Fatty Acid Binding Protein ISOLATION FROM RAT LIVER, CHARACTERIZATION, AND IMMUNOCHEMICAL QUANTIFICATION*

Fatty acid-binding protein (FABP) was identified and isolated from rat liver cytosol by gel filtration, thin layer isoelectric focusing, and affinity chromatogra- phy. FABP (Mr 12,080 +. 80) exists in several immunochemically identical forms differing in isoelectric pH, which may in part reflect differences in their respective complements of bound endogenous ligand. FABP- bound fatty acids accounted for 60% of total cytosolic long chain fatty acids but contained no detectable phos- pholipid; the substantial enrichment of FABP in 18:2 and 20:4 as compared with whole liver homogenate was not influenced by homogenization of tissue in EDTA. The amino acid composition of FABP suggests that it is closely related or identical with certain similar neutral and acidic cytosolic proteins reported from other laboratories. By quantitative radial immunodiffusion, FAJ3P concentration in cytosol from livers of sexually mature female rats exceeded that from mature males (51.7 -C 3.0 versus 39.8 2 4.0 pg/mg of protein, p < 0.05), confirming earlier studies in which sex steroid effects on rates of fatty acid utilization were correlated with FABP concentration as determined by means of a binding assay. The abundance of FABP, its importance in the cytosolic binding of endogenous as well as ex- ogenous fatty acids, and its demonstrated correlation with rates of hepatocyte fatty acid utilization provide additional evidence for its relationship to the cellular metabolism of long chain fatty acids.

Fatty acid-binding protein (FABP) was identified and isolated from rat liver cytosol by gel filtration, thin layer isoelectric focusing, and affinity chromatography. FABP (Mr 12,080 +. 80) exists in several immunochemically identical forms differing in isoelectric pH, which may in part reflect differences in their respective complements of bound endogenous ligand. FABPbound fatty acids accounted for 60% of total cytosolic long chain fatty acids but contained no detectable phospholipid; the substantial enrichment of FABP in 18:2 and 20:4 as compared with whole liver homogenate was not influenced by homogenization of tissue in EDTA. The amino acid composition of FABP suggests that it is closely related or identical with certain similar neutral and acidic cytosolic proteins reported from other laboratories. By quantitative radial immunodiffusion, FAJ3P concentration in cytosol from livers of sexually mature female rats exceeded that from mature males (51.7 -C 3.0 versus 39.8 2 4.0 pg/mg of protein, p < 0.05), confirming earlier studies in which sex steroid effects on rates of fatty acid utilization were correlated with FABP concentration as determined by means of a binding assay. The abundance of FABP, its importance in the cytosolic binding of endogenous as well as exogenous fatty acids, and its demonstrated correlation with rates of hepatocyte fatty acid utilization provide additional evidence for its relationship to the cellular metabolism of long chain fatty acids.
Fatty acid-binding protein (Mr 12,000) is found in the cytosol of many tissues which utilize and/or transport long chain fatty acids (1). A growing body of evidence has implicated FABP' in these processes, but the precise nature of this involvement has not been fully defined. What does seem clear is that in several different circumstances involving nutritional, hormonal, and pharmacologic manipulations, a significant correlation can be demonstrated between cytosolic FABP concentration and rates of fatty acid flux into the cell. Thus, in intestinal epithelium, the influence of dietary and anatomic factomon mucosal FABP concentration was documented by means of a quantitative radial immunodiffusion assay for ' The abbreviation used is: FABP, fatty acid-binding protein.
purified FABP (2). In liver, although significant correlations between FABP concentration and fatty acid utilization have been demonstrated (3-6), quantification of FABP has largely depended on binding assays, and this has hindered the further elucidation of the dynamic regulation of the protein. Moreover, a number of other apparently closely related proteins have been described in liver cytosol (7-13), and because these have been viewed as subserving a diversity of physiological functions, their intrinsic and functional relationship to FABP has remained unclear.
