Structure and Dynamics of the Phospholipase C- (cid:1) 1 Pleckstrin Homology Domain Located at the Lipid Bilayer Surface*

Despite the importance of signal transduction pathways at membrane surfaces, there have been few means of investigating their molecular mechanisms based on the structural information of membrane-bound proteins. We applied solid state NMR as a novel method to obtain structural information about the phospholipase C- (cid:1) 1 (PLC- (cid:1) 1) pleckstrin homology (PH) domain at the lipid bilayer surface. NMR spectra of the alanine residues in the vicinity of the (cid:2) 5/ (cid:2) 6 loop in the PH domain revealed changes in local conformations due to the membrane localization of the protein. We propose that these conformational changes originate from a hydrophobic interaction between the amphipathic (cid:3) -helix located in the (cid:2) 5/ (cid:2) 6 loop and the hydrophobic layer of the membrane and contribute to the membrane binding affinity, interdomain interactions and intermolecular interactions of PLC- (cid:1) 1. Pleckstrin homology (PH) 1 domains are well defined structural modules of about 120 amino acid residues (1, 2) mainly found in proteins involved in cellular signaling and cytoskeletal functions (3–6). It has been proposed that these domains func-tion as mediators of intermolecular interactions analogous to many other structural modules involved in cellular signaling e.g. (cid:2) ; and A118G, 5 (cid:2) -ATCACCAGCTGACGGTC- AGCACTGGGTG-3 (cid:2) The desired point mutation and the sequence flanking the mutagenic primer-annealing site sequence analysis. Protein Expression and Purification— The wild-type and mutated PLC- (cid:1) 1 PH domains (1–140) were expressed as glutathione S -transfer-ase (GST) fusion proteins in Escherichia coli (PR745) (20). Cells were grown in M9 medium (1), which contains 100 mg of each of 20 amino acids but with L -alanine replaced by L -[3- 13 C]alanine and were incu- bated in the presence of 0.1 m M isopropyl-1-thio- (cid:2) - D -galactopyranoside (5 h at 37 °C). After centrifugation, the resulting cell pellets were resuspended in a buffer containing a mixture of protease inhibitors (20 m M Tris-HCl, pH 7.5, 1 m M EDTA, 1 m M dithiothreitol, 0.5 m M phen- ylmethylsulfonyl fluoride, 50 units/ml aprotinin, 2 (cid:4) g/ml leupeptin, 2 (cid:4) g/ml pepstatin A, 0.1 m M banzamidine) and subjected to sonication using a Vibra cell (Sonics and materials). The cell debris was removed by centrifugation (15,000 (cid:3) g for 15 min). The [3- 13 C]Ala-labeled PLC- (cid:1) 1 PH domain-GST fusion proteins were purified using glutathi-one-Sepharose 4B (Amersham Biosciences) affinity chromatography, and the [3- 13 C]Ala-labeled PLC- (cid:1) 1 PH domain was consequently obtained by cleavage of the link between the PH domain and GST using thrombin (Sigma). The amino acid sequence of the wild type PLC-

Pleckstrin homology (PH) 1 domains are well defined structural modules of about 120 amino acid residues (1, 2) mainly found in proteins involved in cellular signaling and cytoskeletal functions (3)(4)(5)(6). It has been proposed that these domains function as mediators of intermolecular interactions analogous to many other structural modules involved in cellular signaling (e.g. SH2 and SH3 domains). Many kinds of inositol lipids and inositol phosphates have been identified as important ligands of PH domains (3)(4)(5)(6), and, in some cases, the PH domains also interact with other proteins and mediate protein-protein interactions (5).
Because functionally important intermolecular interactions of PLC-␦1 with its ligand, PtdIns(4,5)P 2 , or other proteins included in the signal transduction pathways (e.g. transglutaminase II (G␣ h )) take place at the membrane surface (12)(13)(14), structural information of PLC-␦1 at the membrane surface is indispensable for understanding the molecular mechanism underlying the functions of PLC-␦1. The conformation and dynamics of peripheral membrane proteins at the lipid bilayer surface are expected to be different from those in solution, due to intermolecular interactions between the protein and lipids, changes in pH and ionic strength induced by surface charges of the membrane, and drastic changes in the dielectric constant at the lipid bilayer surface.
