Interphotoreceptor retinoid-binding protein (IRBP), a major 124 kDa glycoprotein in the interphotoreceptor matrix of Xenopus laevis Characterization, molecular cloning and biosynthesis

Interphotoreceptor retinoid-binding protein (IRBP) is the major soluble component of the interphotoreceptor matrix (for review see Chader et al., 1986). This important extra-cellular matrix surrounds the outer and inner segments of the photoreceptors and separates the neural retina from the pigment epithelium. IRBP is a large glycolipoprotein (135 kDa in man), which is secreted into this matrix by both rods and cones (Gonzalez-Fernandez et al., 1984, 1992; Hollyfield et al., 1985a; Porrello et al., 1991; Yokoyama et al., 1992; Hessler et al., 1993). In bony fish (Osteichthyes) IRBP is about half the size (67,600 ± 2,700 Da; Bridges et al., 1986, 1984; Hessler et al., 1993) of that in higher ver-tebrates. These observations and the four repeat structure of the mammalian protein suggest that IRBP may have arisen from the quadruplication of an ancestral gene (Borst et al., 1989; Liou et al., 1989, 1991; Si et al., 1989; Fong et al., 1990; Nickerson et al., 1991).

IRBP appears to function as a hydrophobic ligand-bind-ing protein and may have a critical role during retinal devel-opment. IRBP carries endogenous vitamin A in a light-dependent manner as well as six to seven fatty acid equivalents (Liou et al., 1982; Fong et al., 1984b; Bazan et al., 1985; Saari et al., 1982). Possible functions of IRBP include: transport of 11-cis retinal and all-trans retinol between the photoreceptors and pigment epithelium during the visual cycle (Lai et al., 1982; Lin et al., 1989; Flannery et al., 1990; Okajima et al., 1990; Adler and Spencer, 1991; Carlson and Bok, 1992), buffering excess vitamin A in the interphotoreceptor matrix (Ho et al., 1989) and protecting retinoids from degradation (Crouch et al., 1992).

Interactions between the neural retina and pigment epithelium are critical to retinal differentiation and growth (Hollyfield and Witkovsky, 1974; Vollmer and Layer, 1986; Hunter et al., 1992). It is therefore interesting that the expression of IRBP is up-regulated relatively early during the period of retinal differention (Eisenfeld et al., 1985; Johnson et al., 1985; Carter-Dawson et al., 1986; Gonza-lez-Fernandez and Healy, 1990; Liou et al., 1991; Hauswirth et al., 1992; Wang et al., 1992; Gonzalez-Fer-nandez et al., 1993). Its temporal control of expression and the critical interphase that it occupies suggest that IRBP could mediate important interactions between the pigment epithelium and neural retina during development. The appearance of IRBP during retinal development is likely to be linked to one of the proposed functions for IRBP described above. Possible additional roles of IRBP during retinal development could be: (1) to transport from the pig-ment epithelium hydrophobic morphogens such as retinoic acid; (2) to provide a transport vehicle for fatty acids, par-ticularly docosahexanoic acid, an essential fatty acid that is important for normal retinal development, and outer seg-ment structure and function (Scott and Bazan, 1989); or (3) to provide a role unrelated to its hydrophobic ligand-bind-ing properties. Such functions could be adhesion, mainte-nance of structure or cell/cell communication.

Xenopus laevis can provide a unique experimental system to study the role of the interphotoreceptor matrix in devel-opment as well as identify functionally conserved domains within the protein. We have previously shown that mole-cules may be introduced into the future interphotoreceptor matrix of the Xenopus embryo through optic vesicle microinjection (Hollyfield and Ward, 1974; Gonzalez-Fer-nandez and Kittredge, 1992). This technique, which does not disturb the normal relationships between the pigment epithelium and the neural retina, could allow studies of the fate of recombinant IRBP and the effect of immunological inactivation of IRBP. Furthermore, the sequence of non-mammalian IRBPs will be invaluable for identifying functionally important domains and understanding the evolution of its gene.

Despite the potential advantages of lower vertebrates for experimental studies and phylogenetic comparisons, most studies to date have examined only mammalian IRBP. In the present paper, we have characterized the biosynthesis of the Xenopus homologue of IRBP and isolated cDNAs encoding it. Preliminary reports of portions of this work have been published in abstract form (Rayborn et al., 1984; Kittredge et al., 1992).

