Structural Basis for Bile Salt Inhibition of Pancreatic Phospholipase A2

Summary

Bile salt interactions with phospholipid monolayers of fat emulsions are known to regulate the actions of gastrointestinal lipolytic enzymes in order to control the uptake of dietary fat. Specifically, on the lipid/aqueous interface of fat emulsions, the anionic portions of amphipathic bile salts have been thought to interact with and activate the enzyme group-IB phospholipase A2 (PLA2) derived from the pancreas. To explore this regulatory process, we have determined the crystal structures of the complexes of pancreatic PLA2 with the naturally occurring bile salts: cholate, glycocholate, taurocholate, glycochenodeoxycholate, and taurochenodeoxycholate. The five PLA2-bile salt complexes each result in a partly occluded active site, and the resulting ligand binding displays specific hydrogen bonding interactions and extensive hydrophobic packing. The amphipathic bile salts are bound to PLA2 with their polar hydroxyl and sulfate/carboxy groups oriented away from the enzyme's hydrophobic core. The impaired catalytic and interface binding functions implied by these structures provide a basis for the previous numerous observations of a biphasic dependence of the rate of PLA2 catalyzed hydrolysis of zwitterionic glycerophospholipids in the presence of bile salts. The rising or activation phase is consistent with enhanced binding and activation of the bound PLA2 by the bile salt induced anionic charge in a zwitterionic interface. The falling or inhibitory phase can be explained by the formation of a catalytically inert stoichiometric complex between PLA2 and any bile salts in which it forms a stable complex. The model provides new insight into the regulatory role that specific PLA2-bile salt interactions are likely to play in fat metabolism.

Introduction

Lipid metabolism has taken the stage front and center as a human health concern. Metabolic consequences of dietary fat and cholesterol are difficult to evaluate because little is known about how their uptake and secretion is regulated^1,2. It is widely appreciated that the emulsion droplets of such hydrophobic species bound by a phospholipid monolayer pass through the gastrointestinal tract where they sequentially encounter the gastric, pancreatic and intestinal lipases³. Dietary impact of the emulsions is further influenced by bile salts and conjugates cosecreted with the 14 kDa pancreatic group-IB phospholipase A2 (PLA2, EC 3.1.1.4), an sn-2-acyl phospholipid hydrolase. Together, such lipolytic and biophysical changes in the fat emulsion play a role in the solubilization, absorption, and metabolism of the glycerophospholipids, triacylglycerides and cholesterol esters, as well as the bile salts.

Bile salts and conjugates are the non-substrate components of the physiological substrate of pancreatic phospholipase A2 in the gastrointestinal tract⁴. The significance of the variety of bile salts and conjugates (Figure 1) found in mammalian bile is not known. However, the PLA2 dependent rate of hydrolysis of phosphatidylcholine dispersions increases in the presence of added bile salts^5-9. The mechanism of the rate enhancement remains to be established. The confusion arises due to a dependence of the rate of hydrolysis of the phosphatidylcholine dispersions with monodisperse bile salts that is biphasic^5,6,8,9. The rising phase is consistent with the enhanced PLA2 binding and k_cat*-activation by the anionic charge induced by the bile salt anions partitioned onto the surface of a zwitterionic bilayer^10-13. Figure 2 depicts a simple model of how amphipathic bile salts would coat phospholipid emulsions and present a negative charge for PLA2 activation. The sharp decrease at the higher bile salt concentration cannot be explained by the surface dilution of the substrate in the vesicles. The previously accepted explanation for the observed bile salt mediated loss of PLA2 activity was attributed to the disruption of substrate vesicles into mixed-micelles with the bile salt. In this explanation, the turnover is effectively lowered due to a decreased rate of exchange of substrate molecules between smaller particles⁹. However other possibilities cannot be ruled out. For example, it is possible that bile salts bound to PLA2 in the aqueous phase prevent it from binding to the substrate interface.

Figure 1.

The structure, nomenclature and abbreviations of bile salt compounds for which a PLA2-bile salt complex crystal structure is reported. The molecular structure depicted above is conjugated with taurine. Various biles salts are specified by the presence or absence of a hydroxyl group at the C7 and C12 position, as well as the presence or absence of an amino acid conjugated to the carboxyl group.

Figure 2.

Bile salt mediated anionic activation of PLA2. Two amphipathic bile salts are shown embedded on a simplified zwitterionic glycerophospholipid surface of an intestinal fat globule. The bound enzyme is depicted interacting with anions from the bile salts, and would be therefore, optimally poised for catalysis.

In this paper we reexamine the role that bile salts play with PLA2 function. We present 5 crystal structures of PLA2 complexed to 5 naturally occurring bile salts, which are each consistent with an inhibited form of the enzyme. The dual nature of PLA2 interaction with bile salts, both on and off of a zwitterionic interface, suggests a new paradigm for bile salt regulation of lipid metabolism via both activation and inhibition of PLA2.