To begin to address these questions, it seemed essential to purify FABP from rat liver in order to permit its further characterization, develop a suitable immunochemical assay, and better define its possible relationship to other cytosolic proteins of similar molecular size and charge. In the present communication we describe the isolation and identification of FABP from rat liver by gel filtration, isoelectric focusing, and affinity chromatography, establish its amino acid composition and the immunochemical identity of several charge forms, c o n f i i previously observed sex differences in FABP concentration by quantitative radial immunodiffusion, demonstrate that FABP accounts for 4 to 5% of cytosolic protein, and provide evidence suggesting that FABP is very similar or identical with certain proteins reported by other laboratories.
Preparation of Rat Liver Cytosol-Livers obtained from fed adult male Sprague-Dawley rats, 300 to 400 g, were perfused with 0.89% NaCl at 4 "C, homogenized in 0.154 M KCL-0.01 M PO, buffer (pH 7.4, 2 ml/g), and centrifuged for 20 min at 12,000 X g, and the resulting supernatant was centrifuged for 1 h at 105,000 X g (5, 6,14). The clear high speed supernatant fraction was aspirated so as to minimize contamination by floating fat and was used as such for subsequent gel Ntration or binding experiments or, after indicated modifications, for other analytical or preparative procedures. For resolution of cytosolic

Rat Liver Fatty Acid Binding
Protein 7873 protein fractions and/or analysis of fatty acid binding thereto, cytosol or the indicated cytosolic fraction was mixed with ['4C]oleate in 5 pl of propylene glycol prior to subsequent separation by gel filtration. In some experiments, whole cytosol or fractions thereof were delipidated by extraction for 30 min with 2 volumes of di-isopropyl ether (15); the phases were separated, and a second extraction was done for 10 min with an equal volume. After centrifugation, the aqueous phase was obtained, and residual di-isopropyl ether was removed under vacuum at 37 "C for 10 min or by lyophilization. Lipid Analysis-Lipids were extracted from native or delipidated samples by the method of Folch et al. (16) and were separated by thin layer chromatography (0.25 mm Silica Gel 60 in petroleum ether:ethyl ether:glacial acetic acid, W151.5). Appropriate zones were identified by means of standards and were extracted from the chromatogram. Fatty acid methyl esters were prepared by the addition of diazomethane to the fatty acid fraction and were analyzed as previously described (14) by gas-liquid chromatography on a Hewlett-Packard 402 B gas chromatograph employing a 6-foot glass column packed with 10% SP-2330 on 100/120 mesh Chromosorb W AW (Supelco) at 186 "C. Pentadecanoic acid was employed as an internal standard; fatty acid mass was determined by means of a Hewlett-Packard model 3380 digital integrator. Phospholipid was measured in the Folch extraction by the method of Bartlett (17) as modified by Marinetti (18).
Electrophoresis and Isoelectric Focusing-Polyacrylamide disc gel electrophoresis (19) employed 3.5% concentrating (pH 6.4) and 7% separating (pH 8.9) gels buffered with 0.5 M Tris and 0.38 M glycine (pH 8.6, 20 "C), 3 mA/gel for 120 min. Gels were fixed and stained with 1.0% Amido schwarz in 7% acetic acid or 0.2% Coomassie brilliant blue in methanol-water-acetic acid (551) and were destained electrophoretically. In some experiments, gels were exposed to 0.003% 1aniliio-8-naphthalene sulfonic acid, magnesium salt, in 0.1 M NaP04 buffer (pH 6.8) and were examined under ultraviolet light (20). Fluorescent protein bands corresponding to the appropriate zones identified by fixation and staining were sliced from the gel, homogenized, and either extracted repeatedly with KC1-phosphate buffer or a solution of 6 M urea, 1 nm EDTA, and 1 nm dithiothreitol in 0.05 M NaCl (pH 7.2), or used directly for preparation of antiserum.