For the metabolic studies, juvenile Xenopus laevis (Xenopus I, Ann Arbor, MI) 2-3 cm in length were housed under cyclic lighting conditions of 12 h light, followed by 12 h darkness at 19°C for several weeks prior to utilization.

Adult Xenopus laevis were mated, and embryos collected and dejellied as described by Henry and Grainger (1987). Embryos were reared in 20% Steinberg’s solution (Rugh, 1962) containing 50 μg/ml gentamycin sulfate. Embryos were anesthetized with 3-aminobenzoic acid ethyl ester (Sigma) prior to dissection.

For the western blot analysis, interphotoreceptor matrix was extracted from adult Xenopus eyes. The anterior segment including the ciliary body and lens were first removed by a circumferential limbal incision. Under phosphate buffered saline (PBS; 5 mM sodium phosphate, pH 7.4, 0.15 M NaCl) in the presence of 1.0 mM phenylmethylsulfonyl fluoride (PMSF) as a protease inhibitor, the neural retinal was gently detached from the pigment epithelium. The two tissues were completely separated by severing the optic nerve. The retina and eye cup were incubated together for 5 min in PBS on ice and the apical surface of the pigment epithelium was gently irrigated with this saline. The eye cup and neural retina were removed and the PBS extract was centrifuged at 50,000 g for 30 min at 4°C. The supernatant was precipitated with an equal volume of acetone at -20°C, treated with dithio-threitol, sodium dodecyl sulfate (SDS) and fractionated by polyacrylamide gradient (5% to 12%) gel electrophoresis in a discontinuous buffer system (Laemmli, 1970). The proteins were electrophoretically transferred to nitrocellulose and probed with rabbit anti-bovine IRBP immunoglobulins (Fong et al., 1984a) fol-lowed by peroxidase-conjugated goat anti-rabbit IgG as previously described (Gonzalez-Fernandez et al., 1985).

The cDNA library used in this study was prepared from poly(A)+ mRNA from whole stage 42 embryos (swimming tadpoles; refer to staging system of Nieuwkoop and Faber, 1956). A full descrip-tion of the preparation of this library is described elsewhere (Saha and Grainger, 1993). The library was screened with a full-length human IRBP cDNA provided by Dr C. David Bridges (Purdue University). This probe was labeled with [α-32P]dCTP by the method of random primed labeling (Feinberg and Vogelstein, 1983). The bacteriophage λ plaques were immobilized on Nytran membranes (Schleicher and Schuell, Inc., Keene, NH). Duplicate filters were made from each primary master plate so that genuine signals could be distinguished from spurious spots. The λ phage were lysed by steam treatment using the method of G. Struhl as described by Sambrook et al. (1989) and cross-linked to the Nytran paper by ultraviolet irradiation. Prehybridization was carried out overnight at 42°C in 30% formamide, 1 M NaCl, 100 μg/ml dena-tured salmon sperm DNA, 1% SDS, 10 mM Tris-HCl, pH 7.5. Hybridization was carried out overnight under the same conditions with 106 c.p.m./ml-1 denatured probe. Following hybridization, the membranes were washed twice in 2×SSC (1×SSC is 0.15 M NaCl, 15 mM sodium citrate, pH 7.0), 1% SDS twice at room temperature and twice at 50°C. Each wash was 30 min in dura-tion. The filters were exposed to X-ray film (X-AR, Kodak) at –70°C with an intensifying screen. Synthetic oligonucleotides cor-responding to the Xenopus IRBP sequence were used as sequenc-ing primers.

Total RNA was isolated from adult Xenopus tissues or embryos by the method of Chomczynski and Sacchi (1987) as described by Gonzalez-Fernandez and Healy (1990). Glyoxal was used to denature RNA before electrophoresis in 1.0% Seakem GTG agarose (FMC, Rockland, ME). The RNA was transferred to Nytran paper (Schleicher and Schuell) and cross-linked by ultra-violet irradiation. Prehybridization and hybridization were carried out at 42°C in 50% formamide, 5×SSPE (SSPE at a 1×concen-tration is 0.18 M NaCl; 10 mM sodium phosphate, pH 7.7; 1 Mm EDTA), 5×Denhardt’s solution (0.1% Ficoll; 0.1% polyvinylpyrrolidone; 0.1% BSA), 1% SDS, and 100 μg/ml dena-tured salmon sperm DNA. The excised cDNA inserts were gel purified and labeled with [α-32P]dCTP by the random primer method (Feinberg and Vogelstein, 1983). The blots were hybridized with 106 c.p.m. of radiolabeled probe per ml of the above buffer. D.D. Oparian (Brandeis University) provided the bovine opsin cDNA, which was nearly full-length (1.0 kb in size starting at nucleotide number 234; Nathans and Hogness, 1983). Following hybridization, the blots were washed twice at room tem-perature in 6×SSPE, 0.5% SDS for 15 min, twice at 37°C in 1×SSPE, 0.5% SDS for 15 min and finally for 30 min at 65°C in 1×SSPE, 0.5% SDS. The conditions for the northern blot of Fig. 13 (bottom) where RNA probes were used has been described by Gonzalez-Fernandez and Healy (1990). For all northern blots, X-AR film (Kodak) was exposed against an intensifier screen at -80°C.

