HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 22, 20059-20069, May 31, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/22/20059    most recent M200263200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Matoba, Y. Articles by Sugiyama, M. Search for Related Content PubMed PubMed Citation Articles by Matoba, Y. Articles by Sugiyama, M. Social Bookmarking                      What's this?

The Crystal Structure of Prokaryotic Phospholipase A₂^*

Yasuyuki Matoba, Yukiteru Katsube^§, and Masanori Sugiyama^¶

From the  Institute of Pharmaceutical Sciences, Faculty of Medicine, Hiroshima University, Kasumi 1-2-3, Minami-ku, Hiroshima 734-8551 and the ^§ Institute of X-ray Research, Rigaku Co., 3-9-12, Matsubara-cho, Akishima, Tokyo 196-0003, Japan

Received for publication, January 10, 2002, and in revised form, February 27, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

In this study, the x-ray crystal structures of the calcium-free and calcium-bound forms of phospholipase A₂ (PLA₂), produced extracellularly by Streptomyces violaceoruber, were determined by using the multiple isomorphous replacement and molecular replacement methods, respectively. The former and latter structures were refined to an R-factor of 18.8% at a 1.4-Å resolution and an R-factor of 15.0% at a 1.6-Å resolution, respectively. The overall structure of the prokaryotic PLA₂ exhibits a novel folding topology that demonstrates that it is completely distinct from those of eukaryotic PLA₂s, which have been already determined by x-ray and NMR analyses. Furthermore, the coordination geometry of the calcium(II) ion apparently deviated from that of eukaryotic PLA₂s. Regardless of the evolutionary divergence, the catalytic mechanism including the calcium(II) ion on secreted PLA₂ seems to be conserved between prokaryotic and eukaryotic cells. Demonstrating that the overall structure determined by x-ray analysis is almost the same as that determined by NMR analysis is useful to discuss the catalytic mechanism at the molecular level of the bacterial PLA₂.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Phospholipase A₂ (PLA₂¹; EC 3.1.1.4) is an enzyme that liberates fatty acid together with lysophospholipid by hydrolyzing the 2-ester bond of 1,2-diacyl-3-sn-phosphoglycerides. The enzyme, which has been found only by eukaryotic cells, can be broadly divided into two categories, namely secreted and cytosolic types (1). In recent years, a proposal for classification has been made after many types of PLA₂ in each category were found. According to this classification, secreted PLA₂s are grouped into classes I, II, III, V, IX, X, XI, and XII (2, 3). Classes IV, VI, VII, and VIII are included in cytosolic PLA₂ (2). We have discovered a bacterial PLA₂ of the secreted type and propose to establish a new class, which is described in the accompanying paper (41).

All secreted PLA₂s require calcium(II) ion for the expression of enzymatic activity. Interestingly, the kinetic study of the S. violaceoruber PLA₂ with respect to the dissociation constant for calcium(II) ion suggested that the binding affinity of the ion for the prokaryotic PLA₂ is 1 order of magnitude lower than that for the eukaryotic enzyme (41). In addition, the bacterial PLA₂ prefers phosphatidylcholine as a substrate to phosphatidylethanolamine. The primary structure of the prokaryotic PLA₂, except for residues Cys⁶¹-Tyr⁶⁸, is distinct from those of eukaryotic secreted PLA₂s that have already been analyzed for the tertiary structure. As a striking difference, eukaryotic PLA₂s have 6-8 disulfide bonds, whereas the bacterial enzyme has only two disulfide bonds. Over 150 primary structures of PLA₂ have been already determined, and the crystal (4-10) and NMR structures (11-13) of several PLA₂s, such as bovine and porcine pancreatic PLA₂s, snake and bee venom PLA₂s, and human synovial fluid PLA₂, have been solved.

Here, we show the crystal structure of the calcium-free S. violaceoruber PLA₂, which was determined by using the multiple isomorphous replacement methods and refined at a high resolution of 1.4 Å. Furthermore, we determined the crystal structure of the calcium-bound form by the molecular replacement method and refined it at a 1.6-Å resolution. This is the first report disclosing the x-ray crystal structure of prokaryotic PLA₂. The x-ray crystal structures of these two forms of the S. violaceoruber PLA₂ were compared with those by the NMR technique and also with the tertiary structures of eukaryotic PLA₂s. The x-ray crystal structure of the calcium-bound S. violaceoruber PLA₂ is concrete evidence that the catalytic hydrogen-bonding network is essentially flexible even in the crystal structure. We believe that this observation will aid in understanding the molecular mechanism for the catalysis of the secreted PLA₂s.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

X-ray Crystallography of the Calcium-free S. violaceoruber PLA₂-- The S. violaceoruber PLA₂ was purified to homogeneity and lyophilized as described previously (41). Before crystallization, the lyophilized protein, dissolved in a 10 mM Tris-HCl buffer, pH 8.0, containing 20 mM CaCl₂, was adjusted to 20 mg/ml as a final concentration. Initial crystals were obtained by the vapor diffusion method at 297 K using the hanging drop technique (14). Drops containing 5 µl of the protein solution and 5 µl of the reservoir solution were equilibrated against the reservoir solution. The reservoir solution contained 14% (w/v) polyethylene glycol 6000, 0.2 M Li₂SO₄, and 0.1 M sodium cacodylate at pH 6.0. However, the crystals were heavily twinned and dissolved at 292 K. Therefore, after the crystals were completely dissolved, the drops were placed again at 297 K. The resulting single crystals, which grew in 2 weeks to a size of 0.3 × 0.3 × 1.2 mm, were suitable for the diffraction analysis. These crystals belong to the monoclinic space group P2₁ with unit cell dimensions a = 29.4 Å, b = 57.8 Å, c = 31.7 Å, and  = 111.5° with one molecule per asymmetric unit. After the structure determination, it was revealed that no calcium ion bound to the protein.