Results

In order to understand the interaction of PLA2 with bile salts and their conjugates we initiated crystallization screens in the presence of naturally occurring bile salts. Our initial success came in the presence of the bile salt taurochenodeoxycholic acid (TCDC). In the search for suitable crystallization conditions, visually similar crystals were obtained in the absence and presence of the active-site-directed tetrahedral mimic¹⁴ 1-hexadecyl-3-(trifluoroethyl)-rac-glycero-2-phosphomethanol (MJ33). However the diffraction quality of the crystals was better with MJ33 in the crystallization drop. Yet, surprisingly the active site ligand MJ33 was not found in the structure of the PLA2-TCDC complex. It is not unusual for amphipathic molecules to assist in crystallization and yet not be observed in the final structure.

Following the structure solution of the PLA2-TCDC complex, we obtained protein crystals and solved structures of porcine group-IB PLA2 complexed with 4 additional bile salts shown in Figure 1, (CH, GC, TC and GCDC). The crystallographic X-ray diffraction and refinement statistics for these five structures are summarized in Table 1. In each structure presented, the bile salt is bound in a well-ordered stoichiometric manner. Details of the PLA2-TCDC complex will be described initially. The other 4 PLA2-bile salt complex structures will then be compared to the TCDC complex, pointing out conserved features and structural differences of PLA2 interacting with the bile salts.

Table 1.

Summary of PLA2-bile salt complex X-ray diffraction data collection and structure refinement

Bile salt ligand	CH	GC	TC	GCDC	TCDC
Space group	P3121	P3121	P3121	P3121	P3121
a=b (Å)	70.0	69.6	68.7	68.9	68.6
c (Å)	64.5	64.2	65.6	67.0	67.1
Resolution limit (Å)	1.90	1.85	2.20	2.00	2.05
Completeness (%)	96.3	99.7	96.2	98.2	99.8
Rmerge (%)a	4.9	7.3	7.5	5.8	7.3
Refinement resolution (Å)	35.0-1.90	22.8-1.85	27.0-2.20	24.1-2.50	27.1 – 2.20
Rworking b	0.167	0.185	0.199	0.210	0.203
Rfree b	0.201	0.208	0.245	0.242	0.233
RMSD observed
Bond length (Å)	0.005	0.005	0.005	0.006	0.005
Bond angle (°)	1.1	1.1	1.1	1.1	1.1
Total water molecules	193	151	103	93	129
PDB accession code	2azy	2b00	2azz	2b04	2b01

^aR_merge = Σ|Io-Ia|/Σ(Ia), where Io is the observed intensity and Ia is the average intensity, the sums being taken over all symmetry related reflections.

^bR_working = Σ|Fo-Fc|/Σ(Fo), where Fo is the observed amplitude and Fc is the calculated amplitude. R_free is the equivalent of R_working, except it is calculated for a randomly chosen set of reflections that were omitted (10%) from the refinement process.

Stoichiometric complex of PLA2 with TCDC

The PLA2-TCDC complex is shown in Figure 3. The TCDC ligand inserts deeply into the protein interior (Figure 3), causing a complete rearrangement of the 20s-loop from residue G15 to N23. The rest of the PLA2 is relatively undisturbed relative to other group-IB PLA2 structures previously solved. PLA2 contacts the bound bile salt ligand through pairwise van der Waals interactions and hydrogen bonding. These interactions of TCDC and the other 4 bile salt ligand complex structures are listed in Table 2. Also listed in Table 2 is the buried surface area of the protein at the bile salt-protein interface. The total buried surface area at the TCDC-protein interface is 1035 Ų. The total solvent exposed surface area of the TCDC is 771 Ų. A portion of the 20s loop moves downward towards the interface with a largest displacement at position L19 as much as 7 Å. The sterol A-ring (Figure 1) of TCDC is oriented almost perpendicular to the interface binding surface (i-face) of PLA2, whereas sterol rings B, C and D sit at a ∼ 45° angle with respect to the interface binding surface of PLA2. The hydrophobic TCDC binding pocket is made up of 9 residues (L2, F5, R6, I9, I13, F22, Y25, L41 and F106) that are within 4.0 Å from the ligand (Table 2). The steroid ring is sandwiched between two nearly parallel aromatic rings from the side chains of F22 and F106, making CH/π interactions¹⁵. Interestingly, the 3-α, 7-α-hydroxyls point downward toward the interface, in opposition to the expected orientation of the bile salt molecules embedded on the surface of an emulsion droplet in the small intestine (as depicted in Figure 2). The 3-α-hydroxyl makes a weak hydrogen bond with the side chain of R6, whereas the 7-α-hydroxyl makes the strongest hydrogen bond with the main chain carbonyl oxygen of N23. The taurine sulfonic group of TCDC is located near two positive charged C-terminal residues, K113 and K116. No ordered water molecules were identifiable that interact with the sulfonic group.

Figure 3.

The 2.2 Å crystal structure of pancreatic group-IB PLA2 complexed with TCDC. The active site residues H48, D49 and the active site calcium (yellow) are adjacent to the bile salt bind pocket. The ligand TCDC is displayed in a space filling view with atoms colored according to CPK convention. Apparently, the TCDC bound to the i-face partly blocks the access route to the active site. The residues K116, N23, and R6 are shown hydrogen bonded to the taurine-sulfate, 7-hydroxyl, and 3-hydroxyl groups, respectively. Hydrophobic contacts are shown with F5, I9, F22 and F106. (a) The PLA2-TCDC complex is displayed with the i-face of the protein facing the viewer. (b) The i-face of PLA2 is rotated 90° relative to the view displayed in panel-a so that it now faces down. The space filling view of TCDC shows the polar hydroxyl and sulfate groups (red oxygens) facing downward as well. These views were rendered using the programs MOLSCRIPT²⁹, POVSCRIPT³⁰ and POVRAY (www.povray.org).