Despite the importance of the PLC-␦1 structure at the lipid bilayer surface, there is virtually no structural information about the peripheral membrane proteins at the membrane surface, due to a lack of means of investigating the molecular structure of proteins at the lipid bilayer surface at atomic resolution. In this study, we applied solid state NMR as a novel method of gaining insights into atomic level structural information of a peripheral membrane protein at the membrane surface under conditions similar to those of natural membranes. Solid state NMR is a highly suitable technique for this purpose, because it can provide information about the conformation and dynamics of individual amino acid residues in an intact protein under a wide variety of conditions, including those in a protein-lipid vesicle complex suspended in buffer at ambient temperature. By metabolic introduction of carbon-13labeled alanine residues into a protein as NMR probes, the local conformation and dynamics of selectively labeled amino acid residues are readily analyzed (15)(16)(17)(18)(19). Here, we applied this "site-directed" high resolution solid state 13 C NMR technique to the PH domain of PLC-␦1 as the first trial of a high resolution solid state NMR investigation of peripheral mem-brane proteins involved in lipid signal transduction pathways and obtained evidence for the conformational change of the PH domain at the membrane surface.
Protein Expression and Purification-The wild-type and mutated PLC-␦1 PH domains (1-140) were expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli (PR745) (20). Cells were grown in M9 medium (1), which contains 100 mg of each of 20 amino acids but with L-alanine replaced by L-[3-13 C]alanine and were incubated in the presence of 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside (5 h at 37°C). After centrifugation, the resulting cell pellets were resuspended in a buffer containing a mixture of protease inhibitors (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 50 units/ml aprotinin, 2 g/ml leupeptin, 2 g/ml pepstatin A, 0.1 mM banzamidine) and subjected to sonication using a Vibra cell (Sonics and materials). The cell debris was removed by centrifugation (15,000 ϫ g for 15 min). The [3-13 C]Ala-labeled PLC-␦1 PH domain-GST fusion proteins were purified using glutathione-Sepharose 4B (Amersham Biosciences) affinity chromatography, and the [3-13 C]Ala-labeled PLC-␦1 PH domain was consequently obtained by cleavage of the link between the PH domain and GST using thrombin (Sigma). The amino acid sequence of the wild type PLC-␦1 PH domain is shown in Fig. 1. Final preparations of the wild type and mutant PH domains included additional amino acid residues, GSRSTand -ELGPRPNWPTS, at the N and C termini of the natural amino acid sequence, respectively.
Measurement of Solid State 13 C NMR Spectra-High resolution solid state 13 C NMR spectra were recorded on a Varian Infinity 400 spectrometer ( 13 C: 100.6 MHz), using cross polarization-magic angle spinning (CP-MAS) and single pulse excitation dipolar decoupled-magic angle spinning (DD-MAS) methods. The spectral width, acquisition time, and repetition time for CP-MAS and DD-MAS experiments were 40 kHz, 50 ms, and 4 s, respectively. The contact time for the CP-MAS experiment was 1 ms. Free induction decays were acquired with 2,048 data points and Fourier-transformed as 32,768 data points after 30,720 data points were zero-filled. The /2 pulses for carbon and protons were 5.0 s, and the spinning rates were 4 kHz. The dipolar decoupling field strength was 55 kHz unless indicated otherwise in the text. Transients were accumulated 20,000 -40,000 times until a reasonable signal-tonoise ratio was achieved. The 13 C chemical shifts were referenced to the carboxyl signal of glycine (176.03 ppm from tetramethylsilane (TMS)) and then expressed as relative shifts from the TMS value.
