Characterization of a Fatty Acid-binding Protein from Rat Heart*

A fatty acid-binding protein has been isolated from rat heart and purified by gel filtration chromatography on Sephadex G-75 and anion-exchange chromatography on DE52. The circular dichroic spectrum of this protein was not affected by protein concentration, suggesting that it does not aggregate into multimers. Com- puter analyses of the circular dichroic spectrum predicted that rat heart fatty acid-binding protein con- tains approximately 22% a-helix, 45% j3-form and 33% unordered structure. Immunological studies showed that the fatty acid-binding proteins from rat heart and rat liver are immunochemically unrelated. The amino acid composition and partial amino acid sequence of the heart protein indicated that it is structurally related to, but distinct from, other fatty acid-binding proteins from liver, intestine, and 3T3 adipocytes. Using a binding assay which measures the transfer of fatty acids between donor liposomes and protein (Brecher, P., Saouaf, R., Sugarman, J. M., Eisenberg, D., and LaRosa, K. (1984) J. Biol. Chern. 259,13395-13401), it was shown that both rat heart and liver fatty acid-binding proteins bind 2 mol of oleic acid or palmitic acid/mol of protein. The structural and func- tional relationship of rat heart fatty acid-binding protein to fatty acid-binding proteins from other tissues is discussed.

Fatty acid-binding proteins (FABPs') are a class of small but relatively abundant proteins that are generally thought to be involved in the intracellular transport of fatty acids and possibly other organic anions (1). FABPs from liver (2)(3)(4), intestine ( 5 ) , and adipose tissue (6) have been extensively characterized; the primary structure of each has been determined, either by direct protein sequencing or deduced from the cDNA sequence (3,(7)(8)(9)(10)(11). These sequence data have indicated that FABPs from liver, intestine, and adipose tissue are members of a family of proteins derived from a common ancestral gene (8,9).
Early attempts to identify FABP in heart tissue used fatty acid-binding assays to demonstrate that soluble protein fractions from heart homogenates contain a low M, fatty acidbinding protein (1,12,13). Indirect evidence for the existence of a form of FABP in heart tissue was provided by Ockner a d Manning (5) using an antibody directed against jejeunal FABP. More recently, small amounts of a putative FABP mRNA were identified in rat heart tissue using rat liver and rat intestine FABP cDNA probes (14). Additionally, Rose and Goresky (15) perfused dog heart with labeled palmitate and noted that a-bromopalmitate inhibited palmitate uptake and possibly esterification, suggesting that the drug may compete for binding sites on heart FABP. Although these studies indicated that a form of FABP is present in heart tissue, its relation to FABPs in other tissues was not clear.
The first direct evidence for the presence of a specific FABP in heart was provided by Fournier et al. (16), who isolated and partially characterized a low M, fatty acid-binding protein.
Subsequent studies by  have shown that both rat and porcine heart FABP appear to selfaggregate in vitro into four multimeric forms, and it was proposed that this aggregation may occur physiologically and may play a role in the regulation of cardiac energy production. Said and Schulz (20) isolated a fatty acid-binding protein from rat heart and demonstrated that antiserum directed against rat liver FABP does not cross-react with the heart protein. Additionally, they reported that both palmitic acid and oleic acid bind to rat liver FABP, but that only oleic acid is bound by the rat heart fatty acid-binding protein (20). Reers et al. (21) also found differences between heart and liver FABPs with regard to partitioning of palmitate between the protein and unilamellar vesicles. One similarity between rat heart and liver FABP, reported by Glatz et al. (22), is that both proteins appear to display a similar pattern of diurnal variation, although recently, Bass et al. (23) reported that the level of rat liver FABP is not subject to diurnal variations.
The present investigation describes the isolation, characterization, partial amino acid sequence, and fatty acid binding properties of FABP from rat heart.

