Molecular and Cellular Biochemistry 98: 95-99, 1990. © 1990 Kluwer Academic Publishers. Printed in the Netherlands.
Crystal structure of chicken liver basic fatty acid-binding protein at 2.7 A resolution Giovanna Scapin, Paola Spadon, Mario Mammi, Giuseppe Zanotti and Hugo L. Monaco 1
Biopolymer Research Center, Department of Organic Chemistry, University of Padova, 35100 Padova, Italy; i Department of Genetics, University of Pavia, 27100 Pavia, Italy
Key words: fatty acid-binding protein, chicken liver, crystal structure, ~-barrel Abstract
The three-dimensional structure of chicken liver basic fatty acid-binding protein has been determined at 2.7 A resolution by X-ray crystallography. Phases were calculated using the multiple isomorphous replacement procedure and a preliminary model was built. This model, with an initial R-factor of 0.57, was then improved by a cycle of refinement by simulated annealing which brought the R factor down to 0.32. The protein is structured as a compact 10-stranded-[3-barrel which encapsulates a residual electron density that can be interpreted as a fatty acid molecule. The NH2-terminus portion of the molecule contains two short a-helices. The structure of this liver protein appears very similar to that of the Escherichia coli derived rat intestinal FABP recently determined by X-ray diffraction methods.
Introduction
Several proteins that have the property of binding long chain fatty acids have been purified from the cytoplasmic portion of different tissues. They are collectively called fatty acid-binding proteins (FABPs) [1, 2]. Although the precise role in vivo of no member of this family has been strictly defined, there is considerable evidence that different FA BPs may serve different functions. The main ones proposed are the intracellular transport and compartmentation of long chain fatty acids, the modulation of the inhibitory effects of fatty acids and their CoA esters on specific enzyme systems and the protection of the cell from detergent effects [3-5]. The term FABP denotes, thus, a functionally loosely defined group of proteins that in common have the property of binding one type of ligand and therefore are likely to be structurally similar or in some way related. The FABPs characterized so far are small pro-
teins of 127-132 amino acids, reported to have either a single or double ligand binding site for saturated and unsaturated fatty acids [6]. It is this moderate size and simple binding stoichiometry that makes these proteins an attractive system to study fatty acid protein interactions. Sequences have been determined for rat intestinal (I-FABP [7]), liver (L-FABP [8, 9]) and heart (H-FABP [10]) and human intestine [11], liver [12] and heart FABP [13]. The primary structure of other homologous proteins is also known [14]. Crystals, suitable for X-ray structure determinations, have been reported for bovine liver FABP [15] and rat intestinal FABP [16]. The three-dimensional structure of the latter at 2.5 A resolution has also been presented [17]. We have recently reported the purification from chicken liver of a FABP that we have called basic fatty acid-binding protein because it has an isoelectric point of 9.0 [18]. We were also able to grow X-ray diffraction quality crystals of this protein that
96
130
0
~
~'
Fig. 1. Alpha carbon chain trace of the molecule. The positions
of some overlappingatoms are drawn slightlydisplacedto facilitate viewing. Notice the perpendicular arrangement of the strands in the two IS-sheets.
has a significantly different isoelectric point and amino acid composition from those of the chicken liver FABP recently isolated by Sewell et al. [19]. Basic FABP has a molecular weight of 14,000 and is believed to contain one mole of fatty acid bound per mole of protein. Fatty acid analyses after solvent extraction show that most of the endogenous ligand is unsaturated. The physiological role of this molecule is totally unknown, but its amino acid composition shows that it is certainly related to the FABP family of proteins. The primary structure of chicken liver basic FABP has not yet been determined. In this paper we present the three-dimensional structure of chicken liver basic fatty acid-binding protein determined by X-ray diffraction analysis at 2.7,~ nominal resolution.
Experimental procedures Purification and crystallization of chicken liver basic FABP have been described previously [18]. All diffraction data were collected to 2.7 A, resolution using the oscillation method and the resulting photographs were processed using standard procedures. An MIR map, calculated with two
~
"" . ~ h
. i
~
i
f
r
_ chhn tr trCaee~feb;:nC !ttbdt is a photograph from an
heavy atom derivatives, had a mean figure of merit of 0.63. Inspection of the map showed that it was sufficiently clear to allow tracing of the polypeptide chain. A preliminary model was built that gave a crystallographic R factor of 0.57. After one cycle of refinement using simulated annealing [20], the R factor dropped to 0.32. A detailed account of the procedure followed to solve this structure will be published elsewhere.
