Molecular Characterization and Functional Expression of Squid Retinal-binding Protein

The primary structure of squid retinal-binding protein (RALBP) was determined by cDNA and protein sequencing. Squid RALBP contains 342 amino acid residues  in  a single  N-terminal-blocked  chain  with  a molecular weight of 39,111. The N,-blocking group was identified as an acetyl moiety by mass spectrometry. The amino acid sequence revealed that the protein is highly hydrophilic and acidic, but it has several hydrophobic regions that are located mainly in the middle part of the polypeptide chain. It is also predicted that these hydrophobic regions form &sheet structures. The primary structure of RALBP is, however, quite distinct from those of other retinoid-binding proteins, showing that squid RALBP is a novel hydrophobic ligand-binding protein that functions in intracellular retinoid transport. Using the cloned cDNA, squid RALBP was expressed in vitro. By carrying out the translation at 20 "C in reticulocyte lysates, the protein having retinol binding activity was produced.

In vertebrates, five types of retinoid-binding protein have been isolated, and both their primary structures and their functions in retinoid transport have been studied in detail. Retinolbinding protein (RBP)' belongs to a family of plasma proteins that bind and transport extracellular hydrophobic ligands (1,2). RBP transports all-trans-retinol in plasma from the liver to retinal pigment epithelial cells (1,3). Cellular retinol-binding protein (CRBP) and cellular retinoic acid-binding protein (CRABP) belong to a family of cytoplasmic proteins that function in the intracellular transport of hydrophobic ligands (4)(5)(6).
Research on F'rioritykeas (62621001,6362001, and 0162001) (to K. 0.) * This work was supported in part by Grants-in-Aid for Scientific from the Ministry of Education, Science and Culture, Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18  The abbreviations used are: RBP, retinol-binding protein, CRABP, cellular retinoic acid-binding protein, CFtALBP, cellular retinal-binding protein; IRBP, interphotoreceptor retinoid-binding protein; FtALBP, retinal-binding protein, kb, kilobase(s); HPLC, high performance liquid chromatography; PAGE, polyacrylamide gel electrophoresis; bp, base pairs(s).
In the eye, CRBP and CRABP bind cytoplasmic all-trans-reti-no1 and all-trans-rekinoic acid, respectively (7). Cellular retinalbinding protein (CRALBP) is also a cytoplasmic retinoid-binding protein (8). However, CRALBP binds 114s-retinal and 11cis-retinol in the eye (7), and it is distinct from CRBP and CRABP in terms of primary structure (9). Interphotoreceptor retinoid-binding protein (IRBP) is a high molecular weight protein located in the interphotoreceptor matrix of the retina (10,111, and it transports retinoids between the retinal pigment epithelial cells and the photoreceptor cells (12,131. On the basis of the above studies, a possible pathway has been proposed for transport of retinoids to the photoreceptor cells (14)(15)(16). Briefly, all-trans-retinol bound to RBP is transported in plasma from the liver to the retinal pigment epithelial cells, where the retinol is esterified to all-trans-retinyl ester and stored. When necessary, the retinyl ester is directly processed to 11-cis-retinol, followed by oxidation to retinal. Retinoids are carried by CRBP and CRALBP in the pigment epithelial cells, and they are transferred to extracellular IRBP for transport to the photoreceptor cells. In the photoreceptor cells, no retinoid-binding protein has been identified, and the mechanism for retinoid transport in the photoreceptor cells remains to be clarified.
