|Home | About | Journals | Submit | Contact Us | Français|
The first event in light perception is absorption of a photon by an opsin pigment, which induces isomerization of its 11-cis-retinaldehyde chromophore. Restoration of light sensitivity to the bleached opsin requires chemical regeneration of 11-cis-retinaldehyde through an enzymatic pathway called the visual cycle. The isomerase, which converts an all-trans-retinyl ester to 11-cis-retinol, has never been identified. Here, we performed an unbiased cDNA expression screen to identify this isomerase. We discovered that the isomerase is a previously characterized protein called Rpe65. We confirmed our identification of the isomerase by demonstrating catalytic activity in mammalian and insect cells that express Rpe65. Mutations in the human RPE65 gene cause a blinding disease of infancy called Leber congenital amaurosis. Rpe65 with the Leber-associated C330Y and Y368H substitutions had no isomerase activity. Identification of Rpe65 as the isomerase explains the phenotypes in rpe65-/- knockout mice and in humans with Leber congenital amaurosis.
Light perception in vertebrates is mediated by a group of G protein-coupled receptors called the opsins. These opsins are located in the membranous outer segments of rod and cone photoreceptors. Absorption of a photon by an opsin pigment induces 11-cis to all-trans isomerization of its retinaldehyde chromophore. Metarhodopsin, the resulting light-activated form, stimulates the visual transduction cascade (Arshavsky et al., 2002). After a brief period, metarhodopsin decays to yield free all-trans-retinaldehyde (atRAL) and apoopsin. Before light sensitivity can be restored to apoopsin, the atRAL must be chemically reisomerized to 11-cis-retinaldehyde (11cRAL) by an enzymatic pathway called the visual cycle (Figure 1). Most steps in this pathway take place within the retinal pigment epithelium (RPE), a layer of cells adjacent to the photoreceptors. The key step in this pathway is all-trans to 11-cis reisomerization of the retinoid. The enzyme that catalyzes this isomerization reaction has not been identified, but has been shown to use fatty-acyl esters of all-trans-retinol (atROL) as substrate (Gollapalli and Rando, 2003; Mata et al., 2004; Moiseyev et al., 2003). These all-trans-retinyl-esters (atREs) are generated in RPE cells by lecithin-retinol acyl transferase (LRAT), which transfers a fatty acid from the sn-1 position of phosphatidylcholine to atROL (MacDonald and Ong, 1988; Saari and Bredberg, 1989).
Leber congenital amaurosis (LCA) is a severe recessive blinding disease commonly caused by mutations in the RPE65 gene (Hanein et al., 2004; Thompson et al., 2000). Rpe65, the product of this gene, is an abundant protein in RPE cells (Bavik et al., 1991) that has been shown to bind atREs (Gollapalli et al., 2003; Mata et al., 2004). Photoreceptors in rpe65-/- mice contain only apo-opsin (Redmond et al., 1998). These findings suggest that Rpe65 may be the isomerase. Against this hypothesis are the reports that partially purified Rpe65 possessed no isomerase activity (Mata et al., 2004; Znoiko et al., 2002). Also, depletion of Rpe65 from RPE membranes by extraction with saline caused only a minor reduction in isomerase activity (Choo et al., 1998). Finally, the high abundance of Rpe65 in RPE cells (Tsilou et al., 1997) argues against it being an enzyme. Instead, it was suggested that Rpe65 binds insoluble atREs and presents these as substrate to the isomerase (Mata et al., 2004). According to this explanation, Rpe65 is required for retinoid isomerization but has no intrinsic isomerase activity.
In the current work, we performed a cDNA expression screen for the isomerase in bovine RPE. The final clone from this screen contained very high isomerase activity. Sequence analysis revealed it to be Rpe65. We confirmed our identification of the isomerase as Rpe65 by expressing this protein in mammalian and insect cells and demonstrating isomerase catalytic activity.
To undertake a functional screen for the retinoid isomerase, we generated cell lines that stably express LRAT, Rpe65, and cellular retinaldehyde binding protein (CRALBP) alone and in all combinations. Cell line 293T-LRC stably expresses LRAT, Rpe65, and CRALBP (Figure 2).
