Retinoid Binding Properties of Nucleotide Binding Domain 1 of the Stargardt Disease-associated ATP Binding Cassette (ABC) Transporter, ABCA4*

Background: The ABCA4 protein is proposed to transport all-trans-retinal from the outer segment discs of retinal rod and cone photoreceptors. Results: 11-cis-Retinal bound specifically and with high affinity to the NBD1 domain of hABCA4. Conclusion: The NBD1 domain plays further roles in addition to nucleotide hydrolysis. Significance: ABCA4 may play a novel role in the visual transduction cycle involving 11-cis-retinal. The retina-specific ATP binding cassette transporter, ABCA4 protein, is associated with a broad range of inherited macular degenerations, including Stargardt disease, autosomal recessive cone rod dystrophy, and fundus flavimaculatus. In order to understand its role in retinal transport in rod out segment discs, we have investigated the interactions of the soluble domains of ABCA4 with both 11-cis- and all-trans-retinal. Using fluorescence anisotropy-based binding analysis and recombinant polypeptides derived from the amino acid sequences of the four soluble domains of ABCA4, we demonstrated that the nucleotide binding domain 1 (NBD1) specifically bound 11-cis-retinal. Its affinity for all-trans-retinal was markedly reduced. Stargardt disease-associated mutations in this domain resulted in attenuation of 11-cis-retinal binding. Significant differences in 11-cis-retinal binding affinities were observed between NBD1 and other cytoplasmic and lumenal domains of ABCA4. The results suggest a possible role of ABCA4 and, in particular, the NBD1 domain in 11-cis-retinal binding. These results also correlate well with a recent report on the in vivo role of ABCA4 in 11-cis-retinal transport.

The historically accepted model of ABCA4 function has been simply its role in exporting all-trans-retinal from the ROS to the cytoplasm. In the context of the visual transduction, this is important because in order to maintain the retinoid cycle, alltrans-retinal released from light-activated rhodopsin must be recycled from the ROS disc back to the retinal pigment epithelium, where it is enzymatically converted to the 11-cis isomer, to again serve as the chromophore of rhodopsin. Defective ABCA4 function is believed to contribute to the accumulation of cytotoxic lipofuscin that underlies the pathology of the macular diseases leading to photoreceptor cell death.
Excess all-trans-retinal can lead to the formation of all-transretinylidene-phosphatidylethanolamine, a component of lipofuscin (24). This reasoning has helped to support the hypothesis that ABCA4 functions as an exporter of all-trans-retinal. However, using ABCA4(Ϫ/Ϫ) mice, Boyer et al. (25) recently showed that 11-cis-retinal can also support the formation of lipofuscin and proposed that 11-cis-retinal may be imported into the ROS discs by ABCA4. Currently, a mechanism of 11-cis-retinal import into the ROS discs is not known except through membrane diffusion. A transporter for 11-cis-retinal has not yet been identified.
ABCA4 possesses a characteristic ABC protein family structure consisting of two membrane-spanning domains (trans-membrane domains), with each containing six ␣-helices, and two cytosolic nucleotide-binding domains (NBDs) (Fig. 1). The primary sequences of ABC transporter transmembrane domains are highly variable, whereas the NBDs contain the conserved Walker A and Walker B consensus motifs. In addition, the NBDs harbor the LSGGQ signature sequence of ABC proteins, albeit diminished to SGG in the NBD2 domain of ABCA4 (26). Unique to the ABCA subfamily is the presence of large extracellular loops, characteristic of this subfamily (Fig. 1). In ABCA4, these loops are thought to project into the lumen of the rod outer segment discs. Tysbovsky et al. identified the presence of the EAA motif in the transmembrane region of the N-terminal half of ABCA4 (27,28). This motif is characteristic of ABC proteins that are known to be importers (29). The presence of the EAA motif in ABCA4 is unusual, because in general, eukaryotic ABC transporters are exporters. The EAA motif is absent in the corresponding C-terminal half of ABCA4. The significance of the EAA motif in ABCA4 remains unknown; however, it may point to a 11-cis-retinal import function of the protein.
