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

J. Biol. Chem., Vol. 275, Issue 35, 27045-27054, September 1, 2000

This Article Abstract Full Text (PDF) All Versions of this Article: 275/35/27045    most recent M003001200v1 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 Balendiran, G. K. Articles by Sacchettini, J. C. Search for Related Content PubMed PubMed Citation Articles by Balendiran, G. K. Articles by Sacchettini, J. C. Social Bookmarking                      What's this?

Crystal Structure and Thermodynamic Analysis of Human Brain Fatty Acid-binding Protein^*

Ganesaratnam K. Balendiran^§, Frank Schnütgen^¶, Giovanna Scapin^**, Torsten Börchers, Ning Xhong^§§, Kap Lim, Roseline Godbout^¶¶, Friedrich Spener^¶, and James C. Sacchettini

From the  Department of Biochemistry & Biophysics, Texas A&M University, College Station, Texas 77843-2128, the ^¶ Department of Biochemistry, University of Münster, Wilhelm Klemm-Strae 2, D-48149 Münster, Germany, ^** Merck Research Laboratory, Merck & Co, Rahway, New Jersey 07065,  Institute for Chemical and Biochemical Sensor Research, Mendelstrae 7, D-48149 Münster, Germany, ^§§ Department of Chemistry, Queens College of the City University of New York, Flushing, New York 11367, and the ^¶¶ Department of Oncology, Cross Cancer Institute, University of Alberta, Edmonton T6G 1Z2, Canada

Received for publication, April 9, 2000, and in revised form, June 13, 2000

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Expression of brain fatty acid-binding protein (B-FABP) is spatially and temporally correlated with neuronal differentiation during brain development. Isothermal titration calorimetry demonstrates that recombinant human B-FABP clearly exhibits high affinity for the polyunsaturated n-3 fatty acids -linolenic acid, eicosapentaenoic acid, docosahexaenoic acid, and for monounsaturated n-9 oleic acid (K_d from 28 to 53 nM) over polyunsaturated n-6 fatty acids, linoleic acid, and arachidonic acid (K_d from 115 to 206 nM). B-FABP has low binding affinity for saturated long chain fatty acids. The three-dimensional structure of recombinant human B-FABP in complex with oleic acid shows that the oleic acid hydrocarbon tail assumes a "U-shaped" conformation, whereas in the complex with docosahexaenoic acid the hydrocarbon tail adopts a helical conformation. A comparison of the three-dimensional structures and binding properties of human B-FABP with other homologous FABPs, indicates that the binding specificity is in part the result of nonconserved amino acid Phe¹⁰⁴, which interacts with double bonds present in the lipid hydrocarbon tail. In this context, analysis of the primary and tertiary structures of human B-FABP provides a rationale for its high affinity and specificity for polyunsaturated fatty acids. The expression of B-FABP in glial cells and its high affinity for docosahexaenoic acid, which is known to be an important component of neuronal membranes, points toward a role for B-FABP in supplying brain abundant fatty acids to the developing neuron.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Brain tissue contains high amounts of polyunsaturated fatty acids, such as arachidonic, docosahexaenoic acids (DHA),¹ and eicosapentaenoic acid (EPA), in their membrane phospholipids compared with other tissues (1, 2). The central nervous system uses DHA and other long chain polyunsaturated fatty acids during the early postnatal development (3) when cellular differentiation, active synaptogenesis, and photoreceptor membrane biogenesis take place (4-6). The critical role of DHA in the brain has been demonstrated by behavioral studies that have described n-3 deficiency (7). Arachidonic acid and DHA are synthesized by elongation, desaturation, and -oxidation steps of the dietary essential fatty acids 18:2(n-6) and 18:3(n-3), respectively (8). Deficiency of n-3 fatty acid leads to an altered electroretinogram (9), decreased visual activity (10), and impaired learning ability (11). In addition, DHA may be converted into neuroprostanes by free radical-catalyzed peroxidation (12-14) or may modulate the production of oxygenated metabolites from arachidonic acid with important roles in the physiology and pathology of the central nervous system. Similarly, oxidant stress and peroxidation of lipids, which have been implicated in the pathogenesis of a variety of human diseases, including atherosclerosis, cancer, neurodegenerative disorders such as stroke, Alzheimer's disease, and Huntington's diseases (15-21), and oxidative injury of neuroprostanes could lead to impaired neuronal function (12).

Little is known about how fatty acid trafficking and transport are controlled in the brain. Expression of B-FABP in radial glia during the development of the central nervous system is strictly correlated with the differentiation and migration of neurons from these cells (22). The high level of expression of B-FABP during neurogenesis or neuronal migration has implicated B-FABP to play an important role during central nervous system development (23). It is conceivable that it functions as part of a storage/delivery system for either DHA and EPA or the precursors required for DHA production, assuring a steady supply of fatty acids to the developing central nervous system. Alternatively the protein may be involved in transport of fatty acids to specific sties of regulation and may protect DHA from undergoing free radical-catalyzed peroxidation.

