Role of the Ca 2 (cid:1) /Phosphatidylserine Binding Region of the C2 Domain in the Translocation of Protein Kinase C (cid:2) to the Plasma Membrane*

Signal transduction via protein kinase C (PKC) is closely regulated by its subcellular localization. To map the molecular determinants mediating the C2 domain-dependent translocation of PKC (cid:2) to the plasma membrane, full-length native protein and several point mutants in the Ca 2 (cid:1) /phosphatidylserine-binding site were tagged with green fluorescent protein and transiently expressed in rat basophilic leukemia cells (RBL-2H3). Substitution of several aspartate residues by asparagine completely abolished Ca 2 (cid:1) -dependent membrane targeting of PKC (cid:2) . Strikingly, these mutations enabled the mutant proteins to translocate in a diacylglycerol-de-pendent manner, suggesting that neutralization of charges in the Ca 2 (cid:1) binding region enables the C1 domain to bind diacylglycerol. In addition, it was demonstrated that the protein residues involved in direct interactions with acidic phospholipids play differential and pivotal roles in the membrane targeting of the enzyme. These findings provide new information on how the C2 domain-dependent membrane targeting of PKC (cid:2) occurs in the presence of physiological stimuli. Protein kinase

Protein kinase C family (PKC) 1 is known to control many cellular processes including metabolism regulation, receptor signal transduction, cell growth and differentiation, and hormone and neurotransmitter secretion (1,2). This family consists of 10 closely related isoenzymes that can be divided into three groups according to the type of activator they need. For example, conventional PKCs (␣, ␤I, ␤II, and ␥) require the full complement of negatively charged phospholipids, Ca 2ϩ , and diacylglycerol or phorbol esters before they are activated. The novel PKCs (␦, ⑀, , and ) on the other hand do not require Ca 2ϩ , whereas the atypical PKCs (, /) do not require diacylglycerol or Ca 2ϩ (3).
Conventional and novel PKCs share the property of using two membrane-targeting modules for the sensitive, specific, and reversible regulation of their function. Thus, the C1 and the C2 domains are involved in membrane translocation and subsequent enzyme activation (4). Many studies performed to date suggest a sequential mechanism in the activation of PKC␣ by Ca 2ϩ in which membrane association is followed by increased catalytic activity (5)(6)(7)(8)(9). However, it is still not clear whether these domains function in a concerted way or as independent modules.
The first step of this process is probably the Ca 2ϩ -dependent contact of the C2 domain with negatively charged phospholipids at the plasma membrane (6, 8, 10 -12). Recent studies on different C2 domains reveal that, despite their sequential and architectural similarity, they are functionally specialized modules that exhibit distinct equilibrium and kinetic behaviors that are probably optimized for different Ca 2ϩ -signaling applications (13)(14). Most of these studies have been performed with isolated C2 domains and, in the case of PKC␣, the precise molecular mechanism driving the above interaction and the consequences for the general functioning of the enzyme are still not well defined.
Our previous crystallographic studies of the isolated C2 domain of PKC␣ bound to Ca 2ϩ and the short chain lipid 1,2dicaproyl-sn-phosphatidyl-L-serine show that two or three Ca 2ϩ ions bind to the domain, with one of them (Ca1) bridging the protein directly to phosphatidylserine (PS). Moreover, several residues in the domain seem to be involved in a direct interaction with PS (15)(16), and further studies show that both Ca 2ϩ and PS binding sites are important for in vitro enzyme activation (9,17,18). Whether two or three calcium ions are needed for plasma membrane translocation of PKC␣ and how PS and diacylglycerol interact with the regulatory domain to orchestrate membrane targeting and activation of PKC␣ in vivo are key questions that remain to be answered.