The biochemical and kinetic properties of the nucleotide binding domains have been studied in individual recombinant polypeptides as well as in native and recombinant full-length ABCA4 protein (30 -35). Through a variety of independent experimental approaches, the NBD1 and NBD2 domains have been shown to have distinct enzymatic and kinetic properties. NBD1 possesses a low general ribonucleotidase activity, whereas NBD2 has a higher rate of hydrolysis and is specific for adenine nucleotides (36). Thus, NBD1 and NBD2, although both important, probably play distinct mechanistic roles, consistent with that found for other ABC transporters, such as ABCB1 and ABCG2 (37). Although half the size of NBD1, based on its kinetic properties, NBD2 appears to provide the energy required for translocation, whereas NBD1 serves a yet undefined cellular function. Several studies demonstrate that many disease-associated mutations mapping to the NBD1 of ABCA4 do not affect its function as an ATPase (15,28,30,32,38).
Using fluorescence anisotropy, we have shown that the ECD2 domain of ABCA4 interacts specifically and with high affinity to all-trans-retinal (39). Analysis of the interaction of the other soluble domains of ABCA4 (ECD1, NBD1, and NBD2) showed that these domains do not bind all-trans-retinal with appreciable affinity. The ability of all-trans-and 11-cisretinal to stimulate the ATPase of ABCA4 indicates that the protein engages in a physical interaction with both geometric isomers. However, it is not known which domain of ABCA4 mediates interaction with the 11-cis isomer of retinal.
Based on recent studies proposing a physiological role of ABCA4 that may include translocation of 11-cis-retinal across the ROS disc membrane (25) as well as preliminary studies pointing to a specific interaction of NBD1 with 11-cis-retinal (40), we posed the following question. Do the nucleotide binding domains of ABCA4 interact with retinal, and if so, how is this interaction influenced by disease-associated mutations in these domains? In this report, we have investigated the retinal binding properties of the first nucleotide binding domain of ABCA4 and examined the effects of disease-associated mutations on retinal binding.

EXPERIMENTAL PROCEDURES
Nucleic Acids, Enzymes, and Other Reagents-The pRK5 plasmid containing the full-length, wild-type cDNA of the human ABCA4 gene was obtained as a generous gift from Drs. J. Nathans and Michael Dean of Johns Hopkins University (Baltimore, MD) and NCBI (Frederick, MD), respectively. The T7 expression system vector pET30b, Bug Buster protein extraction reagent, Benzonase nuclease, and the S-protein-agarose affinity resin were from Novagen (EMD Sciences, Briggstown, NJ). All-trans-retinal was from Sigma-Aldrich, whereas 11-cisretinal was received through Dr. R. Crouch (Medical University of South Carolina) under the auspices of the NEI, National Institutes of Health, resource program for vision researchers.
In Vitro Site-directed Mutagenesis of the NBD1 Construct-Site-directed mutagenesis was carried out using a PCR-based mutagenesis kit (Stratagene, La Jolla, CA) (35), pET30-NBD1 plasmid as template, and allele-specific primers as described previously (35). The authenticity of the mutations and the absence of other fortuitous mutations were confirmed by DNA sequencing carried out by Eurofins MWG/Operon (Huntsville, AL).
Overexpression of pET30b-NBD1 in E. coli-E. coli cells (strain BL21-CodonPlus(DE3)-RIPL, Stratagene (La Jolla, CA)) harboring pET30b-NBD1 plasmid were used to produce the recombinant NBD1 polypeptide following the manufacturer's instructions. The expressed recombinant polypeptide appeared to be of the anticipated size (62 kDa), as determined by SDS-PAGE.
Extraction and Purification of Recombinant NBD1 Proteins-Extraction and purification of wild-type NBD1 protein carrying the S-tag was performed using immobilized S-proteinagarose affinity resin (EMD Chemicals, Gibbstown, NJ) following the manufacturer's recommendations as described previously (36).
Purification of NBD1 Polypeptide from Solubilized Inclusion Bodies-Introduction of mutations into wild-type NBD1 polypeptide appeared to decrease the solubility of the expressed proteins as determined by SDS-PAGE and a Western blot procedure (data not shown). Consequently, we explored the extraction of recombinant proteins (wild type and mutants) from the inclusion bodies followed by refolding (42). This approach has been shown to be highly successful in the purification of a number of ABC transporters (35,(43)(44)(45).