Table 2.

Close contacts of PLA2 residues with bound bile salts

Bile salt ligand	CH	GC	TC	GCDC	TCDC
van der Waals contacts (Å)a
Leu2				3.8
Phe5		3.9		3.7	3.8
Arg6	4.0	3.6	3.5	3.5	3.4
Ile9	3.6	3.5	3.5	3.4	3.4
Ile13	3.5	3.7	3.7	3.8
Gly15			3.8
His17	3.9			b
Met20	3.9	3.5	3.9	b	3.3
Asp21	3.4	3.5	3.6	b
Phe22				3.2	2.9
Asn23	3.2	3.3	3.3	3.4	3.2
Tyr25	3.6	3.6	3.5	3.6	3.6
Cys29	3.6	3.6	3.8	3.7	3.8
Gly30	3.5	3.6	3.7	3.5	3.5
Leu41	3.9	3.8	3.9	3.9	4.0
Cys45			3.9	3.8	3.8
Tyr69				4.0
Phe106	3.8	3.6	3.8	3.6	3.6
Tyr111	3.8	3.3	3.8	3.2	3.2
Lys113					3.7
Lys116					3.8
Hydrogen bonding interactions (Å) c
Arg6-NE				O3 (3.5)
Arg6-NH2					O3 (3.2)
Met20-O	O7 (3.0)	O7 (3.2)	O7 (3.0)
Asp21-OD1		O3 (3.0)	O3 (3.1)
Asn23-O	O7 (2.7)	O7 (2.7)	O7 (2.8)	O7 (2.9)	O7 (2.7)
Asn23-N					O7 (3.4)
Arg43-NH2					O3 (3.5)
Water-O	O3 (2.5)	O3 (2.5)	O7 (2.8)	O7 (3.4)
	O7 (2.8)	O7 (2.8)	O12 (2.7)	O25 (2.8)	O7 (2.8)
	O12 (2.8)	O12 (2.9)	O12 (2.7)	O28 (2.9)	N26 (3.4)
	O12 (2.7)	O12 (2.9)	O12 (2.7)	O28 (3.1)	O28 (2.8)
	O12 (3.0)	O12 (2.8)	O25 (2.9)
	O25 (2.7)	O25 (3.3)
	O25 (2.8)	O27 (2.8)
	O25 (3.0)	O28 (3.1)
	O26 (2.7)
	O26 (2.8)
Buried surface area (Å2) d	974	1022	1014	977	1047

^aListed are shortest van der Waals contact distances within a 4.0 Å cutoff.

^bResidues 16-21 were not built into the PLA2-GCDC model due to disorder.

^cAtomic numbers for the bile salt ligand indicate position of the H-bond.

^dThe buried surface area of bile salt ligand and protein at the protein-ligand complex interface.

Comparison of PLA2 complexed with 5 naturally occurring bile salts

The pairwise van der Waals and hydrogen bonding interactions between PLA2 and bile salts are listed in Table 2. Overall among these 5 structures, 12-16 protein residues make van der Waals contacts of less than 4 Å with the bound bile salts. The majority of the contacts, as shown in Figure 4, are made between the sterol ring of the bile salt and residues near the N-terminus of the protein, on the 20s-loop and the 30sloop. The aliphatic chain portion of the ligand makes contact only with the protein residues in the C-terminal region. The common contact residues among all ligand complexes are R6, I9, M20, N23, Y25, C29, G30 and L41. The hydrogen bonding pattern for each of the bile salt complexes are similar. In two of three trihydroxy bile salt complexes (TC and GC), the 3-α-hydroxyl makes a hydrogen bond with the carboxyl oxygen of D21. The 7-α-hydroxyl in all three trihydroxy bile salt complexes (CH, TC and GC) makes a hydrogen bond with the main-chain oxygens of M20 and N23. The 12-α-hydroxyl of CH, TC and GC forms hydrogen bonds with three well ordered water molecules. In the 3-α-,7-α-dihydroxy GCDC complex, the 3-α-hydroxyl makes a hydrogen bond with a side chain nitrogen of R6, whereas the 7-α-hydroxyl makes a hydrogen bond with the main-chain oxygen of N23. In the 3-α-,7-α-dihydroxyl TCDC, the 3-α-hydroxyl forms a hydrogen bond with the side chains of R6 and R43, whereas the 7-α-hydroxyl forms a hydrogen bond with the main-chain oxygen and the amide nitrogen of N23. These additional hydrogen bonds with PLA2 may partly explain why TCDC has the highest binding affinity for bile salts currently tested (M. K. Jain, unpublished). As the bile salt is solvated, it is not surprising that a number of water molecules provide hydrogen bond interactions with the bile salts in the complexes. There are 10 bound waters in the PLA2-CH complex, as shown in Table 2. It should be noted that the variation and number of hydrogen bonding water molecules for each bile salt complex is affected by the resolution limits of the X-ray diffraction data and how well the model was refined.