Three types of soluble retinal-binding protein (RALBP) have been reported in the case of invertebrate retinas. Two such proteins were found in the honeybee retina, and they probably function in the isomerization of retinal from all-trans to ll-cis-retinal in the light (17,18). We found another RALBP in the squid retina (19). This protein is distributed throughout the outer and inner segments of the photoreceptor cells (20). Our recent studies in vitro and in vivo have revealed that squid RALBP transports 11-cis-and all-trans-retinals between rhodopsin and retinochrome, and that it regenerates photoproducts of both pigments to maintain the high sensitivity of the photoreceptor cells (21,22). Although it has been suggested that squid RALBP might mediate the exchange of chromophores between retinochrome and rhodopsin, details of the molecular mechanism for such a n exchange still remain unknown. In the present study, we determined the primary structure of squid RALBP and expressed the protein in vitro. This study provides information that may help in efforts to understand the molecular mechanism of the transport and the exchange of retinoid chromophores. eyecups were shaken in 25 m~ histidine/HCl buffer that contained 0.3 M NaCl (pH 6.2). in order to detach the outer segments of the visual cells. The suspended outer segments were then homogenized and centrifuged at 25,000 x g for 1 h. The resultant clear supernatant contained RALBP. After concentration, the supernatant was applied to a polybuffer exchanger column (PBE94, Pharmacia; 10 mm inner diameter x 200 mm) and eluted by 25 m~ histidine/HCI buffer (pH 6.2) with a linear gradient from 0.3 to 1.0 M NaCI. Fractions containing RALBP were pooled and concentrated by ultrafiltration.
In the second purification step, high performance size-exclusion chromatography was camed out on a Hitachi model 635 liquid chromatography system equipped with a size-exclusion column (TSK G-3000, Toyo-Soda; 7.5 mm inner diameter x 600 mm). The column was preequilibrated with 0.2 M phosphate buffer (pH 6.5), and 0.1 ml of the concentrated sample was injected into the column. The sample was chromatographed with the same buffer a t a flow rate of 0.5 mumin a t room temperature, and the separated proteins were monitored by absorbance at 280 or 360 nm. Fractions containing RALBP were pooled and concentrated as described above, and stored a t -20 "C until use.
Preparation of RALBP-specific Antiserum--Rabbits were immunized by the standard procedure for preparation of antisera. Each animal received purified RALBP emulsified with Freund's complete adjuvant. After 3 weeks, a second injection of antigen in Freund's incomplete adjuvant was given. Samples of blood were collected 1 or 2 weeks after the booster injection.
Preparation of Retinal Poly(A)' RNA-RNA was prepared as described by Ullrich et al. (23) with slight modifications. The frozen eyes were cut in half for removal of the anterior part and lens, and the eyecups were homogenized in buffer (5 mueyecup) that contained 6 M guanidine thiocyanate, 5 m~ sodium citrate, 0.1 M p-mercaptcethanol, and 0.5% sodium lauryl sarcosinate using a Hiscotron homogenizer.
After centrifugation a t 80,000 x g for 30 min, the supernatant was mixed with CsCl(0.4 g/mU and layered onto a cushion of 5.7 M CsCI, 0.1 M EDTA. The homogenate was then centrifuged a t 20 "C for 24 h a t 80,000 x g, and the resultant pellet of RNA was dissolved in water. The solution was treated with phenoVchloroform. RNA was recovered by ethanol precipitation and redissolved in water. Poly(A)+ RNA was separated from total RNA by affinity chromatography on a column of oligo(dT)-cellulose, according to the method previously described (24).
Construction and Screening of a cDNA Library-A cDNA library was constructed from poly(A)* RNA using vector-primer DNA, according to the method previously described (25). To prepare the vector-primer, PstI-digested pTTQ18 ( h e r s h a m ) was tailed with TTP by use of terminal transferase (Takara Shuzo), and the T-tailed vector was redigested with XbaI. The T-tailed vector-primer was annealed with the denatured poly(A)* RNA and the first-strand cDNA was synthesized by RAV-2 reverse transcriptase (Takara Shuzo). The second-strand cDNA replacement was performed with RNase H and DNA polymerase I (Takara Shuzo). The double-stranded cDNA with plTQl8 was then blunt-ended and ligated to form circular plasmid. Escherichia coli cells (Epicuriancoli SCS1; Stratagene) were used for the hosts of the plasmids. The library was screened with RALBP-specific antiserum. Protein A-alkaline phosphatase conjugate was used as the second antibody and positive clones were visualized with 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium. Five positive clones were isolated from approximately 1 x lo6 independent recombinants. Three of them remained positive in the second screening. Two positive clones (pTTB1 and pTTB5) contained the identical cDNAinserts of 2.8 kb, as revealed by analysis with various restriction endonucleases. The other clone (pTTB7) contained a shorter (2.0 kb), truncated insert.