The wild-type and mutant NBD1 proteins were extracted from inclusion bodies using a protocol that combines the use of BugBuster protein extraction reagent (Novagen, Madison, WI) to process the insoluble fraction and yield purified inclusion bodies with the method described by Booth et al. (35,43,46). After harvesting the expressed proteins, the cell pellets were resuspended in room temperature BugBuster reagent, and protease inhibitors were added. After incubation on ice for 30 min, the cell suspension was centrifuged to collect purified inclusion bodies. Following cell lysis, the pellet of inclusion bodies was resuspended in buffer B and centrifuged once more. The inclusion body proteins were solubilized in Buffer C. Protein refolding was achieved by rapid dilution in Buffer D. The renatured protein was sequentially dialyzed in Buffer E. After overnight dialysis, proteins were concentrated to ϳ0.5 mg/ml by ultrafiltration (Amicon/Millipore). Overall, the inclusion body protein purification methodology described here yielded highly concentrated, purified, and homogeneous preparations of protein (Fig. 2). The yield of NBD1 protein was Ͼ10 mg from 4 liters of induced cell culture. Purified proteins were stored at Ϫ80°C until use.
Anisotropy Measurements-Fluorescence anisotropy was measured to investigate retinal binding by NBD1 protein in solution. Anisotropy measurements were carried out as described using a steady-state photon-counting spectrofluorometer, PC1, with Vinci software, from ISS Instruments (Champaign, IL) and Fluorolog 3 from Horiba Instruments Inc. (Edison, NJ) (39). The temperature was maintained at 25°C using a Peliter controlled cuvette holder. The excitation wavelength was set at 310 nm, and fluorescence anisotropy was recorded at an emission wavelength of 430 nm. Excitation and emission slits were adjusted to 8 nm to maximize intensity counts. The given retinal isomer was diluted to a concentration of 100 nM and titrated with NBD1 protein within a concentration range of 0.1 nM to 2 M. The sample was incubated for 2 min after each addition. The S.D. for the anisotropy values was ՅϮ0.005 A. Anisotropy at each titration point was measured three times for 10 s and averaged. The total fluorescence intensity did not change significantly (Յ10%) with increase in NBD1 concentration. Therefore, fluorescence lifetime changes or the scattered excitation light did not affect the anisotropy measurements. Anisotropy (A) is defined as follows, where G is the instrumental correction factor for the fluorometer, and it is defined by Equation 2, where I vv , I vh , I hv , and I hh represent the fluorescence signal for excitation and emission with the polarizers set at (0°, 0°), (0°, 90°), (90°, 0°), and (90°, 90°), respectively. The interaction of NBD1 with ligand (all-trans-retinal or 11-cis-retinal) (L) can be represented as follows.
At equilibrium, K a , the equilibrium association constant, can be given as follows.
The fraction of the bound ligand, f, can be represented as follows.
Substituting for [NBD1] and rearranging the equation, we get the following.
Similarly, the equilibrium dissociation constant K d can be expressed, substituting K d for 1/K a , as follows, . Thus, K d can be further defined as the protein concentration at which half of the sites are occupied when the ligand concentration is constant, as in the present case, or the ligand concentration at which half of the sites are occupied when the protein concentration is constant. Analyses of the data were conducted using PRISM (GraphPad Software Inc., San Diego, CA). The K d value (i.e. the concentration of NBD1 required to bind 50% of the ligands) was computed using Equation 9, where A min and A max are the anisotropy values at the bottom and top plateaus, respectively; X represents the log of the NBD1 concentration; X 0 is the X value when the response is halfway between the top and the bottom; and n app is the Hill coefficient.