To investigate the molecular determinants mediating the C2 domain-dependent plasma membrane localization of PKC␣, we generated a construct containing full-length PKC␣ tagged to green fluorescent protein (PKC␣-GFP). Additionally, different constructs containing point mutations at the Ca 2ϩ and PS binding sites in the C2 domain of PKC␣ were produced and transfected into RBL-2H3 cells (a mast cell line that provides a useful model for studying signal transduction mechanisms). The cells are activated by antigens via the tyrosine kinase-dependent high affinity receptor for IgE (Fc⑀RI) to induce the release of secretory granules and the generation of cytokines. Activation of the receptor leads to phospholipase C␥ activation and the consequent generation of inositol 1,4,5-triphosphate/ Ca 2ϩ and diacylglycerol, both of which are regarded as PKC activators (19). Using confocal microscopy combined with measurements of intracellular Ca 2ϩ concentration in timelapse experiments, we have found that both Ca 2ϩ and PS binding sites in the C2 domain play important and differential roles in the spatio-temporal localization of PKC␣, suggesting a very accurate and controlled mechanism driven primarily by the C2 domain.

EXPERIMENTAL PROCEDURES
cDNA Constructions-N-terminal fusions of rat PKC␣ and the different mutants to GFP were generated by inserting cDNAs into the multiple cloning site of the pEGFP-N3 (Clontech Laboratories, Inc., Palo Alto, CA) mammalian expression vector. Briefly, cDNAs encoding PKC␣ and its mutants D187N, D246N/D248N, D187N/D246N/D248N, N189A, R216A, R249A, and T251A were amplified by PCR using the primers 5Ј-ATTCTCGAGCTATGGCTGACGTT and 3Ј-CCGGGTACCT-ACTGCACTTTGCAAGAT. XhoI/KpnI-digested PKC␣ and mutated fragments were ligated with the XhoI/KpnI-digested vector, thus generating the different fusion constructs. Further details on site-directed mutagenesis to generate the different mutants are reported by Conesa-Zamora et al. (17). All constructs were confirmed by DNA sequencing. The stability and viability of the mutated proteins were studied by using specific activity measurements. It was demonstrated that the mutants could be activated in a PS-dependent manner, although not to the same extent as the wild-type protein (9,17). Moreover, earlier studies have shown that a C-terminal GFP tag does not affect the catalytic activity or the cofactor dependence of PKC␣ (20 -22).
Cell Culture-Rat basophilic leukemia (RBL-2H3) cells were cultured at 37°C in a humidified atmosphere of 5% CO 2 in a growth medium of Dulbecco's modified Eagle's medium supplemented with 15% (v/v) fetal calf serum, 50 units/ml penicillin, 50 g/ml streptomycin, and 4 mM glutamine. Cells were prepared for confocal microscopy as described by Bolsover et al. (23). Basically, harvested cells were resuspended in electroporation buffer (120 mM NaCl, 5.5 mM KCl, 2.8 mM MgCl 2 , 25 mM glucose, 20 mM Hepes, pH 7.2) and 30 g of cDNA. Cells were electroporated in a Bio-Rad GenePulser with two 500-V pulses. The cells were immediately placed on ice for 5 min before being plated on glass coverslips and incubated at 37°C for 4 -6 h, after which the growth medium was renewed. Cells were used 24 h later after priming overnight with 500 ng/ml IgE-anti-dinitrophenyl (mouse monoclonal, Sigma). Coverslips were washed with 3 ml of extracellular buffer HBS (120 mM NaCl, 25 mM glucose, 5.5 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 20 mM HEPES, pH 7.2). All added substances were dissolved or diluted in HBS. DiC8 and ionomycin were dissolved in Me 2 SO and diluted to the final concentration with extracellular buffer shortly before the experiment. During the experiment, the cells were not exposed to Me 2 SO concentrations higher than 1%. All the experiments were carried out at room temperature, and unless otherwise stated, on at least four different occasions. In each experiment, recordings were obtained from 2-6 cells.
Imaging of [Ca 2ϩ ] i and PKC-GFP-[Ca 2ϩ ] i was measured using Fura Red (Molecular Probes, Inc., Eugene, OR). Stock solutions (2 mM) of the AM-ester form of the fluorescent Ca 2ϩ indicator were made using a solution of 2.5% (w/v) Pluronic F-127 in absolute Me 2 SO. This stock was diluted 1000-fold in growth medium and applied to cells for 30 min at 25°C and 5% CO 2 . After incubation, the cells were washed with HBS and mounted in a holder attached to the stage of a Leica inverted confocal laser-scanning microscope. During imaging, cells were stimulated with antigen (DNP-conjugated human serum albumin (HSA), Sigma) or other agents as described.
