0.1, (23)). When PIP2 was added, the Hill coefficient decreased slightly but significantly in most cases to values ranging from 1.0 to 1.2 (Table 2). The PKCα C2 domain possesses two Ca2+ binding sites, and the Hill coefficient for Ca2+-triggered membrane docking reports positive cooperativity between these sites during the binding of two Ca2+ ions. The lower Hill coefficients observed for PIP2-containing membranes suggest that PIP2 facilitates the docking of the C2 domain containing only one bound Ca2+ ion, which would in principle exhibit a Hill coefficient of 1.0. This observation has important implications for the mechanism of membrane specificity and docking as further discussed below (see Discussion).
For the cPLA2α-C2 domain, Ca2+ titrations were also carried out using FRET to monitor C2 domain docking to physiological model membrane SUVs (Figure 3B). The resulting [Ca2+]1/2 values are summarized in Table 2. As previously observed, PC is the primary lipid determinant of cPLA2α-C2 membrane docking specificity. Thus, the IM membrane mimic containing 50 mol % PC yielded a [Ca2+]1/2 value of 10 ± 2 μM such that membrane docking occurred at a 3-fold lower Ca2+ concentration than that for any of the PM mixtures containing 6, 2, or 0 mol % PIP2 ([Ca2+]1/2 = 36 ± 3, 31 ± 3, or 29 ± 4 μM, respectively). The PC dependence of this specificity was investigated further using a series of IM mixtures with a low PC content, a high PS content, or a high CH content. A reduction in the PC content of the IM mixture to 13.5 mol % PC resulted in an increase in [Ca2+]1/2 of greater than 3-fold to 36 ± 3 μM, similar to the [Ca2+]1/2 value measured for the PM mixtures. By contrast, increases in the PS (to 21 mol %) or CH (to 25 mol %) content of the IM mixture, yielding levels of these lipids similar to those found in the plasma membrane, had little effect on Ca2+ sensitivity ([Ca2+]1/2 = 9 ± 2 and [Ca2+]1/2 = 11 ± 2 μM). For this C2 domain, the presence of PIP2 in the membrane had no detectable effect on either the [Ca2+]1/2 value or the Hill coefficient, in contrast to the PKCα C2 domain where PIP2 decreased both of these parameters (Table 2). This pattern is consistent with previous observations that unlike the PKCα C2 domain, the cPLA2α-C2 domain possesses no PIP2 binding site (27).
The present findings confirm that in the context of physiological lipid mixtures, membrane PC content has a role in defining cPLA2α-C2 domain-targeting specificity. However, none of the membrane mimics, including the IM mixture, yielded [Ca2+]1/2 values within the micromolar range of Ca2+ concentrations found in cytoplasmic signals. It follows that the conditions of these in vitro experiments did not yet match those experienced by the cPLA2α-C2 domain during intracellular targeting. One possible explanation was that the IM mixture might be missing an important lipid component. Three additional lipid components of internal membranes were tested as potential target lipids: ceramide-1-phosphate, PI(4)P1, and arachidonate-containing PC. As summarized in Table 3, each of these additional lipids yielded less than a 2-fold effect on the [Ca2+]1/2 value, providing strong evidence that these lipids are not specific targets of the cPLA2α-C2 domain. Thus, the evidence to date suggests that the only significant target lipid of this C2 domain is the PC headgroup, although an additional target cannot be ruled out. Another possible explanation for the micromolar values of [Ca2+]1/2 exhibited by the cPLA2α-C2 domain in cells is the much higher local concentration of internal membranes, relative to the concentration of membranes used in the present in vitro study.
All of the above experiments utilized total accessible lipid concentrations (100 μM) that were significantly lower than those estimated for the plasma membrane inner leaflet (400–800 μM) or for internal membranes (>3000 μM) in the cytoplasmic compartment of a living cell (54). In protein-to-membrane FRET studies, the maximum useful accessible lipid concentration is typically about 400 μM because of the excessive light scattering that occurs at higher concentrations. In order to establish the relationship between Ca2+-sensitivity and lipid concentration, [Ca2+]1/2 values were measured for the docking of the PKCα-C2 domain to different concentrations of a PM mixture and for the docking of the cPLA2α-C2 domain to different concentrations of the IM mixture, as shown in Figure 4.