All x-ray data were collected at room temperature. Initially, the native data to 2.0-Å resolution was collected using an R-AXIS IIc imaging plate with graphite monochromated CuK radiation from the rotating anode generator RU-200 run at 40 kV and 100 mA. Each oscillation frame was taken using a rotation of 3° for 20 min at a crystal-to-detector distance of 80 mm. The diffraction intensities were integrated and scaled using the Process program (15). A nearly complete data set (94.6%) to a 2.0-Å resolution with high redundancy was collected from a single crystal. This native data set was used for the phase estimation at lower resolution and the preliminary refinement of the constructed model.

Higher resolution data to 1.4 Å were later collected using the same detector and the same x-ray source. The detector 2 angle was set to 35°, and the distance between its center and the crystal was set to 85 mm. Each oscillation frame was taken using a rotation of 1.5° for 15 min. In the highest resolution bin between 1.45 and 1.40 Å, this data set was 37.4% complete with a signal-to-noise ratio (I/(I)) of 4.52. Details of the data collection statistics are summarized in the top of Table I.

The phase problem was solved to 2.7-Å resolution using the PHASES program (16). Heavy atom derivatives were prepared by soaking the crystal in the reservoir solution containing each heavy atom reagent. Three derivatives (Pt, Au, and Rh) were used for the phasing (middle of Table I). The position of the primary Pt site was easily found in a difference Patterson map. The resulting single isomorphous replacement phases were used to solve other sites of the derivatives. The heavy atom parameters were refined, and a phase set with a figure of merit of 0.74 was obtained. Anomalous scattering information of the Pt derivative was utilized in this process, and anisotropic temperature factors were used for each heavy atom.

A density modification with an extension to a 2.0-Å resolution was performed by the solvent-flattening and histogram-matching method using the SQUASH program (17). The molecular boundary was determined in the electron density map applying a value of the solvent content of 20%, which is smaller than the value calculated on the basis of one molecule per asymmetric unit (33%). The resulting Fourier map enabled us to trace the main chain easily, and the difference map calculated from the native anomalous differences and the obtained phases confirmed the location of two disulfide bonds and one sulfur atom of the Met residue.

An initial model was built into this modified map using the program Xfit in the XtalView software package (18). This model gave an R-factor of 39.8% for the reflections from 10.0- to 3.0-Å resolution. Refinement of the model was performed by a combination of simulated annealing (19) and conventional restrained refinement methods (20) using the X-PLOR program (21). A subset of 10% of the reflections was used to monitor the free R-factor (R_free) (22). The refinement began with data from 10.0- to 3.0-Å resolution, and the upper limit was raised to a 2.0-Å resolution. Each refinement cycle included the refinement of positional parameters and individual isotropic temperature factors, the revision of the model using the omit map, and the addition of the solvent molecules. When the R-factor for the reflections from 10.0- to 2.0-Å resolution fell below 20%, the native data was changed to the 1.4-Å resolution data. Further rebuilding, the addition of the solvent molecules, and the X-PLOR refinement yielded the current model (bottom of Table I). The root mean square deviations from the ideal geometry (23) are 0.008 Å in bond length, 1.9° in bond angle, and 1.13° in improper angle.

X-ray Crystallography of the Calcium-bound S. violaceoruber PLA₂-- For crystallization, the protein solution was prepared as described previously. Initial crystals were obtained by the vapor diffusion at 297 K using the hanging drop method (14). Drops containing 5 µl of the protein solution and 5 µl of the reservoir solution were equilibrated against 1.0 ml of reservoir solution. The typical reservoir solution contained 50-60% (v/v) 2-methyl-2,4-pentanediol and 0.1 M Tris-HCl at pH 8.5. However, the obtained crystals were heavily twinned. To obtain single appropriate crystals, repeated seeding was performed using the crushed crystals as seeds. Single crystals suitable for the diffraction analysis (0.1 × 0.5 × 2.0 mm) grew in 1 week. They belong to the space group P2₁ with unit cell dimensions a = 38.3 Å, b = 54.3 Å, c = 30.6 Å, and  = 90.2° with one molecule per asymmetric unit. These preliminary crystallographic data have been published (24).

X-ray analysis was performed at room temperature using an R-AXIS IIc imaging plate with mirror monochromated CuK radiation from the RU-300 rotating anode generator run at 40 kV and 100 mA. The crystal was mounted with the crystallographic a* axis parallel to the crystal rotation axis. The crystal-to-detector distance was set to 60 mm. Each frame of 2.5° crystal oscillation was taken for 12 min. Diffraction spots in a rotation range of 142.5° were recorded on a total of 57 frames. The diffraction intensities to a 1.6-Å resolution were integrated and scaled by the Process program (15). In the highest resolution bin between 1.7- and 1.6-Å resolution, this data set is 48.0% complete with a signal/noise ratio (I/(I)) of 2.41 and an R_merge of 28.3%. Details of the data collection statistics are shown in Table I (top).

The crystal structure of the calcium-bound S. violaceoruber PLA₂ was solved by the molecular replacement procedure using the programs in X-PLOR (21). The calcium-free form structure was used as search probe in a cross-rotation function against the data in a resolution range of 10.0- to 4.0-Å. A single peak at Euler angles ₁ = 183.3°, ₂ = 72.5°, and ₃ = 30.8° was 8.0  above the mean. A translational search was carried out in two dimensions (x and z). A top peak (x = 0.214, z = 0.425 in fractions of the unit cell) was observed at 7.2  above the mean.