Fluorescence confocal microscopy was used to monitor both the translocation of EGFP fusion constructs and the intensity of the Ca 2ϩ indicator in response to different stimuli. Cells were illuminated using a 488-nm laser line, and emitted light was simultaneously sampled through a 515-540-nm band pass filter (EGFP channel) and 590-nm long pass filter (Fura Red channel). The signals from Fura Red and EGFP could not be completely separated (23). The same photomultiplier voltages were used throughout these experiments, meaning that the measurements made when only one dye was present were also applicable to dual dye readings. Imaging of untransfected cells loaded with Fura Red revealed minimal contamination of the EGFP channel by the Fura Red signal; because the contamination represented less than 1% of the values recorded when EGFP was present, no correction was performed. In contrast, imaging of transfected cells without Fura Red revealed significant contamination of the Fura Red channel by the signal from EGFP. The signals recorded in the Fura Red channel were closely correlated with those recorded in the EGFP channel and had values 24.7% as large. In the dual dye experiments we therefore calculated the net Fura Red signal in any defined area of interest by subtracting 24.7% of the EGFP signal from the overall signal in the Fura Red channel. [Ca 2ϩ ] i was then calculated using where K d is the apparent Ca 2ϩ binding affinity of the Ca 2ϩ indicator, determined by the manufacturer (140 nM), I is the net Fura Red fluorescence intensity of an individual cell, I min is the net Fura Red fluorescence intensity of the cell at zero Ca 2ϩ , and I max is net Fura Red fluorescence intensity of the cell at saturating Ca 2ϩ . I min was replaced by FI 0 , where F ϭ (I min /I 0 ) ϭ (K d ϩ C r )/(C r S ϩ K d ), I 0 is the net Fura Red fluorescence intensity of a resting unstimulated cell, calculated as the average intensity of the first 10 images recorded in each experiment, C r is the Ca 2ϩ concentration in resting unstimulated cells, measured as 136 Ϯ 1 nM in 256 untransfected cells loaded with the ratiometric Ca 2ϩ indicator Fura-2 (see Bolsover et al. (23)), and S ϭ I max /I min , measured as 0.101 on this instrument. When excited at 488 nm, the fluorescence of Fura Red falls on binding Ca 2ϩ . I max was replaced by SI min . Substitution of I min and I max in Equation 1 therefore gives which was used to convert net Fura Red intensities into Ca 2ϩ values. Image Analysis-A series of confocal images were recorded for each experiment at time intervals of 9.3 s. The time series were analyzed using Lucida 4.0 software (Kinetic Imaging, Wirral, UK). An individual analysis of protein translocation for each cell was performed by tracing a line intensity profile across the cell (24). The relative increase in plasma membrane localization of the enzyme for each time point was calculated by using the ratio r ϭ (I mb Ϫ I cyt )/I cyt , where I mb is the fluorescence intensity at the plasma membrane, and I cyt is the average cytosolic fluorescence intensity. Mean values are given ϮS.E.