Results and discussion Since the primary structure of chicken liver basic FABP is not known, the model had to be built using a tentative approximate sequence that was selected making the tacit assumption that the sequence of this protein is homologous to that of other members of the family. Support for this assertion comes from a comparison of the amino acid compositions of the proteins [18]. In more detail, we proceeded in the following way. We started with the primary structure of rat intestinal FABP, one of the two FABPs whose three-dimensional structure is known. When the electron density in the map appeared to fit reasonably well the corresponding amino acid of I-FABP, that amino acid was chosen for the model. If that was not the case, the sequence homology between I-FABP, L-FABP and H-FABP [7, 8, 10] was examined and a different amino acid was selected from another FABP, taking always into account the amino acid composition of the chicken liverprotein. In all, 10 amino acids were substituted. Since, in the end, the chosen
97
/// t~g f~;tTdhee~dct(ofadneoSlietYcianct~dPmretled~eboun;~ig:~d with
Fig. 3. A cartoon showingthe idealizedelements of secondary
structure. sequence does not exactly account for the observed amino acid composition, this sequence must be considered very tentative, and therefore any conclusions regarding particular amino acid interactions must be drawn with great caution. However, the fact that the model built with this primary structure produces a quite acceptable crystallographic R factor, after only one cycle of refinement by simulated annealing, is a good evidence that it is not unreasonable.
Description of the molecule The basic FABP molecule consists of a single globular domain of about 40 ,~ 23 ,~ 30 ~ , composed of 10 strands of anti-parallel B-sheet and two short a-helices. Figure 1 shows the a-carbon chain trace of the molecule, Fig. 2 is a stereo diagram and Fig. 3 is a cartoon showing the idealized elements of secondary structure. The 10 strands of [~-structure, labelled following the notation used by Sacchettini et al. [17] for the structurally very similar rat intestinal FABP run from residues 5-10 (strand A), 37-42 (strand B), 47-52 (strand C), 57-62 (strand D), 68-72 (strand E), 76--83 (strand F), 88-93 (strand G), 99-108 (strand H), 113-118 (strand I) and 123-130 (strand J). The two a-helices run from
residues 14-21 (a-I) and 23-32 (a-II). These assignments are to be considered tentative, because some of the amino acids in the model will have to be changed when the protein sequence is known and because the resolution of this electron density map is modest. A future revision with better resolution data after the primary structure of the molecule is known may change them slightly. The 10 strands of ~-structure are organized into two orthogonal B-sheets: the first includes strands A, B, C, D, E and F, the second strands F, G, H, I and J. Strands F and J are in contact with both sheets, this being made possible by sharp changes in direction that occur in our current model in positions 80 and 128. The two c~-helices I and II are inserted in between strands A and B and have their axes approximately parallel to strands G, H, I and J. Although the hydrogen bond pattern in the beta sheets has not yet been completely determined, it is likely that it may present irregularities. Some alpha carbon - alpha carbon distances appear in this model to be somewhat larger than expected, particularly between strands D and E where they can reach a maximum value of about 9 A. The point through which the ligand can enter or exit its binding site is not obvious in the model and there is at present no structural information whatsoever on any apo-FABP from which it could be inferred. Not surprisingly, the overall architecture of this chicken liver FABP appears very similar to that of rat intestinal FABP [17] and also to that of bovine P2 myelin protein [21]. These three proteins are
98 thus seen to belong to the same structural family, which is likely to include many other members as suggested by the many sequence homologies already found.