We prepared a cDNA library from bovine RPE in the mammalian expression plasmid, pRK5. The library was amplified in 42 independent pools, each containing ~5000 clones. We transfected plasmid DNA from individual library pools into 293T-LRC cells. As a negative control, we also transfected nonrecombinant pRK5. We transferred these library-transfected cells to darkness, added atROL to the medium, and incubated for an additional 18 hr. We then harvested the cells and extracted the retinoids for analysis. Since 11-cis-retinyl esters (11cREs) are difficult to resolve from 9-cis- and 13-cis-retinyl esters by high-performance liquid chromatography (HPLC), we included a saponification step to hydrolyze any retinyl esters formed by LRAT to their cognate retinols before analysis. In the primary screen, we observed variable production of 11cROL by each library pool. Pools #7 and #25 contained approximately 2.8-fold higher levels of 11cROL than the pRK5 control (Figure 3A). To confirm this result, we reanalyzed these library pools for isomerase activity by an alternate approach. As before, we transfected 293T-LRC cells with plasmid from the library pools plus nonrecombinant pRK5. This time, we did not add atROL to the medium. Instead, we prepared cell homogenates from each pool as an enzyme source for in vitro isomerase assays. As before, we performed alkaline hydrolysis of the retinyl esters before analyzing the retinoids by HPLC. Although the absolute levels of 11cROL produced in these in vitro assays were lower, the homogenate from cells transfected with pool #7 synthesized 2-fold more 11cROL than the homogenate from pRK5-transfected cells (data not shown). The homogenate from pool #25 synthesized only 1.5-fold more 11cROL than the pRK5 control (data not shown). We selected library pool #7 for secondary screening.
We transformed E. coli with plasmid from pool #7 and plated at an average density of 650 colonies per dish. We used plasmid DNA from the resulting 30 library subpools to transfect 293T-LRC cells and repeated the in vivo assay for isomerase activity. Subpool #7:22 contained 7-fold more 11cROL than the pRK5 control (Figure 3B). Figure 3C shows the chromatogram of retinoids extracted from subpool #7:22, scaled to reveal the 11cROL peak. As always, we confirmed our identification of 11cROL in the chromatogram by UV-spectral analysis (Figure 3D). Figure 3E shows the chromatogram of retinoids from pRK5-transfected control cells with the same scale. To confirm that subpool #7:22 contains isomerase activity, we retransfected 293T-LRC cells with DNA from multiple subpools including #7:22 plus pRK5 for in vitro analysis. Here again, the pattern of 11cROL production relative to pRK5 was similar to the pattern seen with the in vivo assay (not shown). Thus, subpool #7:22 possessed isomerase activity and was a valid candidate for tertiary screening.
We transformed E. coli with plasmid DNA from subpool #7:22 and plated the cells at an average density of 36 colonies per dish. From each dish we prepared plasmid, which we used to transfect 293T-LRC cells in another set of in vivo assays for isomerase. The amount of 11cROL in subpools #7:22:9 and #7:22:18 was approximately 11-fold higher than the amount produced in pRK5-transfected cells (Figure 3F). As before, we performed in vitro analysis to confirm our identification of subpools at this stage of the screen. The homogenate from subpool #7:22:9 contained only 2-fold higher isomerase activity than the pRK5 control, while subpool #7:22:18 contained 6-fold higher isomerase activity (Figure 3G). The relative differences in 11cROL production between the in vivo and in vitro isomerase assays may reflect the longer incubation time with the in vivo assay (18 versus 2 hr). Unexpectedly, when we cotransfected cells with DNA from subpools #7:22:9 plus #7:22:18, the isomerase activity in the homogenate jumped to 17-fold that of the control (Figure 3G). This suggests that subpools #7:22:9 and #7:22:18 contain cDNAs encoding discrete proteins that act cooperatively to stimulate isomerase activity.
To isolate a clone of the isomerase, we transformed E. coli with plasmid DNA from subpools #7:22:9 and #7:22:18. After plating at low density, we picked individual bacterial colonies and prepared plasmid DNA for a final round of transfection into 293T-LRC cells. We performed in vivo assays for isomerase activity in cell cultures expressing single clones from tertiary subpools #7:22:9 and #7:22:18. As before, we compared the levels of 11cROL produced in each culture dish to the level in pRK5-transfected 293T-LRC cells. Cells transfected with clone #7:22:9:10 contained 12-fold higher 11cROL (Figure 3H), while cells transfected with clone #7:22:18:2 contained 52-fold higher 11cROL than the pRK5-transfected control cells (Figure 3I). Sequence analysis of clone #7:22:9:10 revealed a cDNA identical to the published full-length coding region of bovine LRAT (accession number NM_177503). Sequence analysis of clone #7:22:18:2 revealed a cDNA identical to full-length bovine Rpe65 (accession number NM_174453).
To determine if the increased synthesis of 11cROL in LRAT-transfected 293T-LRC cells was due to endogenous isomerase activity of LRAT or stimulation of the isomerase by synthesis of its substrate, we transiently expressed LRAT in 293T-LC cells. These cells stably express LRAT and CRALBP but not Rpe65 (Figure 2). No 11cROL was synthesized from atROL by 293T-LC cells transfected with pRK5, LRAT-containing subpool #7:22:9, or pLRAT (clone #7:22:9:10) (Figure 3J). However, we observed significant synthesis of 11cROL from atROL by 293T-LC cells transfected with pRPE65 (clone #7:22:18:2) (Figure 3J). These data show that unlike Rpe65, LRAT has no endogenous isomerase activity. The stimulation of 11cROL production with transient expression of LRAT in 293T-LRC cells is likely due to increased synthesis of atREs by LRAT (Figure 3H).