Figure 4.

Stereoviews of PLA2-bile salt complexes with (a) TCDC and (b) CH. The difference electron density (coefficients 2Fo-Fc) surrounding each bile salt is shown. Also displayed are van der Waals contacts < 4 Å and hydrogen bond interactions < 3.5 Å. These stereoviews were rendered using the programs MOLSCRIPT²⁹, POVSCRIPT³⁰ and POVRAY (www.povray.org).

For comparison, the five PLA2-bile salt complex structures with CH, TC, GC, TCDC and GCDC bound to PLA2 were aligned with a PLA2 structure solved in the same space group as the bile salt complexes, P3₁21, but lacking active site or bile salt ligands. The root mean squared deviation (RMSD) of the C_alpha atoms of these structures is displayed in Figure 5. From this comparison, one can see that the three tri-hydroxy bile salt complexes (CH, TC and GC) have PLA2 crystal structures with almost identical conformations, with RMSD values in most regions less than 1 Å. The 5 bile salt complex structures are shown superimposed in Figure 6. In the absence of the 12-α-hydroxyl in both TCDC and GCDC, the aromatic ring of F22 moved in and displays a ring-stacking interaction with the sterol rings of the bile salt (Figure 6(a)). This CH/π interaction¹⁵ is inaccessible for the other three bile salts since the phenyl group of F22 would be in direct steric conflict with the 12-α-hydroxyl group of the bile salt. This difference in ring stacking interaction partly explains the observed higher affinity for TCDC vs. CH towards group-IB PLA2 binding (M. K. Jain, unpublished).

Figure 5.

RMSD comparison of Cα-atoms of PLA2s in complex with bile salts. (a) The structure of PLA2 in a ligand-free state was superimposed with each of the bile salt complexes: CH (blue diamonds), GC (violet squares), TC (yellow triangles), GCDC (green stars) and TCDC (black circles). (b) PLA2 from the PLA2-CH complex was superimposed with each of the other bile salt complexes: GC (violet squares), TC (yellow triangles), GCDC (green stars) and TCDC (black circles).

Figure 6.

A structural overlay of the 5 PLA2-bile salt complexes highlights subtle differences in the structures. (a) The conformation of F22 in complexes with CH (green), TC (yellow), TCDC (orange), GC (cyan) and GCDC (purple). The side-chain positions of F22 are shown as a stick model with their color matching that of the respective ligand from each complex. The C_α-traces of the 20s-loops are shown as a silver coil from residues A12 to Y25. For GCDC, residues 16-21 were not modeled in due to disorder. (b) Conformation of the calcium-binding loop and residue L31 in bile salt bound complexes. The C_α-traces for residues C27-S34 are shown as silver coils relative to the calcium ion from each of the bile salt complex structures (colored as in panel-a). The position of L31 from the CH (green) structure has flipped by nearly 180° relative to its “normal” position. As a result the calcium-loop in the CH complex has assumed an altered conformation, thereby making calcium binding impossible. These views were rendered using the programs MOLSCRIPT²⁹, POVSCRIPT³⁰ and POVRAY (www.povray.org).

The other significant departure from the 5 aligned structures is in the calcium-binding loop region where the RMSD was greater than 2 Å between TC or GC and CH. The origin of this large RMSD mainly resides in a single residue, L31 as shown in Figure 6(b). For a reason that can not be explained by structural observation, the side-chain of L31 in the PLA2-CH complex is flipped by almost 180° from its “usual” position. This altered position of L31 demonstrated a disruption of the conformation of the calcium-binding loop in the protein. As a result, the catalytic calcium ion was not present in the PLA2-CH complex. As described previously, there was a large conformational change of the 20s-loop for the dihydroxyl bile salt complexes (GCDC and TCDC) relative to the trihydroxyl bile salt complexes (CH, GC and TC), with an RMSD of 6 Å in this region.

The effect of bile salt binding on the active site

The Ca-coordination of the PLA2-TCDC complex is similar to that of previously solved PLA2 crystal structures with either a ligand coordinated to the calcium or no ligand present. The calcium coordination is seven coordinate with both the axial and equatorial waters present. The first shell and second shell waters of the PLA2-TCDC structure are located in nearly the same position as in the ligand free high resolution structure of bovine group-IB PLA2¹⁶. It is not surprising that the 60s loop has adopted a different conformation compared with active site bound structures, such as with MJ33 bound¹⁷. However when the 60s-loop position in the PLA2-TCDC structure was compared with another porcine structure in the same space group P3₁21, which lacked active site or bile salt ligands, the conformation of the 60s-loop was the same. This observation is consistent with the conformation of the 60s-loop in the various PLA2 structures determined largely by protein crystal packing constraints.