Structural Models-1DSY was used as the Protein Data Bank identifier for the experimentally determined C2 domain structure (15). The program used to visualize the structures was Swiss-PDB Viewer 3.7 program by GlaxoSmithKline (25). Fig. 1A illustrates a scheme based on the structure of the C2 domain of PKC␣ bound to three Ca 2ϩ and to the short chain lipid 1,2-dicaproyl-sn-phosphatidyl-Lserine (15)(16). The side chains of the five aspartates (Asp-187, Asp-193, Asp-246, Asp-248, and Asp-254) that directly coordinate Ca1 and Ca2 are shown. Simple (D187N), double (D246N/ D248N) and triple (D187N/D246N/D248N) mutants were generated to study the role of these residues in the plasma membrane targeting of the enzyme. D187N was chosen because the crystallographic study showed that this aspartate residue forms a bidentate interaction with Ca1, which is the ion involved in bridging the protein residues with the phosphate group of the phospholipid molecule (Fig. 1A). In the case of the double mutant containing Asp-246 and Asp-248 substituted by Asn, key coordinations with Ca1 and specially with Ca2 were abolished. Double (D246N/D248N) and triple (D187N/D246N/ D248N) substitutions of the aspartate residues by Asn neutralized the highly negative charge exhibited by this region, and it was thought that this might help understand whether this neutralization of charges is the only mechanism required for membrane targeting of the domain or a different mechanism operates this interaction. Fig. 1B shows the side chains of the residues involved in the PS-binding site: Asn-189 (located in loop 1), Arg-216 (located in loop 2), Arg-249 and Thr-251 (both located in loop 3). Thr-251 is a special residue since it has been proposed to be involved in two types of interaction. When the C2 domain only binds two Ca 2ϩ , this residue interacts with the sn-1 acyl group oxygen of the PS (15). However, when three Ca 2ϩ are bound to the domain, the side-chain oxygen of Thr-251 also coordinates the third Ca 2ϩ , which seems to bind to the C2 domain only in the presence of a very short phosphatidylserine analogue (16). Whether this third Ca 2ϩ was necessary for PKC␣ plasma mem-brane translocation was unclear as was the specific role of the threonine residue. Thus, single substitutions by Ala of each amino acidic residue located in the lipid-binding site were generated to study their role in PKC␣ membrane localization. When these mutants were studied, all of them exhibited enzyme activity at saturating Ca 2ϩ , PS, and diacylglycerol concentrations, demonstrating that the mutations did not significantly disrupt the tertiary structural fold of the mutant proteins (9,17).

Mutagenesis Rationale-
The Ca 2ϩ -binding Sites of the C2 Domain Are Critical for PKC␣ Membrane Translocation Induced by the Activation of Cell-surface Receptors-Cells were transfected with PKC␣ -GFP and loaded with Fura Red to monitor protein translocation and [Ca 2ϩ ] i simultaneously. To analyze the kinetics of the two processes, a series of 63 images were taken at 9.3-s intervals. To determine PKC␣ translocation, the relative increase in plasma membrane over cytosolic fluorescence intensity (R) was calculated at each time point. Fig. 2A shows images of antigen-stimulated RBL-2H3 cells expressing PKC␣ -GFP. In unstimulated cells, the protein was evenly distributed throughout the cytosol ( Fig. 2A, 0 s), whereas stimulation with 40 ng/ml DNP-HSA resulted in oscillatory translocations of PKC␣-GFP to and from the plasma membrane ( Fig. 2A, 149, 205, 223, and 279 s). The addition of ionomycin after 381 s also produced plasma membrane translocation ( Fig. 2A, 437).
When the time course of Ca 2ϩ oscillations versus the plasma membrane translocation of PKC␣-GFP was compared (Figs. 2, B and C), intracellular Ca 2ϩ spikes were seen to be highly correlated with the translocation of PKC␣-GFP. Maximum protein translocation occurred between 2 and 26 s after the first Ca 2ϩ spike was generated. It is interesting to note that the first Ca 2ϩ spike was more effective in producing the translocation of the protein to the plasma membrane, whereas the second and third pulses, which showed the same Ca 2ϩ concentrations, were only capable of partially anchoring the protein. The addition of ionomycin in the same experiment produced a further increase in cytosolic Ca 2ϩ and a parallel translocation of PKC␣-GFP that started to revert to the cytosol 79 Ϯ 23 s after ionomycin addition. It is interesting to note that protein started to revert to the cytosol when the intracellular Ca 2ϩ concentration was still compatible with protein translocation. Similar translocation profiles were obtained in control experiments, in which ionomycin was added directly to the cells without previous antigen stimulation (data not shown).