The [Ca2+]1/2 values shown in Figure 4A for the PKCα-C2 domain docking to a PM mixture (PM-[2%PIP2]) showed little change as the total accessible lipid concentration ranged from 100 μM up to 400 μM, where it approached cellular levels. Such findings are consistent with a model in which one or two Ca2+ ions bind initially to the free C2 domain, which then docks rapidly to the target membrane at all tested concentrations of the PM mixture (see Discussion). The limitations of the FRET assay prevented the determination of [Ca2+]1/2 for PKCα-C2 domain docking at IM lipid concentrations approaching those relevant in the cell (>3000 μM). However, the [Ca2+]1/2 measured for the docking of this C2 domain to a total accessible IM lipid concentration of 100 μM was 360 ± 80 μM (Figure 3A, Table 2), a value 300-fold higher than the Ca2+ concentration achieved during a micromolar cytoplasmic Ca2+ signal. Together, these findings indicate that the Ca2+ affinity of the PKCα-C2 domain is high enough to be activated by a physiological Ca2+ signal when the domain is in the vicinity of cellular concentrations of plasma membrane but not in the vicinity of internal membranes. Thus, the results obtained for the binding of the PKCα-C2 domain to complex lipid mixtures in vitro fully explain the observed specificity of this domain for the plasma membrane during micromolar cytoplasmic Ca2+ signals.
In contrast to the findings for the PKCα-C2 domain, the data presented in Figure 4B for the cPLA2α-C2 domain reveal a strong dependence of [Ca2+]½ on the local concentration of the membranes. Notably, the [Ca2+]½ value for cPLA2α-C2 docking to the IM mixture decreased linearly as the total accessible lipid concentration increased from 100 to 400 μM. Overall, [Ca2+]½ decreased over 2-fold from 7.2 ± 1.1 to 3.4 ± 0.4 μM over this range, representing a corresponding increase in Ca2+ affinity. Again, the limitations of the FRET assay prevented the extension of the experiment up to the intracellular concentration of total accessible internal membrane lipids (>3000 μM), but extrapolation of the data to these higher lipid concentrations (Figure 4B, dotted lines) suggests that [Ca2+]½ would enter the physiological micromolar range (Figure 4B, dashed line) at an accessible lipid concentration between 700 and 1000 μM. Assuming that such extrapolation is justified, the intracellular concentration of internal membranes is well above the level needed to generate efficient docking of the cPLA2α-C2 domain during micromolar cytoplasmic Ca2+ signals. For comparison, the Ca2+ sensitivity of the cPLA2α-C2 domain docking to a PM mixture (PM[2%PIP2]) was also measured at an accessible lipid concentration of 400 μM, yielding a [Ca2+]½ value of 20 ± 2 μM. Together, these findings suggest that the Ca2+ affinity of the cPLA2α-C2 domain is high enough to be activated by a physiological Ca2+ signal when the domain is in the vicinity of cellular concentrations of internal membranes but not in the vicinity of the plasma membrane, thereby explaining the specific targeting observed in living cells.
Overall, the present equilibrium binding data indicate that the Ca2+-triggered intracellular targeting of PKCα- and cPLA2α-C2 domains to plasma and internal membranes, respectively, can be fully explained by C2 domain interactions with target lipids. No interactions with other proteins appear to be needed for the inherent targeting specificities and sensitivities to micromolar Ca2+ signals exhibited by these two C2 domains. The plasma membrane specificity and micromolar Ca2+ activation of the PKCα-C2 domain in cells requires target lipids PS and PIP2, which are found primarily on the inner leaflet of the plasma membrane. By contrast, the internal membrane specificity and micromolar Ca2+ activation of the cPLA2α-C2 domain require the extremely high local concentration of PC generated by high internal membrane densities in the cell interior. For the latter C2 domain, a second target in addition to PC cannot be ruled out, although searches for such a second target lipid have been unsuccessful, supporting the conclusion that a high local concentration of PC is the only relevant target.