Refinement was first performed using the X-PLOR program (21). Atomic coordinates, obtained by the molecular replacement method, were refined against the data between 10.0 and 4.0 Å for 20 cycles with the entire molecule as a rigid body, resulting in a crystallographic R-factor of 35.8%. Atomic refinement of the model was performed by using a combination of simulated annealing (19) and conventional restrained refinement methods (20). A subset of 10% of the reflections was used to monitor the R_free (22). The refinement began with data from 10.0- to 3.0-Å resolution, and the upper limit was raised to a 1.6-Å resolution. Each refinement cycle included the refinement of positional parameters and individual isotropic temperature factors. After each refinement step, 2F_o  F_c and F_o  F_c electron density maps were computed. The molecular modeling program Xfit in the XtalView program suit (18) was used on the Silicon Graphics workstations for visualizing and rebuilding the model. Several cycles of the X-PLOR refinement resulted in decreasing both the R-factor and R_free to 20.1 and 26.5%, respectively, for the reflections from 10.0 to 1.6-Å resolution with F > 2 . After the convergence on X-PLOR, further refinement was performed using the SHELXL-97 program (25) against all of the reflections from 5.0- to 1.6-Å resolution with F > 0. Refinement statistics are given at the bottom of Table I. The root mean square deviations from the ideal geometry (23) are 0.008 Å in bond length, 1.9° in bond angle, and 1.13° in improper angle.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Quality of the Electron Density Map and the Current Model-- As shown in Fig. 1, a and b, the final 2F_o  F_c maps of the hydrophobic core region in the calcium-free form and the calcium-binding site in the calcium-bound form structures exhibit very high quality. Almost all aromatic amino acid residues as well as Pro residues have a low density hole through the center of the rings, as expected from the high resolution maps. Mean coordinate errors were estimated by a Luzatti plot (26) to be 0.15 and 0.16 Å for the calcium-free and calcium-bound PLA₂s, respectively (data not shown). Although all side chain atoms in the calcium-free form are included in the final model, those of Gln⁷, Arg²⁸, Gln⁴⁷, Phe⁵³, Lys¹⁰³, and Trp¹¹² have poorly defined electron density. In the calcium-bound form, together with the side chain atoms of these 6 residues, those of Asn²⁹, Glu¹⁰², and Lys¹¹⁹ were poorly defined. The C-terminal main- and side-chain atoms in the calcium-bound form structure are flexible. Mainly, the difference of distribution of the flexible residues between calcium-free and calcium-bound forms seems to be caused by the differences of their crystal-packing effects, as described below.

View larger version (53K): [in this window] [in a new window]   Fig. 1.   A portion of the final 2Fo  Fc electron density map. a, a map of the hydrophobic core region in the calcium-free form structure at 1.4-Å resolution. The map was contoured at 1.2 . b, a map of the calcium-binding site in the calcium-bound form structure at 1.6-Å resolution. The map was contoured at 1.0 .

In both cases, the Ramachandran plot (27, 28) shows that only Leu⁴⁴ is observed in the forbidden regions. The unusual backbone conformation of Leu⁴⁴ is stabilized by a hydrogen bond formed between its backbone amide nitrogen and the Thr⁴² hydroxyl oxygen.

Overall Structure-- On the geometry of class I/II PLA₂s, an anti-parallel -sheet composed of two strands is present together with -helices accounting for 50% of the overall structure (4-7, 9, 10). Class III bee venom PLA₂ consists of three kinds of long -helices and a few -strands (8). However, the S. violaceoruber PLA₂ only has an -helical secondary structure consisting of five -helices and two helical segments (Fig. 2 and Fig. 3a). The current crystal structure model of the bacterial PLA₂ is almost the same as the NMR structure (41). To clarify the tertiary structure of bacterial PLA₂, we conveniently divided the structure into the N- and C-terminal domains; the former domain consists of two -helices (₁ and ₂) and two helical segments. The latter domain consists of three long -helices (₃, ₄, and ₅) and forms a three-helix bundle. The three -helices are tightly packed against each other with an unusual slightly right-handed twist. The angles among the three -helices are 11° (₃-₄), 18° (₄-₅), and 19° (₃-₅), respectively. Both domains are linked to a long loop consisting of residues Glu³⁷ to Phe⁵⁷. The ₃-helix contains a homologous segment consisting of the residues Cys⁶¹ to Tyr⁶⁸, and its orientation is almost anti-parallel to the ₄-helix. The nearly anti-parallel helices are also formed by ₄- and ₅-helices. These three -helices are approximately perpendicular to the ₂-helix.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Ribbon views of the calcium-free (a) and calcium-bound (b) S. violaceoruber PLA2. This model was created by the MOLSCRIPT program (39) and Raster3D (40). Helices or helical segments are in red, and the turn region is in blue. The yellow and green represent the disulfide bonds and the calcium(II) ion, respectively. The current structure has no -strands.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   a, amino acid sequence of the S. violaceoruber PLA2 and structurally homologous segments of typical eukaryotic PLA2s. i), S. violaceoruber; ii), N. naja atra PLA2 (class I/II); iii), bee venom PLA2 (class III). The catalytic and calcium-binding sites are indicated in red and blue, respectively. The secondary structures are defined by the MOLSCRIPT program (39). b, stereoview of the superposition of the -carbon backbone of the calcium-free S. violaceoruber PLA2 (in red) on those of PLA2s that are contained in N. naja atra (in blue) and bee (in green) venoms. The structures of N. naja atra and bee venom PLA2s are described in Refs. 7 and 8, respectively.