In the in vivo studies, L-[6-³H]fucose (30 Ci/mmol; ICN, Irvine, CA) was evaporated to dryness to remove the ethanol carrier and was resuspended in amphibian Ringer’s solution at a concentration of 2.0 mCi/ml. Ten animals were given dorsal subcutaneous injections of approximately 12 μCi. The injected animals were rinsed repeatedly for the next several hours and the eyes were removed the following day. The eyes were placed in a small drop of incubation medium, Ringer’s bicarbonate-pyruvate buffer (RBP; Hollyfield and Anderson, 1982), and the anterior segment removed. Some of the eyes were fixed with the retina and the pig-ment epithelium left intact, while others were dissected free of Mr ×10⁻³each other and fixed after extensive rinses in Ringer buffer. All tissues were fixed for 1.5 h in an ice-cold mixture of 1% OsO₄, 1.65% glutaraldehyde in 0.075 M cacodylate buffer (pH 7.4). Tissues were dehydrated with ethanol and embedded in Epon.

For autoradiography, 1 μm thick sections were cut for light microscopic autoradiography and placed on glass slides. After dip-ping in Kodak NTB liquid emulsion diluted 1:1 with distilled water, the slides were dried and exposed in the dark at 4°C for 5 to 25 days. Electron microscopic autoradiographic methods employed Ilford L4 emulsion and Phenadon developing as described in detail previously (Hollyfield, 1979).

For biochemical analysis, retinas were dissected from the pig-ment epithelium/choroid complex as described above and rinsed extensively in the incubation medium. The proteins in the rinse medium, retinal cytosol, pigment epithelial cytosol, retinal mem-branes and pigment epithelial/choroid membranes were prepared for polyacrylamide gel electrophoresis and fluorography as previ-ously described (Gonzalez-Fernandez et al., 1985).

For the in vitro experiments, retinas and pigment epithelium/choroid from 40 dark-adapted eyes were collected separately in RBP/Wolf-Quimby media (Fliesler et al., 1985) on ice. Metabolic labeling was initiated by placing the two groups of tissue in separate tubes containing 5 ml of medium with 200 μCi/ml L-[6-³H]fucose (30 Ci/mmol). The tissues were maintained in the dark supplied with 95% O₂/5% CO₂ at 21°C with gentle agitation. After 4 h, the incubation medium and the tissues were collected by the addition of 1.0 mM PMSF. The tissues were homogenized in PBS and centrifuged at 100,000 g to obtain soluble and membrane fractions. Aliquots of each fraction were examined by SDS-polyacrylamide gel electrophoresis (SDSPAGE) and fluorography.

The rabbit anti-bovine IRBP immunoglobulins, which cross-react with human and rat IRBPs (Fong et al., 1984a; Gonzalez-Fernandez et al., 1984), recognize a 124 kDa pro-tein in the soluble extract of adult Xenopus interphotore-ceptor matrix. Although smaller than mammalian IRBPs (bovine and rat, 144 kDa; human IRBP, 135 kDa), the Xenopus homolog is similar in size to Rana pipiens IRBP (125 kDa, Fong et al., 1986) and approximately twice the size of IRBP in bony fish (Bridges et al., 1984, 1987; Wagenhorst et al., 1993). The fact that its size is similar to that of IRBPs in higher vertebrates suggests that Xenopus IRBP probably has a four-repeat structure similar to that described for human and bovine IRBPs (Nickerson et al., 1991).