As mentioned above, the best diffracting crystals of the PLA2-TCDC complex were obtained in the presence of the active-site-directed tetrahedral mimic MJ33. However, MJ33 was not found in the structure of the PLA2-TCDC complex. Furthermore, crystals that were grown in similar conditions were partially refined and had occupancy of the active site ligand MJ33, while no bile salt was bound. It appears that the binding of bile salts and an active site mimic, such as MJ33, are mutually exclusive.

Discussion

The most striking implication of the structural models presented is that bile salts can play a specific structural role through their stoichiometric interactions with pancreatic group-IB PLA2. This is in addition to the effects of the bile salt induced anionic charge on the zwitterionic substrate interface. Previously, bile salts have been shown to have a biphasic effect on the PLA2 dependent hydrolysis of zwitterionic vesicles in vitro^5-9, which has direct implications for their in vivo effects in the small intestine. It has been commonly thought that the anionic charge of the amphiphilic bile salt serves to convert, an otherwise zwitterionic interface, to an anionic interface, more suitable for the catalytic action of group-IB PLA2. The decrease of rates consistently^5-9 seen at higher mole ratios of bile salts during in vitro kinetic studies has been explained as either the result of bile salt mediated surface dilution or from the formation of mixed micelles. The structures of the PLA2-bile salt complexes give an alternate explanation for the decreasing activity seen at higher bile salt to substrate molar ratios. Simply put, the occlusion observed of the PLA2 active site in these stoichiometric complexes is direct evidence that these ligands serve as inhibitors of the PLA2 activity.

Bile salts as inhibitors of PLA2

The alignment of PLA2 structures with bound active site inhibitors suggests that the PLA2-bile salt structures represents inhibited complexes because the positions of the alkyl chains of the active site directed mimics overlap with the bound bile salts (Figure 7). Following the superposition of ligand bound structures, both the sn-2 and sn-3 branches of the inhibitor MJ33 are in direct conflict with the A- and B-sterol rings of TCDC (Figure 7(b)). It was not surprising that MJ33 was not observed bound in the active site of the PLA2-bile salt complexes when attempts were made to form a co-complex with both a bile salt and an active site directed ligand. Just as with the MJ33 ligand, a superposition of the PLA2-TCDC complex with the PLA2 crystal structure bound with the active site directed inhibitor MG14¹⁸ shows a steric overlap (Figure 7(d)). In this alignment the sn-2 chain of MG14 clashes with bile salt binding. This comparison of the active site ligand binding pocket and the bile salt binding region of PLA2 provides the structural framework to explain an inhibitory role for specific bile salt interaction with group-IB PLA2.

Figure 7.

Superposition of PLA2s in complex with TCDC with structures of active site inhibitors MJ33 and MG14. (a) Overlay of PLA2-TCDC (orange) and PLA2-MJ33¹⁷ (grey, Protein Data Bank (PDB) code 1fxf) structures. (b) Close-up view of panel-a. (c) Overlay of PLA2-TCDC (orange) and PLA2-MG14¹⁸ (green, PDB code 1mkv) structures. (d) Close-up view of panel-c. These views were rendered using the programs MOLSCRIPT²⁹, POVSCRIPT³⁰ and POVRAY (www.povray.org).

Orientation of bile salt complexed to PLA2 is informative

The specific binding site of the bile salt complexes observed have several surprising features and the ordered interactions appear to have evolved to enable PLA2 to discriminate specific bile salts. The TCDC ligand sits in a hydrophobic pocket of the surface of PLA2 that makes contact with the substrate interface. The total buried surface area at the TCDC-protein interface is 1047 Ų with van der Waals interaction to nine PLA2 residues (L2, F5, R6, I9, I13, F22, Y25, L41 and F106). Hydrogen bonding interactions exist between K116, N23, and R6 to the TCDC taurine sulfate, 7-α-hydroxyl and 3-α-hydroxyl-groups, respectively. These interactions account for the specificity of TCDC binding to a site that is distinct from the catalytic active site or the i-face. As shown in Figures 3 and 7, the active site pocket is partially blocked by the bound bile salt.

The orientation of the bound TCDC in the complex with PLA2 is unexpected, i.e., the polar groups point away from the surface of the protein (Figure 3(b)). This orientation is suited for a complex in the aqueous phase. Amphipathic bile salts are likely to sit on interfaces with their polar groups facing the aqueous phase¹⁹ and therefore also towards an enzyme that is bound at the substrate interface, as depicted in Figure 2. Thus in order to achieve the orientation of TCDC that has been observed in the present PLA2 complex, it is necessary to impose a 180° flip for the bile salt orientation from the interface. Alternatively, PLA2 only binds bile salts in this manner after dissociating from its substrate phospholipid interface. Also TCDC, once bound to the protein in this manner may interfere with the interaction of a PLA2-TCDC binary complex with the interface.