The -carbon backbone of the S. violaceoruber PLA₂ is superimposed on that of PLA₂ from Naja naja atra, the cobra snake (7), by the least squares fit of the catalytic residues (Fig. 3b). The tertiary structural similarity of backbone is limited to two long anti-parallel -helices containing the residues composing the catalytic site. A two-stranded -sheet, commonly found in the class I/II PLA₂s, is absent in the bacterial PLA₂. Although the C-terminal ₅-helix in the S. violaceoruber PLA₂ corresponds to the N-terminal -helix in the class I/II PLA₂s, its orientation is opposite. Similarly, the -carbon backbone of the S. violaceoruber PLA₂ is superimposed on that of PLA₂ from bee venom (8) (Fig. 3b). The orientation of three long -helices in the C-terminal domain of the bacterial PLA₂ is similar to that of PLA₂ from bee venom, whereas the overall structure of bacterial PLA₂ is obviously different from that of the bee venom PLA₂. The overall structure of the S. violaceoruber PLA₂ can be expressed as a shape in which the N-terminal domain of the bacterial PLA₂ is topologically added to a core region consisting of three -helices of the bee venom PLA₂.

Despite the difference of the crystal lattice, the crystal structure of the calcium-free form is identical with that of the calcium-bound form. Fig. 4a shows the superposition of the two forms using the main-chain atoms. The root mean square positional differences between the two structures are 0.93 and 1.10 Å for main-chain atoms and for all protein atoms, respectively. Marked structural deviations occur at the Cys⁴⁵Ala⁴⁸ residues in the long loop (Fig. 4b) and at the Lys¹¹⁹Gly¹²² residues in the C-terminal region (Fig. 4c). The former conformational change may be caused by binding of the calcium(II) ion to the protein, and the latter may reflect the potential flexibility of this region, as described below.

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Stereoviews of conformational differences between the calcium-free (dark gray) and calcium-bound (light gray) forms of the S. voolaceoruber PLA2. a, superposition of the overall structure. Every 10th residue and the N- and C-terminal residues are shown. Marked structural deviations occur at residues Cys45-Ala48 in the long loop (b) and the residues Lys119-Gly122 in the C-terminal region (c).

In the calcium-free S. violaceoruber PLA₂ structure, although calcium(II) ion is contained in a buffer to grow the crystal, the ion is absent in the model. This may be due to the fact that the crystal is grown in the presence of sulfate ion at near pH 6. Interestingly, even without calcium(II) ion, the current model maintains the conformational rigidity. The mean temperature factors are 11.0 Å² for all protein atoms and 9.8 Å² for the main-chain atoms. On the other hand, the mean temperature factors of the calcium-bound form structure increase to 30.1 Å² for all protein atoms and 28.0 Å² for the main-chain atoms. Judging from the fact that V_M (29) values are extremely low (1.85 Å³ Da¹) for calcium-free crystal and moderate (2.35 Å³ Da¹) for calcium-bound crystal, it can easily be speculated that the differences of mean temperature factors are mainly caused by the crystal packing effects. Despite the increase of its mean temperature factor, the calcium-bound form structure has a compact shape and enough rigidity when compared with other protein structures. Since eukaryotic PLA₂s commonly have conformational rigidity, it has been suggested that the rigidity is essential for the expression of catalytic activity (30, 31). This suggestion holds true for the calcium-free S. violaceoruber PLA₂ structure. However, the catalytic hydrogen-bonding network and the substrate-binding site are flexible in the calcium-bound form of the bacterial PLA₂. Interestingly, recent studies of eukaryotic PLA₂s using the NMR method (11-13) showed the flexible feature of the enzyme at a part of catalytic hydrogen-bonding network and the substrate-binding site. The flexibility in the crystal structure of the calcium-bound S. violaceoruber PLA₂ may be associated with these observations in the solution structures of eukaryotic PLA₂s.

The following point of the argument is the number of the disulfide bonds. The conformational rigidity of eukaryotic PLA₂s is partially accomplished by a high content of disulfide bonds (32). However, the S. violaceoruber PLA₂ has only two disulfide bonds. The salt bridges, hydrogen bonds, and hydrophobic interactions seen in the structure of the bacterial PLA₂ may be useful to maintain the rigidity.

Catalytic Site and Surrounding Hydrogen-bonding Network-- The overall structure of the S. violaceoruber PLA₂ obviously differs from those of eukaryotic PLA₂s. However, the geometry of the catalytic site is conserved between eukaryotic and prokaryotic cells (Fig. 5, a, b, and c). The catalytic site of the prokaryotic PLA₂ is stabilized by the formation of hydrogen-bonding networks (Fig. 5, a and b). A marked structural difference between the calcium-free and calcium-bound forms is the size of the catalytic hydrogen-bonding network. In the calcium-free form, the catalytic hydrogen-bonding network extends to the C-terminal region (Fig. 5a). However, the C-terminal region in the calcium-bound form does not participate in the catalytic hydrogen-bonding network (Fig. 5b). We first describe the catalytic site and the surrounding hydrogen-bonding network of the bacterial PLA₂ based on its calcium-free form structure.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Stereoviews of the catalytic site and surrounding hydrogen-bonding network. a and b, the calcium-free and calcium-bound forms of the S. violaceoruver PLA2, respectively; c, N. naja atra PLA2 (7). The hydrogen bonds are shown by black broken lines. The numbers in b and c show the distance (Å) between calcium(II) ion and the His64 N1 (the His48 N1 in N. naja atra).

The catalytic center of the bacterial PLA₂ is His⁶⁴, which corresponds to His⁴⁸ of the class I/II PLA₂s (Fig. 5c). An amino acid adjacent to His⁴⁸ is Asp⁴⁹, and its carboxylate functions as ligands for binding to the calcium ion. In the bacterial PLA₂, the residue adjacent to His⁶⁴ is also Asp⁶⁵, and its carboxylate is revealed to be the calcium ligands. A residue, Asp⁸⁵, found in the ₄-helix of the bacterial PLA₂, is a counterpart of Asp⁹⁹ of the class I/II PLA₂s (Fig. 5c), which forms a hydrogen bond with the His⁶⁴ N2 (His⁴⁸ in class I/II). The Asp⁸⁵ (Asp⁹⁹ in class I/II) may play a role for the neutralization of the positive charge of His⁶⁴ (His⁴⁸), formed during the catalytic reaction cleaving the ester bond of the substrate.