To investigate how the aspartate residues coordinate the different Ca 2ϩ ions and produce the membrane targeting of PKC␣, we generated several mutants that were also fused to GFP, as stated above. As shown in Fig. 3A, all the mutants were localized homogeneously in the cytosol in unstimulated cells. When the cells were stimulated by the addition of 40 ng/ml DNP-HSA, no changes in plasma membrane translocation of the mutant proteins were detected; similar results were obtained when ionomycin was added. A comparison of Fig. 3, B and C, clearly illustrates that although Ca 2ϩ spikes were generated in the cells transfected with the different mutant proteins, these mutations strongly affected the Ca 2ϩ -dependent translocation of the enzyme (Fig. 3C). Additionally, none of the mutant proteins translocated to the plasma membrane, although sustained cytosolic Ca 2ϩ concentrations ranging from 900 to 1400 nM were produced in the cells analyzed by the addition of ionomycin. Taken together these data suggest that the initial physiological translocation of PKC␣ to the plasma membrane is driven primarily in a Ca 2ϩ -dependent manner by the C2 domain and that the residues coordinating Ca1 and Ca2 are critical in the in vivo protein-plasma membrane interaction.
Ca 2ϩ Binding Residues Are Involved in the C1 Domain-dependent Translocation of PKC␣ to the Plasma Membrane-To determine the role of the C1 domain in the Ca 2ϩ -dependent membrane targeting of PKC␣, we examined the direct effect of increasing concentrations of the membrane-permeant diacylglycerol (DiC8) on both cytosolic Ca 2ϩ and the kinetics of PKC␣-GFP translocation. Neither the Ca 2ϩ signal nor PKC␣-GFP targeting of the plasma membrane was produced by increasing the concentrations of DiC8 alone (Fig. 4, A and B). Only when ionomycin was added to the medium did the protein translocate to the membrane in a DiC8 concentration-dependent manner (Fig. 4B). In fact, the higher the DiC8 concentration, the longer the residence time of PKC␣ at the plasma membrane. It is interesting to note that the protein translocation profile obtained after the addition of ionomycin to the medium containing 10 g/ml DiC8 was very similar to that obtained when the cells were stimulated previously with 40 ng/ml antigen (Fig.  2C), suggesting that the concentration of diacylglycerol generated under physiological conditions might also be similar but probably only sufficient to anchor the protein to the plasma membrane for a very short length of time through the C1 domain, as has been suggested in previous reports (8,26).
Because the C1 and C2 domains have always been considered as independent modules, it is striking that the effect of DiC8 is only observed when the concentration of intracytosolic Ca 2ϩ increases. The results obtained with the Ca 2ϩ binding mutants shed light on this issue. When the cells transfected with PKC␣ D187N mutant were stimulated with increasing concentrations of DiC8, no protein translocation was observed (n ϭ 20 cells). However, when ionomycin was added (Fig. 5A) the mutant interacted with the plasma membrane only in 50% of the cells analyzed with a rate (R) of 0.6. Note, also, that after ionomycin stimulation the translocation profiles reflected a slower and more transient process compared with that of the wild-type protein that lasted at least 186 s with the same concentration of DiC8 (100 g/ml). Moreover, maximum protein translocation of PKC␣ D187N mutant was obtained 123 Ϯ 11 s after ionomycin addition, whereas wild-type protein showed maximum translocation after 55 Ϯ 10 s.
Surprisingly, when the cells were transfected with the PKC␣ D246N/D248N mutant (Fig. 5B), the protein started to trans-locate slowly to the plasma membrane after the addition of 50 or 100 g/ml DiC8. This effect was clearly concentration-dependent since 25 g/ml DiC8 only produced a transient translocation after ionomycin stimulation, whereas 50 and 100 g/ml DiC8 promoted the irreversible plasma membrane localization of the protein that was detectable after 115 Ϯ 18 s (n ϭ 8 cells). Note also that measurable levels of PKC␣ D246N/ D248N mutant were recorded after 58 Ϯ 4 s in the presence of ionomycin when the cells were previously stimulated with 25 g/ml DiC8 compared with the 86 Ϯ 4 s needed in the case of PKC␣ D187N mutant. Furthermore, the translocation of PKC␣ D246N/D248N to the plasma membrane became almost independent of ionomycin addition when 50 and 100 g/ml DiC8 were used to stimulate the cells. These results demonstrate a strong correlation between the degree of neutralization of the aspartate residues that coordinate the Ca 2ϩ ions and the accessibility of DiC8 to the C1 domain.