DNA Sequencing and Analysis-Both strands of the cDNA insert of pTTB5 were sequenced by the dideoxy chain termination method using a DeazaGTP Sequencing kit (Nippon Gene). The sequencing strategy is shown in Fig. 1. For sequencing of the coding strand, the cDNA insert of p"B5 was subcloned into the M13 mp19 vector and deletion mutants of the clone were constructed by treatment with exonuclease 111 and mung bean nuclease, using a Kilo-sequence Deletion kit (Takara Shuzo). In order to sequence the antisense strand, the cDNA was subcloned into the M13 mp18 vector (Toyobo) and the deletion mutants were constructed as described above. In addition, the cDNA insert was digested with AatI, AcyI, BamHI, HincII, HindIII, HinfI, andNaeI, and the resultant fragments were subcloned in the M13 mp18 vector for sequencing.
Northern Analysis of Retinal Poly(A)+ RNA-Total RNA (22.3 pg) from the squid retina was subjected to electrophoresis on 1.5% agarose, 6% formaldehyde gels, as described by Maniatis et al. (24); capillarilly transferred to Hybond-N nylon membrane ( h e r s h a m ) in 20 x SSC (1 x SSC contains 150 m~ NaCl and 15 m~ sodium citrate), and fixed on the membrane by irradiation with UV light. The cloned cDNA for RALBP was labeled with [a-32PldCTP by use of a random primer labeling kit (Nippon Gene) for use as a probe for detection of the mRNA for squid RALBP. Hybridization was camed out at 42 "C for 18 h in 50% formamide, 5 x SSC, 0.02% Ficoll, 0.02% polyvinylpyrrolidone, 0.02% bovine serum albumin, and 0.2% SDS. The membrane was then washed successively in 2 x SSC a t 25 "C, in 2 x SSC that contained 1% SDS at 65 "C, and in 0.1 x SSC a t 25 "C, and then it was exposed to x-ray film (X-Omat AR; Kodak).
Amino Acid Analysis and Sequencing-Twenty nmol of purified RALBP were denatured in 6 M urea and cleaved with lysyl endopeptidase. Resultant peptides were fractionated by reverse-phase HPLC on a CIS column (4-mm inner diameter x 300 mm; Nacarai Tesque). Amino acid compositions of the purified peptides were determined with an amino acid analyzer (A-5500; Inca) after hydrolysis by HCI in the presence of thioglycolic acid. Amino acid sequences of peptides were determined by manual Edman degradation.
Mass Spectrometry-The N-terminal blocking group was identified by analysis by fast atom bombardment mass spectrometry of the Nterminal lysyl peptide (K14) obtained by the method described above. The peptide (1.5 nmol; 3 pl) was mixed with glycerol and then placed as a target in the xenon ion beam.
Synthesis of lbtal Retinal Protein in Vitro-Total RNA (6 pg; 6 pl) from the squid retina was mixed with 12 pl of a rabbit reticulocyte lysate (Amersham) and [35Slmethionine (37 TBq/mmol; 10 p~; 2 1.11). The mixture was incubated a t 30 "C for 1 h, and the reaction was terminated by adding 8 pl of 10% SDS to the reaction mixture. RALBP was isolated from other retinal proteins synthesized in vitro, as follows. After the termination of synthesizing reaction by SDS, 30 pl of the reaction mixture was mixed with 8 pl of 20% Triton X-100 and 400 pl of STE buffer (100 m~ NaCI, 1 m~ EDTA, 10 m~ TridHCI; pH 8.0), and the RALBPantibody conjugates were formed by adding 2 pl of RALBP-specific antiserum, with subsequent incubation at 4 "C for 12 h. The conjugates were recovered by batch chromatography on Protein A-Sepharose (1.7