RESULTS
A description of the isolation, spectral properties, isoelectric point, and amino acid sequences of tryptic peptides of rat heart FABP is presented in the Miniprint Supplement under "Results." The amino acid compositions of rat heart and rat liver FABP determined in the present work are compared in Table  I to the amino acid compositions (from cDNA sequences) of FABP from rat liver, rat intestine, and 3T3 adipocytes. All four proteins contain between 122 and 131 residues and are ,id-binding Protein 5585 mg/ml (Fig. 1). The molar ellipticity (8) at all wavelengths examined was unaffected by protein concentration. The CD spectrum has a large negative ellipticity peak a t 216 nm, characteristic of proteins containing a high degree of @-form. An estimate of the content of a-helix in a protein can be obtained from the expression, flZz2 = [-30,300 ( f h ) ] -2340, where f h is the per cent a-helix (24). The molar ellipticity (8) a t 222 nm of rat heart FABP is -9044, from which the calculated content of a-helix is 22.1%.
Computer analyses (25) of the CD spectra of FABP and four reference proteins reveal that the predicted content of ahelix, @-form, and random form in the reference proteins myoglobin, adenylate kinase, papain, and ribonuclease A, are in reasonable agreement with the secondary structure of these proteins based on x-ray diffraction studies (26). Analysis of the CD spectrum of rat heart FABP predicted that the protein contains 22.0% a-helix, 45.0% @-form, and 33.0% random form. These results indicate that heart FABP contains a significant amount of ordered structure. Fig. 2A shows Coomassie Brilliant Blue-stained gels of rat heart high-speed supernatant, FABP containing fractions from Sephadex G-75, purified heart FABP, and purified rat liver FABP. Fig. 2B shows the autoradiograms of immunoblots of identical gels after electrophoretic transfer onto nitrocellulose paper followed by incubation with rabbit anti-rat heart FABP immune serum and '251-labeled protein A. Antiserum directed against rat heart FABP reacted only with heart FABP in the high-speed supernatant, in Sephadex G-75 eluates, or with purified rat heart FABP, but not with rat liver FABP.
In the reciprocal immunoblotting experiment to that shown in Fig. 2, it was found that immune serum directed against rat liver FABP cross-reacts only with rat liver FABP in liver high-speed supernatant, in Sephadex G-75 eluates, and with purified liver FABP, but not with rat heart FABP (data not given). These experiments demonstrate that rat heart and liver FABP are not immunologically related.
The ability of rat heart FABP to bind fatty acids was compared to that of rat liver FABP using a binding assay where each protein was incubated with multilamellar egg lecithin liposomes containing either [l-'4C]oleic acid or [l-14C]palmitic acid (Fig. 3). It was found that both proteins  Fatty Acid Added (&MI FIG. 3. Binding of radiolabeled oleic or palmitic acid to rat heart and rat liver FABP. Varying amounts of liposomes were incubated with FABP (10 PM) for 1 h at 25 "C, and the fatty acid bound to protein was determined as described under "Experimental Procedures." bound 1.8-2.0 mol of fatty acid/mol of protein. No significant differences were observed in the ability of either rat heart or rat liver FABP to bind palmitic or oleic acid, although liver FABP consistently bound slightly more fatty acid under saturating (fatty acid) conditions than did heart FABP. These experiments were performed at an FABP concentration of 10 PM, but the 2:l stoichiometry of bound fatty acid (saturating conditions) to protein was also observed at protein concentrations of 5 or 15 pM (data not given).

DISCUSSION
Recently, a family of proteins has been identified which includes liver (2)(3)(4), intestine (5, 10) and 3T3 adipocyte (11) FABP, myelin P2 protein (27), and cellular retinol-and retinoic acid-binding proteins (28, 29). These cytosolic proteins all have similar M, values and amino acid compositions, and exhibit some sequence homology. The present investigation has shown that rat heart FABP is also a member of this family on the basis of M,, amino acid composition, and partial amino acid sequence (see "Results"). Although it has long been recognized that fatty acid-binding proteins occur in many mammalian tissues (I), the precise biological function of these proteins has not been definitively elucidated. FABP would be expected to play a particularly important role in the metabolism of heart tissue because up to 80% of the energy requirement is derived from oxidation of long chain fatty acids (30).