In order to characterize better the structure of the Xeno -pus homolog of IRBP, cDNAs for this protein were iso-lated. We screened a stage 42 (swimming tadpole) cDNA library under low-stringency conditions (see Materials and Methods) with a human IRBP cDNA. Seven putative Xeno -pus IRBP cDNA clones were isolated, four of which were partially characterized. The sizes of the inserts were determined by digesting the rescued bluescript plasmid with EcoRI. Based on the restriction pattern, the four cloned cDNAs had a similar size (1.2 kb), contained an internal EcoRI site and probably corresponded to a similar portion of the mRNA, presumably the 3′ end. One of the clones, termed XenIRBP.B1, was sequenced.

Fig. 2 shows the translated amino acid sequence of XenIRBP.B1. Synthetic nested oligomers were used to gen-erate the staggered sequence readings. Both strands of the cDNA were sequenced entirely (arrows summarize the sequencing strategy). The 1.2 kb clone consisted of 893 bp of open reading frame (filled bar) followed by 336 bp of 3′-untranslated region (UTR, open bar) ending in a typical signal polyadenlyation motif (AATAAA) located 25 bases upstream from a poly(A) tail (Wickens and Stephenson, 1984). Since the mRNA for Xenopus IRBP is 4.2 kb (major form), the clone corresponds to 29% of the mRNA. The computer program, LAWRENCE (Lawrence and Goldman, 1988), identified homology domains between the Xenopus IRBP cDNA and each of the four repeats of human IRBP. Of the four homology domains, the degree of similarity is greatest within the fourth repeat. In this region, the maxi-mum percentage identity of the nucleic and amino acid sequences between Xenopus and human IRBPs is 70% (Table 1).

Although the overall sequence identity between Xenopus and mammalian IRBPs is low compared to opsin (see Dis-cussion), focal regions are highly conserved. Conserved regions within IRBP probably form domains important to the protein’s function. Fig. 3 shows the amino acid align-ment of XenIRBP.B1 with each of the four repeats of human IRBP. Of the 100 amino acids that are non-identical between the fourth repeats of Xenopus and human IRBP, 75 are conservative substitutions. The boxed regions within this alignment demonstrate a high degree of conservation between the XenIRBP.B1 and each of the four repeats of human IRBP. Conservative substitutions are defined as: I=L=V=M; K=R; D=E. Three regions stand out from this alignment. The first two of these regions have the invariant sequences OGYOROD (residues 113-119) and OOODOR (136-141), where O represents a hydrophobic residue. The fact that these segments contain amino acids perfectly conserved not only in the four repeats of human IRBP, but also in Xenopus, and have a high proportion of hydrophobic residues predicts that these domains may be important for vitamin A or fatty acid-binding. In addition to these two domains, a third region showing a high degree of conservation was also seen (OOGE; 237-240).

The length of the 3′-UTR of Xenopus IRBP mRNA (322 bp) is similar to that of the human IRBP mRNA 3′-UTR (416 bp) and is significantly smaller than that of the 3′-UTR of bovine IRBP mRNA, which is 1,988 bp. There is a 47.4% sequence identity within the last 342 bp of the human 3′-UTR and the Xenopus 3′-UTR. These observa-tions are consistent with the hypothesis that the longer 3′-UTR of bovine IRBP is the result of a large insertion that has been made in the 3′-UTR of the ancestor of the bovine sequence.

Fig. 4 is a northern blot of total RNA from adult Xenopus retina probed with both XenIRBP.B1 and bovine opsin cDNA probes. The mRNA sizes for Xenopus IRBP and opsin are 4.2 kb and 2.1 kb, respectively. A longer expo-sure of the autoradiogram revealed a less-abundant 6.0 kb IRBP transcript but failed to identify additional sizes for the opsin mRNA. Multiple IRBP mRNAs have been observed in a number of mammals (Inouye et al., 1989; Gonzalez-Fernandez and Healy, 1990) and in rat are due to different sizes of the 3′-UTR (Gonzalez-Fernandez et al., 1993). Although multiple forms of the mRNA for rod opsin have been documented in a variety of mammals (Al-Ubaidi et al., 1990), only one form of the opsin mRNA is present Xenopus. The sequence of Xenopus rod opsin has been described by Saha and Grainger (1993).

In the northern blot of Fig. 5, adult Xenopus retina, brain and liver total RNA were probed with XenIRBP.B1 under high-stringency conditions. This blot demonstrates that the XenIRBP.B1 probe does not bind to the upper ribosomal subunit and that brain expresses low levels of the mRNA for IRBP. The source of the mRNA for IRBP in the Xeno -pus brain may be the pineal gland, which is known to express this retinal protein (Chader et al., 1986; Bridges et al., 1987; Gonzalez-Fernandez et al., 1993; Lopez et al., unpublished results).