Other Methods-Routine protein concentrations were determined by the method of Bradford (47)

Structural Domains of ABCA4 and Motifs of Nucleotide
Binding Domain 1-The NBDs are one of the defining features of the ABC family of transporters. Current topological models of human ABCA4 protein indicate the presence of two cytoplasmic loops corresponding to the NBDs. As shown in Fig. 1, NBD1 comprises aa 854 -1375, and the second nucleotide binding domain, NBD2, is localized to the C-terminal half of the molecule at aa 1898 -2273 (49). Both NBD1 and NBD2 contain Walker type A (GXXXXGKT) and type B ((R/K)XXXXGXXXX-LhhhhD) nucleotide binding motifs (26). We have shown earlier that NBD1 binds and hydrolyzes all ribonucleotide triphosphates (rNTPs), but NBD2 is specific for ATP.
ABCA4 possesses two long extracellular/lumenal loops, ECD1 (aa 62-646) and ECD2 (aa 1395-1680); these loops project toward the lumen of the ROS discs. The ECD2 domain interacts specifically with all-trans-retinal (39). Tsybovsky et al. identified an EAA motif, localized to the membrane-spanning region of the N-terminal half of ABCA4 (28). This extended motif has the consensus sequence of EAAXXXGXXXXXXX-IXLP (27). The motif is characteristic of ABC proteins that function as importers, where it contributes to the coupling between the transmembrane helices and the NBDs (29).
Comparison of NBD1 of ABCA4 across the ABC Protein Family-ABC transporters are one of the most abundant and phylogenetically widespread protein superfamilies. Analyses of ABC transporter sequences have suggested that they share an evolutionarily ancient origin and began to specialize even before the division of Archaea, Bacteria, and Eukarya (50). Sequence alignment of NBD1 of ABCA4 from several vertebrates demonstrates that the NBD1 domain remained strictly conserved (Ն96% identity) (supplemental Fig. S2).
To explore the degree of NBD conservation further, the sequences corresponding to the NBD1 domains of several ABCA subfamily members were aligned and compared with that of the NBD1 of ABCA4 (Fig. 3). Although areas of significant sequence conservation were observed, they were localized primarily to the N-terminal and central region of the domains and encompassed the Walker nucleotide binding A and B motifs. The level of homology (identity plus similarity) was approximately ϳ55% across members of the ABCA subfamily (Fig. 3). Clearly, the NBD1 sequences of ABCA4 orthologs (supplemental Fig. S2) are more highly conserved than those of NBD1 domains found in other vertebrate ABCA subfamily proteins (Fig. 3). When comparisons were made between the NBD1 domains of ABCA4 and members of the ABC protein family in general (supplemental Fig. S3), roughly 35% homology was observed; however, the regions of similarity did not appear to be clustered into subdomains but rather appeared randomly distributed throughout. Consequently, despite regions of conservation, such as the ABC signature sequence, Q-loop, and Walker motifs, areas lacking sequence homology are also present within NBDs of ABC and ABCA paralogs. The strict conservation of the NBD1 domain of ABCA4 relative to its related paralogs would seem to suggest a specific functional significance of these residues beyond just that of a role in ATP binding and hydrolysis.