Human B-FABP is a member of the intracellular 14-15-kDa lipid-binding protein family. Members of this protein family have high affinity for amphiphils such as fatty acids, eicosanoids, retinoids, and bile acids. It has been proposed that these proteins are involved in the cellular uptake of lipids, their transport to metabolic pathways (24), and in regulation of lipid transport and metabolizing proteins (25).² Although there is a wide range of variation in the amino acid sequence among members of the family (for example, human B-FABP shows 67% amino acid sequence identity to human H/M-FABP, 28% to human liver FABP), their three-dimensional structures are highly conserved, consisting of ten antiparallel  strands and two short  helices. The ten antiparalled  strands are arranged into two nearly orthogonal 5-stranded  sheets that surround the interior binding cavity. Some of the three-dimensional structures reported to date are bovine myelin P2 (27), rat intestinal FABP (28), chicken liver basic FABP (29), bovine H-FABP (30), adipocyte lipid-binding protein (31), human M-FABP (32), Desert locust Schistocerca gregaria M-FABP (33), Manduca sexta FABP (34), and porcine ileal lipid-binding protein (35). The proteins can be grouped according to sequence homology, which is consistent with the ligand binding characteristics (36). These criteria would separate the proteins into four categories: (i) the intracellular retinoid-binding proteins; (ii) the ileal lipid-binding protein (which binds bile acid) and liver-FABP proteins that accommodate two fatty acids; (iii) intestinal-FABP, which binds a single fatty acid in a linear conformation; and (iv) FABPs with the fatty acid bound in a highly bent or U-shaped conformation. Brain, adipocyte, and the muscle/heart FABPs belong to the last group. Within grouping iv, the carboxylate moiety of the fatty acid is buried within the binding cavity and hydrogen bonds to conserved tyrosine and arginine residues, either directly or through an ordered water molecule. The hydrocarbon tail of the fatty acid forms VDW interactions with hydrophobic residues that line the binding pocket and with ordered water molecules that are in contact with polar residues in the binding pockets (37).

Whereas most of the FABPs bind only a single molecule of fatty acid, different FABPs show variation in their affinity for saturated versus mono- and polyunsaturated fatty acids, and long, medium, and short chain fatty acids, and it has been quite difficult to completely understand the structural basis for these preferences. However, all of the FABPs studied to date exhibit a preference for long chain saturated fatty acids (38). As B-FABP has a higher affinity for long chain polyunsaturated fatty acids, it is an excellent system to begin to explore how structural diversity in the FABPs leads to ligand specificity. Here we report the crystal structures of complexes of human B-FABP with OA and DHA and compare the structural data with the thermodynamic description of fatty acid binding. Our results provide an explanation for the ligand binding preferences of B-FABP based on key protein-ligand interactions.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cloning of Human B-FABP-- Human B-FABP was cloned by screening a human fetal brain  ZAP11 cDNA library (Stratagene) under reduced stringency using a radioactively labeled 600-base pair chicken retina cDNA probe (39). The 3'-ends of the positive clones had a high homology to mouse B-FABP, but they either lacked the 5'-end of the cDNA or they had undergone recombination at their 5'-ends. Sequence information from the section that was homologous to the mouse B-FABP cDNA (23) was employed to clone human B-FABP by rapid amplification of cDNA ends. To obtain the 3'-sequence of the cDNA human fetal brain poly(A)⁺ RNA (CLONTECH) was reverse-transcribed with Moloney murine leukemia virus reverse trancriptase (Amersham Pharmacia Biotech) using a d(T)₁₇-adaptor hybrid primer (5'-TCGGACAGTCGACATCGGTAA(T)₁₇-3'). The 3'-end of the human B-FABP cDNA was amplified using an internal sequence specific primer (5'-CCAACGGTAATTATCAGTCAAGAA-3') and the adaptor primer (5-TCGGACAGTCGACATCGGT-3'). To obtain the 5'-sequence of the cDNA, human fetal brain poly(A)⁺ RNA was reverse-transcribed using a primer corresponding to the predicted 3'-end of the coding part of the cDNA containing an attached BamHI restriction site (5'-CCAGGATCCTTTCTATGCCTTCTC-3'). In this primer the stop codon TAA was exchanged to TAG. The cDNA for B-FABP was then tailed with dATP using terminal transferase (Roche Molecular Biochemicals) and amplified using a nested sequence-specific 3'-primer (5'-CTGAGAGTCCTGATGACCACTT-3') and a mixture of d(T)₁₇-adaptor and adaptor primer. To obtain a continuous coding sequence a specific 5'-primer with an attached NdeI restriction site was designed (5'-CATATGGTGGAGGCTTTCTGTGCTAC-3') and used together with the primer containing the BamHI site in a polymerase chain reaction with the proofreading Pfu DNA polymerase (Stratagene). The 400-base pair polymerase chain reaction product obtained after 35 cycles of amplification was subcloned into a SmaI cut pUC18 vector (SureClone Ligation Kit; Amersham Pharmacia Biotech). lac colonies were analyzed by restriction mapping, and nucleotide sequences were determined by cycle sequencing using thermosequenase (Amersham Pharmacia Biotech) and the GATC 1500 direct blotting electrophoresis system (GATC, Germany). DNA and protein sequences were analyzed using the GCG package.