Direct Interaction of the Ca 2ϩ Binding Region with Phospholipids Is Needed for the Stable Localization of PKC␣-We also generated point mutations of the residues involved in the PSbinding site; Asn-189, Arg-216, Arg-249, and Thr-251 were mutated to Ala and fused to GFP. It is important to note that all these constructs exhibited a very similar degree of Ca 2ϩ -dependent activation to the wild-type protein except Thr-251, which showed a slightly decreased Ca 2ϩ /phospholipid-dependent activation (17). When those constructs were transfected into RBL-2H3 cells, all of them were expressed and uniformly distributed in the cytosol in resting conditions, except the PKC␣ R249A mutant, which was also localized in the nucleus (Fig. 6A). Fig. 6, B and C, illustrate how PKC␣ N189A and PKC␣ R216A mutants partially translocated to the plasma membrane upon antigen stimulation. However, there were two significant differences compared with the wild-type protein.
First, the delay before maximum translocation ranged between 31 and 53 s for PKC␣ N189A and between 30 and 74 s for PKC␣ R216A, whereas the wild-type protein only needed 2-26 s; second, the mutated proteins only translocated once after the first Ca 2ϩ spike, none of the other Ca 2ϩ spikes producing plasma membrane translocation of these two mutants. Furthermore, when ionomycin was added, intracellular Ca 2ϩ concentrations increased to 1000 nM or above in all cells (Fig. 6B), and the localization profiles of PKC␣ N189A and PKC␣ R216A showed a very transient process compared with that seen in the case of the wild-type protein. In particular, the half-maximal plasma membrane dissociation time of PKC␣ N189A mutant was 73 Ϯ 9 s and of PKC␣ R216A mutant was 76 Ϯ 16 s compared with the wild-type PKC␣ of 108 Ϯ 12 s, suggesting that these mutations produce a slight alteration in the Ca 2ϩdependent membrane targeting of the enzyme. PKC␣ R249A and PKC␣ T251A mutant did not translocate to the plasma membrane even after ionomycin addition (Fig.  6C). Because both these proteins interact with one of the carbonyl group of the fatty acyl chains of the phospholipid (15), these results suggest that the protein might use these residues to anchor to the membrane interface, resulting in greater stabilization of the protein-membrane complex.
When RBL-2H3 cells transfected with these phospholipid binding mutants were stimulated with increasing concentrations of DiC8, no plasma membrane translocation was observed until ionomycin was added to the stimulation medium (Fig. 7). The fact that both PKC␣ N189A and PKC␣ R216A mutants translocated to a similar extent to the wild-type protein sug-gests that the C1 domain can overcome the slight lack of function of these C2 domain mutants (Fig. 7, A and B). In contrast, the PKC␣ R249A mutant did not translocate to the plasma membrane either in the presence of DiC8 or ionomycin (Fig. 7C). The PKC␣ T251A mutant targeted the plasma membrane in a transient manner (Fig. 7D), and the half-maximal plasma membrane dissociation time of 114 Ϯ 14 s (n ϭ 10 cells) was much shorter than that exhibited by wild-type PKC␣, PKC␣ N189A, and PKC␣ R216A. These results suggest that the role played by Arg-249 and Thr-251 in the C2 domain membrane anchorage cannot be overcome by DiC8 acting through the C1 domain and that the interactions existing between these residues and the plasma membrane might be essential for the membrane targeting of PKC␣.