We have isolated an FABP from rat heart by sequential gel filtration on Sephadex G-75 and anion-exchange chromatography on DE52, based on a previously reported method for purification of liver FABP (31). Two reproducible protocols were developed for the anion-exchange step, one employing batchwise adsorption and stepwise elution (Method 1) and the other employing gradient elution (Method 2) (Figs. 5 and 6 in Miniprint). Both methods consistently yielded approximately 4-8 mg of purified heart FABP from 50 frozen rat hearts. Neither Method 1 nor Method 2 requires a delipidation step, as was used by , thereby avoiding possible alterations in protein structure produced by exposure to organic solvents. In addition, cation-exchange chromatography, as used by Said and Schultz (20), was not required, thereby avoiding prolonged exposure to acidic pH (less than 6.0), which has been shown to cause instability and irreversible precipitation of rat liver FABP (2).
Although rat heart FABP is related to FABPs from other tissues, it was shown to be a distinct protein by several criteria. The amino acid composition of rat heart FABP, reported here for the first time, shows that this protein is enriched with respect to Asx, Glx, and lysine (Table I) like other FABPs, but is different in that it contains 2 residues of tryptophan, lacks cysteine, and contains significantly more threonine than FABPs from other tissues. Although the obtained amino acid sequences of rat heart FABP tryptic peptides represent only a portion of the primary structure (40 of 131 residues), alignment of these sequences (with gaps to maximize homology) with the amino acid sequence of rat liver FABP showed that partial homology exists between the two proteins, even though their primary structures are not identical. This is not surprising in view of the fact that rat liver and rat intestine FABP are members of the same family of proteins, even though only 34 of 127 amino acid residues are the same (10). Immunoblotting experiments ( Fig. 2) revealed that immune serum directed against rat heart FABP does not cross-react with rat liver FABP. This is in agreement with the results of Said and Schultz (20), who showed that immune serum directed against rat liver FABP does not cross-react with rat heart FABP.
Perhaps the most important difference between our findings and those reported by others is related to the aggregation behavior, or lack thereof, of FABPs. Fournier and co-workers (17-19) postulated that porcine and rat heart FABP aggregate into at least four distinct molecular species based on CD and fatty acid-binding isotherms derived from electron spin resonance measurements. We found no evidence for aggregation of rat heart FABP by CD under conditions comparable to those of 18), in that the molar ellipticity (8) did not vary as a function of protein concentration from 0.22 to 1.8 mg/ml ( Fig. 1). It is possible that the delipidation and lyophilization steps used during purification of porcine and rat heart FABP (16-19) may have altered the properties of those proteins, resulting in aggregation which may not occur under physiological conditions.
In our studies, rat heart and rat liver FABP bound both oleic or palmitic acid when these fatty acids were incorporated into liposomes and equilibrated with these proteins (Fig. 3). By incorporating fatty acids into artificial membranes we were able to use relatively high concentrations of fatty acid, which would normally be in the form of an acid soap if added directly to the aqueous phase (32). Perhaps more importantly, the fatty acid-phospholipid complex may provide a somewhat more physiological orientation for the fatty acid, comparable to that in a cellular membrane, as opposed to an aqueous dispersion of a fatty acid ligand. Using our binding assay no major distinctions between oleic and palmitic binding were found. These observations are different from those described previously (20), which indicated that rat liver FABP binds both oleic and palmitic acid, whereas rat heart FABP binds oleic acid but not palmitic acid unless delipidated with butanol. Since conditions used for direct measurement of fatty acid binding in this work differed greatly from those reported elsewhere, comparisons between results are difficult to make. It seems likely that under physiological conditions, both pal-mitate and oleate would interact similarly with heart FABP if such interactions were required for normal fuel metabolism. Reers et al. (21) used fluorescence to monitor the movement of (9'-antrolyloxy)palmitic acid between vesicles and bovine liver and heart FABP. It was found that bovine heart FABP will only donate the ligand to vesicles, whereas bovine liver FABP removes the ligand from vesicles, suggesting different affinities of the respective proteins for C-16 fatty acids. Again, the conditions used in these studies differed considerably from ours, in which radiolabeled fatty acids rather than fluorescent analogues were used, and in which movement of fatty acids from synthetic bilayers to both rat heart and rat liver FABP was demonstrated.