Expression and Purification of Human B-FABP-- The human B-FABP coding region of the cDNA was cloned into NdeI- and BamHI-restricted pET-3 expression vector (Novagen) and introduced into Escherichia coli BL21(DE3)pLysS (40). Terrific broth with cultured cells was induced with 0.4 mM isopropyl-1-thio--D-galactopyranoside after they reached an optical density of 1 in a 2-liter and 5 in a 10-liter scale fermentation. They were then harvested by centrifugation and lysed by one freeze thaw cycle and sonication. Nucleic acids were partially digested by Benzonase (Merck). The supernatant of the lysate was desalted on Sephadex G-25, and recombinant human B-FABP was isolated by anion exchange chromatography using Q-Sepharose. The protein was further purified by gel filtration on Fractogel EMD BioSec 650 (S), and the purity was analyzed by 15% SDS-polyacrylamide gel electrophoresis (41); sandwich enzyme-linked immunosorbent assay was used to detect human B-FABP in fractions during purification.

Delipidation of purified human B-FABP by the Lipidex procedure (42) resulted in only 30% delipidated protein as later became evident upon titration calorimetry analysis. Therefore, human B-FABP was further delipidated by extracting 1 ml of protein solution three times with 0.4 ml of diisopropylether:n-butanol (3:2) for 30 min with gentle shaking (43). The protein was then dialyzed by ultrafiltration (centricon with 10-kDa cut-off, Amicon) against calorimetry buffer (50 mM Tris/HCl, pH 8.0, 1 mM -mercaptoethanol, 0.05% sodium azide) to remove n-butanol. M/H-FABP used in this study was cloned, expressed, purified, and delipidated following previously described procedures (44).

Crystallization and Data Collection of Human B-FABP·OA Complex-- Crystals of the human B-FABP·OA complex were obtained using the hanging drop vapor diffusion method, 1.8-2.2 M (NH₄)₂SO₄ in 100 mM Tris, pH 8.5, precipitant solution. The crystals diffracted to a maximum resolution of 2.8 Å, an 82.0% (11,489 total/5444 unique reflections) complete data set to 2.8 Å was collected using a Rigaku RU-200 rotating anode x-ray source operating at 55 kV and 85 mA and a Siemens Multiwire area detector. Data processing, scaling, and merging were done using the XENGEN software (45) (Table I). The unit cell parameters were determined to be a = b = 93.6 Å, c = 46.7 Å,  =  = 90°, and  = 120° with the space group P6₅ containing one molecule of protein/asymmetric unit corresponding to the V_m ratio of 3.5 Å³/Da.

Crystallization and Data Collection of Human B-FABP·DHA Complex-- Delipidated human B-FABP (42, 43) was incubated with a 2-fold molar excess of DHA for 15 min at 37 °C prior to crystallization. Crystals of human B-FABP with bound DHA were obtained at 16 °C with the hanging drop vapor diffusion method by mixing 3 µl of a 10 mg/ml protein solution with an equal amount of precipitant solution containing 20% polyethylene glycol 4000, 0.1 M Tris, pH 8.1, and 50 mM NaOAc and allowed to equilibrated against 1 ml of the same precipitant solution. Long and thin crystals of the B-FABP·DHA-bound protein complex appeared in a week and reached the dimensions of 0.02 × 0.01 × 0.5 mm. The crystals diffracted to a maximum resolution of 1.8 Å. An 84% (102,823 total/21,863 unique reflections) complete data set to 2.1 Å with 5-fold redundancy was collected from two crystals at room temperature using a Rigaku RU-200 rotating anode x-ray source operating at 55 kV and 85 mA and a Nonius/Macscience dip2030 image plate. Data processing, scaling, and merging was done using the DENZO and SCALEPACK software package (46) (Table I). The unit cell parameters were determined to be a = 37.2 Å, b = 87.9 Å, c = 46.8 Å, and  = 113.4^o with the space group P2₁ containing two molecules of protein/asymmetric unit with the corresponding V_m ratio of 2.4 Å³/Da.

Structure Solution and Refinement of Human B-FABP·OA-- The three-dimensional structure of the human B-FABP·OA complex was solved using molecular replacement routines implemented in XPLOR (47, 48), with a model of human M-FABP (43) without bound fatty acid or water molecules as the search model. The highest peak obtained from the cross-rotation search using 15.0-4.0 Å data with a 2  cutoff on F corresponded to the Eulerian angles 45°, 121.5°, and 27°. A clear unique solution was found (9.2  above the second best solution) for the translation search with data between 10.0 and 4.0 Å with a 2  cut-off on F. Fifty cycles of rigid body refinement for data between 10.0 and 3.0 Å lowered the R-factor to 39.1%. The coordinates were then subjected to simulated annealing procedure (48), with data between 8.0 and 3.0 Å, which lowered the R-factor to 35.2%. At this point, the correct amino acid sequence was built into the structure, and the refinement was continued with alternating cycles of simulated annealing and manual rebuilding of the model. Several "omit" maps were used throughout the model building to confirm that the main chain and side chain atoms were assigned correctly. The resolution range was extended to include all data between 20.0 and 2.8 Å. The R-factor at this stage was 20.4%, and the model has conserved a good geometry (r.m.s. deviation for bond lengths and bond angles were 0.015 Å and 1.95^o, respectively) and stereochemistry (49). Inspection of ||F_o|   |F_c||_calc electron density maps at this stage allowed the positioning of the bound fatty acid and 25 ordered water molecules. Two more cycles of model building and refinement including oleic acid and water molecules reduced the R-factor and R_free to 19.0 and 26.1%, respectively (Table I).