Fatty acid binding While building the protein molecule model, it became apparent that there was in the map a very clear electron density, not connected to the polypeptide chain, that could be interpreted as a fatty acid molecule. Figure 4 shows the appearance of this electron density in the map. To this electron density we have fitted an oleic acid molecule, because the cis-bond in position 9-10 gives us a good fit to the density and because we know from fatty acid analysis that oleic acid is one of the ligands bound to the protein in our preparation [18]. In the model we have chosen to place the carboxyl group facing residue 29 and the hydrocarbon tail close to residues 8, 60, 70, 71, 72, 78, 83 and 103. Since the sequence of basic FABP is not known, we do not know the identity of these residues, in particular of residue 29, but an examination of sequence homologies shows that residue 29 is a highly conserved arginine. The electron density we observe close to the position of this residue is compatible with the presence of an arginine in basic FABP. If our interpretation, based only on the characteristics of the electron density observed and subject to the caveats we have mentioned above, is correct, the binding of the ligand would be, in this case, different from that observed in the case of the two other FABPs whose three-dimensional structure is known. In rat intestinal FABP the carboxyl group of the ligand has been placed close to Arg 127 [17], whereas in the P2 myelin protein the residues believed to be involved in the carboxyl binding are Arg 106 and 126 [21]. Higher resolution data and sequence information, the two prerequisites to a more accurate refinement of the model, are essential to either confirm or disprove our hypothesis which, if corroborated, would suggest alternative binding modes of analogous ligands to highly similar macromolecules.
Acknowledgements We thank Dr. J. Priestle for the program RIBBON that was used to draw a preliminary version of Fig. 3. The technical assistance of M. Tognolin and R. Pavan is gratefully acknowledged. This work was supported by grants from the Italian National Research Council and the Ministry of Education, Rome.
References 1. Glatz JFC, Veerkamp JH: Intracellular Fatty Acid-Binding Proteins. Int J Biochem 17: 13-22, 1985 2. Sweetser DA, Heuckeroth RO, Gordon JI: The metabolic significance of mammalian fatty acid-binding proteins: abundant proteins in search of a function. Ann Rev Nutr 7: 337-359, 1987 3. Grinstead GF, Trzaskos JM, Billheimer JT, Gaylor JL: Cytosolic modulators of activities of microsomal enzymes of cholesterol biosynthesis. Effects of Acyl-CoA inhibition and cytosolic Z-Protein. Biochim Biophys Acta 751: 41-51, 1983 4. Peeters RA, Veerkamp JH, Demel RA: Are fatty acidbinding proteins involved in fatty acid transfer? Biochim Biophys Acta 1002: 8--13, 1989 5. Spener F, Boerchers T, Mukherjea M: On the role of fatty acid-binding proteins in fatty acid transport and metabolism. FEBS Lett 244: 1-5, 1989 6. Cistola DP, Walsh MT, Corey RP, Hamilton JA, Brecher P: Interactions of oleic acid with liver fatty acid-binding protein: a carbon-13 NMR study. Biochemistry 27: 711717, 1988 7. Alpers DH, Strauss AW, Ockner RK, Bass NM, Gordon JI: Cloning of a cDNA encoding rat intestinal fatty acidbinding protein. Proc Natl Acad Sci USA 81: 313-317, 1984 8. Gordon JI, Alpers DH, Ockner RK, Strauss AW: The nucleotide sequence of rat liver fatty acid-binding protein mRNA. J Biol Chem 258: 3356-3363, 1983 9. Takahashi K, Odani S, Ono T: Isolation and characterization of the three fractions (DE-I, DE-II and DE-III) of rat-liver Z-protein and the complete primary structure of DE-II. Eur J Biochem 136: 589-601, 1983 10. Sacchettini JC, Said B, Schulz H, Gordon JI: Rat heart fatty acid-binding protein is highly homologous to the murine adipocyte 422 protein and the P2 protein of peripheral nerve myelin. J Biol Chem 261: 8218-8223, 1986 11. Sweetser DA, Birkenmeier EH, Klisak IJ, Zollman S, Sparkes RS, Mohandas T, Lusis A J, Gordon JI: The human and rodent intestinal fatty acid-binding protein genes. J Biol Chem 262: 16060-16071, 1987 12. Chan L, Wei C-F, Li W-H, Yang C-Y, Ratner P, Pownall
99
13.
14.
15.
16.