K7 G A G M C A C A G A C G A C T T T G M A C M T G G A M C M T C A C C G T T G G M G T G G C G A C M G A T T T A T G T A G M T A T G~T T G~T G M M T E N T D D P E T M E T I T V G S G D K I Y V E Y E I E N E N K11 A C C T A C A T C A M T G G G M T A C M G A C T G M G M C A T G A C A T T G G T T T C G G T C T C T T C A G~G M C G G T G A T G M T G G G M G M G T C G T A T Y I K W E Y K T E E H D I G P G L P R K N G D E W E E V V K5
K12 K13

G G C C G G A G A G T G A T A T M A G A C G A C A M G A C G G A G A M C T G A T T G M G T G M T C A T T G T A T T A T A G M A C M C C C G C C G G C C A C A M T A C 1440 M C T A T T A T A T C C T C A C C M T T A T C C A T M T T G M T A M T C A T G M T M T T T T A A M M T M T T A A M G T T A C M T T A T T M T~T 1530
CGTATTGGTGMTGTGTATMTTTCTTTTCCGMTTATCAMGTTTTTTTCTAMCTTTCATTGGATATTCTGTTAGATMTCCCTATM 1620 ACGTGATCGTCTCTATMlLMMTTGGTGCCATMGTTTTCACGTTGMMGCTTCTGTTCTGTTTTTTTTTCCATTATATT~TTTCCTA 1710

A T G G A G T T T T G T T T T G A M T A G M C M G A G T T M C G G C A A M T G T A T T T T T A M G M T G A T A T M G C M T A T T A T C T T T A T M G T A T T T T 1800 CAGAMTTAGTACCGGTTTCCTTACATCATCATATTACTTACAMTACATCTCTATATTTTTTTTCATCTTTMCAMTTAG~CAGATATG lago T A C T T T C A T A T C A T G T A T T C T T C A T M C T T T T T A T T T A T C T G T~T T~T C A A M T M G T A T T T T G T T T M C~T A C C M 1980
ATMTTlVLMTATTGCGTGATCTGMCAGCGCGCACTMGTGGGTATAGTAGAGCAGTACTCTTGGGAGCAMTAGGCTTTATGMTCCG 2070 GCACGCCGATTTCATTTTATGTGATGTTAGGCCAMTCGGCATTTATGGTGTTCCTTCTTTTCATGGTCGTCMCCTCCTTGAMCATTCTG 2160 mg). After through washing of the RALBP-antibody-Protein A-Sepharose complex with STE buffer, proteins were solubilized in SDS-containing buffer for SDS-PAGE. In order to check the specificity of the antiserum and the possibility of the nonspecific binding of synthesized proteins to the antibody-Protein A-Sepharose complex, a similar experiment was carried out by use of the antiserum previously incubated with purified RALBP.

A G T T G C A C C A M T A T A T A T M C A C T C T C M C A G A G C C T T T G G A A M C C A T A T T T C C A G A C T G A M T M C A G G C C T A T C T M C A T M C M C 2250 C M T C A T T T T C T T T A G M T C G T C T T T T A M T T C A T M M G G G A T T C G T T G A G M C C M T C G A C A T G T T G T T C M C T T C T C G C G T G A M C
Synthesis of Cloned U B P in Vitro-In order to synthesize capped mRNA of RALBP in vitro, full-length cDNA was subcloned in the pBS vector (Stratagene) with the T7 RNApolymerase promoter (pBSB1). The plasmid was linearized with SphI, treated successively with proteinase K and phenol/chloroform, and recovered by ethanol precipitation. RNA with a 5'-cap structure was synthesized from the linearized DNA by use of an mCAP mRNA Capping kit (Stratagene). T7 RNA polymerase was used for RNA synthesis. After incubation at 37 "C for 1 h, template DNA was digested with deoxyribonuclease. The synthesized capped RNAwas recovered by ethanol precipitation and dissolved in water. The pBS vector, carrying 5'-end-truncated RALBP cDNA(pBSBlD1X was constructed as follows (Fig. 7A). The plasmid pBSBl was doubledigested with EcoRI and StyI, blunted with mung bean nuclease (Takara Shuzo), and ligated with T4 DNA ligase. As a result of this treatment, the complete 5'-untranslated region and the first 134 nucleotides of the translated region were deleted. After the truncation, it was anticipated that translation would be initiated at Met-50, resulting in the deletion of 49 N-terminal amino acids. Capped RNA from pBSlDl was also synthesized in the same manner as described above.