In order to better define the molecular mechanisms underlying the targeting of C2 domains to specific membranes, the kinetics of membrane docking were investigated. Previous kinetic studies of membrane association and dissociation have been carried out for both PKCα- and cPLA2α-C2 domains but only at super-physiological Ca2+ concentrations in the 100 to 1000 μM range (10, 12, 23, 45). At these very high Ca2+ concentrations, the equilibrium binding data (Figure 2) indicate that the membrane specificity of the PKCα-C2 domain is weak such that the domain prefers target PM lipid mixtures but exhibits significant docking to nontarget IM mixtures as well. By contrast, at low Ca2+ concentrations, this domain exhibits strong specificity for its target PM lipid mixture. To pursue the molecular basis of these equilibrium results, membrane association and dissociation rates were measured for the PKCα-C2 domain binding to various lipid mixtures at different Ca2+ concentrations in a stopped-flow fluorescence spectrometer (see Materials and Methods). Association kinetics were measured by rapidly mixing a solution containing the C2 domain with a suspension of membranes, where both components were pre-equilibrated with the desired Ca2+ concentration before mixing. Dissociation kinetics were measured by mixing the preformed Ca2+–protein–membrane complex with an EDTA solution. Following rapid mixing, the approach to equilibrium was monitored by protein-to-membrane FRET for both the association and dissociation reactions.
The kinetic results for the PKCα-C2 domain are shown in Figure 5 and Table 4. Strikingly, at a super-physiological Ca2+ concentration of 1 mM, the association kinetics were nearly identical for PKCα-C2 domain docking to target (PM[2%PIP2] or PM[6%PIP2]) and nontarget (IM, PM[6%PIP2](−)PS, or PM[0%PIP2]) membranes, varying no more than 1.7-fold. Similarly, although the kinetics of PKCα-C2 domain dissociation from different membranes were measurably different, the differences were too small to yield specific membrane docking. Thus, relative to the dissociation of the C2 domain from PM target membranes (PM[2%PIP2] or PM[6%PIP2]), the removal PS or PIP2 sped the dissociation 3- to 4-fold, whereas the simultaneous removal of both PS and PIP2 sped the dissociation 8-fold, and the use of IM membranes sped the dissociation 5-fold. Overall, the moderate effects of lipid composition on the association and dissociation kinetics of the PKCα-C2 domain at 1 mM Ca2+ explains the measurable docking of this domain to nontarget membranes at this superphysiological Ca2+ concentration (Figure 3A). At the lower Ca2+ concentration of 5 μM, approaching the level of a micromolar Ca2+ signal, the poor equilibrium binding of the C2 domain prevented kinetic measurements for nontarget membranes. However, at this lower Ca2+ concentration, kinetic measurements were successfully carried out for the C2 domain docking to its PM target membrane. The association reaction revealed a new kinetic component exhibiting a rate constant 13- to 17-fold smaller than that previously observed at high Ca2+ concentrations. Such slow association can be attributed to the membrane docking of C2 domains that are only partially, rather than fully, occupied by Ca2+. These observations support a mechanism (see Discussion) in which micromolar Ca2+ concentrations load the C2 domain with only one Ca2+ ion, which then docks to membranes in a reaction that exhibits a strong target specificity, in contrast to the weaker target specificity observed for the docking of the domain loaded with two Ca2+ ions at super-physiological Ca2+ concentrations.