H, Gotto AM Jr, Smith LC: Human liver fatty acid-binding protein cDNA and amino acid sequence. J Biol Chem 260: 2629-2632, 1985 Offner GD, Brecher P, Sawlivich WB, Costello CE, Troxler RF: Characterization and amino acid sequence of a fatty acid-binding protein from human heart. Biochem J 252: 191-198, 1988 Walz DA, Wider MD, Snow JW, Dass C, Desiderio DM: The complete amino acid sequence of porcine gastrotropin, an ileal protein which stimulates gastric acid and pepsinogen secretion. J Biol Chem 263: 14189-14195, 1988 Paehler A, Maslowska M, Parge HE, Schneider M, Steifa M, Saenger W, Keuper HJK, Spener F: X-ray studies on triclinic crystals of fatty acid-binding protein. Example of an extremely X-ray-resistant protein. FEBS Lett 184: 185187, 1985 Sacchettini JC, Meininger TA, Lowe JB, Gordon JI, Banaszak L J: Crystallization of rat intestinal fatty acid-binding protein. Preliminary X-ray data obtained from protein expressed in Escherichia coli. J Biol Chem 262: 5428--5430, 1987
17. Sacchettini JC, Gordon JI, Banaszak LJ: The structure of crystalline Escherichia coli-derived rat intestinal fatty acidbinding protein at 2,5,~ resolution. J Biol Chem 263: 58155819, 1988 18. Scapin G, Spadon P, Pengo L, Mammi M, Zanotti G, Monaco HL: Chicken liver basic fatty acid-binding protein (pI = 9.0). Purification, crystallization and preliminary Xray data. FEBS Lett 240: 196--200, 1988 19. Sewell JE, Davis SK, Hargis PS: Isolation, characterization and expression of fatty acid-binding protein in the liver of Gallus domesticus. Comp Biochem Physiol 92B: 509-516, 1989 20. Bruenger AT, Kuriyan J, Karplus M: Crystallographic R factor refinement by molecular dynamics. Science 235: 458460, 1987 21. Jones TA, Bergfors T, Sedzik J, Unge T: The three-dimensional structure of P2 myelin protein. The EMBO J 7: 1597-1604, 1988
Address for offprints: H.L. Monaco, Dipartimento di Genetica, Sezione di Cristallografia, via Taramelli, 16 27100 Pavia, Italy
Crystal structure of chicken liver basic fatty acid-binding protein at 2.7 A resolution Giovanna Scapin, Paola Spadon, Mario Mammi, Giuseppe Zanotti and Hugo L. Monaco 1
Biopolymer Research Center, Department of Organic Chemistry, University of Padova, 35100 Padova, Italy; i Department of Genetics, University of Pavia, 27100 Pavia, Italy
Key words: fatty acid-binding protein, chicken liver, crystal structure, ~-barrel Abstract
The three-dimensional structure of chicken liver basic fatty acid-binding protein has been determined at 2.7 A resolution by X-ray crystallography. Phases were calculated using the multiple isomorphous replacement procedure and a preliminary model was built. This model, with an initial R-factor of 0.57, was then improved by a cycle of refinement by simulated annealing which brought the R factor down to 0.32. The protein is structured as a compact 10-stranded-[3-barrel which encapsulates a residual electron density that can be interpreted as a fatty acid molecule. The NH2-terminus portion of the molecule contains two short a-helices. The structure of this liver protein appears very similar to that of the Escherichia coli derived rat intestinal FABP recently determined by X-ray diffraction methods.