A water molecule (Wat) or a hydroxyl ion, which forms a hydrogen bond with the His⁴⁸ N1 in the class I/II PLA₂s, is inferred to attack nucleophilically the carbonyl carbon of the substrate. Wat²⁶⁰, observed in the calcium-free form of the bacterial PLA₂ structure, seems to be a nucleophilic water molecule. Thus, the geometry among Wat²⁶⁰, His⁶⁴, and Asp⁸⁵, which is bridged in line by the hydrogen bonds, is very similar to the catalytic triad of serine proteases. This geometry is commonly observed not only in the class I/II PLA₂s but also in the class III enzyme. The existence of the conserved catalytic geometry suggests that it may be essential for the expression of the catalytic activity in all secreted PLA₂s.

The hydrogen-bonding network extends from the catalytic residues and may help to stabilize the catalytic residues. The importance of the catalytic hydrogen-bonding network and the resulting stable conformation of the catalytic site and its surroundings is pointed out from the crystal structure analyses of other PLA₂s (30, 31). Around the catalytic site of Class I/II PLA₂s, Tyr⁵², and Tyr⁷³ are present, and their phenolic hydroxyls interact with the Asp⁹⁹ carboxyl oxygens (Fig. 5c). The residues Tyr⁵² and Tyr⁷³ interact with a carboxyl oxygen, forming a hydrogen bond to His⁴⁸ and with the other carboxyl oxygen, respectively. Residue Tyr⁶⁸ in the bacterial PLA₂, corresponding to the class I/II Tyr⁵², forms a hydrogen bond with Asp⁸⁵. A stereochemical analogous position of class I/II Tyr⁷³ hydroxyl oxygen is occupied by Wat²⁶³ linking to the C-terminal carboxylate and the Lys¹¹⁹ amino group of the S. violaceoruber PLA₂. In the crystal structures of class I/II PLA₂s, the catalytic hydrogen-bonding network extends to the surface loop through the N-terminal amino group. In the S. violaceoruber PLA₂, the amino group of Lys⁸¹ is linked by a salt bridge to the carboxyl oxygen of Asp⁸⁵. The positively charged Lsy⁸¹ amino group bonds to the carboxylate of the C-terminal Gly¹²² by a salt bridge and forms a hydrogen bond with the backbone carbonyl oxygen of Val¹¹⁸, which has the negative electrostatic potential generated by the -helical dipole effect, at the C terminus of the ₅-helix. In addition, the Asp⁸⁵ carboxyl oxygen links to the Lys¹¹⁹ amino group by the mediation of one water molecule (Wat²⁶³) (Fig. 5a). These interactions seem to be useful to stabilize the structure around the catalytic site.

Second, we describe the conformational difference at the catalytic site and its surroundings between calcium-free and calcium-bound form structures.

In the calcium-free form, a nucleophilic water molecule certainly exists and makes a linear hydrogen bond with His⁶⁴ N1 (Fig. 5a). However, near the position of the nucleophilic water molecule observed in the calcium-free form structure, there are no water molecules within hydrogen-bonding distance from the His⁶⁴ N1 in the calcium-bound form structure. Wat²⁰¹, which is one of the calcium ligands, exists at the analogous position to the nucleophilic water, although it cannot make a hydrogen bond with His⁶⁴ N1. Instead, Wat²⁵⁶, having a large temperature factor, forms a nonlinear hydrogen bond with His⁶⁴ N1 (Fig. 5b). Probably, Wat²⁵⁶ nucleophilically attacks the sn-2 carbonyl carbon of the substrate, and the resulting oxyanion of the intermediate ligates to the calcium ion at the analogous position of Wat²⁰¹.

In the calcium-bound form structure, the Asp⁸⁵ carboxyl oxygen forms hydrogen bonds with the Tyr⁶⁸ hydroxyl oxygen and the Wat²³⁷ and salt bridges with the Lys⁸¹ amino group. The Lys⁸¹ amino group further forms hydrogen bonds with the Tyr⁶⁸ hydroxyl oxygen and Val¹¹⁸ carbonyl oxygen (Fig. 5b). The C-terminal Lys¹¹⁹ and Gly¹²² residues in the calcium-free form participate in the hydrogen-bonding network (Fig. 5a). However, in the calcium-bound form, the C-terminal region is highly flexible, having large temperature factors, and does not interact with the catalytic residues. Similarly, the NMR structure of the bacterial PLA₂ indicates that the C-terminal region is highly flexible and makes no hydrogen bonds with the catalytic residues regardless of the presence or absence of the calcium ion (41). Therefore, the crystal structure of the calcium-bound form is thought to be identical with the NMR structure. All x-ray analyses for other eukaryotic PLA₂s have shown conformational rigidity by the catalytic hydrogen-bonding network. The structure of the calcium-bound form of the bacterial PLA₂ is the first evidence that the catalytic hydrogen-bonding network of the secreted PLA₂ is essentially flexible even in the crystal structure.

In the solution structure of the porcine pancreatic PLA₂, a research group has observed that the N-terminal region and surface loop are highly flexible. The formation of a hydrogen-bonding network extended from the catalytic site is incomplete when compared with the crystal structure (11). However, the solution structure of the porcine PLA₂ complexed with a competitive inhibitor and micelles showed that the N-terminal region and surface loop take stable conformations and form the complete hydrogen-bonding network, as seen in the crystal structure (12). From these results, the research group has concluded that the conformational change is caused by the existence of the aggregated substrates and results in the formation of the complete hydrogen-bonding network to stabilize the catalytic residues, as observed in the crystal structure (13). Furthermore, they suggested that this conformational change may explain the phenomenon that the activity of PLA₂ is much higher on aggregated substrates than on monomolecularly dispersed phospholipids, so-called interfacial activation (33).