The experiments in this section utilized biochemical, and light and electron microscopic autoradiographic analysis of [³H]fucose in vitro and in vivo radiolabelling to localize IRBP and further define its site of synthesis. Fucose was selected for these studies because it is present in the oligosaccharide of mammalian IRBP is not metabolized to other compounds, and is not incorporated into gly-cosaminoglycans. We separately incubated isolated adult neural retinas and eye-cups (pigment epithelium/choroid) in the presence of [³H]fucose for 4 h. The incubation medium, cytosolic and membrane fractions from the neural retinas and pigment epithelium/choroid were analyzed by SDS-PAGE and fluorography (Fig. 6). A radioactive 124 kDa glycoprotein was observed in the incubation medium of the neural retina, but not in the medium of the pigment epithe-lium/choroid. A faint 124 kDa band was identified in the neural retinal cytosol. This experiment, coupled with the immunoblotting data presented in Fig. 1, demonstrates that Xenopus IRBP is fucosylated and secreted by the neural retina and not the pigment epithelium.

In order to show that the radiolabeled IRBP secreted into the incubation media normally accumulates in the inter-photoreceptor matrix, we performed a similar experiment in vivo. [³H]fucose was injected intraperitoneally and the eyes were removed after 24 h. The tissues were either fixed for light and electron microscopic autoradiography or ana-lyzed biochemically. For the latter studies, the neural retina was gently detached from the eye-cup, and the saline washes of the neural retina and apical surface of the pig-ment epithelium were combined and subjected to SDS-PAGE and fluorography. Fig. 7 shows that radiolabelled IRBP could be identified in the interphotoreceptor matrix extract. Besides IRBP, an additional radioactive band at approximately 183 kDa was noted. This band probably does not represent a matrix component secreted into the sub-retinal space because it was not detected in the in vitro experiment of Fig. 6. This band more likely represents a labeled protein that leaked into the matrix extract from neural retinal cells damaged as this tissue layer was mechanically separated from the pigment epithelium. Con-sistent with this interpretation is the fact the retina cytosol normally contains a prominently radiolabelled protein of this size (compare with Fig. 6, lane 2).

Light and electron microscopic autoradiography were performed in order to confirm the compartmentalization of the radiolabelled protein. The light microscopic autoradi-ographs in Fig. 8 show the distribution of silver grains across the outer retina from one of the recovered eyes. The silver grains were present over the inner segments of the photoreceptors, in the interphotoreceptor matrix and exten-sively over the pigment epithelium. When the retina was slightly teased from the pigment epithelium and immedi-ately fixed, an extensive cloud of silver grains was present over the extracellular compartment between the retina and pigment epithelium. Our interpretation is that these silver grains represent radiolabelled matrix glycoconjugates that have become dispersed throughout the expanded volume occupied by the matrix at the detachment site. This preparation was not rinsed but the retina and pigment epithelium were slightly separated at the margin of the eye cup before fixation.

In the lower panel of Fig. 8, an isolated retina is shown that was rinsed prior to fixation. Little radioactivity is associated with the interphotoreceptor matrix. Electron microscopy was used to study the localization of the radioactivity still remaining in the retina. Fig. 9 shows that at the outer limiting membrane, radioactivity is associated with both the apical termination of the villous processes and the cytoplasm of the Muller cells. Additional radioac-tivity is associated with what appears to be Golgi appara-tus in the inner segments of the two side-by-side photo-receptors. Fig. 10 demonstrates that the flocculent interphotoreceptor matrix material is also radioactive, as indicated by the silver grains associated with this extracel-lular compartment. In this figure, a band of radioactivity is evident at the base of one of the rod outer segments shown. This represents one of the minor rods, which has a very rapid rate of membrane turnover. Hollyfield et al. (1984) has shown that there are two populations of rods in Xeno -pus, which have different rates of outer segment renewal and utilization of fucose. The absence of the band in the right photoreceptor in this figure suggests that this outer segment belongs to a principal rod and the presence of a band on the photoreceptor to the left probably reflects this minor rod. Fig. 11 shows that when the pigment epithelium is rinsed and prepared for autoradiography, silver grains are distributed over the cytoplasm of the pigment epithelium and the apical microvilli of these cells. In view of the fact that IRBP is synthesized by the photoreceptors (at least in mammals) the silver grains over the outer segments, pig-ment epithelium, and Muller cells probably represent gly-coconjugates newly synthesized by these cells. The possi-bility that some of the radiolabel may represent uptake of IRBP cannot be excluded (see Hollyfield et al., 1985b).