Role of Ca 2ϩ Ions in the C2 Domain-dependent Translocation of PKC␣-
The results obtained in this work suggest that a very sensitive mechanism exists whereby the Ca 2ϩ binding region of the C2 domain drives PKC␣ membrane targeting in vivo. Substitutions of Asp-187 and Asp-246/248 by Asn had a dramatic effect on the Ca 2ϩ -dependent membrane targeting of the enzyme, confirming previous biochemical studies in which similar substitutions produced a severe impairment of Ca 2ϩ -dependent phospholipid binding (8,9,28). It is important to note that Asp-187 mainly coordinates Ca1, and the single substitution of an aspartate residue by asparagine was not expected to produce such a damaging effect, since the other four aspartate residues involved remain intact and could coordinate at least one Ca 2ϩ ion. Thus, these results confirm the hypothesis proposed in the crystallographic model whereby Ca1 serves as a bridge between some of the aspartate residues located in the protein and the phosphate group of the PS (15,16). Similar function has been attributed to Ca1 in the case of synaptotagmin I (C2A domain) (29,30). In contrast, Ca1 seems not to be essential for the membrane binding of cytosolic phospholipase A 2 (31). Thus, the varied results obtained with the different proteins suggest that the subtle structural differences existing among C2 domains are probably the basis of the different functional mechanisms exhibited by the proteins that hold these domains.
Additionally, it was observed that the Ca 2ϩ binding mutant proteins associated with the plasma membrane in a DiC8-dependent manner was strongly correlated with the number of aspartate residues neutralized in each mutant (Fig. 5). This effect could well correspond to the neutralization of charges that occurs after Ca2 binds to the C2 domain (13,32,33) and which probably produces a conformational change that in turn allows diacylglycerol access to the C1 domain. This hypothesis is supported by several studies that have shown through different techniques that the full-length enzyme undergoes a general conformational change upon Ca 2ϩ binding (6, 12, 34 -36). Whether or not this conformational change also produces the liberation of the pseudosubstrate region from the catalytic domain before C1 domain binding to diacylglycerol cannot be easily answered, but previous works in our lab have shown that a significant activation of the enzyme can occur in the only presence of PS and Ca 2ϩ , further supporting this hypothesis (9,17).
The need of a third Ca 2ϩ ion (Ca3) to anchor PKC␣ to the plasma membrane is still under debate. In this work, no Ca 2ϩdependent translocation of PKC␣ T251A was observed, suggesting that if there is a need for a third Ca 2ϩ to anchor the enzyme to the plasma membrane, this ion has to bind to the domain before PKC␣ interacts with the phospholipid bilayer. In addition, the reduced plasma membrane association exhib- ited by PKC␣ T251A mutant in the presence of DiC8 implies that this residue might be involved in stabilizing C2 domain membrane docking before diacylglycerol binds to the C1 domain. Taking into account the low affinity exhibited by this C2 domain for a third Ca 2ϩ in solution (14,16), it is more likely that this stabilization might be due to a direct inter-  action between Thr-251 and the phospholipid bilayer, but the mediation of a third Ca 2ϩ in this process cannot be completely ruled out.
Role of the Phosphatidylserine Binding Region in the C2 Domain-dependent Translocation of PKC␣-Another important issue addressed in this work was the role of the amino acidic residues at the Ca 2ϩ binding region of the C2 domain of PKC␣ involved in direct binding with the phosphatidylserine at the membrane surface. In this case, we observed that mutation to Ale of the side chains of Asn-189 and Arg-216 had a different effect on the Ca 2ϩ -dependent translocation of PKC␣ than did the same substitutions in Arg-249 and Thr-251. In the first two cases, the mutations resulted in a weaker membrane association of the proteins. However, after the addition of diacylglycerol and ionomycin, these mutants bound to the membrane in a similar way to the wild-type protein, suggesting that the interactions exerted by these two residues located in loops 1 and 2 are important for stabilizing the protein-membrane complex, although they are not crucial for the protein to anchor to the membrane.