Interaction of the NBD1 Domain with Retinoids-Earlier, we have shown that of four soluble domains of ABCA4, ECD2 bound all-trans-retinal, and not 11-cis-retinal, with appreciable affinity. We have developed a direct retinal binding assay using the fluorescence anisotropy of all-trans-and 11-cis-retinal. Highly purified homogeneous preparations of recombinant NBD1 proteins were used to analyze retinal interaction in these domains (Fig. 2). Fluorescence anisotropy was measured using 100 nM 11-cis-retinal with the wavelength set at 310 nm for excitation and 430 nm for emission. The fluorescence anisotropy of 11-cis-retinal was 0.06 Ϯ 0.005 A, which increased upon titration with NBD1 protein as shown in Fig. 4. A sigmoidal semilog plot was obtained; saturation binding was observed with a maximum anisotropy of 0.115 Ϯ 0.005 A. The data shown in Fig. 4 were analyzed by nonlinear regression using Equation 9 as described under "Experimental Procedures." The binding parameters were determined from the regression analysis. Based on this investigation, a dissociation constant (K d ) of 8.0 Ϯ 1.3 ϫ 10 Ϫ8 M was obtained for NBD1 binding to 11-cis-retinal (Table 1). Thus, the NBD1 domain bound with reasonable affinity to 11-cis-retinal. To explore the specificity of retinal interaction, an analogous titration of NBD1 protein in the presence of 100 nM all-trans-retinal was carried out. The data shown in Fig. 4 were analyzed by nonlinear regression analysis as described above. Although some initial changes in anisotropy with increasing concentrations of all-trans-retinal were seen, saturation binding was not observed in the concentration range studied; the K d for alltrans-retinal was Ն2.0 ϫ 10 Ϫ6 M. Thus, NBD1 interaction appeared specific for 11-cis-retinal. . Sequence alignment of NBD1 domains derived from several human ABCA subfamily members. The amino acid sequences corresponding to the NBD1 domains of several human ABCA subfamily members (ABCA4, ABCA1, ABCA13, and ABCA12) were obtained from NCBI and aligned using ClustalW2 (available on the EMBL-EBI Web site). The amino acid residue number for each protein is indicated at the beginning of each sequence. The alignment is color-coded to indicate chemical characteristics of a given amino acid. Red, basic; blue, hydrophobic; green, hydrophilic; orange, neutral; pink, acidic; light green, proline. The Walker A and Walker B nucleotide binding motifs are enclosed in the consensus sequence by a red rectangle.

Inhibition of 11-cis-Retinal Binding in the Presence of
Nucleotides-NBD1 is an rNTPase and binds ribonucleotides; thus, we have analyzed possible effects of nucleotide binding on the interaction of NBD1 with 11-cis-retinal. Binding was measured, as shown in Fig. 5, in the presence and absence of ADP and AMP-PNP, a non-hydrolyzable analog of ATP. AMP-PNP was used in lieu of ATP to prevent hydrolysis of the effector during anisotropy measurements. Fluorescence anisotropy isotherms in the presence of AMP-PNP and ADP are shown in Fig.  5, A and B. In the absence of nucleotide, a sigmoidal semilog plot was obtained indicating equilibrium saturation binding of 11-cis-retinal by NBD1, similar to that observed in Fig. 4. The interaction of 11-cis-retinal with NBD1 was significantly influenced in the presence of nucleotides. Binding was markedly attenuated in the presence of AMP-PNP, and a lesser inhibition was observed in the presence of ADP (Fig. 5). Nonlinear regression analysis of the data in Fig. 5, A and B, gave a dissociation constant (K d ) of 3.6 Ϯ 0.8 ϫ 10 Ϫ7 M for 11-cis-retinal binding in the presence of ADP and 4.1 Ϯ 0.6 ϫ 10 Ϫ6 M for 11-cis-retinal binding in the presence of AMP-PNP (Table 1). Thus, 11-cisretinal binding was notably attenuated in the presence of AMP-PNP and ADP.
Influence of Stargardt Disease Mutations on NBD1 Interaction with 11-cis-Retinal-Genome-wide association studies have provided us with a unique roadmap to explore the structure-function relationships between mutations and visual dis-   ease (51)(52)(53)(54). Several macular degenerative disease-associated mutations map to the NBD1 domain of ABCA4 (16,(55)(56)(57)(58). Three ABCA4 mutations associated with Stargardt disease (R943Q, P940R, and G863A) were chosen for analysis in this study. These missense mutations have been observed to occur in patients as simple and/or compound heterozygotes. The missense mutations in NBD1 examined in this study result in modest decreases in the ATPase activity of NBD1, as reported earlier (Table 1) (35). All three of the mutations correspond to strictly conserved amino acids in vertebrate ABCA4 proteins (supplemental Fig. S2).
We have explored the effects of missense mutations R943Q, P940R, and G863A on 11-cis-retinal interaction with the NBD1. The plots of fluorescence anisotropy changes of 11-cisretinal with wild type and mutant NBD1 protein titrations are shown in Fig. 6, A-C. The binding of mutant G863A was significantly attenuated as indicated by the drastic right shift of the binding isotherm as well as its inability to achieve saturation in the presence of a high concentration of protein (Fig. 6A) Table 1). As a consequence of Stargardt disease mutations, 100-fold (R943Q) to 50-fold (P940R) decreases in the binding affinity of the NBD1 domain for 11-cis-retinal were observed. The retinal binding of the G863A mutant was severely attenuated, and the K d was Ն1.0 ϫ 10 Ϫ5 M.