The kinetic results for the cPLA2α-C2 domain are presented in Figure 6 and Table 4. At super-physiological Ca2+ concentrations of 1 mM or 50 μM, the association kinetics for C2 domain docking to target (IM) and nontarget (PM or IM with reduced PC levels) membranes were nearly identical, varying no more than 2.3-fold. The dissociation kinetics similarly varied no more than 4-fold among these different membranes. These findings explain the equilibrium binding results (Figure 3B) for the cPLA2α-C2 domain at 1 mM Ca2+, where the domain exhibited significant docking to both target and nontarget membranes at this super-physiological Ca2+ concentration (Figure 3B). When the Ca2+ concentration was reduced to the near physiological levels of 10 or 5 μM, sufficient equilibrium binding to both target and nontarget membranes was obtained to again carry out kinetic studies. Notably, the association kinetics still varied by no more than 5-fold between target (IM) and nontarget (PM or IM with reduced PC levels) membranes, whereas the dissociation kinetics varied only 4-fold. Such results explain the significant equilibrium binding of the cPLA2α-C2 domain to nontarget membranes at these lower Ca2+ concentrations (Figure 3B). Finally, the observed rate constant for target (IM) membrane association remained nearly the same, within 2.4-fold when the Ca2+ concentration was reduced from 1 mM to 5 μM. It follows that the mechanism of Ca2+-dependent membrane docking is essentially the same at both micromolar and super-physiological Ca2+ concentrations, indicating that for both of these conditions, Ca2+ activation is driven by the binding of two Ca2+ ions to the C2 domain prior to membrane docking, as previously described (45). Overall, the kinetic analysis of the cPLA2α-C2 domain confirms the observation of the equilibrium analysis: this C2 domain exhibits significantly less discrimination between target and nontarget membranes than the PKCα-C2 domain.
The present studies of Ca2+-stimulated intracellular targeting in the RAW macrophage cell line confirm the target membrane specificities previously observed for the PKCα- and cPLA2α-C2 domains in other cell types (12, 13, 27, 46, 57). Upon Ca2+-activation, the PKCα- and cPLA2α-C2 domains specifically target to the plasma and internal membranes, respectively, with no detectable overlap in their targeting specificities. During typical intracellular Ca2+ signals, this remarkably specific targeting occurs as the Ca2+ concentration increases from a basal level of 0.1 μM up to a peak concentration of 0.5 to 0.9 μM in the bulk cytoplasm (53) and rarely exceeds 1 μM. The resulting findings for Ca2+-triggered C2 domain docking to synthetic membrane vesicles significantly extend the current mechanistic understanding of specific C2 domain targeting at physiological Ca2+ concentrations. As predicted by previous models (10, 12, 45), the new results confirm that membrane specificity is dominated by C2 domain docking to the membrane lipids. Furthermore, these results clarify the lipid compositions and concentrations needed for such targeting specificity and shed light on the mechanism of targeting to specific membranes in the complex intracellular environment.
Two different mechanisms can be proposed for specific intracellular targeting arising from the coincidence detection of a global second messenger and a localized target molecule, such as the specific targeting of a C2 domain by coincidence detection of a global Ca2+ signal and localized target lipids. The messenger-activated-target-affinity (MATA) mechanism is characterized by a targeting protein that possesses an apo state with a high affinity for the second messenger and a low affinity for the target molecule. Specifically, in its apo state, the KD for messenger binding is less than or approximately equal to the peak messenger concentration during the signal, whereas the KD for target binding is significantly greater than the intracellular target concentration. In such a system, the binding of the second messenger activates the targeting protein by triggering a large increase in the affinity for the target molecule via their coupled binding equilibria. During a second messenger signaling event, the targeting protein will be activated everywhere because of its innate high affinity for the messenger and will subsequently dock to all accessible target molecules. It follows that the targeting protein will be recruited to any and all regions of the cell where the global messenger signal extends and where accessible target molecules exist. This MATA mechanism is likely to be operating in second messenger pathways wherein the goal of activation is to drive docking to all available target molecules throughout the cell. However, this mechanism will only yield specific targeting when the target molecules are limited to specific locations within the cell.