Several previous studies, including one from this laboratory (22, 33, 34), have reported that FABPs bind an equimolar amount of fatty acid in experiments where the protein concentration was determined by either the Lowry et al. (35) or Bradford (36) procedure. In the present work the concentration of FABP in the binding assays was determined by quantitative amino acid analysis, and it was consistently found that both rat heart and rat liver FABP bind 1.8-2.0 mol of fatty acid/mol of protein (Fig. 3). It is recognized that quantitative amino acid analysis is the best way to determine accurately the concentration of a protein in solution. It is also well known that spectrophotometric ( A~B~) and colorimetric assays using bovine serum albumin as the standard can frequently underestimate or overestimate the concentration of a particular protein (37). Therefore, when it is desirable to determine binding stoichiometries between proteins and ligands, caution should be exercised when protein concentrations are determined by spectrophotometric or colorimetric methods.
It is noteworthy that our binding assay measures the partitioning of a fatty acid between donor liposomes and acceptor protein after equilibrium has been reached, and that proteinbound and free (unbound) fatty acids are separated by centrifugation. In other fatty acid-binding assays described in the literature, free and bound fatty acids were separated by gel filtration 134) or using Lipidex (22, 38). These procedures may result in removal of weakly bound fatty acids, thereby lowering the apparent number of fatty acids bound to the protein. Consequently, the method selected for measurement of protein concentration and the procedure used for measuring fatty acid binding may explain why we observe that rat heart and rat liver FABPs bind approximately 2 mol of fatty acid/ mol of protein, whereas earlier studies (22, 33, 34) have indicated that FABPs bind only 1 mol of fatty acid. Finally, examination of the primary structure of rat and human liver FABPs has shown that these proteins are comprised of two homologous half-molecules, each of which contains two tandemly repeated sequences (8). This suggests the possibility that both half-molecules may contain a ligand binding domain. The latter observations may provide a structural basis for our finding that rat heart and rat liver FABPs bind 2 mol of fatty acid/mol of protein.

33.
Taylor, J. M. (1985) J. Biol. Chem. 260,1995-1998 Acta 533 (1251);protein A was obtained from hrsham. Protein stan-obtained from Sigma Chemical Co. Freund's complete and incomplete adjuvants were dads for isoelectric focusing were f r m S i 9 and electrophoresis reaq?nts were from Bio-Rad. Trypsin (L-1-tosyl-amido-2-phenylethyl chloromethyl ketone treated1 was obtained from Worthington, citrawnic anh.rlride was fran Aldrich chenical C a , and 4 N methanesulfonic acid w a s fram Pierce Qmnical Co. sequencer Chenicals were puchased from Beckman InStKWentS or from Burdick and Jackson. Seprating gels contained 15% acrylamide, and after electrophoresis were fixed and stained with Co-sie B r i l l i a n t Blue. Imuncblotting of gels was performed using a Txans-Blot C e l l q p r a t u s @io--) essentially by the procedure of Burnette (41). Transfer onto n i t m e l l u l o s e ~a p r was carried out a t d i e a t t m w rature for 16 h in a buffer mntaining 20 mM Tris base, 150 mEl glycine, and 20% (v/v) methanol. lbe nitrocellulose papr was dried and thea inabated for 3 h in a buffer containing 0.01 M phosphate, 0.11 M NaC1. 0.05% Tween-20, 0.25% g e l a t i n , pH 7.4 (Buffer A). The buffer was changed twice during this period. The n i t r o c e l l u l o s e paFer was then incubated w i t h 10 m l of Buffer A containing a 1:300 dilution of imInme serum d i r e c t e d a g a i n s t e i t h e r r a t h e a r t or r a t l i v e r FABP f o r 2.5 h. followed by 3 successive 1 h washes with buffer A and usually stored overnight a t 4W in Buffer A.