DISCUSSION
The ABCA4 protein is an essential component of the visual transduction cycle of vertebrate retina. A distinctive feature of ABCA4 is its four large soluble domains, NBD1, NBD2, ECD1, and ECD2. The NBDs of all ABC transporters share a common role in providing the energy required for translocation of their substrates through the hydrolysis of ATP. In ABCA4, ATP hydrolysis to support transport is believed to be carried out at the NBD2 domain (59,60). The general ribonucleotidase activity of NBD1 and its low rate of hydrolysis have led to the speculation that this nucleotide binding may serve a regulatory function rather than having a role in energy transduction. Recent studies have suggested that the NBDs of ABC transporters do play additional roles, such as coupling to the transmembrane ␣-helical domains required for ligand transport (61,62). In the ABCG5/ABCG8 sterol transporter, the NBD1 and NBD2 domains are not functionally equivalent. The ABCG5/ABCG8 NBD1 domain plays an important role in substrate selectivity and binding during sterol transport from hepatocytes into bile, whereas the NBD2 domain is the engine that drives the translocation process (63). However, other function(s) of NBD1 of ABCA4, in addition to its low level ribonucleotidase activity, has not been shown or proposed.
Sequence conservation is indicative of an evolutionarily conserved function. To gain insights into the role(s) and relative degree of sequence conservation of the NBD1 domains, we aligned (i) the NBD1 domains of ABCA4 proteins from several vertebrate species (supplemental Fig. S2), (ii) the NBD1 domains from several members of the ABCA subfamily (Fig. 3), and (iii) the NBD1 domains from several members of the ABC  DECEMBER 28, 2012 • VOLUME 287 • NUMBER 53 transporter family (ABCA4-ABCB1-ABCG2) (supplemental Fig.  S3). Together, these alignments demonstrated extensive conservation of the NBD1 domain only in ABCA4 members. The domain is not conserved to the same degree either in the ABCA subfamily or the general ABC family of transporters. The extensive sequence conservation in ABCA4 suggests that the NBD1 domain may play an important and specific role in protein function, beyond that of the generalized ribonucleotidase activity.

Retinal Binding by the First Nucleotide Binding Domain of ABCA4
We have demonstrated earlier that the lumen-facing ECD2 domain of ABCA4 binds all-trans-retinal preferentially over 11-cis-retinal; we further examined possible binding of alltrans-or 11-cis-retinal to other soluble domains of ABCA4. We did not find any significant retinal binding by either the ECD1 or NBD2 domains. On the other hand, the NBD1 domain strongly and preferentially bound the 11-cis-isomer ( Fig. 4 and Table 1). The currently accepted topological model of ABCA4, where the NBD domains project outwards from the ROS disc membrane toward the cytoplasm, creates a scenario in which the NBD1 domain is in close proximity with the retinal pigment epithelium-generated 11-cis-isomer, essential for the regeneration of rhodopsin localized in the ROS disc membrane. This finding is similar to that observed with the NBD1 domain of the ABCG5/ABCG8 sterol transporter, where the NBD1 domain was shown to play an important role in high specificity substrate binding during sterol transport from hepatocytes into bile (63). Thus, it is possible that the NBD1 domain of ABCA4 plays a similar task in mediating the directionality of retinal transport. Our current findings are consistent with recent studies by Boyer et al. (64), which suggest that ABCA4 may play a role in sequestration or transport of 11-cis-retinal or its complex. Using ABCA4(Ϫ/Ϫ) mice as a model system, these investigators demonstrated that free 11-cis-retinal is involved in the formation of lipofuscin and its component all-trans-retinylidenephosphatidylethanolamine, as opposed to all-trans-retinal, and postulate that 11-cis-retinal may be imported into the ROS discs by ABCA4 (64). Our report, demonstrating the specific interaction of the NBD1 domain of ABCA4 with 11-cis-retinal, now provides direct mechanistic evidence for such a role.
