The Primary Structure of Rat Liver Cellular Retinol-binding Protein*

The complete amino acid sequence of a cellular re-tinol-binding protein (CRBP) has been determined for the first time. The primary structure of rat liver CRBP was elucidated by analyses of cyanogen bromide frag- ments and peptides obtained by tryptic and thermolytic digestions. The single polypeptide chain of rat CRBP consists of 134 amino acid residues. Under reducing conditions, CRBP exists as a monomer, but, in the absence of reducing agents, dimers and multimers of the protein emerge. This is explained by the observation that CRBP contains 3 cysteines, one of which seems to be highly reactive. Whether CRBP contains a disulfide bond is not yet established. The present data extend the previously described homology between CRBP and a family of low molecular weight proteins, all members of which may bind hydrophobic ligands. Since some of these proteins ap- parently display intracellular transport functions, a similar role for CRBP is envisaged.

Vitamin A is transported in plasma from its storage sites in the liver to various epithelial tissues by retinol-binding protein (RBPl). The transport of vitamin A in plasma has been extensively studied from both biochemical and physiological aspects (for reviews, see Refs. 1 and 2).
In contrast, only limited information is available about the storage and metabolism of vitamin A in the liver and in epithelial cells. The presence of two vitamin A-binding proteins with ligand specificities for retinol (cellular retinolbinding protein (CRBP)) and for retinoic acid (cellular retinoic acid-binding protein (CRABP)) has been demonstrated in the cytosol of several types of cells (3). Both proteins have been isolated and found to have molecular weights of approximately 15,000 (4)(5)(6)(7)(8). The NHz-terminal sequences of CRBP and CRABP were found to be highly homologous to each other (9,10). Unexpectedly, a similar degree of homology was also found between these two proteins and the myelin protein P2 (11), a protein without any known connection to vitamin A. More recently, the fatty acid-binding Z-protein was shown to be a member of this protein family (12).
The intracellular vitamin A-binding proteins have been suggested to mediate, directly or indirectly, the effects of the * This work was supported by a grant from the Swedish Medical Research Council and by Grant 5 ROI Ey 02417-02 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
4 To whom correspondence should be sent.
The abbreviations used are: RBP, retinol-binding protein; CRBP, cellular retinol-binding protein; CRABP, cellular retinoic acid-bindingprotein; HPLC, high pressure liquid chromatography; PTH, phenylthiohydantoin; TFA, trifluoroacetic acid. vitamin (3). Further studies of the physiological role of these proteins would be facilitated if more were known about their molecular characteristics. We are therefore trying to elucidate the primary and tertiary structures of the intracellular binding proteins. We report here the primary structure of rat CRBP.

RESULTS
NH2-terminal Sequence Analyses of CRBP and CNBr Fragments Thereof-Automated NH2-terminal sequence analysis of reduced and S-carboxymethylated CRBP allowed the identification of the first 45 amino acid residues (Figs. 1 and 9). One of the 5 methionines present in CRBP (cf. Table I) was located at position 10.
Cleavage with CNBr resulted in two large and three small fragments (CB1-CB5) which were isolated by gel chromatography (Fig. 2). The amino acid compositions of the fragments are given in Table I together with that of the intact CRBP chain. A comparison of the amino acid compositions of the CNBr fragments with the NH2-terminal sequence of CRBP showed that CB1 constitutes the NHZ-terminal 10 residues of CRBP.
CB2-CB5 were subjected to NHz-terminal sequence analyses. Amino acid residues were identified in a total of 91 positions ( Fig. 1, Table 11, and Fig. 9). The sequence determinations demonstrated that CB1 is followed by the long fragment CB2 in the sequence.
Isolation and Analyses of Tryptic and Thermolytic Peptides of CRBP. Deduction of the Primary Structure of CRBP-Five peptides, Tl-T5, were isolated from a tryptic digest of maleylated CRBP by gel and ion-exchange chromatography as well as by HPLC (Figs. 4 and 5, and Table 111). The NH2terminal part of T2 and virtually the complete sequences of T1 and T3-T5 were determined by automated Edman degradation (Table JV). The sequence information on T1 and T2 enabled the alignment of CNBr fragments CB3 and CB4, and confirmed the order of all CNBr fragments, as outlined in Fig. 9 (see below). The sequence determination of T3 revealed the presence of a Met-Thr bond (residues 83-84). The difficulty to cleave such a bond with CNBr (15) explains why five  CNBr fragments were obtained and not six, as expected from the number of methionines present in CRBP (cf. Table I).
The amino acid sequence of the COOH-terminal part of CB4 was obtained by analyses of the tryptic peptides T3 and T4. Sequence determination of peptide T5 confirmed the sequence of CB5 and extended it up to the COOH-terminal residue of CRBP.
A number of thermolytic peptides of CRBP were isolated by gel chromatography followed by HPLC (Figs. 6-8) and subjected to amino acid analyses. Two peptides, TL1 and TL2, were chosen for amino acid sequence determinations (Tables V and VI). The sequences of TL1 and TL2 overlapped those of the tryptic peptides T3-T5, and thus provided the information needed to establish the primary structure of CRBP (Fig. 9).