For expression in vitro of full-length and truncated RALBP, a solution of capped RNA (3 pg; 3 pl) and [Wlmethionine (37 TBq/mmol; 10 p~; 3 pl) were added to 19 pl of a rabbit reticulocyte lysate (Amersham). The mixture (25 pl) was then incubated at 20 "C for 2 h for synthesis of RALBP. For control experiments, 3 pl of water were added to the reaction mixture instead of the solution of capped RNA. In order to measure retinol binding activity of RALBP, 2 pl of r3H1retinol in a 3% solution of bovine serum albumin plus 1 pl of unlabeled L-methionine ( 1 1 1 1 ) were used instead of 3 pl of radiolabeled methionine. used directly for analysis by SDS-PAGE. SDS-PAGE was performed on slab gels by the method of Laemmli (26). The concentration of acrylamide was 5% in the stacking gel and 12% in the running gel. f i r electrophoresis, the proteins were stained with Coomassie Brilliant Blue R-250, incubated in enhancer solution (EN3HANCE; Du Pont-New England Nuclear), and exposed to x-ray film (X-Omat AR; Kodak). For the retinol-binding assay, 10 pl of SDS-solubilized sample were added to 10 pl of scintillation mixture (Clear Sol; Nacarai Tesque), and photons in the 3H-window were counted in a liquid scintillation counter (LS-9000; Beckman) for 10 min.

RESULTS
Cloning and Seqwncing of Squid U B P cDNA-Three clones recognized by polyclonal RALBP-specific antibodies were isolated by screening 1 x lo5 recombinant plasmids. Analysis with restriction endonucleases indicated that two of these clones (p' I"l' B1 and p'l"B5) contained a n identical cDNAinsert of about 2.8 kb, and another clone (p'lTB7) contained a 2.0-kb insert that lacked the 5"region (0.8 kb) of the longer insert of p'ITB1 and P'ITB5. Therefore, we sequenced the cDNA insert of p'l"B5.  (Fig. 1B) showed that the RALBP mRNA consists of at least two classes of transcripts, a major 2.8-kb and a minor 1.7-kb transcripts, suggesting that the AATAAA sequence at nucleotides 1474-1479 and/or nucleotides 1521-1526 is also used as a polyadenylation signal.
The RALBP mRNA is characterized by a long 3"untrans- Separation of peptides derived from squid RALBP by digestion with lysyl endopeptidase. As described in the text, RALBP that had been purified from the squid retina was digested with the enzyme, and the digest was fractionated by reversed-phase HPLC. Material in the peptide peaks indicated (Kl-Kl6) was collected and subjected to amino acid analysis. Sequence analysis was also carried out on peptides K4 and K9. K14 was subjected to mass spectrometory (Fig. 4).  M p o o 2 u K ) a B a o 2 8 0 0 3 M x ) (25,27). Molecular Structure of Squid RALBP-From the nucleotide sequence of FWLBP cDNA, we deduced the amino acid sequence of the encoded polypeptide (Fig. 2). In order to confirm the deduced sequence, we determined the amino acid composition of 16 peptides (Kl-Kl6) derived from purified RALBP by digestion with lysyl endopeptidase (Fig. 2, underlined; Fig. 3). Partial amino acid sequences of K4 and K! 3 peptides were also determined by Edman degradation. Of the 16 peptides, 15 peptides (Kl-K13, K15, K16) had amino acid compositions identical to those of corresponding peptides in the deduced sequence (Table I). Amino acid sequences of K4 and K9 were also consistent with those of the deduced sequence. K10 was expected to be the C-terminal peptide, since it was the only fragment without a lysine residue. In fact, the amino acid composition of K10 corresponded to that expected from the deduced sequence of the C-terminal peptide. The amino acid composition of the K14 peptide corresponded to that of the N-terminal fragment of the deduced sequence except that 1 of 2 methionine residues was missing. Since this result suggested that the first methionine of RALBP might have been removed by post-translational modification, we tried to determine the N-terminal sequence of purified RALBP. However, since no amino acid could be detected by Edman degradation, it appeared that RALBP has an N-terminal blocking group. Therefore, we carried out mass spectrometric analysis of the N-terminal lysyl peptide (K14). As shown in Fig. 4, the ion peak showing the exact mass ([M + HI' ) was observed at mlz 2617.3, indicating that the molecular weight of the original peptide was 2616.3. This value coincides with the theoretical weight of the K14 peptide, if 1 methionine residue is removed from the peptide and the peptide is acetylated. From these results, we concluded that RALBP was subjected to posttranslational modifications in which the N-terminal methionine is removed and the second residue, serine, is acetylated.