Fig. 12 compares the expression of the mRNAs for IRBP and opsin during development. In the top panels of this figure, the mRNA for both proteins at stage 43 (swimming tadpole) are shown to be restricted to the head, absent from the body and have a similar electrophoretic mobility to that of the mRNAs for these proteins in the adult retina. In the bottom panel of this figure, 23 μg of total RNA from whole embryos at various developmental stages from neurula to tadpole were subjected to northern blot analysis. This blot was first probed with antisense Xenopus IRBP transcripts generated from the bacteriophage T7 promoter using BamHI-linearized plasmid (the BamHI site is contained within the multiple cloning segment 5′ to the insert). The transfer was then reprobed with a Xenopus opsin antisense probe. The mRNAs for both IRBP and opsin could be detected first at stage 40 (3 days old, recently hatched). At this stage photoreceptor outer segments are just beginning to form (Kinney and Fisher, 1978b). Between stages 40 and 45/46 there was a marked up-regulation of both IRBP and opsin mRNA expression. During this period there is further growth of the outer segments (outer segments attain their adult length at stage 53/54; Kinney and Fisher, 1978b).

Our long-term goals are to determine IRBP’s structural requirements for ligand-binding and role during retinal development. Xenopus laevis will provide a valuable system in which to study both of these questions through phylo-genetic comparisons of protein sequence and embryologi-cal manipulation of the interphotoreceptor matrix possible in this species. Preliminary data from other groups have led us to believe that a protein homologous to IRBP probably exists in the Xenopus retina. Using antisera raised against bovine or primate IRBPs, IRBP-like proteins have been identified in all six major vertebrate classes, emphasizing its importance to visual function (Bridges et al., 1986). Bovine and human IRBP cDNAs hybridize with the mRNA for IRBP in a variety of mammals but not birds (Inouye et al., 1989; Liou et al., 1991). In contrast, IRBP has been identified in the chicken retina by western blotting (Bridges et al.,1986). This apparent discrepancy is probably due to the greater evolutionary constraints on the amino acid sequence than on the nucleic acid sequence because of evo-lutionary pressure to preserve functionally important domains in the protein. Another possibility is that some non-mammals may utilize another protein in the interpho-toreceptor matrix rather than IRBP to transport vitamin A. In this regard, it has been noted that the chicken IPM con-tains purpurin, a retinol-binding protein with no apparent homology with IRBP (Schubert et al., 1986). Fong et al. (1986) showed that the N-terminal sequence of a 125 kDa glycoprotein from the soluble extract of the interphotore-ceptor of Rana pipiens is highly conserved compared to IRBPs from nine other vertebrate species (see also Schnei-der et al., 1986; Wood et al., 1984). Liou et al. (1991) were able to detect in the retina but not liver of Rana pipiens a 4.4 kb band by probing a northern blot of poly(A)+ RNA with a human cDNA under reduced stringency conditions (5×SSC, in 50% formamide at 37°C).

In order to isolate a cDNA for Xenopus IRBP, we uti-lized the technique of low-stringency screening. Based on the size of the mRNA for IRBP in Xenopus, and the homol-ogy with the fourth repeat of human IRBP, the cDNA iso-lated in this study represents all but the eight N-terminal amino acids of the fourth repeat. Because the present cDNA spans virtually the entire fourth repeat of mammalian IRBP, we were able to align the translated XenIRBP.B1 with the four repeats of human IRBP. Although XenIRBP.B1 is most homologous with the fourth repeat, at least two regions within the clone were highly conserved between all four repeats of human IRBP. 70% of the nucleotides and amino acids between Xenopus and human IRBPs are identical. In contrast, the nucleotide and amino acid sequences of Xeno -pus rod opsin have percentage identities of 76% and 85%, respectively with human rodopsin (Saha and Grainger, 1993). Although the overall percentage identity between Xenopus and human IRBPs is low compared to opsin, focal regions are highly conserved. This supports the idea that specific regions within IRBP participate in the formation of functionally important domains. Two regions in particular, which were noted by Liou et al. (1991) to be conserved between each of the four repeats of human IRBP, stand out when Xenopus and human IRBPs are compared. These regions have the invariant sequences OGYOROD and OOODOR, where O represents a hydrophobic residue. The fact that these segments contain amino acids perfectly con-served not only between the four repeats of human IRBP, but also in Xenopus, and have a high proportion of hydrophobic residues suggests that these regions are impor-tant to the formation of the ligand-binding domains. Evi-dence that these domains have hydrophobic binding activity comes from the observation that they are present in Tsp, a tail-specific protease that selectively degrades proteins with nonpolar C termini in Escherichia coli (Silber et al., 1992).