Isoelectric focusing of the M, 12,000 FABP fraction was carried out on a thin layer of Sephadex G-75 superfine added to a solution of 11.4 l l l~ lysine and 9.6 mM arginine in 30 ml of H20 to which was added 4 ml of LKB ampholytes (pH 3.5 to 10) and 6 ml of sample protein in 0.13 M glycine (21); correspondingly higher volumes were used in preparative runs. In some experiments, [*4C]methyloleate was added to the protein sample so as to provide an electrically neutral marker of long chain acyl-binding fractions. Samples were subject to focusing at 400 V (4 "C) and allowed to stabilize for at least 18 h. Prints of the focused gels were made by brief application of a strip of special print paper (Desaga 12-20-27, 120 g/m'), followed by fixation and removal of ampholytes in 10% trichloroacetic acid and staining of the strip in a solution of 0.2% Coomassie brilliant blue R-250 ip methanol-wateracetic acid ( 5 5 1 ) (21). Appropriate zones (1-cm intervals in ["C] methyloleate experiments or specific protein bands in preparative runs) were then scraped and either assayed directly for radioactivity or pH, or proteins were eluted for subsequent analysis of electrophoretic, immunochemical, or fatty acid-binding properties, or amino acid composition after removal of ampholytes by Sephadex G-50 chromatography in KCl-PO, buffer.
Polyacrylamide gel electrophoresis in sodium dodecyl sulfate was performed according to the method of Shapiro and Maize1 (22). Ten per cent gels were run at 8 mA/gel for 4-5 h after which they were fixed and stained in a solution of 0.2% Coomassie brilliant blue in methano1water:acetic acid (5:51). The relative mobility of purified FABP was compared with that of protein standards to determine molecular weight.
Affinity Chromatography-This was employed to identify those components of the M, 12,000 FABP fraction which exhibited fatty acid-binding activity. Oleic acid was complexed to Bio-Rad Affi-Gel 101 by the method of Peters et al. (23). The partially purified FABP fraction was applied to the column, and protein fractions which passed immediately through or which were bound and subsequently eluted with 25% ethanol (pH 6) were analyzed by disc gel electrophoresis.
Immunochemical Studies-Protein fractions (approximately 200 pg) separated by polyacrylamide gel electrophoresis or thin layer isoelectric focusing were mixed with an equal volume of complete Freund's adjuvant and administered in multiple intradermal sites to New Zealand white rabbits. Specificity of antiserum and immunochemical identity of antigens in various protein fractions were as-sessed by the Ouchterlony double immunodiffusion method (24). Concentration of FABP in whole rat liver cytosol and in partially purified FABP fraction was determined by the quantitative radioimmunodiffusion method of Mancini et al. (25).
Amino Acid Analysis-The protein was precipitated from samples of FABP purified from the M, 12,000 fraction by isoelectric focusing (Band 5, see below) with 5% trichloroacetic acid. Trichloroacetic acid was then removed with an acetone wash. Samples were hydrolyzed in duplicate in 6 N hydrochloric acid (Ultrex, J. T. Baker Chemical Co.) at 105 "C for 22, 48, and 72 h. Analyses were made on a Beckman model 121" amino acid analyzer using a two-column technique.
Values for serine and threonine were extrapolated to zero time, and values for leucine, isoleucine, and valine were taken from the 72-h hydrolysate; these analyses were performed in duplicate. All other values were means of six determinations. Cysteine was determined as cysteic acid and methionine as methionine sulfone using the technique of Hirs (26). Tryptophan was measured following hydrolysis with mercaptoethanesulfonic acid (27,28). Miscellaneous Analytical Techniques-In column eluates, protein concentration absorbance was estimated at 280 nm and in these and other samples by the method of Lowry et al. (29). Actual FABP aminoacyl mass was calculated on the basis of the amino acid analysis and found to be only 0.547 of the mass as determined by the Lowry method. This discrepancy reflects both the 10.54% water of hydration of the albumin standard2 and the differences between albumin and FABP in the relative abundance of Lowry-reactive residues in the two proteins. In this communication, stated values for purified FABP protein (but not impure fractions) include this correction. Radioactivity was measured in a Beckman LS-250 liquid scintillation system employing an automatic external standard for quench monitoring and automatic quench correction; aqueous samples were suspended by a 10% solution of Bio-Solv BBS-3 (Beckman Instruments) in Liquifluor (New England Nuclear) and toluene. Statistical significance of differences between experimental groups was determined by the Student's t test.