Measurement of Dynamic Light Scattering-Dynamic light scattering of the wild-type PH domain-Ins(1,4,5)P 3 complex in 10 mM MES buffer (pH 6.5) containing 25 mM NaCl, 10% glycerol, and 0.025% NaN 3 was measured at 20°C using a Dyna Pro dynamic light scattering/ molecular sizing instrument (Protein Solutions). The molecular weights of the monomeric or oligomeric PLC-␦1 PH domain particles were estimated from the particle sizes by assuming the particles had a spherical shape. Fig. 2 (A and B) shows NMR spectra of the [3-13 C]Ala-labeled PH domain-Ins(1,4,5)P 3 complex in solution measured using the single pulse excitation dipolar decoupled-magic angle spinning (DD-MAS) method using different amplitudes of magnetic fields for dipolar decoupling: 19 and 55 kHz. Five narrow peaks resonating at 14.55, 15.35, 15.77, 17.49, and 18.46 ppm in Fig. 2A, which are unaffected by dipolar decoupling field strength, are ascribed to the 13 C-labeled side-chain methyl groups of five alanine residues in the monomeric or oligomeric PH domain. The dipole-dipole interactions between 1 H and 13 C nuclei that cause serious increases in the line width of 13 C signals are decoupled by fast isotropic rotational motions of the small PH domain particles. A peak at 11.41 ppm, indicated by an asterisk, is thought to be an artifact because of the lack of reproducibility. An obvious increase in line width of the peak at 16.79 ppm under the weaker decoupling field ( Fig. 2A) indicates that this signal arises from larger complexes that undergo slow rotational motions with a frequency lower than 10 5 Hz, that of 1 H-13 C dipolar interactions. Dynamic light scattering measurement indicated that the [3-13 C]Ala-labeled PH domain-Ins(1,4,5)P 3 complex forms two components with different particle sizes in solution. A major component consists of particles with a hydrodynamic radius of 1.5-5 nm and average molecular mass between 19 and 50 kDa. A minor component consists of particles with a hydrodynamic radius greater than 10 nm and an average molecular mass higher than 5 MDa. The former component is thought to comprise the monomeric, dimeric, and/or trimeric PH domains, and the latter aggregated represents clusters of PH domains consisting of a large number of molecules. These results support the above-mentioned assignments of the signals to particles with different sizes. Although the large aggregated clusters undergo slow rotational motions, these motions are sufficiently fast to eliminate most of the dipole-dipole interaction between 1 H and 13 C nuclei that is required to form magnetization through cross-relaxation between 1 H and 13 C nuclei, because no signal was observed using the cross polarization-magic angle spinning (CP-MAS) method (data not shown).

NMR Spectra of [3-13 C]Ala Residues in the PLC-␦1 PH Domain-Ins(1,4,5)P 3 Complex-
Studies of synthetic polypeptides, structural proteins, and membrane proteins have shown that the chemical shift of the C ␤ carbon in the Ala residue in high resolution solid state 13 C NMR is primarily determined by the torsion angles (,) of the main chain of the residue itself (21)(22)(23). The narrow line widths and the symmetric line shape of the peaks from the smaller particles in Fig. 2A reveal that the conformation of the Ala residues in the monomeric PH domain and dimer and/or trimer, if any, observed by NMR are identical, showing no structural variation that causes displacement of chemical shifts. The buffer composition of the [3-13 C]Ala-labeled PH domain-Ins(1,4,5)P 3 complex solution was the same as the crystallization buffer used in the x-ray diffraction study in which the three-dimensional structural model of the PH domain-Ins(1,4,5)P 3 complex was determined (11). Because the secondary structures of the PH domain-Ins(1,4,5)P 3 complex in solution is expected to be similar to those of the complex in the crystal, the narrow peaks observed in the 13 C NMR spectra of the PH domain-Ins(1,4,5)P 3 complex ( Fig. 2A) were assigned on the basis of the conformational dependence of chemical shift and the three-dimensional model structure of the PH domain-Ins(1,4,5)P 3 complex. The chemical shifts of the C ␤ carbons of Ala residues contained in the typical ␣-helix (14.9 ppm) and ␤-sheet (19.9 ppm) of poly L-alanine and random coil (16.9 ppm) of the C terminus of bacteriorhodopsin (16,24) Fig. 