As shown in the PLA2-MJ33 crystal structure¹⁷, the positive electrostatic potential provided by the side-chains of R6 and R10 is critical to the binding of three phosphate molecules (Figure 8(b)). These primary phosphate binding sites near the N-terminus, which were observed in this previous structure, are disrupted in the PLA2-TCDC complex as shown in Figure 8. The positive potential no longer exists due to the conformational change of the two arginine residues. Secondly, the rearrangement of the 20s-loop upon bile salt binding directly abolished one of the three phosphate binding sites (the upper right phosphate as shown in Figure 8(b) and Figure 8(d)). It is fair to say that all three phosphate binding sites are abolished as a result of bile salt binding by examination of the electrostatic potential in Figure 8(a). Consistent with this assessment, there were no observed phosphates in this crystal structure. We have examined the crystal structure obtained from a condition containing 0.2 M ammonium sulfate, the TCDC bile salt and active site ligand MJ33. The difference fourier electron density map (coefficients Fo-Fo) clearly showed the presence of three sulphate ions and the ligand MJ33, but no bile salt ligand was bound. This comparison is consistent with a conclusion that bile salt and phosphate/sulfate binding are mutually exclusive.

Figure 8.

Interfacial binding surface comparison of porcine group-IB bound with the bile salt TCDC versus bound with the active site directed inhibitor MJ33. (a) Electrostatic potential of TCDC bound PLA2. The potential surface shown in panel-a and panel-b were rendered in a mesh representation in order to show the complete ligand molecule. The electrostatic potentials were rendered with a range from blue-positive to red-negative [−6kT, 0, 6kT] (b) Electrostatic potential of MJ33 and 3 phosphates bound to PLA2 of the anion assisted dimer subunit-A¹⁷ (PDB code 1fxf). (c) Superposition of the PLA2-MJ33 (subunit-A, PDB code 1fxf) and PLA2-TCDC complexes. The molecular surface of the PLA2-MJ33 structure is shown in curvature representation, and the PLA2-TCDC model is rendered in a stick model. (d) Superposition of molecular surfaces of the PLA2-MJ33 structure (white, PDB code 1fxf) and the PLA2-TCDC (blue) structure. Also shown in a stick model are the bile salt ligand TCDC (gold), as well as the ligands MJ33 (dark grey) and 3 phosphate anions from the PLA2-MJ33 structure. These views were rendered using the program GRASP³¹.

The rearrangement of the 20s-loop also significantly altered the footprint of the interface binding surface. As shown in Figures 8(c) and (d), the PLA2 20s-loop of the bile salt complex protrudes out of the plane of the i-face, when compared to the active site ligand bound form of the enzyme. This large difference has a potentially significant effect on how the enzyme would bind to the lipid interface, and therefore could change the K_d of interface binding significantly. Ultimately a difference in the K_i values for the PLA2-bile salt complex in the aqueous phase would have kinetic consequences for the partitioning of the enzyme between the inactive E-BS (solution) or E*BS (bile salt and interface bound) versus the active E* form of the enzyme that can carry out interfacial turnover.

Model for the effect on function of a PLA2-bile salt complex

The PLA2 catalytic rate for the hydrolysis of zwitterionic vesicles is significantly modified in the presence of anionic amphiphiles like bile salts⁹. However, the effect of added bile salts is biphasic. The initial increase in the rate is followed by a decrease above 0.05 mole fraction. The simplest explanation is that above this mole fraction the interface becomes saturated, and the bile salt then begins to accumulate in the aqueous phase. Bile salts, such as cholate and deoxy cholate have been shown to have high CMC values of 16 mM and 6 mM, respectively²⁰. It is therefore unlikely that the biphasic kinetics can be explained by bile salt micelles partitioning PLA2 away from its substrate interface. Furthermore, mixed micelles of bile salts and phosphatidylcholine dispersions do not start forming until much higher mole ratios^5,6,8,21 than observed for the PLA2-bile salt biphasic kinetics that we are now addressing. Figure 9 shows a schematic of our model in which free E complexes with the bile salt to form the 1:1 EB complex as seen in the PLA2-bile salt structures. In this scheme, the PLA2-bile salt complex does not bind the interface, and is therefore inactive. Alternatively, it is possible that bile salts exert their inhibitory role with the PLA2-TCDC complex still associated with the interface.

Figure 9.

Bile salt mediated regulation of PLA2 activity on a phospholipid decorated fat globule. The interface bound form, E*S reacts with dialkyl phospholipids to form a free fatty acid and a lyso-phospholipid product. Bile salts, which bind to the surface with their polar groups pointed to the aqueous phase, activate PLA2 via anionic charge allostery. However, above a mole fraction that saturates the interface surface, the bile salt would begin to accumulate in the aqueous phase. At this point, free E complexes with the bile salt to form the 1:1 EB complex as seen in the PLA2-bile salt structures. This complex may or may not bind the interface. Regardless, the occlusion of the active site renders the enzyme inactive.

Regardless, as ordered stoichiometric modes of binding, the PLA2-bile salt complexes directly suggest an evolved regulatory role of dietary fat absorption by the bile salt modulation of PLA2 activity in the small intestine. This activity would depend not only on the relative stabilities of the individual PLA2-bile salt complexes, but also on the composition and the phase properties of different bile salts. Recall that the physiological form of the substrate interface for PLA2 is an emulsion with an organized monolayer surface of phospholipids and a hydrophobic core of unorganized triacylglycerides and cholesterol esters. The PLA2 catalyzed hydrolysis of this phospholipid monolayer in the small intestine has a direct effect on the susceptibility of the fat particles to a variety of hydrolytic interfacial enzymes in the gastrointestinal tract. This access by PLA2 will depend not only on the composition of the dietary fat, but also on the molecular structure and the composition of the bile salts and bile conjugates. The regulatory model now supported as a result of the bile salt structures, with previously observed biphasic activation and inhibition effects, offers a way for understanding and controlling gastrointestinal lipid metabolism by controlling the ratio of phospholipid and the composition of the bile salt conjugates.