Identification of Principal Fatty Acid Binding Protein Fraction in Rat Liver Cytosol-Although it
has been well documented that exogenous 14C-labeled long chain fatty acids are almost entirely bound to one or more proteins in the M, 12,000 range, the distribution of endogenous long chain fatty acids among cytosqlic proteins has not been defined. To address this question, whole rat liver cytosol was fractionated by Sephadex G-75 gel filtration; fractions were analyzed separately for protein (absorbance, 280 n m ) and were combined into four larger fractions for analysis of fatty acid and Lowry protein content (Fig. 1). It can be seen that fractions 30-34, the midpoint of which approximated the elution volume of M, 12,000 globular proteins, contained 60.3% of cytosolic long chain fatty acids (13.5 nmol of fatty acid/mg of protein) and far exceeded the other fractions in the amount of fatty acid per mg of protein. Moreover, even further erlrichment was observed in the M, 12,000 fraction as purified by two consecutive gel filtrations through Sephadex G-50 ( Table I). The FABP correction of the Lowry determination (see "Experimental Procedures") was not applied to experiments involving whole cytosol or M, 12,000 fraction; its application to the M , 12,000 fraction in Fig. 1 and Table I would have increased even further the differences in fatty acid:protein ratio between FABP and other fractions of cytosol.
Homogenization of liver in a buffer containing 5 mM EDTA did not significantly affect the fatty acid composition of the FABP fraction, suggesting that its apparent enrichment in polyunsaturated fatty acids did not reflect artifactual phospholipase Az-mediated fatty acid release during preparation of the sample. Furthermore, this fraction was found to contain virtually no detectable phospholipid (i.e. less than 0.01 m o l / mg of protein), suggesting that it is not likely t o be involved Rat Liver Fatty Acid Binding Protein to a substantial extent in phospholipid transfer or exchange. Delipidation of the FABP fraction with di-isopropyl ether, as described under "Experimental Procedures," removed virtually all of these noncovalently bound long chain fatty acids (Table I). Following delipidation, the protein in this fraction was stable for several weeks.
Isolation of FABP-The partially purified FABP ( M , 12,000) fraction obtained by Sephadex G-50 gel fdtration of rat liver cytosol was subjected to thin layer isoelectric focusing in the presence of tracer ['4C]methyloleate to identify fractions with significant affinity for long acyl chains. Native and delipidated preparations were analyzed similarly in simultaneous side-by-side runs. In Fig. 2, distribution of [I4C]methyloleate radioactivity and the pH of the gel are plotted directly above the corresponding focused patterns of native and delipidated FABP fractions. It can be seen that these preparations were resolved into a large number of components. For both preparations, however, recovered radioactivity was principally localized to a prominent protein band with an isoelectric pH of approximately 6.9 (Band 5). Smaller amounts of radioactivity were associated with proteins of isoelectric pH between 5 and 6 and between 7 and 8. The significantly greater association of radioactivity with the pH 6.9 band in the delipidated preparation, as compared with native, corresponded to an increased overall recovery of applied radioactivity (51.6 uersus 31.3%). This consistent difference in ['4C]methyloleate association with the pH 6.9 band may reflect increased binding capacity of the delipidated material and/or may reflect conversion of more acidic species to the pH 6.9 protein during delipidation. In the isoelectric focusing patterns, it can be seen that the delipidated preparation appears to contain less of the acidic bands and more of the pH 6.9 band.