A striking feature of these two conserved regions is the presence in both of an arginine residue that is conserved between Xenopus IRBP and each of the four repeats of human IRBP. This amino acid, in other hydrophobic ligand-binding proteins, confers specificity for fatty and retinoic acids by providing its α--guanidinium group to stabilize the carboxyl moiety of these ligands (Cheng et al., 1991; Stump et al., 1991). The presence of two arginines per repeat suggests that the whole protein should be able to bind eight fatty acids. Bazan et al. (1985) found that IRBP carries four fatty acids noncovalently. The number of endogenous fatty acids may be lower than our predicted stoichiometry because the native protein used in those studies may not have been saturated with ligand. An alternative model would be that more than one arginine participates in the stabilization of the carboxyl group of each bound fatty acid. Although the relationship between IRBP’s retinol and fatty acid-binding sites is not known, preliminary studies suggest that one but not both of IRBP’s retinol-binding sites can be blocked by palmitic acid (Hazard et al., 1991). Perhaps the third domain, OOGE, which does not contain arginine, is more important to retinol than to fatty acid-binding. Muta-geneis studies aimed at understanding the role of these con-served arginines are in progress in our laboratory (Van Niel et al., 1993). In any event, we predict that the above three conserved regions form the ligand-binding domain(s) and the conserved arginines in the first two regions are impor-tant for fatty acid-binding.

The smaller size of the 3′-UTR of human IRBP mRNA (416 bp) compared to the 3′-UTR of the bovine IRBP mRNA (1,988 bp) is due either to a large insertion into the ancestor of the bovine gene or a large deletion from the ancestor of the human gene (Si et al., 1989). We found that the final 342 bp of the human 3′-UTR has a 46% sequence identity with the 3′-UTR of the Xenopus IRBP mRNA. This high degree of conservation, which is not present between the 3′-UTRs of Xenopus and human rod opsins (Saha and Grainger, 1992), may suggest an importance of the IRBP 3′-UTR in promoting mRNA stability and translational con-trol. The sequence homology, taken together with the sim-ilar size of the of Xenopus IRBP 3′-UTR (322 bp), suggests that the mechanism responsible for the larger bovine IRBP 3′-UTR was an insertion into the bovine ancestor rather than a deletion from the human ancestor gene. The physi-ological significance of the longer 3′-UTRs of mammalian IRBPs is unknown.

Two forms of the mRNA for Xenopus IRBP were noted, a major form at 4.2 kb and a less-abundant 6.0 kb tran-script. Two forms of the mRNA have been observed in sev-eral species, particularly rat, which displays clear bands at 5.2 kb and 6.4 kb (Gonzalez-Fernandez and Healy, 1990; Inouye et al., 1989). Different lengths of the 3′-UTR are responsible for the two forms of the rat IRBP mRNA (Gon-zalez-Fernandez et al., 1993). Opsin displayed only one size for its mRNA, although multiple forms of the mRNA for rod opsin are common in mammals and have been shown to be due to multiple signal polyadenylation sites in the 3′-UTR (Al-Ubaidi et al., 1990).