The crystal structure of the bacterial calcium-free PLA₂ is significantly different at the C-terminal region with a solution structure (41). The crystal structure expects the C-terminal region to participate in the catalytic hydrogen-bonding network. These observations for the S. violaceoruber PLA₂ are quite similar to the results of porcine pancreatic PLA₂ (11-13). Therefore, we think that flexibility in the C-terminal region of the S. violaceoruber PLA₂ is caused by the lack of aggregated substrates, and that, in the case of the calcium-free form, the obtained crystal structure may be derived from the preferential crystallization of one protein conformer resembling the interfacial activated enzyme. However, in the calcium-bound form structure, the C-terminal region is highly flexible and does not interact with catalytic residues, as seen in its solution structure (41). Why does preferential crystallization not occur in the case of the calcium-bound form? The difference of the crystal packing effects between two crystal structures of calcium-free and calcium-bound forms may be a primary reason. In the calcium-free form structure, the C-terminal region has close interactions to the symmetry-related molecules, whereas, in the calcium-bound form, it protrudes to the bulk solvent region with no interactions to the symmetry-related molecules. Another possible explanation could be found in the difference of the crystallization pH. Atomic resolution analysis (our unpublished data)² of the calcium-free form structure revealed that a catalytic residue His⁶⁴ exhibits a positive charge in the crystal grown at pH 6.0. This protonation state is analogous to the intermediate state immediately after the nucleophilic water attacks the sn-2 carbonyl carbon of the substrate. However, in the calcium-bound form crystals, which were grown under pH 8.5 conditions, the His⁶⁴ residue must have no charge. It is reasonable that, when the charge condition of the catalytic residues resembles an intermediate state, a catalytic hydrogen-bonding network forms more easily. Therefore, the charge effect appears to cause the conformational difference at the C-terminal region between the calcium-free and calcium-bound form crystal structures.

Calcium-binding Site-- The S. violaceoruber PLA₂ possesses no calcium-binding sequence useful for the formation of a calcium-binding loop, X-Cys-Gly-X-Cys, which was found in the primary structure of eukaryotic PLA₂s (Fig. 3a). We could not even find any structure corresponding to the calcium-binding loop. The calcium-binding affinity of the bacterial PLA₂ is 1 order of magnitude lower than those of eukaryotic PLA₂s, suggesting that the lower affinity may be due to the absence of a calcium-binding loop. This also suggests that the bacterial PLA₂ must have a calcium-binding mechanism different from that of eukaryotic PLA₂s.

As shown in Fig. 1b, calcium(II) ion was clearly defined in the electron density. Judging from the relatively lower temperature factor of the calcium ion, it is thought that calcium ion is tightly bound to the protein. Calcium(II) ion is heptacoordinated by a pentagonal bipyramidal cage of oxygens (Fig. 6b) as well as other extracellular PLA₂s (6-9). However, its binding manner is quite different from that of other PLA₂s. In other PLA₂s, calcium ligands are two carboxyl oxygens of an invariant Asp residue, three backbone carbonyl oxygens of the highly conserved "calcium-binding loop," and two water molecules (6-9). Fig. 6c shows the calcium-binding site of N. naja atra PLA₂ (7) as an example. It is noted that two axial ligands are the Tyr²⁸ carbonyl oxygen and one of two bound waters (Wat²⁰²). On the other hand, in the S. violaceoruber PLA₂, the structural counterpart of the "calcium-binding loop" does not exist. As shown in Fig. 6b, the calcium-bound form structure reveals that calcium ligands are both carboxyl oxygens of Asp⁶⁵, one carboxyl oxygen of Asp⁴³, a carbonyl oxygen of Leu⁴⁴, and three water molecules, and that the axial ligands are both Asp⁴³ carboxyl oxygen and one of three bound waters (Wat²⁰¹).

View larger version (24K): [in this window] [in a new window]   Fig. 6.   Stereoviews of the calcium-binding site. a and b, the calcium-free and calcium-bound forms of the S. violaceoruver PLA2, respectively; c, N. naja atra PLA2 (7). A green ball represents the calcium(II) ion. The hydrogen bonds and coordination bonds to the calcium (II) ion are shown by black broken lines.

The residue Asp⁶⁵ is adjacent to the catalytic residue His⁶⁴. The His-Asp primary sequence is conserved among all of the extracellular PLA₂s, and the Asp residue at this position contributes two carboxyl oxygens to the calcium ligation. Therefore, it was easily speculated from the amino acid sequence that Asp⁶⁵ of the S. violaceoruber PLA₂ contributes to the calcium ligation. The current model of the calcium-bound form shows clearly that two carboxyl oxygens of Asp⁶⁵ function as the calcium ligands. In the current calcium-bound form structure, the side chain of Asp⁶⁵ is stabilized by two hydrogen bonds, as in the calcium-free form structure. One of two proton donors is the hydroxyl of Thr⁴², and the other is the backbone amide of Asp⁴³. Probably due to these hydrogen bonds, the torsion angle of the Asp⁶⁵ side chain of the S. violaceoruber PLA₂ is considerably different from that of the corresponding Asp residue of other PLA₂s.

Similarly, because of the positional relation to Asp⁶⁵, it was easily speculated even from the calcium-free form structure that the Leu⁴⁴ carbonyl oxygen is one of the calcium ligands. The current calcium-bound form structure shows that this idea is sure. However, the Leu⁴⁴ carbonyl oxygen of the S. violaceoruber PLA₂ is an equatorial ligand to the calcium(II) ion, whereas the carbonyl oxygen of the corresponding residue of other PLA₂s (Tyr²⁸ for N. naja atra PLA₂ (7)) is an axial ligand.