DISCUSSION
CRBP is composed of 134 amino acid residues. The calculated molecular weight, 15,700, agrees with that determined by physical methods. Thus, Ong and Chytil(4) estimated the molecular weight of CRBP to be 14,600 by averaging the results from three different methods, whereas Liou et al. (8) measured the apparent molecular weight of CRBP from a sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis at 15,900.
CRBP contains 3 cysteine residues, but it is not known if a disulfide bond exists in the native molecule. CNBr fragmentation of unreduced CRBP shows that the most COOHterminal cysteine residue can form a disulfide bond (Fig. 2B). This bond could, however, have been formed during preparation of the protein.
Under nonreducing conditions, newly isolated CRBP migrates as a monomer on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Upon prolonged storage or repeated thawing and freezing, dimers and multimers of the protein are formed. An interchain disulfide is likely to be involved in this process, since the dimer disappears upon red~ction.~ This indicates that at least 1 of the cysteine residues is exposed. Another indication that 1 cysteine residue is located at the surface of the molecule is the observation that the organomercurial compound p-chloromercuriphenylsulfonic acid reacts with native CRBP (16).
p-Chloromercuriphenylsulfonic acid treatment abolishes the binding of retinol to CRBP (16). The abolished interaction is reversible, as a reducing agent restores the binding capacity. Although it is possible that 1, or more, of the cysteine residues is located at the retinol-binding site, it is also possible that the inhibition of retinol-binding by p-chloromercuriphenylsulfonic acid is due to steric hindrance or to an induced conformational change in the CRBP molecule. Further work is needed to clarify the status of the cysteine residues and their possible involvement in the binding of retinol to CRBP.
Partial sequence data showed that CRBP is homologous to CRABP and protein P2 (9)(10)(11). Later, Takahashi et al. (12) noted that the Z-protein belongs to the same protein family. The data now available allow a more comprehensive analysis of the relatedness of these proteins. In the following paper (17), describing the complete amino acid sequence of bovine CRABP, such an analysis is presented. One should point out, however, that no significant homology was found between CRBP and plasma RBP.
The molecular functions of the four proteins, CRBP, CRABP, protein P2, and the Z-protein, are not clearly defined. However, it is well established that the Z-protein binds fatty acids (18) and it has been suggested that it interacts with several microsomal enzymes, thereby influencing the biosynthesis of cholesterol (19). Protein P2 is a peripheral nerve membrane protein, which most probably interacts with lipid components in the myelin layer (20). Accordingly, all hitherto identified members of this protein family bind hydrophobic low molecular weight substances and it seems rea-

Primary
Structure of Rat CRBP sonable to suggest that CRBP and CRABP might also be able to interact with membranes as do protein P2 and possibly the Z-protein. No continuous stretch of hydrophobic amino acid residues longer than a few residues is present in either the CRBP sequence or in that of protein P2 (Fig. 9). However, a hydrophobic surface might very well be created by the threedimensional folding of the peptide chain. This will be further clarified by the determination of the three-dimensional structure of CRBP, well under way (21).
It might be inferred from the sequence comparisons with protein P2 and the Z-protein that CRBP has evolved from an ancestor cytosol lipid-binding protein with the ability to interact with membrane structures.
One biological role of CRBP might accordingly be to store vitamin A in the cytoplasm and to transport the vitamin between cellular organelles. A role of CRBP as a cellular storage and transport protein for vitamin A implies the ligand saturation of CRBP to be high under normal nutritional conditions. This seems to be the case. CRBP isolated from bovine pigment epithelial cells is completely saturated with retinol (22

COOH-Terminal determinations
The COOH-terminal sequemes of CNBr-fragment CB5 and oi i n t a c t .
reduced and alkylated CRBP were analyzed by digestions with carboxypeptidase A ag Outlined i n Fig. 3. Due to the very rapid release or both hlstldlne and Valine could not the sequential order of these amino a i d s be detcmlned from t h l s experiment. Sequence deterninatlon of the tryptic peptide T5 (Table IV)