Structure Solution and Refinement of Human B-FABP·DHA-- Refined human B-FABP·OA complex structure without the fatty acid and water molecules was used as the search model for molecular replacement (47, 48). The top peaks obtained from the cross-rotation search, using 12.0-4.5 Å data with a 3  cutoff on F, corresponded to the Eulerian angles 202.5°, 32.5°, and 277.5° for the first molecule and 110.0°, 80.0°, and 332.5° for the second molecule. Each of these rotation solutions gave a unique translation solution, between 9.0 and 4.0 Å with a 3  cut-off on F. The subsequent refinement was carried out using CNS (50), and strict noncrystallographic symmetry constraints were imposed. Ninety cycles of rigid body refinement, using 10.0 and 3.0 Å data, lowered the R-factor to 40.5%. The resulting model was subjected to annealing procedure (CNS), 9.0 and 3.0 Å, that lowered the R-factor to 34.2%. The refinement was continued with alternating cycles of simulated annealing and manual rebuilding of the model. Four series of omit electron density maps were used to guide the model building of the main chain and side chain atoms of both the molecules. The data were extended to include all reflections between 20.0 and 2.5 Å. The R-factor at this stage was 21.6%, and the model conserved a good geometry (r.m.s. deviation for bond lengths and bond angles was 0.015 Å and 2.01^o, respectively) and stereochemistry (49). Inspection of ||F_o|  | F_c||_calc electron density maps at this stage allowed the positioning of the bound fatty acid and 140 ordered water molecules. Three cycles of model building and refinement reduced the R-factor and R_free to 18.0 and 25.7%, respectively, without noncrystallographic symmetry constraints (Table I).

Calorimetric Binding Assays-- Titration calorimetry was carried out using a Microcal Omega Isothermal Titration Microcalorimeter (Microcal, Northhampton, MA). After delipidation and dialysis the protein concentration in calorimetry buffer (50 mM Tris-HCl, pH 8.0, 1 mM EtSH, 0.05% NaN₃) was adjusted to about 60 µM for all experiments and titrated with the ligands up to two-fold molar excess. The ligand was weighed into a 1.5-ml microreaction tube and dissolved in 250 mM KOH to give a concentration of 100 mM. This stock solution was adjusted to the final concentration with the calorimetry buffer. In a typical experiment this solution with the final concentration of 1.68 mM was injected in 25 aliquots (4 µl each) in 4-min intervals at 30 °C. Raw data were processed by the Origin software package supplied by the manufacturer. After the heat of dilution was subtracted from the heat of binding, data were fitted to a model assuming one binding site, and values for enthalpy and binding affinity were directly obtained (51) (Table II).

Lipidex Binding and Competition Assays-- The binding of OA to delipidated recombinant human B-FABP was analyzed by the Lipidex assay (42) essentially as described for mouse B-FABP (52). Various amounts of [1-¹⁴C]OA (Amersham-Buchler, 53 Ci/mol, [L]₀) were incubated with 0.6 µM human B-FABP, and data for protein-bound ligand ([PL]) were fit for binding of a ligand to a single class of binding sites. From the dissociation constant K_d, the apparent concentration of binding sites on the protein, [P]₀ and B_max = [P]₀/0.6 µM were obtained.

For competition assays, delipidated recombinant human B-FABP (3.4 µM) was incubated with equimolar amounts of [1-¹⁴C]OA at 37 °C for 15 min. To 75 µl of this solution a 25-µl solution with increasing competitor concentration was added. After 15 min at 37 °C 50 µl of a 50% (v/v) Lipidex-1000 suspension (Packard) was added with subsequent vortexing, and the mixture further incubated for 30 min on ice. Raw data for the construction of binding curves were obtained after separation of protein bound and free fatty acid (Lipidex bound) by centrifugation (4 min, 15,000 × g, 4 °C) and scintillation counting of 50 µl of solution. Curves describing the concentration of B-FABP·OA complex [PL] depending on the concentration of competitor [C]₀ were fit with the assumption that the free protein concentration is negligible in the binding equilibrium (Table II). With known values for the binding constant of the B-FABP·OA complex (K_L), B-FABP concentration [P]₀ and OA concentration [L]₀, the fit directly resulted in values for the binding constant K_C of the B-FABP competitor complex.