In contrast, interactions established by the side chain of Arg-249 were very critical since its mutation inhibited the ability of the protein to interact with the plasma membrane even in the presence of saturating Ca 2ϩ and diacylglycerol. This effect might be because of the loss of certain interactions with the phospholipid moieties since this mutant exhibited very similar Ca 2ϩ -dependent activation to wild-type PKC␣, as demonstrated in previous work (17). Moreover, PKC␣ is not the only case that has been reported since it has been shown recently (37) that the Ca 2ϩ binding affinity of a homologue residue mutated in the C2A domain of synaptotagmin I does not differ from observations made in wild-type protein. In the crystallographic model of the C2 domain of PKC␣ (15), the guanidinium group of Arg-249 has been proposed to interact through hydrogen bonding to the sn-2 acyl group oxygen and additionally to the ester oxygen of the sn-1 acyl chain. Because the Arg-216 side chain establishes hydrogen bonding only with the carbonyl group of the sn-2 acyl chain, it seems that the interactions occurring between Arg-249 and the phospholipid in the C2 domain membrane anchorage are more critical than the corresponding interaction of Arg-216. It should also be mentioned that in the crystallographic model, extra hydropho-bic interactions were established between the aliphatic carbons of Arg-249 and the acyl chains of 1,2-dicaproyl-sn-phosphatidyl-L-serine, and this is supported by the drastic effect observed in the membrane targeting of the enzyme when Arg-249 was substituted by Ala. Although it is not clear that the location and orientation of these acyl chains of the phospholipid shown in the crystallographic study are the same as their corresponding location in a bilayer, it is possible that hydrophobic interactions further increase the importance of Arg-249 for the C2 domain membrane docking and orientation. It is important to take into account that there are hydrophobic residues (Trp-245 and Trp-247) surrounding Arg-249 in loop 3 that have been implicated in the partial membrane penetration of PKC␣ (6). Similar evidence has been proffered in the case of the C2A domain of synaptotagmin I, where direct interactions of residues located in loop 3 and loop 2 with a soluble phosphatidylserine were reported (38 -39).
Conclusions-Together, these results provide a deeper molecular understanding for the sequential mechanism of PKC␣ activation driven by the C2 domain (Fig. 8). In our model, increase in intracellular Ca 2ϩ produces the binding of Ca1 and Ca2 when the protein is still in the cytosol, leading to the membrane targeting of the enzyme through the C2 domain (Fig. 8b). Ca1 is responsible for bridging the protein to the specific phospholipid molecules, which are also recognized with the help of Asn-189 and Arg-216, whereas Ca2 is responsible for keeping Ca1 in its proper location and for inducing a conformational change in PKC␣, which partially penetrates and orientates in the phospholipid bilayer through loop 3 (Arg-249 and Thr-251). Only after this has occurred can the C1 domain find the diacylglycerol generated in the membrane surface, enabling the protein to become fully activated (Fig. 8c). Whether or not the C2 domain-dependent translocation of PKC␣ itself is able to transduce signals downstream, representing a distinct, diacylglycerol-independent signaling mode is still not known and will have to be further studied.
The present study provides new insights into the molecular mechanism underlying the dynamic nature of PKC␣ redistribution in vivo. Our results have dissected a sequential mechanism that controls the function of the C2 domain of PKC␣. Because the different PKC isoenzymes have been implicated in the malignant transformation and proliferation of cells, they FIG. 8. Proposed mechanism for C2 domain-dependent translocation of PKC␣ to the plasma membrane. In resting conditions, where cytosolic Ca 2ϩ is ϳ160 nM, the protein is found in the cytosol (a). When the Ca 2ϩ level increases upon cross-linking of IgE receptors, at least two Ca 2ϩ ions bind to the C2 domain to form a 2-Ca 2ϩ -PKC␣ complex with high membrane binding affinity and which undergoes a conformational change (b). After that the C2 domain can penetrate the interface of the bilayer through loop 3 when it orientates itself, leading to a more stable interaction that, in turn, allows the C1 domain to access the diacylglycerol (DAG) generated in the membrane, fully activating the enzyme (c). Note that C1a and C1b have been placed based on the model described by Cho and Medkova (27). R, relative plasma membrane translocation; Cat, catalytic domain.
have become a very attractive target for anticancer drug design (40). One of the main problems in using this strategy is the ubiquity and the lack of specific inhibitors/activators for each isoenzyme. For this reason, an accurate definition of the residues involved in the Ca 2ϩ /lipid-dependent translocation is a very useful tool for the design of new drugs that can act specifically in this critical activation step.