2A, similar to the chemical shifts of the typical ␣-helix attributed to three Ala residues, Ala-21, Ala-116, and Ala-118, contained in the ␣ helices at the N and C termini of the PH domain. The peak at 18.46 ppm can be ascribed to the Ala-112 located at the C terminus of the ␤7 strand of the PH domain based on the chemical shift similar to that of the ␤-sheet. The peak at 17.49 ppm, whose chemical shift does not correspond to any secondary structure, is ascribed to Ala-88 contained in the loop between the ␤5 and ␤6 strands (the ␤5/␤6 loop). These assignments of the peaks were found to be consistent with the assignments of the NMR spectra of the PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complex based on site-directed replacements of the Ala residues as described below. The chemical shift of the peak arising from the large aggregated particle, 16.79 ppm, agrees with the chemical shift of the C-terminal random coil of bacteriorhodopsin. This signal presumably comes from the mobile random coil moiety of the large denatured aggregate.
High Resolution Solid State NMR Spectra of the PH Domain-PtdIns(4,5)P 2 Complex- Fig. 3 (A and B) show DD-MAS and CP-MAS NMR spectra of the [3-13 C]Ala-labeled PH domain forming a complex with PtdIns(4,5)P 2 in the PtdCho/ PtdIns(4,5)P 2 liposome suspended in buffer solution at 20°C. The chemical shifts of the signals from the [3-13 C]Ala-labeled PH domain-Ins(1,4,5)P 3 complex are shown by vertical bars at the bottom of the spectra. The signals from the lipid molecules are indicated by asterisks. The relative intensity of the lipid signals in the CP-MAS spectrum (Fig. 3B) is lower than that in the DD-MAS spectrum (Fig. 3A) because of the lower efficiency of magnetization formation through the cross-polarization process due to the high mobility of the lipid molecules. Chemical shifts of the peaks at 14.41, 15.37, and 15.83 ppm in the CP-MAS NMR spectrum (Fig. 3B) Fig. 2A). In contrast to these signals, the signals resonated at 17.49 and 18.46 ppm in the spectra of the PH domain-Ins(1,4,5)P 3 complex ( Fig. 2A) show an upfield displacement to 16.99 ppm and a downfield displacement up to 19.1 ppm, respectively. To assign these signals, mutant PH domains in which Ala residues are replaced by Gly or Leu residues are prepared. Fig. 4 (A and B) show the CP-MAS NMR spectra of the [3-13 C]Ala-labeled A112G mutant PH domain in which Ala-112 is replaced by Gly and A88G mutant PH domain in which Ala-88 is replaced by Gly forming complexes with PtdCho/PtdIns(4,5)P 2 vesicles, respectively. The signals resonating between 18.5 and 19.1 ppm (Figs. 3B and 4A) were assigned to Ala-112 based on the disappearance of these peaks in the spectrum of A112G. The replacement of Ala-112 by Gly induced downfield displacement of the peak at 16.99 ppm in the CP-MAS spectrum of the wild type PH domain (Fig. 3B) to 17.50 ppm (Fig. 4B). The peaks at 15.37 and 15.83 ppm in the CP-MAS NMR spectrum of the wild type PH domain (Fig. 3B) were also shifted in the range of 15.51 and 16.40 ppm (Fig. 4B). Fig. 4C shows the CP-MAS NMR spectrum of the [3-13 C]Ala-labeled A88G mutant PH domain. The peak at 16.99 ppm in the CP-MAS spectrum of the wild type PH domain (Figs. 3B and 4A) was assigned to Ala-88, because this peak disappeared in the spectrum of A88G (Fig. 4C). Fig. 5 (A-C) shows the DD-MAS NMR spectra of A116L, A118G, and A21L mutant PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complexes, respectively. The signals resonating at 15.83 and 15.40 ppm were ascribed to Ala-116 and Ala-118 based on the disappearances of the peaks indicated by arrows in the spectra of A116L (Fig. 5A) and A118G (Fig. 5B), respectively. As shown by a closed triangle in Fig. 5B, a strong suppression of the signal of Ala-112 was induced by the replacement of Ala-118 by Gly. This suppression would be caused by decrease in an efficiency of the dipole decoupling due to an interference between the frequency of the dipole decoupling field (55 kHz) and a newly induced thermal motion of the Ala-112 residue at a frequency around 10 4 -10 5 Hz (25). The removal of the side-chain methyl group of Ala-118 could facilitate such a thermal motion of Ala-112, because the side chain of Ala-118 is in van der Waals contact with the side chain of Ala-112 in the three-dimensional model structure. Fig. 5C shows the DD-MAS spectrum of