The target-activated-messenger-affinity (TAMA) mechanism is quite different. This mechanism is characterized by a targeting protein that possesses an apo state with a low affinity for the second messenger as well as a low affinity for the target. Specifically, in its apo state, the KD for messenger binding is significantly higher than the peak messenger concentration during the signal, and the KD for target binding is significantly greater than the intracellular target concentration. During a second messenger signaling event, the affinity of the targeting protein is too low to bind the second messenger except in those regions of the cell where the local concentration of the target is high. In such regions, the high target concentration drives an increase in the effective affinity for the second messenger because of the thermodynamic effect of the coupled binding equilibrium. Thus, targeting proteins will be successfully activated and targeted by the second messenger only when they lie within a cellular region that both senses the second messenger signal and possesses a high local concentration of target molecules. This TAMA mechanism is useful in second messenger-activated pathways wherein the goal of activation is to drive docking only in regions of the cell possessing large pools of the target.
For the present C2 domains, TAMA is the logical mechanism because essential target lipids (PS and PC) are found in membranes throughout the cell but are significantly enriched in the target membranes (plasma and intracellular membranes). In the absence of target molecules, the PKCα and cPLA2α C2 domains exhibit low Ca2+ affinities ([Ca2+]½ of 35 and 14 μM, respectively) (10, 23, 45) that would allow only minor activation by a physiological micromolar Ca2+ signal. For the PKCα C2 domain, the target lipid PS is found in significant concentrations in internal membranes (4 mol %) but is enriched in the inner leaflet of the plasma membrane (21 mol %) (12, 60, 61). For the cPLA2α C2 domain, the target lipid PC is present at significant levels in the plasma membrane inner leaflet (10 mol %) but is substantially enriched in internal membranes (50 mol %) (12, 60, 61). Thus, for both of these C2 domains, TAMA is used to optimize the specificity of membrane targeting by limiting Ca2+ activation to the regions of the membrane containing the largest pools of common target lipids.
For conventional PKCα, β, and γ C2 domains, the TAMA mechanism is further enhanced by positive cooperativity in the binding of two or more PS molecules to the C2 domain (23) and by the exclusive plasma membrane localization of a second target lipid, PIP2. During a Ca2+ signal, the combined TAMA effects of PS and PIP2 enable conventional PKC C2 domains to drive highly specific Ca2+-activated docking to the plasma membrane surface, where important protein substrates of the PKC kinase domain are located.
For the cPLA2α C2 domain, it is not yet possible to rule out the existence of a second target lipid highly localized to the internal membranes (analogous to PIP2 for PKC C2 domains) or the existence of a local, higher-than-bulk Ca2+ concentration in the vicinity of the target membrane. However, the present findings suggest that the high local density of PC found in internal membranes is essential to explain the observed specific targeting to these membranes during a Ca2+ signal. Because of extensive invagination and the dense packing of adjacent membrane structures, the local density of internal membranes in the cell interior is at least 10-fold higher than the membrane density in the vicinity of the plasma membrane (54). This high membrane density, together with the 5-fold higher mole percent of target PC found in internal membranes (12, 60, 61), ensures that the cPLA2α C2 domain will experience at least a 50-fold higher local PC concentration in the vicinity of internal membranes relative to that of the plasma membrane. Such a large density of target lipid is proposed to drive TAMA targeting of the cPLA2α C2 domain to internal membranes, thereby recruiting the associated phospholipase domain to these membranes, which possess the highest mole percent of the substrate arachidonate-containing phospholipids found anywhere in the cell.
In other types of Ca2+ signaling pathways, examples of both the MATA and TAMA mechanisms are evident. An interesting case is calmodulin, which possesses distinct N-and C-terminal domains, both of which are regulated by Ca2+ binding (66). This protein docks to a wide array of effector proteins using a diversity of different docking mechanisms (67, 68). The KD for Ca2+ binding to the C-terminal domain is in the low micromolar range (69), and this domain can dock independently to certain protein targets (67, 68). Thus, the C-domain is an example of MATA targeting because a physiological Ca2+ signal substantially loads and activates the C-domain even in the absence of the target. By contrast, the KD for Ca2+ binding to the N-terminal domain is in the tens of micromolar range (69) such that efficient Ca2+ activation of this domain requires the presence of target proteins that lower its effective Ca2+ KD via the coupled binding equilibrium (67, 68). Thus, calmodulin targeting events that require N-domain activation are best described by the TAMA mechanism. One advantage of this two-domain system is that it enables a single regulatory protein to utilize both the MATA and TAMA mechanisms, depending on which of its two domains is the limiting factor in target docking.