The amount of 11cROL produced in 293T-LRC cells transiently transfected with pLRAT or pRPE65 was much greater than the amount produced in 293T-LRC cells transfected with pRK5 (Figures 3H and 3I). A possible explanation is that expression of LRAT and Rpe65 was higher following transient transfection. To test this possibility, we performed immunoblot analysis of 293T-LRC homogenates before and after transient transfection with pRPE65 or pLRAT. Rpe65 was approximately 10-fold higher in 293T-LRC cells transiently transfected with bovine Rpe65 compared to nontransfected 293T-LRC cells (Figure 4A). LRAT was more than 10-fold higher in transiently transfected versus nontransfected 293T-LRC cells (Figure 4A). As expected, neither protein was detectable in 293T cells.
We measured the synthesis of 11cROL by homogenates of 293T-RC and 293-LRC cells transiently transfected with pRPE65 or pRK5. Ten micromolar atROL or 10 μM all-trans-retinyl palmitate (atRP) were used as substrates in this experiment. No 11cROL was synthesized from atROL by homogenates of 293T-RC cells, which lack LRAT (Figure 4B). However, significant 11cROL was synthesized from atRP by homogenates of 293T-RC cells transfected with Rpe65. In contrast, homogenates from 293T-LRC cells transfected with pRPE65 synthesized 11cROL from both atROL and atRP substrates (Figure 4B). These data confirm that atREs are the substrate for the isomerase. Further, these data suggest that LRAT is not required for isomerase activity except to synthesize atREs from atROL.
The results presented above are consistent with two possible explanations: (1) Rpe65 is the isomerase (necessary and sufficient for activity), or (2) Rpe65 is not the isomerase but is required for the isomerase to exhibit activity (necessary but not sufficient). The second explanation requires that the “true” isomerase be expressed in 293T-LRC cells in an inactive form. To distinguish between these possibilities, we expressed several proteins including Rpe65 in baculovirus-infected Sf9 cells and assayed for production of 11cROL from atROL or atRP substrate. No 11cROL was produced from atROL added to the culture medium by Sf9 cells expressing green fluorescent protein (GFP), CRALBP, Rpe65, CRALBP plus LRAT, or CRALBP plus Rpe65 (Figure 5A). However, significant 11cROL was produced by Sf9 cells expressing LRAT and Rpe65, and by cells expressing LRAT, Rpe65, and CRALBP (Figure 5A). When atRP substrate was added to the culture medium, we observed 11cROL synthesis in all cells that expressed Rpe65 (Figure 5B). Much less 11cROL was synthesized with atRP, probably reflecting inefficient delivery of this insoluble substrate to the isomerase-containing internal membranes. The presence of CRALBP increased synthesis of 11cROL with both substrates. These data suggest that Rpe65 possess intrinsic isomerase activity.
To confirm these observations, we prepared membranes from baculovirus-infected Sf9 cells expressing GFP, LRAT, Rpe65, or LRAT plus Rpe65. These membranes were used for in vitro isomerase assays with atROL or atRP substrate. As before, we observed no significant 11cROL synthesis from atROL by membranes from cells expressing GFP, LRAT, or Rpe65 alone. However, membranes from Sf9 cells expressing LRAT plus Rpe65 synthesized significant 11cROL (Figure 5C). With atRP substrate, we observed synthesis of 11cROL by Sf9 membranes expressing only Rpe65 and by membranes expressing LRAT plus Rpe65. No significant 11cROL was produced by GFP- or LRAT-expressing membranes (Figure 5D). Representative chromatograms for the in vitro Rpe65 and GFP isomerase determinations (Figure 5D) are shown in Figures 5E and 5G. Figure 5F shows the UV spectrum from the 11cROL peak in Figure 5E.
To analyze the kinetics of Rpe65-catalyzed isomerization, we prepared membranes from Sf9 cells infected with recombinant baculoviruses for Rpe65 or GFP. We incubated these membranes in assay buffer containing different concentrations of atRP substrate and measured the initial rates of 11cROL formation by HPLC. Rpe65 exhibited classical Michaelis-Menten kinetics for a single-substrate enzyme (Figure 5I). No synthesis of 11cROL was observed by membranes from Sf9 cells expressing GFP. Eadie-Hofstee transformation of the data for Rpe65 yielded the kinetic parameters Vmax and KM of the isomerase for atRP substrate (Figure 5J).