A21L mutant PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complex. A remarkably intense signal at 16.9 ppm indicated by an open triangle was ascribed to the random coil structure of the denatured PH domain included in the aggregated cluster similar to that observed for the PH domain-Ins(1,4,5)P 3 complex (Fig. 2). The replacement of Ala-21 might reduce a stability of the PH domain. Because the peaks at 15.83 and 15.40 ppm remained intact in the spectrum of A21L, the peak resonating at 14.41 ppm in the CP-MAS spectrum of the wild type PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complex (Fig. 3B) could be ascribed to Ala-21, although an effort of direct assignment from a removal of the peak by the replacement of Ala-21 was unsuccessful due to the intense signal of the methyl carbon of the lipid resonated at 14.20 ppm either in the DD-MAS or CP-MAS spectra of the A21L mutant PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complex.  (Fig. 2, A and B). The broad signal resonating between 18.5 and 19.4 ppm in the DD-MAS and CP-MAS NMR spectra of the PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complex (Fig. 3, A and B), corresponding to the peak of the PH domain-Ins(1,4,5)P 3 complex at 18.46 ppm ( Fig. 2A), is attributed to the coexistence of a variety of different conformations of Ala-112 at the C terminus of the ␤7 strand. Notably, the higher relative signal intensity of the peak at 18.82 ppm in the CP-MAS NMR spectrum (Fig. 3B) compared with that in the DD-MAS spectrum (Fig. 3A) indicates that the Ala-112 taking the conformation corresponding to this peak is highly immobile. Fig. 6 (A and B) shows the DD-MAS and CP-MAS NMR spectra of the PH domain-PtdCho/PtdIns(4,5)P 2 vesicle complex, respectively, at 4°C (solid trace) and 20°C (dotted trace). In the CP-MAS NMR spectrum at 4°C, a peak at 14.27 ppm originating from the phospholipid methyl groups is enhanced due to the improved cross-polarization efficiency caused by decrease in mobility of the lipid molecules at 4°C (Fig. 6B). In the DD-MAS spectrum at 4°C, the relative intensity of the Ala-112 signal increased to form a new peak at 18.78 ppm (Fig.   6A). In the CP-MAS spectrum at 4°C, the line width of the Ala-112 signal resonating at 18.77 ppm is narrower than that in the spectrum at 20°C (Fig. 6B). These changes indicate that the conformation of Ala-112 corresponding to the chemical shift of 18.77-18.78 ppm, which is most immobile at 20°C, becomes dominant at 4°C. DISCUSSION In this study, we aimed to gain insights into the structural features of the PH domain at the lipid bilayer surface when the domain forms a complex with PtdIns(4,5)P 2 . The PH domain of PLC-␦1 forms a high affinity complex with either PtdIns(4,5)P 2 or Ins(1,4,5)P 3 with comparable binding constants (10) through a specific interaction between the phosphoinositol group and the side chains located at the basic surface of the PH domain (e.g. the ␤1/␤2 and ␤3/␤4 loops) (11). There is a possibility, however, that additional interactions between the PH domain and the lipid bilayer contribute to the characteristics of the PH domain at the membrane surface, such as binding affinity, conformation, mobility, and orientation. Such nonspecific interactions might be responsible for the dependence of binding affinity of the PH domain to inositol compounds on assay conditions, such as pH, ionic composition of buffer, or lipid composition of membrane (4). Possible candidates for such interactions are electrostatic interactions between the positively charged surface of the PH domain, which is not involved in the specific interaction with PtdIns(4,5)P 2 and the lipid head groups, and hydrophobic interactions between the hydrophobic surfaces of the PH domain and the hydrophobic inner layer of the lipid bilayer. These interactions, if any, would modify the conformation, dynamics, and orientation of the PH domain at the lipid bilayer surface, which in turn, would influence lipidprotein and protein-protein interactions at the membrane surface involved in signal transduction pathways. In the case of multidomain proteins such as PLC-␦1, which contains the PH domain, EF-hands, an active site domain and a C2 domain, intramolecular interaction between the domains would also be affected by the conformational characteristics of the domains at the membrane surfaces.