Introduction
Several proteins that have the property of binding long chain fatty acids have been purified from the cytoplasmic portion of different tissues. They are collectively called fatty acid-binding proteins (FABPs) [1, 2]. Although the precise role in vivo of no member of this family has been strictly defined, there is considerable evidence that different FA BPs may serve different functions. The main ones proposed are the intracellular transport and compartmentation of long chain fatty acids, the modulation of the inhibitory effects of fatty acids and their CoA esters on specific enzyme systems and the protection of the cell from detergent effects [3-5]. The term FABP denotes, thus, a functionally loosely defined group of proteins that in common have the property of binding one type of ligand and therefore are likely to be structurally similar or in some way related. The FABPs characterized so far are small pro-
teins of 127-132 amino acids, reported to have either a single or double ligand binding site for saturated and unsaturated fatty acids [6]. It is this moderate size and simple binding stoichiometry that makes these proteins an attractive system to study fatty acid protein interactions. Sequences have been determined for rat intestinal (I-FABP [7]), liver (L-FABP [8, 9]) and heart (H-FABP [10]) and human intestine [11], liver [12] and heart FABP [13]. The primary structure of other homologous proteins is also known [14]. Crystals, suitable for X-ray structure determinations, have been reported for bovine liver FABP [15] and rat intestinal FABP [16]. The three-dimensional structure of the latter at 2.5 A resolution has also been presented [17]. We have recently reported the purification from chicken liver of a FABP that we have called basic fatty acid-binding protein because it has an isoelectric point of 9.0 [18]. We were also able to grow X-ray diffraction quality crystals of this protein that
96
130
0
~
~'
Fig. 1. Alpha carbon chain trace of the molecule. The positions
of some overlappingatoms are drawn slightlydisplacedto facilitate viewing. Notice the perpendicular arrangement of the strands in the two IS-sheets.
has a significantly different isoelectric point and amino acid composition from those of the chicken liver FABP recently isolated by Sewell et al. [19]. Basic FABP has a molecular weight of 14,000 and is believed to contain one mole of fatty acid bound per mole of protein. Fatty acid analyses after solvent extraction show that most of the endogenous ligand is unsaturated. The physiological role of this molecule is totally unknown, but its amino acid composition shows that it is certainly related to the FABP family of proteins. The primary structure of chicken liver basic FABP has not yet been determined. In this paper we present the three-dimensional structure of chicken liver basic fatty acid-binding protein determined by X-ray diffraction analysis at 2.7,~ nominal resolution.
Experimental procedures Purification and crystallization of chicken liver basic FABP have been described previously [18]. All diffraction data were collected to 2.7 A, resolution using the oscillation method and the resulting photographs were processed using standard procedures. An MIR map, calculated with two
~
"" . ~ h
. i
~
i
f
r
_ chhn tr trCaee~feb;:nC !ttbdt is a photograph from an
heavy atom derivatives, had a mean figure of merit of 0.63. Inspection of the map showed that it was sufficiently clear to allow tracing of the polypeptide chain. A preliminary model was built that gave a crystallographic R factor of 0.57. After one cycle of refinement using simulated annealing [20], the R factor dropped to 0.32. A detailed account of the procedure followed to solve this structure will be published elsewhere.
Results and discussion Since the primary structure of chicken liver basic FABP is not known, the model had to be built using a tentative approximate sequence that was selected making the tacit assumption that the sequence of this protein is homologous to that of other members of the family. Support for this assertion comes from a comparison of the amino acid compositions of the proteins [18]. In more detail, we proceeded in the following way. We started with the primary structure of rat intestinal FABP, one of the two FABPs whose three-dimensional structure is known. When the electron density in the map appeared to fit reasonably well the corresponding amino acid of I-FABP, that amino acid was chosen for the model. If that was not the case, the sequence homology between I-FABP, L-FABP and H-FABP [7, 8, 10] was examined and a different amino acid was selected from another FABP, taking always into account the amino acid composition of the chicken liverprotein. In all, 10 amino acids were substituted. Since, in the end, the chosen
97
/// t~g f~;tTdhee~dct(ofadneoSlietYcianct~dPmretled~eboun;~ig:~d with
Fig. 3. A cartoon showingthe idealizedelements of secondary
structure. sequence does not exactly account for the observed amino acid composition, this sequence must be considered very tentative, and therefore any conclusions regarding particular amino acid interactions must be drawn with great caution. However, the fact that the model built with this primary structure produces a quite acceptable crystallographic R factor, after only one cycle of refinement by simulated annealing, is a good evidence that it is not unreasonable.