We used site-directed mutagenesis to generate expression plasmids encoding bovine Rpe65 with the aminoacid substitutions C330Y and Y368H, associated with Leber congenital amaurosis in humans (Hanein et al., 2004; Thompson et al., 2000; Yzer et al., 2003). To study the effect of these substitutions on isomerase activity, we transiently transfected 293T-LC cells with wild-type and mutant Rpe65 plasmids. As before, we added at-ROL to the culture media and measured synthesis of 11cROL by these transfected cells following overnight incubation. Cells transfected with plasmid for wild-type Rpe65 synthesized significant 11cROL (~9 pmol per dish). However, no detectable 11cROL was synthesized by cells transfected with the plasmids for C330Y- or Y368H-substituted Rpe65 (Figure 6A). We assayed for expression of Rpe65 by immunoblotting. Similar levels of wild-type, C330Y- and Y368H-substituted Rpe65 were present in transfected 293T-LC cells (Figure 6B). Thus, nonproduction of 11cROL by these expressing cells was due to reduced catalytic activity and not decreased stability of the substituted Rpe65.
The goal of this project was to isolate a cDNA for the previously unidentified retinoid isomerase in RPE. To this end, we developed a screening assay for isomerase activity in a mammalian cell line. Prior work on the visual cycle suggested that coexpression of three additional proteins would be required to detect isomerase activity in the planned screen. LRAT would be required to synthesize atRE substrate from atROL added to the culture medium (MacDonald and Ong, 1988; Saari and Bredberg, 1989). Rpe65 would be required to bind atREs and present these as substrate to the isomerase (Mata et al., 2004). Finally, CRALBP, which specifically binds 11cROL (Saari and Bredberg, 1987), would prevent product-inhibition of the isomerase (Winston and Rando, 1998). We generated 293T-LRC cells, which stably expresses LRAT, Rpe65, and CRALBP (Figure 2) as a host line for the screen. This screen was unbiased with regards to identity of the isomerase. Nontransfected (data not shown) and pRK5 control-transfected 293T-LRC cells (Figure 3) synthesized small amounts of 11cROL from atROL added to the medium. At the time, we interpreted this background synthesis of 11cROL as an indication that 293T cells endogenously express the RPE isomerase at a low level. This erroneous interpretation was supported by previous reports that the related cell line, HEK-293S, possesses endogenous retinoid processing activities (Brueggemann and Sullivan, 2002; Chen et al., 2003; Ma et al., 1999).
As expected, synthesis of 11cROL increased at each step of the screen (Figure 3). Two subpools containing isomerase activity emerged in the third round of screening (Figure 3F). The 11cROL-producing clone in subpool #7:22:9 turned out to be LRAT. To rule out that LRAT possesses intrinsic isomerase activity, we transiently transfected subpool #7:22:9 and clone #7:22:9:10 (pLRAT) into 293T-LC cells, which stably express LRAT and CRALBP but not Rpe65 (Figure 2). No 11cROL was produced by these transfected cells (data not shown). In contrast, abundant 11cROL was produced by 293T-LC cells transiently transfected with pRPE65 (Figure 3J). Thus, LRAT does not possess intrinsic isomerase activity. The higher ester-synthase activity in cells transiently transfected with the LRAT clone suggests that isomerase activity was limited by the availability of atRE substrate during the screen. Substrate starvation also explains the in vitro potentiation of 11cROL synthesis in homogenates of 293T-LRC cells cotransfected with subpools #7:22:9 plus #7:22:18 (Figure 3G), and why the isolation of clone #7:22:9:10 from subpool #7:22:9 was not accompanied by an increase in 11cROL production (Figures 3F and 3H). Finally, the limited substrate availability in 293T-LRC cells explains the nonproportional increase in 11cROL synthesis with the degree of clonal enrichment observed at each stage of the screen. During the last step of the screen, 293T-LRC cells transfected with clone #7:22:18:2 synthesized 54-fold more 11cROL than pRK5-transfected cells. Identification of clone #7:22:18:2 as Rpe65 came as a surprise.