The bile salt binding site of PLA2 is likely to provide a basis for the design of inhibitors by coupling the structural features for the bile salt binding to those of the active site directed ligands. Such inhibitors are of particular interest for pancreatic group-IB PLA2, whose inhibition has been shown to reduce the intestinal cholesterol uptake¹ by 30%, and is implicated in insulin resistance, glucose sensitivity and glycogen synthesis via its lyso-phospholipid product^22,23. More recently, it was shown in mice that the reduction of lysophospholipid absorption via the targeted inactivation of group-IB PLA2 enhanced insulin-mediated glucose metabolism and was protective against postprandial hyperglycemia²⁴. A broad implication of our current model of specific bile salt complex regulation predicts a role in the tight coupling of bile salt distribution and release in lipid homeostasis.

Modeling human group-X PLA2 binding to TCDC

The question arises as to whether stoichiometric bile salt interactions are unique to the group-IB PLA2. In order to explore whether such interaction may exist in other secreted PLA2 family members, we performed ligand binding modeling of the human group-X PLA2 structure solved in our lab²⁵. Coordinates of human group-X PLA2 (subunit-A, PDB code 1le6) were used as the starting model. As shown in Figure 10, the 20s-loop was rebuilt in order to allow the TCDC molecule to bind into the human group-X PLA2 model. Here, the 20s-loop was manually rebuilt using the conformation of the 20s-loop from the group-IB PLA2-TCDC crystal structure. The group-X model of bile salt bound was then minimized by gradient minimization with the program CNS²⁶. The minimization converged after 200 cycles. The ligand TCDC was then built into this group-X PLA2 model, followed by another 200 cycles of gradient minimization to obtain the final model of the group-X PLA2-TCDC complex. As shown in Figure 10, the position of the TCDC ligand moved only slightly from its initial position. In this model, the hydrophobic pocket for bile salt binding was largely preserved. Overall the interaction of group-X with TCDC is predicted to be weaker than that observed in the group-IB PLA2-TCDC complex crystal structure. Future studies are directed at the experimental characterization of possible bile salt interactions, like those observed here for group-IB PLA2, with the other PLA2 family members.

Figure 10.

Modeling of bile salt binding to human group-X PLA2. (a) Superposition of human group-X PLA2 starting model (red), group-X PLA2-TCDC final model (silver) and group-IB PLA2-TCDC crystal structure (orange). Overall, the conformation of group-X PLA2 did not change appreciably except in the 20s-loop, where it has moved away from and is now clear of the TCDC binding site. (b) Hydrophobic residues that are predicted to line a potential group-X PLA2 TCDC binding pocket. Six out of ten bile salt contacting residues of PLA2 residues are conserved with similar residues: L2, I9, F22, Y25, L41 and Y111 in group-IB PLA2 are replaced by I2, V9, Y20, Y23, I39 and Y103 in human group-X PLA2, respectively. Non-conserved substitutions are F5 to L5, R6 to A6, I13 to G13, and F106 to L98 respectively. Overall the interactions of group-X PLA2 with TCDC are predicted to be weaker than those observed in the group-IB PLA2 complex. These views were rendered using the programs MOLSCRIPT²⁹, POVSCRIPT³⁰ and POVRAY (www.povray.org).

Materials and Methods

Crystallization and X-ray data collection

The bile salts CH, GC, TC, GCDC, and TCDC (Figure 1) were purchased from Sigma and used without further purification. The preparation of other materials used in this study has been described elsewhere^17,25. Of the 5 PLA2 bile complexes reported, the TCDC complex was the initial complex, and therefore its crystallization, data collection and structure solution will be reported in more detail below. The conditions to crystallize and collect X-ray diffraction data of the other 4 bile salt complexes are nearly identical to the PLA2-TCDC complex, except where noted below.

TCDC bound PLA2 crystals were obtained by use of the hanging drop method at ambient temperature. In this experiment 1 μl of a pre-equilibrated protein aqueous solution containing 15 mg/ml porcine group-IB PLA2 protein, 10 mM CaCl₂, 3 mM MJ33, and 3 mM taurochenodeoxy cholate sodium salt was mixed with an equal volume of the protein crystallization well solution. Protein crystals of high visual and X-ray diffraction quality were obtained in several crystallization conditions. Crystal quality was assessed for bile salt complexes in the absence and presence of the active-site-directed tetrahedral mimic MJ33 present in the crystallization mixture. The quality of the crystals was better with MJ33 in the crystallization drop, but to our surprise the MJ33 was not found in the structure of the PLA2-TCDC complex. The pre-equilibrated protein solution was mixed with the optimal crystallization condition that contained 0.2 M CaCl₂, 0.1 M HEPES pH 7.5 and 28% v/v PEG400. The PLA2-TCDC complex crystals appeared in one week with a near cubic shape and were allowed to grow for two months to a final edge dimension of 0.15 mm.