An Asp⁴³ carboxylate makes salt links to the Arg⁶⁹ guanidino group in the calcium-free form structure (Fig. 6a). However, the Asp⁴³ carboxylate does not make such salt bridges in the calcium-bound form structure. Moreover, the rearrangement of the Asp⁴³ side chain occurs by the rotation around the bond between C and C atoms, and one of the Asp⁴³ carboxyl oxygens participates in the calcium binding (Fig. 6b). On the other hand, we tried to construct a model for the phosphatidylcholine, major substrate for this enzyme, into the current protein structure by manual fitting, based on the information of the crystal structures of other PLA₂s in a complex with the substrate analogue (6, 8, 9). The resulting model strongly suggests that the sn-3 choline group of the substrate may be packed with the methylene unit of Arg⁶⁹, whose guanidino group salt-links to Asp⁴³ carboxylate in the calcium-free form structure, and with that of Lys⁷² by hydrophobic interaction. Therefore, the conformational change of the Asp⁴³ side chain is very attractive, and it is suggested that there may be a correlation between calcium binding and substrate recognition (i.e. calcium binding may cause a conformational change suitable for substrate recognition and vice versa.

As described above, three water molecules bind to the calcium(II) ion. These waters are buried inside the protein, and two out of three form hydrogen bonds to the protein atoms. Wat²⁰¹ forms a hydrogen bond with the Cys⁶¹ carbonyl oxygen, and Wat²⁰² forms a hydrogen bond with the Thr⁴⁶ carbonyl oxygen. In the calcium-free form, the Cys⁴⁵-Ala⁴⁸ residues form a type I -turn, and the Thr⁴⁶ carbonyl oxygen protrudes toward the solvent region (Fig. 6a). However, in the calcium-bound form, the Cys⁴⁵-Ala⁴⁸ residues form a type II -turn, and the Thr⁴⁶ carbonyl oxygen protrudes toward the protein interior (Fig. 6b). This backbone-conformational change enables the formation of a hydrogen bond between the Thr⁴⁶ carbonyl oxygen and the Wat²⁰² molecule. As a result, Wat²⁰² is considerably stabilized and has a relatively lower temperature factor (27.1 Å²). Since NMR structure analysis indicated that a long loop region including residues Cys⁴⁵-Ala⁴⁸ is highly flexible in the absence of the calcium(II) ion (41), it is thought that a calcium-free form structure may be derived from the preferential crystallization of one protein conformer, in which residues Cys⁴⁵-Ala⁴⁸ form a type I -turn. However, judging from the appropriate conformation of the calcium-binding site, the current calcium-bound crystal structure should be correctly determined and also reflect the structure in solution.

The calcium(II) ion is essential for both substrate binding and catalysis. The x-ray analyses of other PLA₂s in a complex with the substrate analogue (6, 8, 9) revealed that two waters bound to calcium(II) ion are replaced by the pro-S nonbridging oxygen of the sn-3 phosphate of the substrate and the sn-2 oxyanion of the putative tetrahedral intermediate. When comparing the S. violaceoruber PLA₂ with the complexed form of other PLA₂s, the position of Wat²⁰¹ corresponds to the sn-2 oxyanion of the substrate analogue, and the position of Wat²⁰³ corresponds to the sn-3 phosphate oxygen. Because of the positional relation of the bound waters, we anticipate that Wat²⁰¹ is replaced by the oxyanion of the putative tetrahedral intermediate and that Wat²⁰³ is replaced by the sn-3 phosphate oxygen of the substrate during catalysis of the S. violaceoruber PLA₂.

Substrate-binding Site-- As mentioned above, we believe that the calcium-free form is more similar to the activated structure at the interface than the calcium-bound form. Therefore, we discuss the substrate-binding site of the bacterial PLA₂ based on its calcium-free form structure.

In class I/II PLA₂s, a hydrophobic channel is formed by the apolar side chains of the invariant or highly conserved residues (6, 9). This channel accommodates the aliphatic chains of the sn-1 and sn-2 substituents of the substrate and plays a role to keep the polar residues participating in the catalytic hydrogen-bonding network from the solvent region. Although only one crystal structure complexed with a substrate or an inhibitor can provide an atomic model for the protein-substrate interactions, the current structure allows us to identify the hydrophobic channel. The hydrophobic channel of the S. violaceoruber PLA₂ consists of Cys⁴⁵ and Pro⁴⁹ in the long loop; Cys⁶¹ in the ₃-helix; Phe⁸⁸ and Met⁹² in the ₄-helix; Tyr¹¹⁴, Ala¹¹⁷, and Val¹¹⁸ in the ₅-helix; and Ile¹²⁰ and Phe¹²¹ (Fig. 7). In these residues, a disulfide bond is formed between Cys⁴⁵ and Cys⁶¹. It is very interesting that residues Ile¹²⁰ and Phe¹²¹, which are speculated to form a hydrophobic channel, have a stable conformation in the calcium-free form but not in the calcium-bound form. This may imply that the conformational change by interfacial activation results in the stabilization of the substrate-binding site as well as the catalytic hydrogen-bonding network.

View larger version (22K): [in this window] [in a new window]   Fig. 7.   Stereoview of the proposed hydrophobic channel of the S. violaceoruber PLA2. The conserved catalytic residues His64 and Asp85 and the invariant residue Tyr68 are shown in red. They are connected through a hydrogen bond to the nucleophilic water molecule (also in red). The side chains of the residues forming the proposed hydrophobic channel are shown in orange. The side chains of Arg69 and Lys72 forming the putative choline-receiving pocket are shown in purple. This model is based on the calcium-free form structure, and the calcium(II) ion (green) is put in position estimated from the calcium-bound form.