The results of this expression screen are consistent with an alternative explanation. Rpe65 may not itself be the isomerase but may be required for isomerase activity. This explanation requires that 293T cells endogenously express the true isomerase, which further must be catalytically inactive without coexpression of Rpe65. To test this alternate hypothesis, we expressed Rpe65 in baculovirus-infected Sf9 cells, derived from larvae of the Spodoptera frugiperda moth. Unlike the vertebrate, invertebrate metarhodopsin does not spontaneously hydrolyze to yield apo-opsin and atRAL following photoisomerization (Hamsdorf, 1979; Hubbard and St George, 1958; Schwemer, 1985). Instead, light sensitivity is restored to a bleached invertebrate pigment by absorption of a second photon, causing “reverse” photoisomerization of the coupled retinaldehyde chromophore in a process called photoregeneration. Thus, invertebrates, including insects, do not employ an enzymatic visual cycle and hence do not possess retinoid isomerase. Further, the insect genome does not contain an ortholog of Rpe65. It is therefore exceedingly unlikely that Sf9 cells endogenously express an inactive retinoid isomerase waiting to be “uncovered” by coexpression of vertebrate Rpe65. If expression of Rpe65 confers retinoid isomerase activity to Sf9 cells, this activity must be intrinsic to the Rpe65. We tested the alternative hypothesis by expressing several proteins including Rpe65 in Sf9 cells. The results of these experiments were unambiguous. With atROL as substrate, both LRAT and Rpe65 were required for synthesis of 11cROL (Figure 5A). With atRP as substrate, solitary expression of Rpe65 in Sf9 cells resulted in significant 11cROL synthesis (Figure 5B). Similar results were seen with the in vitro isomerase assay, using baculovirus-infected Sf9 cell membranes as an enzyme source (Figures 5C and 5D). Finally, Rpe65 exhibited classic Michaelis-Menten kinetics for synthesis of 11cROL from atRP (Figures 5I and 5J). The KM of Rpe65 was 7.1 μM. Collectively, these results confirm that Rpe65 possesses intrinsic isomerase activity.
Mice homozygous for a null mutation in the rpe65 gene show massive accumulation of atREs in the RPE and contain no detectable 11-cis-retinoids in the retina or RPE (Redmond et al., 1998). Partial loss of Rpe65 in rpe65+/- heterozygous and rpe65 L450M-substituted homozygous mice caused several-fold slowing of rhodopsin regeneration (Van Hooser et al., 2000; Wenzel et al., 2001). These observations are all consistent with Rpe65 functioning as the isomerase.
Humans with homozygous or compound-heterozygous mutations in the RPE65 gene are afflicted with the blinding disease Leber congenital amaurosis (Hanein et al., 2004; Thompson et al., 2000). If Rpe65 is the isomerase, disease-associated mutations in its gene should affect isomerase activity. We tested this prediction by measuring the activity of C330Y- and Y368H-substituted Rpe65. Both disease-associated substitutions abolished the catalytic activity of Rpe65 (Figure 6A) but had little effect on its stability in 293T-LC cells (Figure 6B). Recently, it was shown that Rpe65 undergoes S-palmitoylation at residues C231, C329, and C330 (Xue et al., 2004). The palmitoylated form has high affinity for atRP while nonpalmitoylated Rpe65 does not (Xue et al., 2004). These observations suggest that the membrane-associated form of Rpe65 is catalytically active. However, the complete loss of isomerase activity in C330Y-substituted Rpe65 suggests that residue C330 may play a role beyond modulating the affinity of Rpe65 for membranes. Rpe65 is a homolog of apocarotenoid-oxygenase (ACO). The structure of ACO from cyanobacterium was recently solved (Kloer et al., 2005). ACO was found to form a seven-bladed β propeller. Four histidine residues at the propeller hub, which are conserved in Rpe65, coordinate an Fe2+ ion that is required for catalytic activity. Also, residue C330 in Rpe65 is conserved in ACO (C322). ACO effects oxidative cleavage at the 15,15′ double bond of several all-trans apo-carotenoids to yield atRAL. Interestingly, binding of these substrates to ACO resulted in formation of 13,13′-di-cis-isomers (Kloer et al., 2005).
How has Rpe65 so long escaped functional identification? One reason is its low catalytic activity. Vmax for the isomerase in bovine RPE microsomes was 43 pmol per minute per mg protein using atROL as substrate (Winston and Rando, 1998). In contrast, Vmax for LRAT in bovine RPE microsomes was 103-199 nmol per minute per mg (Saari and Bredberg, 1988; Shi et al., 1993). Therefore, LRAT is several thousand-fold more active than the isomerase in similar membrane preparations. Rpe65 has been estimated to account for 10% of total microsomal proteins in bovine RPE (Bavik et al., 1992). Although the abundance of LRAT has not been published, it undoubtedly represents less than 1% of total microsomal proteins. Thus, the specific activity of LRAT is at least 25,000-fold higher than that of Rpe65. RPE cells probably compensate for this low catalytic activity by increasing the abundance of Rpe65. With a nonlimiting substrate, such as atREs in RPE membranes, the “strategy” of upregulating expression of an enzyme to compensate for its low catalytic is kinetically valid. Hence, the high abundance of Rpe65 is probably a reflection of its low activity as an enzyme. Another reason why assigning a catalytic function to Rpe65 has proven so difficult is that this enzyme loses activity during purification (Mata et al., 2004). The water insolubility of atREs further complicated analysis of the isomerase. Use of atRP as substrate results in synthesis of very little 11cROL (Gollapalli and Rando, 2003; Mata et al., 2004; Moiseyev et al., 2003). For this reason, most investigators use atROL as substrate, relying on endogenous LRAT to form atREs in situ (Winston and Rando, 1998). The dependence on LRAT to synthesize substrate for the isomerase has thwarted its purification and hence identification. An important observation in the current study is that addition of sodium cholate to the isomerase buffer permits efficient utilization of atRP substrate. For example, using membranes from Rpe65-expressing Sf9 cells and otherwise identical assay conditions, we observed 9.2-fold greater synthesis of 11cROL from atRP in the presence of 3 mM sodium cholate (Figure 5H). This permitted us to assay Rpe65-isomerase using its true substrate in the absence of LRAT.