Protein fractions were eluted from the isoelectric focusing gels and subjected to disc gel electrophoresis in a nondenatur- Distribution of endogenous long chain long chain f a t t y acids a m o n g rat liver cytosolic proteins. Rat liver cytosol was prepared as described in "Experimental Procedures"; 280.8 mg of cytosolic protein (4 ml) was applied to a Sephadex G-75 column (2.5 x 34 cm, 4 "C, 0.5 d / m i n ) , and 3.6-d fractions were collected and grouped as indicated. Aliquots were assayed for protein and fatty acid content as described.
ing buffer. In Fig. 3, the electrophoretic pattern of the native FABP fraction is compared with isoelectric focusing Bands 1 through 6. It can be seen that Bands 3 , 4 , and 5 exhibit nearly identical mobility in this system and correspond to the prominent uppermost broad band of the native partially purified M , 12,000 FABP fraction. B a n d 2 corresponds to the lower band, while Band 1 corresponds to the most rapidly moving components of the fraction. Band 6 was recovered in insufficient amounts in this experiment to be easily detected in the gel under these conditions but showed the same mobility as Bands 3, 4, and 5. Bands which corresponded to more basic Thin layer isoelectric focusing of ["C]methyloleate with the FABP (Mr 12,000) fraction of rat liver cytosol. Native and delipidated FABP fraction (16 mg) partially purified from whole cytosol by Sephadex G-50 gel filtration and 0.64 nmol of ['"Clmethyloleate in 1% glycine were added to separate suspensions each containing 2 g of Sephadex G-75 (superfine), 50 mg each of lysine and arginine, and 4 ml of carrier ampholytes (pH 3.5-10) in 30 ml of H20. Each mixture was subjected simultaneously and side-by-side to isoelectric focusing as described in "Experimental Procedures." The photographs of the stained offprints are aligned with each other and with the total radioactivity recovered in each I-cm zone along the entire length of th'e gel. The FABP fraction (M, = 12,000) was partially purified from rat liver cytosol by gel filtration twice in succession on Sephadex G-50 and was extracted and analyzed for free fatty acid by gas-liquid chromatography (see "Experimental Procedures"). The "EDTA" sample was prepared after homoeenization of tissue in 5 mM EDTA. The "delipidated" sample was extracted (after partial purification) with diisopropyl ether (see "Experimental Procedures").
proteins in the focusing gel and which did not bind significantly in these experiments did not enter the disc gel, as would be expected. The material from the lower band on disc gel (corresponding to Band 2B in isoelectric focusing) was employed as an antigen for preparation of an antiserum to FABP. This antiserum produced a single line of immunoprecipitation against the partially purified M, 12,000 fraction (Fig. 4, A and B, top and  bottom wells). It can be seen that there were reactions of identity between this line and that formed with Bands 2A, 2B, 3, 4, and 5. The same pattern was observed with antiserum prepared against Band 5." The very weak reaction of Band 1 (Fig. 4A) reflects contamination of this fraction by one of the acidic immunochemically identical forms of FABP as shown on overloaded gels3 Band 6 (not shown) also showed a reaction of identity, whereas the more basic fractions separated by isoelectric focusing failed to react with the antiserum. Of significance, the major FABP bands (Figs. 2 and 3) thus share immunochemical identity but differ from one another in charge and in isoelectric pH; this may in part reflect the amount of bound endogenous long chain fatty acid (as suggested by the effect of delipidation in Fig. 2) or differential binding of ampholines (30). Such an influence of noncovalently bound ligand on charge behavior could account for the observation that repeated electrophoresis or isoelectric focusing of an apparently single species results in the "reappearance" of the other major forms of the protein (9).2 Other factors which may contribute, e.g. proteolysis or covalent modification, are not excluded, although the failure to observe lower molecular weight forms among the various isoelectric focusing fractions suggests that proteolysis is not likely to be of major significance.
That isoelectric focusing Bands 2-5 represent the principal fatty acid binding components of this material is further demonstrated by affinity chromatography as shown in Fig. 5. In this experiment, the disc gel electrophoretic pattern of the native M, 12,000 fraction is compared with those components which fail to bind to the column and those which did bind and were readily eluted. The latter (i.e. the fatty acid-binding components) consists of 2 major bands and an interposed relatively minor band, corresponding to isoelectric focusing Bands 3-5 (upper) and 2B (lower). The interposed band was inconsistently separable from the upper band and was not further characterized.