Description of the molecule The basic FABP molecule consists of a single globular domain of about 40 ,~ 23 ,~ 30 ~ , composed of 10 strands of anti-parallel B-sheet and two short a-helices. Figure 1 shows the a-carbon chain trace of the molecule, Fig. 2 is a stereo diagram and Fig. 3 is a cartoon showing the idealized elements of secondary structure. The 10 strands of [~-structure, labelled following the notation used by Sacchettini et al. [17] for the structurally very similar rat intestinal FABP run from residues 5-10 (strand A), 37-42 (strand B), 47-52 (strand C), 57-62 (strand D), 68-72 (strand E), 76--83 (strand F), 88-93 (strand G), 99-108 (strand H), 113-118 (strand I) and 123-130 (strand J). The two a-helices run from
residues 14-21 (a-I) and 23-32 (a-II). These assignments are to be considered tentative, because some of the amino acids in the model will have to be changed when the protein sequence is known and because the resolution of this electron density map is modest. A future revision with better resolution data after the primary structure of the molecule is known may change them slightly. The 10 strands of ~-structure are organized into two orthogonal B-sheets: the first includes strands A, B, C, D, E and F, the second strands F, G, H, I and J. Strands F and J are in contact with both sheets, this being made possible by sharp changes in direction that occur in our current model in positions 80 and 128. The two c~-helices I and II are inserted in between strands A and B and have their axes approximately parallel to strands G, H, I and J. Although the hydrogen bond pattern in the beta sheets has not yet been completely determined, it is likely that it may present irregularities. Some alpha carbon - alpha carbon distances appear in this model to be somewhat larger than expected, particularly between strands D and E where they can reach a maximum value of about 9 A. The point through which the ligand can enter or exit its binding site is not obvious in the model and there is at present no structural information whatsoever on any apo-FABP from which it could be inferred. Not surprisingly, the overall architecture of this chicken liver FABP appears very similar to that of rat intestinal FABP [17] and also to that of bovine P2 myelin protein [21]. These three proteins are
98 thus seen to belong to the same structural family, which is likely to include many other members as suggested by the many sequence homologies already found.
Fatty acid binding While building the protein molecule model, it became apparent that there was in the map a very clear electron density, not connected to the polypeptide chain, that could be interpreted as a fatty acid molecule. Figure 4 shows the appearance of this electron density in the map. To this electron density we have fitted an oleic acid molecule, because the cis-bond in position 9-10 gives us a good fit to the density and because we know from fatty acid analysis that oleic acid is one of the ligands bound to the protein in our preparation [18]. In the model we have chosen to place the carboxyl group facing residue 29 and the hydrocarbon tail close to residues 8, 60, 70, 71, 72, 78, 83 and 103. Since the sequence of basic FABP is not known, we do not know the identity of these residues, in particular of residue 29, but an examination of sequence homologies shows that residue 29 is a highly conserved arginine. The electron density we observe close to the position of this residue is compatible with the presence of an arginine in basic FABP. If our interpretation, based only on the characteristics of the electron density observed and subject to the caveats we have mentioned above, is correct, the binding of the ligand would be, in this case, different from that observed in the case of the two other FABPs whose three-dimensional structure is known. In rat intestinal FABP the carboxyl group of the ligand has been placed close to Arg 127 [17], whereas in the P2 myelin protein the residues believed to be involved in the carboxyl binding are Arg 106 and 126 [21]. Higher resolution data and sequence information, the two prerequisites to a more accurate refinement of the model, are essential to either confirm or disprove our hypothesis which, if corroborated, would suggest alternative binding modes of analogous ligands to highly similar macromolecules.
Acknowledgements We thank Dr. J. Priestle for the program RIBBON that was used to draw a preliminary version of Fig. 3. The technical assistance of M. Tognolin and R. Pavan is gratefully acknowledged. This work was supported by grants from the Italian National Research Council and the Ministry of Education, Rome.