Enzyme-linked Immunosorbent Assay Procedure-- Polyclonal antibodies against recombinant human B-FABP generated in White New Zealand rabbits were affinity purified using recombinant human B-FABP coupled to CH-activated Sepharose 4B (Amersham Pharmacia Biotech). To remove antibodies cross-reacting with human M/H-FABP, this eluate was applied to a column with immobilized human M/H-FABP. A fraction of these antibodies was biotinylated by incubation with D-biotin--aminocaprylic acid N-hydroxysuccinimide ester (Roche Molecular Biochemicals) for 4 h at 4 °C followed by dialysis against phosphate-buffered saline, pH 7.4. The sandwich enzyme-linked immunosorbent assay was essentially performed as described for human H-FABP (44). The calibration curve was linear from 0.5 to 35 ng/ml.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Description of Human B-FABP Structure-- The crystal structures of human B-FABP with OA (Fig. 1a) and DHA (Fig. 1b) were solved and refined against 2.8 Å and 2.1 Å x-ray diffraction data, respectively. The structure of the human B-FABP·OA complex contains one polypeptide chain (131 residues), one molecule of OA, and 25 water molecules/asymmetric unit, and the structure of the B-FABP·DHA complex consists of two polypeptide chains, two molecules of DHA, and 140 water molecules/asymmetric unit. The r.m.s. difference between the two molecules in the asymmetric unit of B-FABP·DHA is 0.41 Å over 131 C atoms, and these molecules are related by a rotation of 179.6° about the noncrystallographic 2-fold axis. The r.m.s. difference between the human B-FABP·OA complex and human B-FABP·DHA complex is 0.54 Å (molecule 1) and 0.51 Å (molecule 2) over 131 C atoms.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   a, ribbon representation of the three-dimensional structure of the human B-FABP·OA complex. The bound OA molecule found in the binding cavity is shown in ball and stick. b, ribbon representation of the three-dimensional structure of the human B-FABP·DHA complex. The bound DHA is shown in ball and stick. Figs. 1a, 1b, 3a, and 3b are prepared using Molscript (64) and Raster3D (65). Secondary structure elements and the corresponding residues are -strand B1 (Cys5-Leu10), -helix A1 (Phe16-Ala22), -helix A2 (Gly26-Val35), -strand B2 (Lys37-Glu45), -strand B3 (Lys48-Ser55), -strand B4 (Lys58-Gln65), -strand B5 (Glu69-Thr74), -strand B6 (Asp77-Asp87), -strand B7 (Lys90-Trp97), -strand B8 (Lys100-Ile108), -strand B9 (Gly111-Phe119), and -strand B10 (Val122-Glu129). c, superposition of human B-FABP·DHA complex and human B-FABP·OA complex is shown using SPOCK (66), where the deviation between the C atoms is represented by the width of the worm. OA is shown in cyan, and DHA is shown in orange. Deviation seen away from the cavity corresponds to the different crystal packing and that around the cavity is attributed to the differences caused by DHA binding.

The general structure of human B-FABP is very similar to that of the other fatty acid-binding proteins reported to date (53). Eighty seven of the 131 residues in human B-FABP form 10 antiparallel -strands arranged into two perpendicular -sheets and two short -helices (A1, 7 residues, and A2, 10 residues) are located between -strands B1 and B2. Backbone hydrogen bonds typical of antiparalleled -pleated sheets are found between all strands, with the exception of -strands 4 and 5. These strands are separated, and the space between these two strands is filled with protein side chain atoms. -Strands B1, B2, B3, B4, B5, and B6 form the first -sheet and B6, B7, B8, B9, and B10 from the second -sheet. An exaggerated bend in B6 allows it to act as a linker between two -sheets, where the first five residues (Asp⁷⁷-Lys⁸¹) of the -strand hydrogen bond to the first -sheet, whereas the remaining six residues (Ser⁸²-Asp⁸⁷) of B6 hydrogen bond to main chain atoms of B7, part of the second -sheet.

OA Binding in U-shaped Conformation-- After the initial refinement with only atoms of the protein, difference electron density maps calculated to 2.8 Å showed a U-shaped positive density between the two -sheets, in the upper portion of the internal cavity as viewed of Fig. 2a. A good fit for the electron density was achieved by introducing a C16 fatty acid (C16:1, n-7) as there was no electron density for the last two carbon atoms of the OA (C18:1), the fatty acid that was incubated with the protein prior to crystallization. The fatty acid was in a bent conformation with an eclipsed bond between C5 and C6 atoms and a cis-double bond between C9 and C10 atoms (the position corresponding to the double bond in the OA). In the final steps of molecular refinement the fatty acid was included with torsional constraints only to the torsional angle corresponding to the cis-double bond. In addition to the fatty acid, five ordered water molecules were located within the fatty acid binding cavity.

View larger version (36K): [in this window] [in a new window]   Fig. 2.   a, difference Fourier electron density map, calculated with ||Fo|  |Fc||calc coefficients using 2.8 Å data, and contoured at 2.0 , for the human B-FABP·OA complex. b, difference Fourier electron density map, calculated with ||Fo|  |Fc||calc coefficients using 2.0 Å data and contoured at 2.0 , for the human B-FABP·DHA complex. a and b are prepared with the program SETOR (26).

O1 of the carboxylate of bound OA hydrogen bonds to the hydroxyl group of Tyr¹²⁸ (2.8 Å) and to an ordered water molecule (3.0 Å), which in turn hydrogen bonds to the side chain oxygen of Thr⁵³ and the guanidinium group of Arg¹⁰⁶ (3.2 and 3.3 Å, respectively). The other oxygen is within direct hydrogen bonding distance to the guanidinium group of Arg¹²⁶ (2.9 Å). Carbon atoms of the fatty acids' aliphatic chain form VDW interactions with several residues of the protein (Fig. 3a) and ordered water molecules. The phenyl ring of Phe¹⁶ appears to play a pivotal role in ligand binding as it is within 4.5 Å from carbons C4, C7, C12, and C13 of OA's aliphatic chain. In addition, the side chains of residues Tyr¹⁹, Met²⁰, Leu²³, Thr³⁶, Pro³⁸, Val⁴⁰, Thr⁵³, and Arg⁷⁸ are positioned around the outer surface of the binding site forming a set of interactions that appear to stabilize the folded conformation of the fatty acid.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Active site of the fatty acid (a) OA and (b) DHA are shown as found in the structures of their complexes with human B-FABP. Only important residues in the active site are shown. The twelve residues, which form VDW contacts, are Phe16, Val25, Thr29, Gly33, Ser55, Phe57, Lys58, Thr60, Ala75, Asp76, Phe104, and Met115 in the OA complex. The twenty one residues, which form VDW contacts are Phe16, Tyr19, Met20, Leu23, Val25, Thr29, Val32, Gly33, Thr36, Ser55, Thr60, Ile62, Glu72, Thr73, Thr74, Ala75, Asp76, Arg78, Gln95, Phe104, and Leu117 in DHA complex.