Like the case of the TCDC complex described above, protein crystals of CH, GC, GCDC and TC complexes were obtained by first pre-equilibrating a 15 mg/ml stock solution of PLA2 with 10 mM CaCl₂, 3 mM MJ33 and the appropriate bile salt at a concentration of 3 mM. Protein crystals were optimized with the various bile salts by screening of conditions that both bridged the conditions described above for TCDC, as well as conditions identified from commercial protein screens (Hampton Research). The final crystallization conditions that lead to the best diffracting crystals were the following. The complex with GCDC and TC were both crystallized using the identical crystallization condition described above for TCDC (0.2 M CaCl₂, 0.1 M HEPES pH 7.5 and 28% v/v PEG 400). The complex with CH was crystallized using a crystallization solution of 0.1 M Tris-HCl, pH 8.5 and 8% w/v PEG 8000. The complex with the bile salt GC was crystallized using a crystallization solution of 0.1 M sodium cacodylate, pH 6.5 and 1.4 M sodium acetate.

An X-ray diffraction data set was collected from a single protein crystal of a complex with each of the 5 bile salts on a Rigaku-RU300 rotating anode generator with a RAXIS IV image plate area detector. In each case, the crystal was flash frozen in the −180 °C cryo stream after using the protein crystallization well solution as the sole cryoprotectant. The Programs DENZO and SCALEPAK²⁷ were used for data processing and scaling. The PLA2-bile salt complex crystals were all found to belong to the spacegroup P3₁21 with similar cell dimensions (Table 1).

Crystal structure refinement

Statistics for X-ray diffraction data and refinement of the 5 bile salt complexes are listed in Table 1. The structure of the PLA2-TCDC complex was the first structure solved by the molecular replacement method using the program AMoRe²⁸. A porcine group-IB PLA2 crystal structure (subunit-A, PDB code 1fxf) was used as the search model. Both rotational and translational searches yielded unambiguous solutions. The final correlation factor and molecular replacement R-factor values were 0.48 and 0.42, respectively. Structural refinement was carried out using the program CNS²⁶. Several rounds of positional, temperature B-factor and simulated annealing refinement were carried out to a resolution limit of first 2.8 Å, then to 2.5 Å, and the values of R_free and R_working were lowered to 0.362 and 0.321, respectively. Two surface loops (15–31 and 61-66) were temporarily deleted from the model to remove bias. The catalytic calcium ion was fit into the model. After two rounds of refinement with the resolution extended to 2.3 Å, the R_free and R_working were improved to 0.332 and 0.306, respectively. At this point the difference fourier showed clear electron density for the TCDC bile salt, which could be unambiguously fit. One additional round of refinement lowered the values of R_free and R_working to 0.327 and 0.296, respectively. The six residues in the 60s loop were re-built into the model. Finally the residues 20-23 were also built into the model. After several rounds of refinement with 90 waters built into the model, the values of R_free and R_working were 0.244 and 0.213, respectively. In addition to the catalytic calcium, two additional Ca²⁺ ions and five Cl⁻ ions were modeled into the structure. Extending the refinement to include diffraction data to a resolution of 2.2 Å, the electron density for the 20s loop was improved to allow building of the main chain atoms for residues 15-19 and a total of 129 waters. The final values of R_free and R_working were 0.232 and 0.205, respectively with good geometry (Table 1).

The structures of PLA2 complexed with CH, GC, TC and GCDC were solved using the crystal structure of the PLA2-TCDC complex as the molecular replacement search model. In each case, the ligand was built into the model at an intermediate stage of the refinement as electron density clearly showed the location of the ligand.

Protein Data Bank accession codes

Coordinates and structure factors have been deposited at the RCSB Protein Data Bank. The accession codes are 2azy, 2b00, 2azz, 2b04, and 2b01 for the crystal structures of PLA2 complexed with CH, GC, TC, GCDC, and TCDC bile salts, respectively.

Cover Illustration: Model of bile salt mediated regulation of phospholipase A2 activity on a phospholipid and bile salt decorated fat globule. (bottom left) The interface bound form E*S hydrolyzes a dialkyl phospholipid. (top left) The aqueous enzyme form complexes with a bile salt to form an inhibited bile salt complex. (bottom right) Electrostatic potential of the phospholipase A2 interface binding surface complexed with a substrate mimic and 3 phosphates (PDB code 1fxf). (top right) Electrostatic potential of a bile salt bound to phospholipase A2 (PDB code 2b03). See article by Pan & Bahnson in this issue, pp. ###-###.

Abbreviations used

CH

cholic acid

GC

glycocholic acid

TC

taurocholic acid

GCDC

glycochenodeoxycholic acid

TCDC

taurochenodeoxycholic acid

i-face

interface binding surface of enzyme

MJ33

1-hexadecyl-3-(trifluoroethyl)-sn-glycero-2-phosphomethanol

PDB

Protein Data Bank

PLA2

14 kDa secreted phospholipase A2

RMSD

root mean squared deviation

R_free

free R-factor

R_working

working R-factor