From the above results, the molecular weight of RALBP, including the N-terminal blocking group, was calculated to be 39,111. This value is somewhat smaller than that of purified RALBP, as estimated by SDS-PAGE (48,000; Fig. 6).' However,  since the molecular weight of RALBP expressed in vitro from both the total retinal mRNA (Fig. 6) and the cloned RALBP cDNA (Fig. 7B) was also estimated to be 48,000 by SDS-PAGE, it is most likely that SDS-PAGE provided an overestimate of the molecular weight of RALBP. Possibly, this overestimation might be due to the highly acidic nature (pK = 4.55, calculated from the sequence) of RALBP.
Considering both the nucleotide sequence and the amino acid sequence, we searched for the sequences homologous to squid RALBP in the GenBank, EMBL, PIR, and SWISS-PROT data libraries. However, no sequences with significant homology to squid RALBP were found. We next examined more closely the sequence homology between squid RALBP and other known retinoid-binding proteins (RBP, CRBP, CRABP, CRALBP, and IRBP), but found no homology. These data strongly suggest that squid RALBP is a novel retinoid-binding protein, which does not belong to any of the families of the hydrophobic ligandbinding proteins reported to date.
The analysis of hydropathicity of the amino acid sequence with Kyte and Doolittle (28) showed that squid RALBP is, on the average, hydrophilic but possesses several hydrophobic regions, each of which is composed of 5-10 amino acid residues ( Fig. 5; residues 10-15, 89-95, 110-119, 130-136, 137-146,  148-152,159-164,195-200,220-226,277-282, and 294-2991, and might be able to have hydrophobic interaction with retinoid. Predictions of secondary structure with Chou and Fasman (29) indicated that most of these hydrophobic regions are concentration of polyacrylamide and the nature of molecular weight marker proteins. Under the conditions for SDS-PAGE used in the present study, the molecular weight of RALBP purified from squid retina was calculated to be 48.000. associated with a higher probability than other parts of the RALBP polypeptide of adopting the 0-sheet structure (Fig. 5).
Expression of Squid RAL,BP in Vitro-%tal mRNA extracted from the squid retina was translated in vitro using a rabbit reticulocyte lysate system. As shown in Fig. 6, RALBP was one of the major components of translated products (lane 3), and it was precipitated by forming complex with RALBP-specific antiserum and Protein A-Sepharose (lane 1 ). Since no labeled protein was recovered when such immunoprecipitation was carried out using the antiserum that was pretreated with the purified squid RALBP (lane 2), it was confirmed that RALBP, synthesized in vitro, was specifically bound to the RALBP-specific antibodies and was recovered by immunoprecipitation. This experiment also showed that the molecular weight of RALBP synthesized in vitro coincides with that of the mature RALBP ex-tracted from the squid retina (48,000; lane 4). This result suggests that RALBP is not derived from a larger precursor.
Expression and functional analysis of the cloned wild-type and mutant RALBP may be useful for studies of the relationship between the molecular structure and the function of the protein.
We first tried to express RALBP in E. coli. Although we could generate a fair amount of the RALBP peptide in such a system, all the desired product was denatured, with resultant aggregation and loss of retinoid binding activity (data not shown). Since it is likely that RALBP is thermally unstable at 37 "C, we next used a reticulocyte lysate as an in vitro expression system, since it can function at a lower temperature (20 "C). Using this system, we achieved expression of a wild-type and an N terminally truncated RALBP, and we compared retinol binding activity of mutant RALBP with that of wild-type RALBP. As shown in Fig.  7E, full-length RALBP had an apparent molecular weight of 48,000 on SDS-PAGE, a value similar to that of RALBP isolated from the squid retina (Fig. 6). By contrast, corresponding to the deletion of 49 N-terminal amino acids, the molecular weight of the truncated RALBP was reduced to 42,000. On these two proteins, we measured retinol binding activity using 3H-labeled retinol. As shown in Fig. 7C, radioactivity of the sample that contained the full-length RALBP (pBSB1, mRNA+) was significantly higher than that of the sample of the control experiment (pBSB1, mRNA-). Therefore, this result indicated that the fulllength RALBP, synthesized in uitro, could bind L3H1retinol added to the reaction mixture. By contrast, radioactivity of the sample that contained the truncated RALBP was as low as that of the control experiment. This result suggests that the N-terminal region of the protein is essential for forming a functional RALBP. Moreover, this result also excludes the possibility that retinol binds nonspecifically to the synthesized protein, and confirms the previous result that full-length RALBP, expressed in this system, has activity to bind retinol.