In order to examine whether nucleotide binding would influence the interaction of NBD1 with 11-cis-retinal, binding was measured in the presence and absence of AMP-PNP, a nonhydrolyzable analog of ATP, as well as ADP (Fig. 5). Unfortunately, the binding in the presence of ATP could not be measured because of the hydrolysis of ATP to ADP by NBD1. The affinity of NBD1 was highest in the absence of nucleotide, whereas saturation binding of retinal was not observed in the presence of the ATP analog, AMP-PNP (Fig. 5, A and B, and Table 1). The affinity of NBD1 was also attenuated in the presence of ADP. This alteration in affinity could be related to the nucleotide binding-regulated interactions of the NBDs, which are believed to occur as part of the transport mechanism (59,60,63,65,66). However, a different role for modulation of binding by ATP could not be ruled out.
Three missense mutations in NBD1 were examined in this study, R943Q, P940R, and G863A. These mutations have small effects on the kinetics of ATP hydrolysis (Table 1) (35). Therefore, we have explored the effects of these mutations, if any, on 11-cis-retinal interaction with the NBD1, as shown in Fig. 6. 11-cis-Retinal binding was attenuated in all three NBD1 mutants examined. The most significant change was observed with G863A, in which saturation binding was no longer observed. The mutation P940R seemed to affect 11-cis-retinal interaction the least and led to a ϳ50-fold decrease in binding affinity, whereas the R943Q mutation led to a 100-fold decrease in binding affinity. These results perhaps suggest the significant contributions of these amino acid residues to retinal binding and nucleotide hydrolysis. Attenuated binding of 11-cis-retinal to these mutants may indicate that this binding is important in the pathogenesis of the associated macular degenerative diseases.
In addition to its possible role in facilitating 11-cis-retinal import into the ROS discs, an additional consequence of NBD1 binding to this retinal isomer is the sequestration of 11-cisretinal or its adduct by ABCA4. One may argue that sequestration of 11-cis-retinal from the cytosolic milieu is important for channeling 11-cis-retinal toward the ROS disc membrane, either as a step in an import mechanism mediated by ABCA4 itself or as a process that simply facilitates the diffusion of 11-cis-retinal across the ROS membrane. In addition, sequestration mechanisms may have evolved to protect the cytosolic machinery from the reactive aldehyde, given that 11-cis-retinal is not processed by the rod outer segment retinol dehydrogenase.
The results described provide additional details about the mechanism of transport by ABCA4. Ligand interaction with the NBD1 domain may provoke a conformational change, which could in turn promote translocation of 11-cis-retinal or its adduct from the cytosolic to the lumenal side of the ROS disc or facilitate its diffusion across the membrane. The "alternating access" model of ABC protein ligand transport proposes that the transmembrane domains alternate between two conformational states, thereby creating two different binding sites for the substrate, the net result of which is translocation of the ligand without ever fully opening a channel from one side of the lipid bilayer to the other (67-69). The differential affinity of the lumenal (ECD2) and cytoplasmic (NBD1) domains of ABCA4 for all-trans-and 11-cis-retinal are consistent with the alternating access model but also raise the possibility that ABCA4 may function to regulate the direction of flux of transport ( Fig. 7 and supplemental Fig. S4) (39,40). Recent in vivo studies utilizing ABCA4(Ϫ/Ϫ) mouse models are consistent with ABCA4 translocation away from the cytoplasm (25). Given the large size of ABCA4 and its pivotal role in vision, a complex transport mechanism seems plausible.
In summary, the results presented here demonstrate that the NBD1 domain interacts preferentially and with high affinity with the 11-cis isomer of retinal. Nucleotide binding to NBD1 is not required for 11-cis-retinal binding. Stargardt disease-associated mutations lead to variable loss-of-function attenuation of NBD1 binding to 11-cis-retinal. Together, these results point to a role of ABCA4 in 11-cis-retinal translocation. Furthermore, the findings support the notion that the generation lipofuscin, which underlies the pathogenesis of Stargardt disease, may arise from defects in the import of 11-cis-retinal to the ROS discs.