DHA Binding in a Helical or Looped Conformation-- The binding of DHA to B-FABP is facilitated by interactions of nonconjugated 6 cis double bonds found in the fatty acid (Fig. 2b). The carbon atoms of the acyl chain are in a somewhat helical conformation with a diameter of about 6 Å (the distance between C10 and C18) and a rise of about 5 Å, giving a distance of 9 Å between C1 and C22. All six double bonds (between C4=C5, C7=C8, C10=C11, C13=C14, C16=C17, and C19=C20) of DHA are in cis configuration. Rotational flexibility of the saturated C atoms (C2, C3, C6, C9, C12, C15, C18, C21, and C22) located between each double bond permit the ligand to attain a suitable conformation that can fit in the cavity. The arrangement of the double bonds in the bound conformation also allows formation of intra-fatty acid  interactions between unsaturated C atoms of DHA.

The carboxylate moiety of DHA binds to the hydroxyl group of Tyr¹²⁸ (3.1 Å) and the guanidinium group of Arg¹²⁶ (2.6 Å) in a fashion similar to that observed for hydrogen bonding for OA. The double bonds in DHA form extensive - interactions with side chains of the protein. For example, - interactions are formed between the C4=C5 double bond and the benzene ring of Phe¹⁰⁴ as well as the sulfur of Met¹¹⁵. It is important to note that residues Phe¹⁰⁴ and Met¹¹⁵ are structurally replaced by Leu¹⁰⁴ and Leu¹¹⁵ in M-FABP, and Ile¹⁰⁴ and Val¹¹⁵ in ALBP. In addition, double bonds C4=C5, C7=C8, C10=C11, and C13=C14 are arranged around Phe¹⁶ to permit additional  stacking interactions.

Fatty Acid Binding Studies-- Representative titrations of human B-FABP with DHA and arachidonic acid are shown in Fig. 4. The binding isotherm was derived from integration of the peaks and fitting to a model assuming one binding site. Dissociation constants, binding enthalpies, and calculated entropies obtained for B-FABP with different ligands are shown in Table II. The binding stoichiometry in all cases was very close to 1:1, and the binding entropies near zero, indicating that binding was primarily an enthalpy driven event. As with other FABPs, medium chain fatty acids such as lauric acid bind with relatively low affinity. Moreover, the binding of longer saturated fatty acids could not be accurately assayed by titration calorimetry, because of their inherent insolubility, so a second type of binding assay was used (see below). Monounsaturated n-9 fatty acids bound to human B-FABP with high affinity (K_d 41-47 nM), as did the long chain polyunsaturated n-3 fatty acids (27-53 nM). The protein also showed a preference for n-3 polyunsaturated fatty acids compared with those of the n-6 group (K_d 115-206 nM).

View larger version (38K): [in this window] [in a new window]   Fig. 4.   Titration of human B-FABP with DHA (A) and arachidonic acid (B). Binding isotherms were derived from the raw data shown in the insets, with 60 µM human B-FABP at 30 °C.

To confirm our isothermal titration calorimetry results and to test the binding of long chain saturated fatty acids, we used another accepted method for measuring fatty acid binding, referred to as the Lipidex assay (42), a variant of the classical charcoal assay (Table III). The binding of OA to human B-FABP gave a K_d of 100 nM using the Lipidex assay (Fig. 5 and Table II), in the same order of magnitude as those obtained by titration calorimetry. The binding affinity of saturated fatty acids (palmitic and stearic acids) to human B-FABP were two orders of magnitude weaker than that of OA, and the binding of retinoic acid was as weak as that for the saturated fatty acids. As the Lipidx assay provides relative K_d values, we normalized the results to the binding constant for OA obtained by titration calorimetry (Table II and Fig. 6). Although our binding constants are different from those reported by Xu et al. (54) for murine B-FABP (K_d OA = 0.44 µM, K_d arachidonic acid = 0.25 µM, and K_d DHA = 10 nM) using a modified Lipidx assay, there is good agreement with the high affinity binding of DHA for both the murine and human proteins. In addition, their work supports our findings that long chain saturated fatty acids display only weak binding to B-FABP. Together these binding studies show that fatty acid solubility is not as critical a determinate for the binding of long and very long chain fatty acids to B-FABP, as proposed for other FABPs (38). Binding affinity for B-FABP is more dependent on the number and position of double bonds in the fatty acids' hydrocarbon tail.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Binding of OA to recombinant human B-FABP. Analysis of [1-14C]OA bound to B-FABP (0.6 µM) by the Lipidex assay.