The present studies of the PKCα-C2 domain show that the local densities of target lipids PS and PIP2 on the inner leaflet of the plasma membrane are sufficient to drive Ca2+-activated membrane docking during a physiological Ca2+ signal. At these micromolar Ca2+ levels, the data support a molecular mechanism in which the free C2 domain binds a single Ca2+ ion and then docks to PIP2 on the membrane prior to associating with another Ca ion and PS. Equation 5 summarizes this multistep mechanism
where the target lipids PIP2 and PS are located on the surface of the target membrane, and the asterisk indicates the membrane-bound protein. This proposed mechanism of specific PKCα-C2 domain docking is further strengthened by a molecular analysis of the individual binding steps. It has previously been shown that the empty C2 domain containing no bound Ca2+ ions cannot dock to the membrane because of the charge repulsion between the negative charges in the Ca2+ binding site and the negative surface charge of the membrane (7). The present model proposes that the binding of only one Ca2+ ion neutralizes enough of the protein negative charge to allow binding to PIP2 on the membrane surface but not to PS, which is believed to require two Ca2+ ions for its association with the Ca2+ binding site (7, 19). The binding of the C2 domain to PIP2 would facilitate the subsequent binding of a second Ca2+ ion and one or more PS headgroups to the Ca2+ binding site. This mechanism explains the low Hill coefficient observed for Ca2+-triggered docking to membranes containing PIP2 because the binding of a single Ca2+ ion to one of the two Ca2+ binding sites would exhibit minimal positive cooperativity and thereby yield a Hill coefficient approaching 1.0, as observed (Table 2). The mechanism also explains the slow kinetic component observed in the membrane docking reaction at micromolar levels of activating Ca2+ (Table 4). The slow component is proposed to represent the docking of the singly Ca2+-occupied C2 domain, which becomes undetectable at higher Ca2+ concentrations when the domain is loaded with two Ca2+ ions and docks more rapidly to the membrane. The model fully accounts for the synergistic effects of Ca2+, PIP2, and PS on membrane binding because high-affinity membrane docking is achieved only after the C2 domain, two Ca2+ ions, PIP2, and at least one PS molecule form a stable complex. A key element of the model is the binding of the singly Ca2+-occupied C2 domain to PIP2 on the membrane surface, which greatly enhances the Ca2+ affinity of the second Ca2+ binding site because of the enhanced proximity to its PS ligand.
By contrast, physiological Ca2+ signals are unable to drive the docking of the PKCα-C2 domain to internal membranes. Because internal membranes lack accessible PIP2 and have low mole fractions of PS relative to that of the plasma membrane, two Ca2+ ions must first bind to the C2 domain before it docks to PS on the membrane surface. However, the [Ca2+]1/2 value for the binding of two Ca2+ ions to the free PKCα-C2 domain is known to be 35 μM (23); thus, the intrinsic Ca2+ affinity is much lower than that needed for a physiological Ca2+ signal to drive activation and membrane docking in the vicinity of internal membranes. Instead, the equilibrium favors the unbound state of PKCα-C2 molecules located near internal membranes during a physiological Ca2+ signal. Thus, the ability of the singly Ca2+-occupied PKCα-C2 domain to bind to PIP2 explains the exquisite specificity of intracellular targeting to the cytosolic leaflet of the plasma membrane, where virtually all of the accessible PIP2 is located and which contains high PS concentrations as well. For full length PKCα, the binding of C1 domain to diacylglycerol increases the lifetime of the bound state (31), and interactions with other proteins such as RACK also occur (70). Overall, however, the in vitro findings strongly suggest that the C2 domain interactions with PS and PIP2 dominate plasma membrane targeting.