Our in vitro and in vivo [³H]fucose-labelling studies demonstrate that IRBP is synthesized by the neural retina and secreted into the interphotoreceptor matrix. Fucose was used for these studies because it is present in the oligosac-charide component of mammalian IRBP (Taniguchi et al., 1986). Although this has not been formally demonstrated for IRBPs from other vertebrate classes, Rana pipiens IRBP has been shown to bind concanavalin A (Bridges et al., 1986). In the present study, we used in vitro [³H]fucose-labelling to show that the isolated Xenopus neural retina, but not the pigment epithelium, synthesizes and secretes a pro-tein that has identical electrophoretic mobility to the 124 kDa interphotoreceptor matrix protein, which crossreacts with anti-bovine IRBP immunoglobulins (Fig. 1). We showed by in vivo [³H]fucose-labelling studies employing light and electron microscopic autoradiography that this gly-coprotein is normally secreted into the extracellular matrix surrounding the rod outer and inner segments. These meta-bolic studies, taken together with the western blot analysis, suggest that the homolog of IRBP in Xenopus laevis is syn-thesized by the neural retina and secreted into the interpho-toreceptor matrix. The cell types within the neural retina that border the interphotoreceptor matrix are the photoreceptors and the Muller cells. Which of these two cells is responsi-ble for the synthesis of IRBP cannot be determined from the data presented here. Studies using immunohistochemistry (Chader et al., 1986; Rodrigues et al; 1986; Carter-Dawson and Burroughs, 1992; Gonzalez-Fernandez et al., 1992), mutant rats with photoreceptor specific degeneration (Gon-zalez-Fernandez et al., 1985), cell culture systems (Holly-field et al., 1985a) and in situ hybridization (Veen et al., 1986; Porello et al., 1991; Hessler et al., 1993; Bukelman et al., 1993; Hessler et al., 1993; Wagenhorst et al., 1993) indicate that the photoreceptors and not the Muller cells are responsible for IRBP synthesis.

During development, the mRNAs for IRBP and opsin were first detected at stage 40 and increased markedly by stage 45/46. The developing Xenopus neural retina segre-gates into layers at stage 33/34. The postmitotic cells find themselves in different microenvironments, leading to adoption of distinct cellular fates (Holt et al., 1988). Outer segments first develop at stage 37/38-40 and reach their adult length by stage 53–54 (Kinney and Fisher, 1978). The up-regulation of these genes coincides with the emergence of the visual cycle in the Xenopus embryo (Azuma et al., 1990; Bridges et al., 1987). In the developing rodent retina, the mRNA for IRBP is up-regulated before that of opsin (Gonzalez-Fernandez and Healy, 1990; Gonzalez-Fernan-dez et al., 1993; Wang et al., 1992). In the present study, differences in the temporal expression of these two genes may not have been apparent because our Northern blot includes only two time points with photoreceptor gene expression. Furthermore, cellular differentiation is less syn-chronous in the amphibian retina compared to that of rodents because the former continues to grow throughout life by continual cell division. In situ hybridization studies, in progress in our laboratory, should help to address these issues.

We have demonstrated that Xenopus laevis will provide useful information towards understanding the structural requirements of vitamin A-binding and allow future studies to exploit the potential of this system to study the role of IRBP in development. Using a combination of biochemi-cal, genetic and [³H]fucose-radiolabeling techniques, we have demonstrated that the Xenopus neural retina secretes into the interphotoreceptor matrix a soluble 124 kDa gly-coprotein homologous to mammalian IRBPs. Our study demonstrates that low-stringency screening may also be useful to isolate cDNAs for proteins homologous to IRBP in species distant from man, and that comparative studies could provide clues to the location of the protein’s ligand-binding domains. Future studies from our laboratory will utilize the cDNA described here to express Xenopus IRBP in order define the specific amino acids involved in creat-ing hydrophobic ligand-binding pocket(s). Xenopus will be particularly valuable in studying the role of IRBP in devel-opment, since optic vesicle microinjection provides a way of introducing molecules into the embryonic subretinal space (Gonzalez-Fernandez and Kittredge, 1992). This approach will allow us to study the fate of IRBP and the effect of immunological inactivation of IRBP on eye devel-opment.

This work was supported by The Thomas F. Jeffress and Kate Miller Memorial Trust (F.G.-F.); grant IN149H from the Ameri-can Cancer Society (F.G.-F.); an Ophthalmology Research Grant from The Knight’s Templar Eye Foundation Inc. (F.G.-F.); grant T32 NS 7236 from the NINCDS (F.G.-F.); a post-doctoral fel-lowship from the Fight for Sight Research Division of the National Society to Prevent Blindness (K.L.K.); a development grant from Research to Prevent Blindness; a gift from Elizabeth Jones to the University of Virginia for pediatric eye research; NIH grant EY-06675 (R.M.G.) and NSF grant DCB9005468 (R.M.G.); NIH grant EY02633 (J.G.H.); a Retina Research Foundation Grant (J.G.H.), a Senior Investigator Award from Research to Prevent Blindness (J.G.H.) and an award from Alcon Research Institute (J.G.H.).

In situ hybridization studies performed in our laboratory demonstrate that in the Xenopus eye, IRBP is synthesized uniquely by both the rod and cone photoreceptors (Bukel-man et al., 1993; Hessler et al., 1993).