FIG. 3. DD-MAS (A) and CP-MAS (B) NMR spectra of the [3-13 C]Ala-labeled PLC-␦1 PH domain-PtdIns(4,5)P
As shown in Fig. 3, changes in the 13 C NMR spectra of the [3-13 C]Ala-labeled PLC-␦1 PH domain clearly revealed conformational changes induced by localization of the domain at the surface of the PtdCho/PtdIns(4,5)P 2 vesicle. The changes in the PH domain structure were found to occur in the vicinity of the ␤5/␤6 loop containing Ala-88 and the C terminus of the ␤7 strand containing Ala-112. In contrast, the N-and C-terminal ␣-helices located at the surface opposite to the ligand binding site of the PH domain showed virtually no conformational change.
The side chains of Leu-84 and Ala-88 in the ␤5/␤6 loop and Ile-111 in the ␤7 strand form a hydrophobic cluster between the ␣-helix in the ␤5/␤6 loop (␣2-helix; Fig. 7A) and the ␤7 strand according to the three-dimensional model structure of the PH domain-Ins(1,4,5)P 3 complex (11). The proximity of Ala-88 and Ala-112 in the model structure suggests a relationship between these residues in the conformational changes. The origin of the upfield displacement of the Ala-88 signal is ascribed to a direct interaction between the ␣2-helix and the membrane, considering the close location of the ␤5/␤6 loop to the positively charged lipid-binding surface of the PH domain (Fig. 7A). Taking into account the highly amphipathic nature of the ␣2-helix (as shown in Fig. 7B by a helical wheel), it is plausible to predict a hydrophobic interaction between the hydrophobic face of the ␣2-helix and the hydrophobic inner layer of the lipid bilayer. In fact, x-ray diffraction studies of membrane binding states of synthetic model peptides (26,27) have suggested that such a hydrophobic interaction facilitates the orientation of an am- phipathic ␣-helix parallel to the membrane plane at the interface between the layer of the polar head groups and the inner layer of the membrane. The reorientation of the ␣2-helix is thought to cause an opening of the hydrophobic cluster between the ␣2-helix and the ␤7 strand (Fig. 7C). As a result, hydrophobic interaction between the newly exposed hydrophobic surface and the inner layer of the lipid bilayer may induce tilted orientation of the PH domain (Fig. 7C, right). The downfield displacement of the Ala-112 signal to the chemical shift of the typical ␤ sheet, together with the reduction of mobility at the membrane surface, suggests the formation of an additional hydrogen bond at the C terminus of the ␤7 strand between the carbonyl group of Ala-112 and the amide group of Arg-95. The formation of a new hydrogen bond elongates the anti-parallel ␤ sheet structure of the ␤6 and ␤7 strands and stabilizes it (Fig.  8). This conformational change is consistent with the abovementioned "open and tilt" model. The opening of the ␤5/␤6 loop would result in the breakage of a hydrogen bond between the carbonyl group of Pro-92 and the amide group of Arg-95 contained in the ␤ turn connecting the ␤5/␤6 loop and the ␤7 strand. As shown in Fig. 8, the rupture of this hydrogen bond is expected to facilitate formation of a hydrogen bond between the amide group of Arg-95 and the carbonyl group of Ala-112. Because only changes of three torsion angles, of Asp-94 and and of Arg-95, are required for the reorientation of the amide group of Arg-95, this reconstitution of the hydrogen bond is expected to occur readily. The chemical shift of the Ala-112 signal at 4°C (18.77-18.78 ppm; Fig. 6) reflects the conformation with the lowest energy