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Competition of ligands with [1-14C]OA. Recombinant human B-FABP (2.55 µM) was incubated with 2.55 µM [114C]OA and increasing amounts of nonradioactive competitors OA (), palmitic acid (), all-trans retinoic acid (), and DHA (). Individual data points obtained in a Lipidex assay are the mean of three determinations.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Binding of OA and DHA to Human B-FABP-- OA binds to B-FABP in the so-called U-shaped conformation, which has been observed in several of the group iv FABP structures determined to date (55). DHA (C22:6, n-6) assumes a significantly different conformation in the B-FABP binding cavity, resembling a turn of helix. Analysis of the binding properties and tertiary structure of B-FABP with bound lipid and a comparison to similar data for other FABPs provides a structural understanding for the high binding affinity observed for n-3 unsaturated fatty acids. As reported for other binary complexes of group iv FABPs, the carboxylates of OA and DHA form hydrogen bonds with conserved Arg¹²⁶ and Tyr¹²⁸ residues in B-FABP (Table II). These electrostatic interactions are not entirely fixed within the cavity, as the orientation and precise interactions of the carboxylate of bound fatty acids varies significantly between OA and DHA bound to B-FABP (Table II). Similar variations in the orientation of the fatty acid's carboxylate and the conserved Arg and Tyr have been observed in other members of this group of FABPs.

The hydrocarbon tail of OA forms a large number of VDW interactions with side chains within the binding cavity and the single double bond between C9=C10 interacts with the conserved benzene ring of Phe¹⁶, which is found in all group iv FABPs. The six double bonds in DHA provide extensive - interactions between the unsaturated groups and polar and aromatic side chains of the protein. In fact, as the conformation of OA in the cavity appears to be dominated by VDW interactions between the acyl chain and hydrophobic residues of B-FABP, the path of the acyl chain of DHA aligns the unsaturated carbons of the acyl chain more directly with polar groups in the protein.

A unique feature of the human B-FABP·DHA complex is - interactions, which occur between Phe¹⁰⁴ and the C4=C5 double bond of DHA. Residue Phe¹⁰⁴ is conserved in human B-FABP and chicken retina FABP. In murine (44, 52, 54, 55) and rat (56) B-FABP the comparable residue is a Cys. The sulfur in Cys¹⁰⁴ of murine and rat B-FABP may also form sulfur- interaction with the C4=C5 double bond of DHA in a similar fashion to the known sulfur- interactions as observed in eye lens protein (57).

The helical conformation of DHA is further stabilized by the internal - interactions between the double bonds C4=C5 and C7=C8, C10=C11 and C13=C14, and C16=C17 and C19=C20 (Table IV and Fig. 3b). It can be expected that as the number of double bonds decreases the internal  interactions will also decrease, and as a consequence the propensity of the ligand to be stable in the observed helical conformation will also decrease. This scenario will lead to a reduction in affinity of B-FABP for very long chain fatty acids without the correct number and arrangement of double bonds. The helical conformation adopted by DHA (and probably EPA) is likely to be the low energy conformation of DHA in solution and therefore not a conformation imposed by binding to the protein. Alternatively, B-FABP may force polyunsaturated fatty acids to accommodate this conformation to prevent them from undergoing free radical-catalyzed peroxidation.

Comparison to Other FABP Proteins-- The three-dimensional structure of human B-FABP has been compared with the crystal structures of M/H-FABP (called heart FABP (H-FABP), muscle FABP (M-FABP), and cardiac FABP (C-FABP)) and the adipocyte lipid-binding proteins ALBP (31, 58). Human B-FABP has 66.2% identity (79.7% similarity) with human M-FABP and 56.8% identity (73.5% similarity) with mouse ALBP, 43.5% identity. The similarities in the primary sequences are reflected in the conservation of their three-dimensional structures, as seen by the root mean square deviation of 0.72 and 0.63 Å for the C positions for H/M-FABP(1hms), ALBP(1adl1), respectively. The "U-shape" fatty acid binding mode of OA to human B-FABP is very similar to what has been reported for H/M-FABP with OA, palmitic acid, and elaidic acid as well as ALBP with bound lipid (31, 58). In H/M-FABP and ALBP the carboxylate of the fatty acid interacts with the conserved residues Arg¹²⁶ and Tyr¹²⁸, although the relative orientation of the carboxylate to these groups on the protein are not identical, the distances of the hydrogen bonds are well maintained.

A comparison of the binding sites of human B-FABP·DHA, human B-FABP·OA, H/M-FABP·OA, ALBP·palmitate, and ALBP·arachidonate shows that fatty acids of different length were all found to fit within the confines of the binding cavity. Interestingly, the volume does not adjust significantly in response to fatty acid length or degree of saturation. The binding cavity is about 14 Å wide at the widest point between residues 75 and 117 in all FABPs. The volume of the cavity of the B-FABP·DHA complex (after removal of the fatty acid and ordered waters) is ~914 ų. This is about 2 Å³ larger than the B-FABP·OA complex, and the cavity for the M-FABP·OA complex is about 25 Å³ more spacious than B-FABP·DHA complex. Therefore, the amount of space within the binding cavity appears to be constant and thus, has little effect of fatty acid specificity.