DISCUSSION
In the present study, we determined the primary structure of squid RALBP, and it was clear from this structure that the protein is a novel hydrophobic ligand-binding protein. Although we previously reported that IRBP-specific antibodies reacted with squid RALBP (30), we could not find any sequences that were present in common in W B P and IRBP. This result suggests that the antibodies might possibly have recognized the higher structure of the two proteins. Using mass spectrometry, we characterized the N terminus of squid RALBP. We found that the first residue, methionine, corresponding to the initiation codon, had been removed and the second residue, serine, had been acetylated in squid RALBP. Unlike IRBP, RALBP did not include a signal sequence. This result is consistent with the fact that W B P is not a secretory protein but functions in the photoreceptor cells to transport intracellular retinoid between rhodopsin and retinochrome molecules (22).
Squid RALBP can bind retinol as well as retinal (19). This property suggests that not formation of a Schiffs base but, rather, hydrophobic interactions are essential for binding of retinoid to RALBP. In the present study, we found that squid RALBP contains several hydrophobic regions which might adopt a P-sheet structure. The studies on crystalline structure of RBP had revealed that the protein contains a p-barrel structure in which hydrophobic ligands are trapped (31). The hydrophobic regions in RALBP might form a similar tertiary structure for binding of retinoid.
By using a rabbit reticulocyte lysate system, and carrying out translation at a relatively low temperature, we succeeded in expressing RALBP that retained the ability to bind retinol. Although the result on retinol binding activity of expressed RALBP is statistically significant and reproducible, we took concern about high background radioactivity observed in the control experiments. Since nonspecific binding of retinol to Protein A-Sepharose or other contaminants seemed to increase the background radioactivity, we added either much excess of bovine serum albumin or a diluted solution of detergent to the reaction mixture in order to block nonspecific binding. This treatment greatly suppressed not only background activity but also retinol-binding to RALBP. This result suggests that interaction between expressed RALBP and retinol is very weak. Such weak interaction is, however, observed in the native squid RALBP, too, and is favorable for exchanging retinal with rhodopsin and retinochrome in the photoreceptor cells (21,22). We thus made an attempt to confirm retinol binding activity of expressed RALBP by examining retinol-binding to an N terminally truncated RALBP similarly expressed in. uitro. The result showed that the radioactivity in the truncated RALBP stayed at background level, strongly suggesting that the binding of retinol to the expressed full-length RALBP does not result from nonspecific binding of retinol to RALBP polypeptides. In addition, the above results also suggest that the missing N-terminal region of the protein forms the retinoid-binding site directly, or more likely, that the region is essential for formation of the correct configuration of W B P required to bind retinoid. We also tried to express RALBP in E. coli, but RALBP peptides synthesized in E. coli had completely lost the ability to bind retinol, although much more RALBP peptide could be obtained from E. coli than from the in uitro system. A possible reason for our inability to obtain active RALBP in the E. coli expression system is that squid RALBP is less stable to heat than other vertebrate retinoid-binding proteins. We tried growing E. coli at a lower temperature (20 "C) for expression of RALBP, but the bacteria did not synthesize any RALBP at all at the lower temperature. Using the reticulocyte lysate system, we succeeded in expressing active RALBP at 20 "C. However, since the amount of RALBP expressed in this system was still very low, the development of a more efficient system is now necessary if we are to synthesize sufficient RALBP for further molecular analysis.