I E 4~n f~~~
The paper was then incubated for 2 h with 1.0 uCi of (lZ51)-Protein A i n 1 0 m l Of Buffer A. washed briefly with water and then washed for 2 hours with Buffer A. The p p r was then air dried and exposed t o Kcdak -5 x-ray film at -7C" I n I Q u l Q l " or l i v e r FABP mixed w i t h an equal volums of Freund's conplete adjuvant. Four w & S Rabbits were irmunized by suhcutanecus injection of 75 ug (0.5 ml) of rat heart a f t e r the initial injection, rabbits were injected subcutaneously a t four w e & interv a l s w i t h 30 ug protein mixed with Freund's incomplete adjuvant. After the third booster injection, animals were b l e d p r i c d i c a l l y f r a n the ear vein. lmrune sem directed against rat liver FABP was stored a t -7 W mtil used. m e serm directed against rat heart FABP was routinely passed through a nyoglcbin affinity colmm prgared by W p l i n g r a t h e a r t nycglabin ( Binding was determined by the method described by Brecher et a1 (33). Multilanr e l l a r liposanes mntaining egg lecithin, cholesterol and lateled fatty acid (20:7:1 mol/mol) were formed i n 0.01 K Tris, 0.1 M NaC1. pA 7.4 and mixed with a designated m u n t of protein in a t o t a l volume of 150 u l using mnical incubation tubes w i t h a capacity of 1.5 ml. Following incubation a t 25W for 60 m i n , the tubes were centri-fug& a t 13,000 x g for 2 m i n in a B e c k -microfuge Over 97% of the phospholipid, cholesterol and fatty acid were pelleted under these coditions in the absence of r a t the s u p m t a n t following centrifugation bund t o protein) was c a l c u l a t d from the l i v e r or heart FABP. In the presence of FABPs, the quantity of fatty acid present in original specific radioactivity (1100 dpn/nnol).  Overall, the yield of purified FABP from 50 frozen rat hearis ranged from 4-8 mg using either MetiEd 1 or Methcd 2. n e a s u r e m e n t n f m s i g n i f i c a n t l y d i f f e r e n t from that of r a t l i v e r FABP (data not given), primarily The electronic absorption spectrum i n t h e u l t r a v i o l e t of r a t h e a r t FABP w a s because t h e former contains 2 residues of tryptophan and t h e l a t t e r h a s none (See Table I). The molar extinction coefficient (e) a t 280 nm of rat heart FABP was 1.54 x 10 4 and that of l i v e r FABP was a98 x 10 4 a s s h g the K, of both proteins was 14,200 Omsed on polyacrylamide gel electrophoresis using globular protein standards).

necessary t o accurately determine the wncentration of p r o t e i n m t i t a t i v e mino
In order to determine the binding stoichiometry of fatty acid to protein, it i s acid analysis is probably the mst precise rnW for detennining protein wncentration but this procedure is time consuming and n o t p r a c t i c a b l e when l a r g e numbers samples are processed rwtinely. Ihus, it seemed useful t o find a reliable procedure Isoelectric focusing of rat heart FABP shwd a single band with an a w r e n t PI between 5.5-5.8 (Figure 7). I n c o n t r a s t , r a t l i v e r FABP displayed 3 bands, each of which w a s m r e basic than heart FABP. ?his is consistent with the observation that r a t l i v e r FbBP is not retained on DE-52 at pH 9.0 (3). whereas r a t h e a r t FABP was retained on DE52 lmder these d i t i o n s (Figure 5, 6 ) .

a r d i n l & @ . i m B f i d~Q f B i l t i i e a r . t E B B p
tion. No FTH derivative could be i d e n t i f i e d a t any of t h e 5 cycles, and it was Pat heart FABP (20 m o l ) was subjected t o 5 cycles of autansted Erfrw degraiacorcluded that the protein had a blocked aminc-terminal amlno acid. and the resulting peptides separated by gel filtration on Bio-Cel P 1 0 (Figure 8).
FABP, chenically msdified with citramnic anhydride, was di-ed with trypsin Peptides in p o l e d fractions here recavered by lyophilization ard subjected t o autonatd FBmn deqradation (Table 11).
peptide T-1 in peak A, was sequenced for 19 steps with 3 blanks. Peak B contained a mixture of 2 peptides, T-2 and T-3, whose sequences muld be deduced u r a n b i w s l y because ~2 was present in a greater armunts than T-3. Peak 5 contained a 5 residue peptide, T-4, i n which l y s i n e was the carwith citraconic anhydride had a x u r r e d Ihe sequences of portions of these tryptic boxyl-terminal amino acid, Indicating that i n c q l e t e m d i f i c a t i o n of lysine residues peptides could be aligned (with gaps) in the sequences of r a t l i v e r (7). rat intestine 110) ard 3 T 3 adipxyte FABP (11) (data mt given).