Binding Characteristics of Native and Delipidated Proteins-To assess the relative binding affinity of the native and delipidated M, 12,000 fractions as well as the isolated FABP (Band 5), a Sephadex G-25 binding assay was employed using [I4C]oleate as ligand, as previously described (5). As shown in Table 11, the delipidated M, 12,000 fraction exhibited somewhat greater binding of [ 14C]oleate than did the native material, consistent with the findings on isoelectric focusing (Fig.  2). However, it also can be seen that the binding of [I4C]oleate by purified FABP (Band 5) was less than that for the native or delipidated partially purifed fraction. This may reflect residual binding of ampholytes to purified FABP after isoelectric focusing, despite the fact that efforts were made to remove them by means of Sephadex G-50 chromatography. The fact that exposure to ampholytes during isoelectric focusing did not appear to substantially diminish binding of ["C] methyloleate to Band 5 (Fig. 2) may reflect the fact that in this instance protein and lipid were added together prior to their exposure to ampholytes and reduced pH, whereas the isolated Band 5 (Table 11) had already been so exposed.  4. A and B, immunodiffusion analysis of FABP (Mr 12,000) fraction of rat liver cytosol and individual bands separated by isoelectric focusing. In each gel, the center well contained 25 pl of antiserum to FABI', and each peripheral well contained 0.5 pg of the indicated protein band, eluted after separation by isoelectric focusing, and 1.0 pg of M, 12,000 fraction.

FIG.
binding protein (isoelectric focusing Band 5) was found to be 12,080 k 80 ( n = 5) as determined by sodium dodecyl sulfatepolyacrylamide gel electrophoresis. In this system, Band 5 regularly migrated as a single band, even on overloaded gels; isoelectric focusing Band 2 occasionally contained much smaller quantities of a larger protein ( M , -17,000).

Age and Sex Differences in Rat Liver FABP Concentra-
tion-Employing the antiserum to purified FABP in a quantitative radial immunodiffusion assay, the concentration of FABP in whole cytosol was determined in livers from sexually immature and mature female and male Sprague-Dawley rats. As shown in Table 111, the immature (31-day) females and males did not differ significantly. With maturation, FABP concentration increased in both, but more so in females, so that by the time of sexual maturation (62 days) there was a significantly higher concentration in females than in males. These findings are in essential agreement with earlier measurements of rat liver FABP concentration by means of a binding assay (6). Moreover, the studies indicate that FABP accounts for 3.98 and 5.17% of cytosolic protein in adult male and female rat liver, respectively, and more than 50% of the partially purified M, 12,000 FABP fraction.
Amino Acid Composition of FABP-The amino acid composition of FABP (Table IV) FABP (M, 12,000) fraction was fractionated as described in "Experimental Procedures." Fractions were analyzed by disc gel electrophoresis. A, proteins which passed directly through the fatty acid affinity column. B, proteins which bound to the column and which were subsequently eluted with 25% ethanol in 0.075 M POn buffer (pH 6.0). C, residual bound proteins, eluted from the column with 0.05 M NaOH-EtOH (1:l).

Binding of ["CJoleate by rat liver FABP
Measurement of ["C]oleate binding to the indicated fraction was performed as described previously (see "Experimental Procedures" and Ref. 5). Protein (0.42 mg of FABP fraction, 0.23 mg of purified FABP) was mixed with 20 nmol of [''C]oleate in 8 pl of propylene glycol and applied in 1.0 ml to Sephadex G-25 (medium grade) in a column (0.9 X 21 cm) at 4 "C. Effluent fractions (1.2 ml) were assayed for radioactivity. "C eluting in the void volume was regarded as Drotein hound.

TABLE I11
Age and sex differences in rat liver FABP concentration FABP concentration in whole rat liver cytosol was measured by quantitative radial immunodiffusion (see "Experimental Procedures"). n = 4 (31 days) or 6 (62 days) mean k S.E. For 62-day females versus males. D < 0.05. Gaylor (10). Our analyses for tryptophan using mercaptoethanesulfonic acid, which gives higher yields of tryptophan, confirm the observation of Dempsey et al. (11) that tryptophan is absent from this protein. Our analyses of cysteine gave higher values than those reported for the other proteins. The content of methionine was also appreciably higher in FABP than those reported for the "A" and "Z" protein (9,31) and the "Band C" protein (10). Because detection of both cysteine and methionine is improved by the performic acid technique, these differences between our analyses and those reported from the other laboratories probably reflect losses of these amino acids in the other analytical techniques.