Whereas M-FABP, B-FABP, and ALBP have highly conserved residues within their respective fatty acids binding sites, only B-FABP has a phenylalanine residue, Phe¹⁰⁴, in close proximity to the C4=C5 double bond found in both the DHA structure and a model we have constructed of B-FABP·EPA, using Insight II (Molecular Simulations). As discussed above, we believe that the - interactions between the side chain of Phe¹⁰⁴ and the C4=C5, is a primary determinate for binding specificity. The comparable residue to Phe¹⁰⁴, in M/H-FABP and ALBP is Leu and Ile, respectively, which do not have a propensity to form  interactions and this may explain why M-FABP and ALBP do not show strong affinity for EPA and DHA (Table II). However, all members of the group have a comparable residue to Phe¹⁶, which is in a location that could support - interactions with the C9=C10 double bonds found in OA, linoleic, and -linolenic acid. It is important to note that these fatty acids all bind with high affinity to the group iv FABPs.

Thermodynamics of Binding-- The binding of fatty acids to human B-FABP is driven mainly by the enthalpic term. Similar results have been reported for the binding of OA to locust muscle FABP (59) and of (1, 8)-anilinonaphtalenesulfonate to intestinal FABP (60). Enthalpic contributions are attributable to noncovalent bonding interactions such as electrostatic and hydrogen bonding interactions, which are observed in the B-FABP crystal structures between residues Thr⁵³, Arg¹⁰⁶, Arg¹²⁶, and Tyr¹²⁸ and the carboxylate group of DHA and OA. Despite the significant difference in conformation of DHA and OA, the enthalpic components to the binding energy are similar. The enthalpic contribution of binding observed for OA, linoleic acid, and -linolenic acid is comparable, but the entropic contribution decreases with the number of double bonds. This may be because of decreased flexibility caused by additional double bonds.

We have previously proposed that the affinity of a given fatty acid for an FABP can be classified into three primary components: 1) the energetic strain imposed on the ligand in the bound state, 2) the specific interactions of the protein with the bound ligand, and 3) the desolvation of the binding cavity and the ligand upon binding (55). The relatively low entropic contribution to fatty acid binding that have been observed in several studies of different FABPs has lead to the suggestion that desolvation may be less important (61). The structural and thermodynamic analysis of B-FABP suggests that this protein has a binding pocket that is well suited to bind n-3 fatty acids by creating an environment within the cavity that permits binding of these fatty acids in a low energy conformation and provides specific interactions of the protein with conserved double bonds in the fatty acid. Comparison of B-FABP and M-FABP indicates that the number and positions of the double bonds appear to be a primary determinant of binding affinity. For example, whereas M-FABP's binding cavity is large enough to accommodate DHA in its low energy conformation and the atoms that compose the binding cavities are nearly identical, the presence of Phe¹⁰⁴ in B-FABP permits additional - interactions and therefore the binding of DHA to B-FABP (K_d = 53 nM) is significantly tighter than to M-FABP (K_d = 4100 nM).

The specificity of human B-FABP for n-3 fatty acids reflects the importance of this protein for brain development. In the brain, polyunsaturated long chain fatty acids are synthesized by astrocytes and cerebral endothelial cells (62), which may secrete these fatty acids into the cerebrospinal fluid. B-FABP may act as a sink for these fatty acids in glial cells, thereby facilitating their uptake. It may then deliver n-3 fatty acids to the enzymes of phospholipid synthesis. Recently, it was shown that liver FABP and intestinal FABP protect intracellular fatty acids against peroxidation by lipoxygenase and may modulate the availability of polyunsaturated fatty acids to intracellular oxidative pathways by differential binding (63). By analogy, B-FABP may also protect DHA and its precursors but not n-6 fatty acids from competing metabolic pathways in glial cells. In other cells of the brain, H/M-FABP, which has a low affinity for n-3 fatty acids, may be involved in normal transport of fatty acids for lipid metabolism in these cells. The presence of different FABPs in the brain represents an example of how these proteins may exert specialized functions in one organ.

	ACKNOWLEDGEMENTS

We thank Drs. J. Eads and R. Feldbrügge for help with microcalorimetry and fermentation, respectively. Many thanks also to H. vom Bruch for his help in large scale chromatography and to R. Franke for excellent technical assistance.

	FOOTNOTES

^* This work was supported by the Robert A. Welch Foundation, Grant SFB 310/A4 from the Deutsche Forschungsgemeinschaft, and Grant GM45859 from the National Institutes of Health.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 structure factors (code 1FDQ) 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.: 409-862-7641; Fax: 409-862-7638; E-mail:balendra@reddrum.tamu.edu.

Present address: Institut de Genetique et de Biologie Moléculaire et Cellulaire, CNRS-INSERM-ULP, Collège de France, BP 163, 67404 Illkirch-Cedex, France.

Work by F. S. was carried out in partial fulfillment for the requirements of his Ph.D examination.

Published, JBC Papers in Press, June 14, 2000, DOI 10.1074/jbc.M003001200

² Wolfrum, C., Borrmann, C. M., Rolf, B., Börchers, T., and Spener, F. (1999) Biochim. Biophys. Acta 1437, 194-201.

	ABBREVIATIONS

The abbreviations used are: DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; B-FABP, brain fatty acid-binding protein; VDW, van der Waals; OA, oleic acid; r.m.s., root mean square; E-FABP, epidermal fatty acid-binding protein; H-FABP, heart fatty acid-binding protein; M-FABP, muscle fatty acid-binding protein; K_d, dissociation constant; ALBP, adipocyte lipid-binding protein.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us