Still another issue that has confounded identification of Rpe65 as the isomerase is the report that bovine RPE microsomes stripped of most Rpe65 protein retained isomerase activity (Choo et al., 1998). Rpe65 has been shown to exist in two forms that differ in molecular mass by approximately 1 kDa (Ma et al., 2001). The higher-mass form is associated with membranes while the lower-mass form is present in the cytosol (Ma et al., 2001). When we repeated the saline-extraction experiment with addition of 0.1% CHAPS detergent, we observed a proportional decrease in Rpe65 immunoreactivity and isomerase catalytic activity in the membrane pellet (M.J. and W.N.M., unpublished data). Thus, when the differential solubility of palmitoylated and nonpalmitoylated Rpe65 (Xue et al., 2004) was taken into account, isomerase activity was directly correlated with Rpe65 immunoreactivity. Interestingly, human Rpe65 expressed in Sf9 cells was predominantly associated with membranes (Ma et al., 2001). This explains the relatively high isomerase activity observed in membranes from Sf9 cells expressing Rpe65 (Figure 5). The biochemical mechanism for palmitoylation of Rpe65 in Sf9 cells is unknown. However, we can rule out LRAT as the obligate acyltransferase, since expression of Rpe65 without LRAT did not reduce isomerase activity (Figure 5D). In fact, isomerase activity was slightly higher when Rpe65 was expressed in 293T or Sf9 cells without LRAT (Figures (Figures4B,4B, 5B, and 5D). This effect may be due to depletion of atRP substrate by LRAT through the reverse estersynthase reaction (Saari et al., 1993). We conclude that LRAT is not required for isomerase activity except to synthesize atRE substrate from atROL. In contrast, coexpression of Rpe65 with CRALBP stimulated the synthesis of 11cROL (Figures 5A and 5B). This observation is consistent with the proposed role of CRALBP as an 11-cis-retinoid binding-protein (Saari and Bredberg, 1987) that frees the isomerase from inhibition by its 11cROL product (Winston and Rando, 1998).
In summary, we identified Rpe65 as the retinoid isomerase in RPE using an unbiased expression screen. We confirmed this identification by demonstrating isomerase activity in Sf9 cells that express only Rpe65. It is possible that Rpe65 requires for catalytic activity an auxiliary protein endogenously present in both 293T and Sf9 cells. This newly assigned function for Rpe65 is consistent with the phenotypes in animals and humans bearing mutations in the RPE65 gene.
We subcloned the coding region of bovine Rpe65 (Hamel et al., 1993) into pTriEx-Bsd, which we prepared from pTriEx-1.1-Neo (Novagen) by replacing the neomycin resistance gene with the blasticidin resistance gene. We subcloned the coding region of bovine LRAT (Ruiz et al., 1999) into pTriEx1.1-Hygro (Novagen), which contains the hygromycin resistance gene. Finally, we subcloned human CRALBP (Intres et al., 1994) into pIRESpuro3 (BD Biosciences), which contains the puromycin resistance gene. HEK293T (293T) cells were grown in D-MEM supplemented with 10% heat-inactivated fetal bovine serum (FBS) and antibiotics (100 units per ml penicillin G and 100 μg per ml streptomycin). The 293T cells were transfected with pCRALBP-puro, pLRAT-hygro, or pRPE65-Bsd using PolyFect transfection reagent (Qiagen) and the manufacturer’s procedure. Stably transformed lines were established by maintaining these transfected cells in media containing the appropriate antibiotics (10-25 μg per ml puromycin, 100-300 μg per ml hygromycin B, and/or 10-25 μg per ml blasticidin) for 1-2 weeks. Individual antibiotic-resistant colonies were expanded and tested for expression of the integrated plasmid(s) by immunoblotting (Figure 2). To generate cells that stably express pairs of proteins, we transfected cells from lines that express LRAT (293T-L), Rpe65 (293T-R), or CRALBP (293T-C) with a second plasmid and cultured under selection with the appropriate pairs of antibiotics. This resulted in the cell lines 293T-LR, 293T-RC, and 293T-LC, which stably express LRAT plus Rpe65, Rpe65 plus CRALBP, and LRAT plus CRALBP, respectively. Finally, to generate a cell line that expresses all three proteins (293T-LRC), we transfected 293T-LC cells with pRPE65-Bsd and cultured under selection with puromycin, hygromycin B, and blasticidin. The resulting seven stable cell lines were maintained in the same media without antibiotics and split (1:4) every two days.