Amino acid composition of related liver cytosolic proteins
Rat liver cytosolic FABP (Band 5, Fig. 2) was analyzed for amino acid composition as described in "Experimental Procedures." Values shown for FABP are the means of 6 determinations for each amino acid. Values for A (form II), Band C, sterol carrier protein (SCP), and Z (pH 7.2) are taken from Refs. 9-11 and 31, respectively. A growing body of evidence suggests that the cellular utilization and metabolism of relatively nonpolar poorly watersoluble substances involves the participation of intracellular low molecular weight cytosolic proteins. With certain exceptions these proteins appear not to exhibit enzymatic activity in the usual sense, i.e. in directly catalyzing the formation or cleavage of covalent bonds. Rather, they have been identified largely as the result of, and are principally characterized by, their binding of more or less closely related groups of ligands. These proteins are viewed as facilitating the intracellular transport or exchange of their respective ligands or exerting a permissive or enhancing effect on the entry of these ligands into reactions catalyzed by membrane-bound enzymes (32). Among these proteins are the several sterol carrier (10-12) and phospholipid exchange proteins (13), retinol and retinoic acid-binding proteins (33), the glutathione transferases (the only class of such proteins which in fact appear to be enzymes) (34), the amino azo dye-binding (9) and Z proteins (31), and FABP (1, 5, 6, 14, 35). These proteins in many cases differ substantially in regard to molecular weight, charge, and substrate specificity but share in common the fact that the relationship between their properties in vitro and their function in vivo remains somewhat unclear. Furthermore, it now is evident that a number of these proteins, viewed differently by various investigators in terms of their biological function and significance, and designated accordingly, may in fact be the same (10, 11, 31). The subject of the present communication falls into this category.
The earlier description of FABP as a M, 12,000 acidic protein is confirmed in the present studies. It is also evident that the isolated protein exists in several immunochemically identical and a t least partially interconvertible forms, differing in isoelectric pH, and possibly in the amount of bound endogenous ligand. That bound ligand may influence the charge of the carrier molecule is shown by the well documented effect of bound fatty acid on the isoelectric pH of serum albumin (36). Other possible explanations for the existence of charge isoforms of FABP are not excluded. Since these different charge forms of the protein are seen as identical by the antiserum employed in these experiments, it has not been possible to quantify each independent of the others. It is noteworthy, however, that in the aggregate they comprise approximately 5% of liver cytosolic protein and a major portion of the protein mass in the M, 12,000 fraction. Moreover, the important influence of sex and maturation of the animal on the concentration of hepatic FABP as previously demonstrated by binding assay (5,6) is confirmed immunochemically in the present study.
On the basis of the physical properties and the amino acid composition of the protein, it seems highly probable that FABP is in fact the same protein designated by Billheimer and Gaylor (10) and Dempsey (11) as sterol carrier protein, by Ketterer as amino azo dye-binding protein A (9), and by Arias and colleagues as Z (31). These workers have attributed to the protein a role in the metabolism and/or intracellular transport of cholesterol, heme, carcinogens, and cholephilic anions such as bilirubin, sulfobromophthalein, and indocyanine green, respectively, although there is no direct and conclusive indication that any or all of these putative roles is valid in vivo. On the other hand, an increasingly compelling body of circumstantial evidence supports a relationship of this protein to cellular transport and utilization of long chain fatty acids. This evidence can be considered in two broad categories.