In the class I bovine pancreatic PLA₂, a choline group esterified to the sn-3 phosphate is held in a region designated a choline-receiving pocket, consisting of the methylene units of the side chains of Lys⁵³ and Lys⁵⁶ (34, 35). The corresponding positions on the S. violaceoruber PLA₂ are Arg⁶⁹ and Lys⁷², which also have a long methylene unit. The Arg⁶⁹ side chain makes a hydrogen bond to the Ser⁴¹ hydroxyl oxygen and a salt bridge to a calcium ligand, the Asp⁴³ carboxylate oxygen, in the calcium-free form structure, and the latter salt bridge disappears in the calcium-bound form, suggesting that a correlated mechanism for the calcium binding and the substrate recognition may be present. The presence of these residues corresponding to the choline-receiving pocket may be a possible reason for the substrate preference to the phosphatidylcholine of this enzyme.

Interfacial catalysis of PLA₂ involves interfacial binding prior to the catalytic step. PLA₂s have an interfacial binding site, which surrounds the opening of the hydrophobic channel, to bind the aggregated substrate. The positively charged residues at the interfacial binding site mainly contribute to the binding to the anionic phospholipids (36), whereas the aromatic residues mainly contribute to the binding to the zwitterionic ones (37). The S. violaceoruber PLA₂ prefers zwitterionic phospholipids, although the molecular surface on the bacterial enzyme structure is mainly occupied with hydrophilic residues. In this case, a few aromatic residues, such as Trp¹¹² and Phe⁵³, found in the putative interfacial binding site, are thought to play a role as a linker for binding to the interface of zwitterionic phospholipids.

Conformational Rigidity-- The enzyme activity of the S. violaceoruber PLA₂ is not lost under incubation at 50 °C for 10 min (41). This thermostable PLA₂ has only two disulfide bonds, but its structure is highly rigid, like that of eukaryotic PLA₂s having several difulfide bonds. The lack of some disulfide bonds may be supplemented with the polar and apolar interactions. Since any interaction between residues adjacent on the primary structure is not useful in discussing the conformational rigidity of the bacterial PLA₂, the hydrogen bonds and salt bridges forming between residues separated from each other on the primary sequence are listed in Table II.

The hydrophobic interactions predominantly contribute to the structural rigidity of the N-terminal domain. The aromatic and aliphatic side chains of this domain and certain aromatic side chains in the C-terminal domain (Phe⁶⁶ and Tyr⁷¹ in the ₃-helix) are tightly packed by the hydrophobic interaction. This hydrophobic core lies near the catalytic hydrogen-bonding network and prevents the solvent molecules from invading the catalytic site or the hydrophobic channel. In addition to the apolar interactions, some polar interactions are present between the N- and C-terminal domains. The carboxyl end of the ₁helix, having a negative electrostatic potential by the helix dipole effect, is stabilized by the positively charged residues, such as Arg⁶³ and Arg⁸³, in the C-terminal domain.

A disulfide bond between Cys⁴⁵ in the long loop and Cys⁶¹ in the ₃-helix of the bacterial PLA₂ contributes to stabilize the long loop that has no secondary structures. Interestingly, this disulfide bond is present also in the eukaryotic PLA₂s. Perhaps this conserved bond might be helpful for the linkage of the catalytic site to the calcium-binding site. In addition, the current models have several polar interactions, such as five or six hydrogen bonds, between the long loop and the ₃-helix (Table II).

In eukaryotic PLA₂s, the anti-parallel -helices containing catalytic residues are held by two disulfide bonds, which are conserved at both ends. However, the S. violaceoruber PLA₂ does not have the disulfide-bond between the ₃- and ₄-helices. In general, the anti-parallel helices are known to be structurally unstable in terms of energy (38). This disadvantage may be overcome by a polar interaction between the Arg⁶³ guanidino group and the Asp⁹¹ carboxylate at the center of the -helices and by the presence of the N-terminal domain lying on both helices.

Another pair of anti-parallel -helices, the ₄- and ₅helices, has a disulfide bond between Cys⁹⁶ and Cys¹⁰⁷ at one end. Several hydrophobic side chains, derived from these -helices and the long loop, are observed around the disulfide bond. These side chains, packed by the hydrophobic interaction, are useful to keep the conformational rigidity. As mentioned above, at the other end of these anti-parallel -helices, there is a hydrogen bond between the Lys⁸¹ amino group and Val¹¹⁸ carbonyl, which may be useful to stabilize the conformation around the catalytic site. A salt bridge between the Lys⁸¹ amino group and Gly¹²² terminal carboxylate exists only in the calcium-free form and not in the calcium-bound form.

	ACKNOWLEDGEMENTS

We are grateful to Dr. A. Yamano (Institute of X-ray Research, Rigaku Co., Japan) for valuable discussion about x-ray protein crystallographic analysis. We thank Dr. T. Koike and K. Ohtani (Institute of Pharmaceutical Sciences, Faculty of Medicine, Hiroshima University) for valuable discussion about the catalytic mechanism of the bacterial PLA₂.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The atomic coordinates and the structure factors (code 1FAZ and 1KP4) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).

^¶ To whom correspondence should be addressed. Tel.: 81-82-257-5280; Fax: 81-82-257-5284; E-mail: sugi@hiroshima-u.ac.jp.

Published, JBC Papers in Press, March 15, 2002, DOI 10.1074/jbc.M200263200

² Y. Matoba and M. Sugiyama, manuscript in preparation.

	ABBREVIATIONS

The abbreviations used are: PLA₂, phospholipase A₂; Wat, water molecule.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

CiteULike

Complore