We isolated RPE from dark-adapted cattle eyes obtained fresh from a local slaughter house, and prepared poly(A)+ RNA using established procedures (Sambrook et al., 1989). This RNA was used as a template for synthesis of double-stranded cDNA using the cDNA synthesis system for the lambda ZAP vector following the manufacturer’s procedures (Stratagene). The resulting cDNA, which contained an XhoI site at the 3′ end and was methylated on one strand, was ligated to EcoRI adapters. After size-fractionation by gel filtration, cDNA larger than 1 kb was digested with EcoRI and XhoI and directionally cloned into pRK5-SK (Stratagene) linearized with EcoRI and XhoI. The library plasmid was electroporated into Epicurian Coli (Stratagene) and plated onto 42 plates with approximately 5000 colonies per plate. The plates were overlaid with medium, and plasmid DNA was prepared from the suspended cultures. These 42 pools of plasmid comprised the primary bovine RPE library.
For the first round of screening, DNA from the 42 primary library pools were transfected into 293T-LRC cells grown in 100 mm dishes using the PolyFect reagent. Nonrecombinant pRK5 was used as a negative control. For the second round of screening, plasmid DNA from pool #7 was transfected into 293T-LRC cells grown in 60 mm plates. For the third round of screening, plasmid DNA from subpool #7:22 was prepared and used to transfect 293T-LRC cells growing in 6-well culture plates. For the final round of screening, plasmid DNA corresponding to individual clones from subpools #7:22:9 and #7:22:18 were transfected into 293T-LRC cells grown in 12-well culture plates. For each round of screening, cells were prepared for transfection from a master mixture. The density of cells varied less than 10% between dishes within each round of screening. Thirty hours after transfection, cell medium was replaced with fresh medium containing 15% FBS, 0.5% bovine serum albumin (BSA), and 25 mM HEPES buffer (pH 7.5). We added atROL in ethanol to the medium (5 μM final concentration) under dim red light and incubated the cells for an additional 18 hr. Cells were collected from the plates, pelleted by low-speed centrifugation, and lysed in 0.5 ml lysis buffer (0.2% SDS in 10 mM HEPES buffer [pH 7.5]). The cell lysates were mixed with 1 ml methanol and 150-550 μl of 6 M KOH (depending on the size of culture dish) and incubated for 15 min at 55°C to saponify retinyl esters. After 0.5 ml of dH2O was added, retinoids were extracted with 24 ml hexane. The retinoid-containing organic phase was collected and evaporated to dryness under a stream of argon for HPLC analysis (see below).
Retinoids were analyzed by normal phase HPLC as previously described (Mata et al., 2002). In brief, retinoids dissolved in 200 μl hexane were separated on a silica column (Supelcosil LS-SI 5-mm, 4.6 × 250 mm ID) using 10% dioxane in hexane as mobile phase at a flow rate of 1.0 ml per minute in an Agilent 1100 liquid chromatograph equipped with a UV photodiode-array detector. Identified peaks were confirmed by spectral analysis and by coelution with authentic retinoid standards.
Cloned cDNAs for GFP, human CRALBP, and bovine LRAT and Rpe65 were subcloned into the plasmid pBAC-1 (Novagen). These baculovirus transfer vectors were cotransfected into Sf9 cells with BacVector-2000 triple-cut virus DNA (Novagen) using Cellfectin (Invitrogen) to generate recombinant baculoviruses expressing GFP, CRALBP, LRAT, and Rpe65. High-titer virus stocks were made by amplifying the recombinant viruses in Sf9 cells. Expression of recombinant proteins was verified by immunoblotting (data not shown). For the in vivo assays, Sf9 cells were grown to confluence in 12-well plates and infected with the indicated high-titer baculovirus stocks. Two days postinfection, media were replaced with fresh media containing 15% FBS and atROL (Fluka) or atRP (Sigma) in ethanol was added to yield a final concentration of 5 μM. Cells were incubated 18 hr in the dark, and retinoids were extracted with hexane and analyzed by HPLC as described above.
Two days posttransfection of the indicated 293T cell line, or three days postinfection of Sf9 cells with recombinant baculovirus, cells were collected by centrifugation and resuspended in ice-cold 10 mM HEPES buffer (pH 7.5) containing 100 mM NaCl, 1 mM MgCl2, 1 mM CaCl2, and protease inhibitor cocktail. Cells were homogenized in a glass-to-glass tissue grinder. The resulting cell homogenates were assayed directly for the 293T cell lines. For the Sf9 cells, membranes were prepared as described (Ma et al., 2001).