First, the concentration of FABP correlates with overall rates of cellular fatty acid uptake, transport, and utilization in several circumstances. In the intestinal epithelium, the concentration of a similar but nonidentical FABP (2) is greater in mucosa from villi than crypts, in jejunum than ileum, and in mucosa from animals fed a high fat diet than those fed a low fat diet. In liver, the influence of sex steroids on cytosolic FABP concentration noted above is closely paralleled by sex steroid effects on rates of total fatty acid utilization and triglyceride biosynthesis in rat hepatocytes in suspension (5, 6) and culture4 and in net uptake in isolated single pass perfused rat liver (37). Similarly, clofibrate administration increases both hepatic FABP concentration and fatty acid uptake (3, 4); also, binding activity of hepatic FABP reportedly is increased in the Zucker obese rat (38), in which total free fatty acid flux is increased. FABP is also present in a wide variety of tissues such as myocardium, adipose tissue, skeletal muscle, and kidney which utilize free fatty acid (1,39). Finally, we recently have found that a M, 12,000 FABP increases dramatically in the cytosol of 3T3-Ll mouse embryo fibroblasts as they differentiate into adip~cytes.~ The second line of evidence suggesting a relationship of FABP to cellular fatty acid utilization consists of studies in several laboratories demonstrating an influence of FABP in uitro on the activity of a number of enzyme reactions involved in fatty acid metabolism. Thus, FABP has been shown to enhance the activity of hepatic mitochondrial and microsomal acyl-CoA synthetase (14), peroxisomal fatty acid oxidation (41), microsomal glycerophosphate acyltransferase (14, 42) and diglyceride acyltransferase (43), and acetyl-coA carboxylase (35). In the intestine, enhancement of microsomal acyl-CoA synthetase (2) and diglyceride acyltransferase (43) has been demonstrated. In several of these studies, the FABP effect was concentration dependent and suggested that variations in FABP concentration within the physiologically relevant range in vivo could modulate enzyme activity.
It should be noted that other evidence is consistent with the possible relationship of these proteins to the metabolism of cholesterol (10, ll), heme (9, lo), and other ligands (9, 31), but in general this evidence is less extensively developed. Even if it is accepted, however, that this closely related or identical group of proteins (FABP, sterol carrier protein, A, Z) is principally related to cellular fatty acid utilization, the nature of that relationship remains unclear. As noted, in vitro evidence is consistent with an FABP effect solely on the interaction of fatty acid substrate with particulate enzymes. In this model, FABP could be viewed either as a vehicle for the efficient transfer of fatty acid to the active site of the enzyme or as a relatively nonspecific "solubilizer." The generally similar effects of albumin in this regard (14) would tend to support the latter interpretation, but the former is by no means excluded. It is also possible that FABP facilitates desorption of fatty acid from the cytosolic aspect of the plasma membrane during uptake, a concept suggested by evidence that desorption may be rate limiting in transport of amphipathic ions such as fatty acids (44-46) and for which preliminary direct experimental support is already available (47). Or, formation of the fatty acid-FABP complex could facilitate the movement of fatty acid through the cytosol, achieving a greater overall net fatty acid flux rate despite a reduced diffusion coefficient by virtue of a greatly increased concentration (14). In regard to these postulated physical effects on the movement of fatty acid within the cell, it may be of particular significance that the abundance of FABP mass in cytosol suggests that expression of its function requires a relatively high molar concentration (estimated from the present data to be 0.4 m~) compared to that of most other cytosolic proteins and that it is involved in processes characterized by high flux rates of relative nonpolar molecules. The flux of long chain fatty acids through liver dwarfs that of other amphipathic substances, as discussed previously (14), and supports the concept that FABP is primarily related to the metabolism of fatty acids rather than that of less abundant moieties such as bilirubin, bile acids, or xenobiotics.
The foregoing considerations notwithstanding, much remains unknown about this protein. Although its cytosolic concentration can be modulated, apparently reflecting changes in synthesis (48), the nature of the effective ''signal'' for these responses is not defined. Thus, on the one hand, it is equally possible that FABP concentration controls cellular fatty acid flux or, on the other hand, that it is responsive to fatty acid flux. Furthermore, the physical and structural properties of the protein which are essential in its function remain to be defiied, as does its possible significance in states of abnormal fatty acid metabolism. These and other fundamental questions about this abundant but still poorly understood protein species are under investigation and only now may be on the threshold of being answered.