The present results suggest that the specific targeting of the cPLA2α-C2 domain to internal membranes is driven by the extremely high local density of the target lipid PC associated with these membranes. This high target concentration in the cell interior lowers the [Ca2+]1/2 value for membrane docking into the range accessible to cytosolic Ca2+ signals. The available evidence suggests a two-step molecular mechanism in which the free C2 domain binds two Ca2+ ions and then docks to membrane-bound PC as summarized in eq 6
where the target lipid PC is located on a membrane surface, and the asterisk indicates the membrane-bound protein. This mechanism is supported by a molecular analysis of the microscopic docking events. Previous findings have shown that the empty C2 domain is initially prevented from membrane docking by its negatively charged Ca2+ binding site (7). During a Ca2+ signal, the domain binds two Ca2+ ions with positive cooperativity (45), which neutralize the charge of the Ca2+ binding site and allow the binding to the PC on the surface of internal membranes (7). In the absence of membranes, the free C2 domain exhibits a [Ca2+]1/2 value of 14 μM for the binding of two Ca2+ ions (45), but the high local concentration of PC in the cell interior pulls this binding equilibrium toward the membrane-docked state and decreases the effective [Ca2+]1/2 value, thereby approaching the micromolar range accessible to physiological Ca2+ signals. The high local concentration of PC needed to facilitate such Ca2+-triggered docking is provied by the extremely dense, highly invaginated membrane distributions exhibited by the nuclear, Golgi, and ER, which all contain high mole fractions of PC (54). In contrast, physiological Ca2+ signals are unable to drive the binding of the domain to the plasma membrane, where the local concentration of PC is too low to bring the [Ca2+]1/2 value for membrane docking into the physiological range.
Overall, the target-activated-messenger-affinity (TAMA) mechanism is proposed to drive the highly specific intracellular targeting of the PKCα and cPLA2α C2 domains purely on the basis of protein–lipid interactions. For each C2 domain, the [Ca2+]1/2 value for membrane docking varies with its location in the cell: when the C2 domain is far from its target membrane, the [Ca2+]1/2 value is too high for successful Ca2+ activation, but in the vicinity of the target membrane, high local concentrations of the target lipid decrease [Ca2+]1/2 into the range accessible to physiological Ca2+ signals. More broadly, the TAMA targeting mechanism and the distinct messenger-activated-target-affinity (MATA) mechanism are each expected to be utilized in different signaling pathways wherein targeting is controlled both by a global second messenger and by one or more additional target ligands.
Finally, the present findings emphasize the importance of carrying out in vitro studies of signaling proteins at physiological ligand concentrations. In principle, super-physiological concentrations can drive interactions even when an important cofactor is missing, thereby slowing the identification of key cofactors. For many years, the importance of PIP2 to the intracellular targeting of the PKCα-C2 domain to plasma membrane was obscured by in vitro studies (carried out in our laboratory and in others) using super-physiological concentrations of Ca2+ to drive membrane docking. At these Ca2+ concentrations, the C2 domain docks quite well to membranes lacking PIP2. However, the present findings demonstrate that PIP2 is essential for efficient membrane docking at physiological Ca2+ concentrations.
We thank Dr. Christina Leslie (National Jewish Medical and Research Center, Denver, CO) for the critical reading of this manuscript and for providing the construct expressing the cPLA2α C2 domain. We also thank Dr. Jae-Won Soh (Inha University, Incheon, Korea) for providing the construct expressing the PKCα full length protein.
†Financial support was provided by NIH Grant GM R01-063235.
1Abbreviations: cPLA2α, the α-isoform of cytosolic phospholipase A2; PKCα, the α-isoform of protein kinase C; PC, phosphatidylcholine; PS, phosphatidylserine; PIP2, phosphatidylinositol-4,5-bisphosphate; MATA, messenger-activated target affinity; TAMA, target-activated messenger-affinity; CFP, cyan fluorescent protein; RFP, red fluorescent protein; YFP, yellow fluorescent protein.