The isomerase assay mixtures contained 25 mM HEPES (pH 7.5), 100 mM NaCl, 2 mM CaCl2, 2 mM MgCl2,20 μM leupeptin, and 5% BSA. For all experiments except the one represented in Figure 5H, 3 mM sodium cholate was added to the assay mixtures. The atROL or atRP substrate was dissolved in dimethylsulfoxide and added to yield a final concentration of 10 μM. Cell homogenates (293T-LRC cells) or membrane fractions (Sf9 cells) were added to yield final protein concentrations of 1.0-2.0 mg per ml. Reaction volumes were 100-200 μl. After incubation for 2 hr in dark at 37°C, the reactions were quenched by adding 0.2% SDS and two volumes of methanol. Saponification of retinyl esters, as described above, was done on all retinoid samples derived from atROL by cells expressing LRAT. Extraction of retinoids with hexane and HPLC analysis were done as described above.
Membrane fractions of Sf9 cells infected with recombinant baculoviruses expressing Rpe65 or GFP were incubated with increasing concentration of atRP (0.125 to 125 μM) in assay buffer (25 mM HEPES, 100 mM NaCl, 2 mM CaCl2, 2 mM MgCl2, 20 mM leupeptin, 6 mM sodium cholate, and 5% BSA) for 30 min at 37°C in the dark. Samples were quenched and retinoids extracted for HPLC analysis as described above. The quantified data were then fitted to the Michaelis-Menten equation using the Enzyme Kinetics Module 1.1 for Sigmaplot version 8.0. Vmax and KM values were obtained from the rate data by the Eadie-Hofstee transformation.
Cell homogenates containing 10 μg total protein were heated at 100°C for 2 min in Laemmli sample buffer, separated by electrophoresis in a 12% polyacrylamide-SDS gel, and transferred to an Immobilon-P membrane (Millipore). The membrane was incubated in blocking buffer (pH 7.4 PBS, 5% nonfat milk) for 2 hr at 37°C, then with a rabbit polyclonal antibodies against CRALBP (Crabb et al., 1991), LRAT (Bok et al., 2003), or Rpe65 (Mata et al., 2004) overnight at 4°C. After washing in PBS for 30 min, the membrane was incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories) for 1 hr and washed again. The immunoreaction was visualized with the enhanced chemiluminescence (ECL) system (Amersham Biosciences).
DNA sequencing on both strands was done using a DYEnamic ET Terminator Cycle Sequencing Kit (Amersham Biosciences) with an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems/Hitachi). Vector-specific and insert cDNA-specific primers were designed using Vector NTI software (Invitrogen) and are available on request. Database searches were performed in the internet server http://www.ncbi.nlm.nih.gov/BLAST.
Point mutations resulting in the C330Y and Y368H amino-acid substitutions were introduced into the bovine RPE65 cDNA using QuikChange II site-directed mutagenesis kit (Stratagene). The resulting constructs, pRK-C330Y and pRK-Y368H, were confirmed by DNA sequence analysis. We transfected similar numbers of 293T-LC cells (from the same master cell suspension) in 12-well tissue culture dishes with nonrecombinant pRK5, pRK-C330Y, pRK-Y368H (1,000 ng each), or wild-type pRPE65 (250 ng) plasmid. Following overnight incubation, we performed the in vivo isomerase assay with atROL added to the culture media as described above. We also prepared protein homogenates from representative culture dishes for immunoblot analysis, as described above.
We gratefully acknowledge Nathan Mata and Roxana Radu for their valuable insights and suggestions. We thank Alberto Ruiz and Dean Bok for their gift of antisera to LRAT. We thank Rosalie Crouch for her gift of 11cRAL. We thank John Crabb for his clone of CRALBP and antisera to CRALBP. This work is supported by grants from the National Eye Institute and the Foundation Fighting Blindness. G.H.T. is the Charles Kenneth Feldman and Jules & Doris Stein Research to Prevent Blindness Professor.
(I)In vivo analysis of individual clones from subpool #7:22:18. Sequence analysis of plasmid from clone #7:22:18:2 showed it to be Rpe65.
(J)In vivo isomerase assay of LRAT and Rpe65 in 293T-LC cells, which do not express Rpe65. Histograms show 11cROL peak areas from HPLC chromatograms acquired at 318 nm. Error bars show standard deviations (n = 4). Note the absence of 11cROL synthesis in these cells transfected with pRK5, LRAT-containing subpool #7:22:9, or pLRAT (clone #7:22:9:10). Also note the synthesis of 11cROL in these same cells transfected with pRPE65 (clone #7:22:18:2). The substrate for all assays represented in this figure was atROL. Gray bars on the histograms denote in vivo while black bars denote in vitro assays.