References 1. Glatz JFC, Veerkamp JH: Intracellular Fatty Acid-Binding Proteins. Int J Biochem 17: 13-22, 1985 2. Sweetser DA, Heuckeroth RO, Gordon JI: The metabolic significance of mammalian fatty acid-binding proteins: abundant proteins in search of a function. Ann Rev Nutr 7: 337-359, 1987 3. Grinstead GF, Trzaskos JM, Billheimer JT, Gaylor JL: Cytosolic modulators of activities of microsomal enzymes of cholesterol biosynthesis. Effects of Acyl-CoA inhibition and cytosolic Z-Protein. Biochim Biophys Acta 751: 41-51, 1983 4. Peeters RA, Veerkamp JH, Demel RA: Are fatty acidbinding proteins involved in fatty acid transfer? Biochim Biophys Acta 1002: 8--13, 1989 5. Spener F, Boerchers T, Mukherjea M: On the role of fatty acid-binding proteins in fatty acid transport and metabolism. FEBS Lett 244: 1-5, 1989 6. Cistola DP, Walsh MT, Corey RP, Hamilton JA, Brecher P: Interactions of oleic acid with liver fatty acid-binding protein: a carbon-13 NMR study. Biochemistry 27: 711717, 1988 7. Alpers DH, Strauss AW, Ockner RK, Bass NM, Gordon JI: Cloning of a cDNA encoding rat intestinal fatty acidbinding protein. Proc Natl Acad Sci USA 81: 313-317, 1984 8. Gordon JI, Alpers DH, Ockner RK, Strauss AW: The nucleotide sequence of rat liver fatty acid-binding protein mRNA. J Biol Chem 258: 3356-3363, 1983 9. Takahashi K, Odani S, Ono T: Isolation and characterization of the three fractions (DE-I, DE-II and DE-III) of rat-liver Z-protein and the complete primary structure of DE-II. Eur J Biochem 136: 589-601, 1983 10. Sacchettini JC, Said B, Schulz H, Gordon JI: Rat heart fatty acid-binding protein is highly homologous to the murine adipocyte 422 protein and the P2 protein of peripheral nerve myelin. J Biol Chem 261: 8218-8223, 1986 11. Sweetser DA, Birkenmeier EH, Klisak IJ, Zollman S, Sparkes RS, Mohandas T, Lusis A J, Gordon JI: The human and rodent intestinal fatty acid-binding protein genes. J Biol Chem 262: 16060-16071, 1987 12. Chan L, Wei C-F, Li W-H, Yang C-Y, Ratner P, Pownall
99
13.
14.
15.
16.
H, Gotto AM Jr, Smith LC: Human liver fatty acid-binding protein cDNA and amino acid sequence. J Biol Chem 260: 2629-2632, 1985 Offner GD, Brecher P, Sawlivich WB, Costello CE, Troxler RF: Characterization and amino acid sequence of a fatty acid-binding protein from human heart. Biochem J 252: 191-198, 1988 Walz DA, Wider MD, Snow JW, Dass C, Desiderio DM: The complete amino acid sequence of porcine gastrotropin, an ileal protein which stimulates gastric acid and pepsinogen secretion. J Biol Chem 263: 14189-14195, 1988 Paehler A, Maslowska M, Parge HE, Schneider M, Steifa M, Saenger W, Keuper HJK, Spener F: X-ray studies on triclinic crystals of fatty acid-binding protein. Example of an extremely X-ray-resistant protein. FEBS Lett 184: 185187, 1985 Sacchettini JC, Meininger TA, Lowe JB, Gordon JI, Banaszak L J: Crystallization of rat intestinal fatty acid-binding protein. Preliminary X-ray data obtained from protein expressed in Escherichia coli. J Biol Chem 262: 5428--5430, 1987
17. Sacchettini JC, Gordon JI, Banaszak LJ: The structure of crystalline Escherichia coli-derived rat intestinal fatty acidbinding protein at 2,5,~ resolution. J Biol Chem 263: 58155819, 1988 18. Scapin G, Spadon P, Pengo L, Mammi M, Zanotti G, Monaco HL: Chicken liver basic fatty acid-binding protein (pI = 9.0). Purification, crystallization and preliminary Xray data. FEBS Lett 240: 196--200, 1988 19. Sewell JE, Davis SK, Hargis PS: Isolation, characterization and expression of fatty acid-binding protein in the liver of Gallus domesticus. Comp Biochem Physiol 92B: 509-516, 1989 20. Bruenger AT, Kuriyan J, Karplus M: Crystallographic R factor refinement by molecular dynamics. Science 235: 458460, 1987 21. Jones TA, Bergfors T, Sedzik J, Unge T: The three-dimensional structure of P2 myelin protein. The EMBO J 7: 1597-1604, 1988
Address for offprints: H.L. Monaco, Dipartimento di Genetica, Sezione di Cristallografia, via Taramelli, 16 27100 Pavia, Italy