at the membrane surface that might include the hydrogen bond between Arg-95 and Ala-112. Another reason for the origin of the conformational change at the C terminus of the ␤7 strand is interaction between the ␤6/␤7 loop located at the positively charged surface of the PH domain and the polar head groups of the lipids. Because the ␤6/␤7 loop does not contribute to the specific interaction between the PH domain and Ins(1,4,5)P 3 in the model structure, except by an indirect hydrogen bond between the carbonyl group of Thr-107 and the 4-phosphate of Ins(1,4,5)P 3 , nonspecific electrostatic interactions between the charged side chains of this loop (Lys-102, Asp-103, and Arg-105) and the polar head groups of phospholipids are likely to occur at the membrane surface. A variety of these nonspecific interactions might induce a variety of conformations of Ala-112 at the C terminus of the ␤7 strand through conformational change of the main chains of the ␤6 and ␤7 strands.
The nonspecific hydrophobic or electrostatic interactions described above would contribute to the affinity of the PLC-␦1 PH domain to PtdIns(4,5)P 2 in the membrane. These auxiliary mechanisms for membrane binding are probably susceptible to assay conditions, such as the structures of the hydrophobic acyl-chains of the lipids, surface charges of the membrane, pH value, and ionic strength of the buffer (4). For example, although it has been reported that the K d value of the PLC-␦1 PH domain-PtdIns(4,5)P 2 interaction measured using dimyristoyl phosphatidylcholine (DMPC)-PtdIns(4,5)P 2 vesicles is 1.66 M, eight times larger than that of the PLC-␦1 PH domain-Ins(1,4,5)P 3 interaction (210 nM) (10), the affinity of the PLC-␦1 PH domain to PtdIns(4,5)P 2 in the natural membrane could be different and might be higher than that observed for the DMPC vesicle system due to the presence of unsaturated acyl chains.
Although more detailed structural information about the membrane binding state of the PH domain is required to judge the validity of these models, the results of the solid state NMR experiments clearly indicate that the PLC-␦1 PH domain has an unique conformation and dynamics at the lipid bilayer surface, which are different from those in solution. The structural information at the membrane surface is indispensable for gaining insights into the molecular mechanisms of the functions of peripheral membrane proteins. Our results also proved that FIG. 7. The possible conformational change of the PLC-␦1 PH domain at the membrane surface. A, a schematic representation of the PLC-␦1 PH domain-Ins(1,4,5)P 3 complex, based on the three-dimensional model structure determined by x-ray diffraction study (11). The ␤7 strand in contact with the ␣2-helix is highlighted in yellow. The positions of the Ala residues are indicated as green circles. B, the ␣2-helix is represented as a wheel viewed from the C terminus. Residues with hydrophobic and polar side chains are colored blue and red, respectively. C, a schematic model of the changes in conformation and orientation of the PLC-␦1 PH domain induced at the membrane surface. The hydrophobic and hydrophilic surfaces of the amphipathic ␣2-helix are colored blue and red, respectively. The ␤7 strand is colored blue. The lipid bilayer is colored a purple shade. The re-orientation of the ␣2-helix induces hydrophobic interaction between the hydrophobic surfaces of the ␣2-helix and ␤7 strand and the hydrophobic inner layer of the membrane that might cause a tilt of the PH domain.