Nuclear Trafficking of Photoreceptor Protein Crx: The Targeting Sequence and Pathologic Implications

Abstract

purpose. To identify the targeting sequence controlling the nuclear transport of the photoreceptor-specific transcription factor cone–rod homeobox (Crx) protein and to address the question of whether disease-causing Crx mutations disrupt the nuclear trafficking of the Crx protein.

methods. A series of cDNA fragments encoding Crx protein with deleted C termini were generated from mouse Crx cDNA by polymerase chain reaction (PCR). Point mutations were introduced into Crx coding sequence through PCR-based, site-directed mutagenesis. These mutated Crx fragments and the wild-type Crx were fused to cDNA encoding the jellyfish green fluorescent protein (GFP) and were transiently expressed in human embryonic kidney (HEK) 293T cells. Twelve to 48 hours after transfection, the living cells were counterstained with the red fluorescent nucleic acid dye SYTO 59 and examined with epifluorescence and confocal microscopy to determine the subcellular localization of Crx fusion proteins.

results. GFP expressed without a fusion partner was distributed evenly throughout the cells, whereas the wild-type Crx protein fused to GFP was localized only in the nucleus. GFP-tagged Crx proteins truncated at residues 107 or 165, demonstrated exclusive nuclear localization. In contrast, Crx fusion proteins truncated at residues 88, 79, 44, and 36, were located equally in both the cytoplasm and the nucleus. These results demonstrate that the nuclear localization signal (NLS) of Crx appears to reside in the amino acids between residue 88 and 107, which is surprising because the putative NLSs identified by prosite search are at residues 36 to 43 and 116 to 122. Further, a Crx fusion protein truncated at residue 99 was localized within the nucleus in the majority of the transfected cells, and two point mutations at residues 88 (K88T) and 98 (R98L) disrupted the nuclear localization, which indicates that the sequence between 88 and 98 in the C terminus of the Crx homeodomain contains a NLS that is essential for targeting Crx to the nucleus. However, the fusion protein truncated at residue 99 did not produce a complete nuclear localization in every transfected cell, suggesting that the Gln-rich domain at residues 99 to 106 is also required for the full accumulation of Crx protein in the nucleus. Two point mutations of Crx, R41W and E80A, that cause cone–rod dystrophy in humans and lie within the homeodomain but outside the NLS did not disrupt the nuclear localization of Crx protein, but a R90W mutation of Crx that causes human Leber congenital amaurosis (LCA) and resides within the NLS resulted in the fusion protein localized in both nuclei and cytoplasm in majority (51% to 69%) of the transfected cells.

conclusions. The wild-type Crx protein is localized within the nucleus. The putative NLSs of Crx at residues 36 to 43 and 116 to 122 are not essential. The minimal NLS necessary for the nuclear transport of Crx protein is located at residues 88 to 98 in the C terminus of the homeodomain. The R90W mutation of Crx found in LCA disrupts the nuclear transport of the mutant protein. The defective nuclear trafficking of Crx protein may be a part of the molecular mechanism of this early-onset retinal degeneration.

In eukaryotic cells, the nucleus is separated from the cytoplasm by the nuclear envelope. Most RNAs transcribed in the nucleus are exported to the cytoplasm for protein synthesis, whereas proteins required for nuclear functions are synthesized in the cytoplasm and imported into the nucleus. ^{1 2} This selective trafficking of macromolecules between the nucleus and the cytoplasm is critical for maintaining cellular function. The pathway of nuclear protein transport to the nucleus is mediated by nuclear localization signal (NLS) sequences that are characterized by one or more clusters of basic amino acids. ³ Understanding how nuclear proteins are targeted to and retained within the nucleus can provide insights into both the function and regulation of the proteins and the molecular pathogenesis of genetic diseases caused by mutant gene products. 

The cone–rod homeobox (Crx) protein is a highly conserved, photoreceptor-specific homeodomain transcription factor that plays a crucial role in photoreceptor differentiation and development. ^{4 5 6} Crx is capable of transactivating some photoreceptor genes both in transfected cultured cells ^{4 5 7 8} and in transgenic ⁸ and knockout ⁹ mice. Because nuclear localization is a key feature of transcription factors, the nuclear trafficking and localization of Crx protein would be crucial for Crx to function as a transcription factor. Although Crx mRNA has been localized to the photoreceptor cells of developing and adult mammalian retinas by in situ hybridization, ^{4 5 6} the subcellular localization of the Crx protein is unknown but presumed to be within the nucleus. Mutations in Crx gene have been identified to be the cause of several forms of human retinal degenerations. ^{6 10 11 12 13} Targeted disruption of the Crx gene in mice produces a retinopathy. ⁹ However, the molecular mechanism by which the Crx mutations lead to retinal degenerations remains largely unclear. Because the trafficking of Crx protein from the cytoplasm to the nuclei of the photoreceptor cells represents a critical step for activating its biological function, we hypothesized that Crx is a nuclear protein that has a NLS responsible for its nuclear transport and that mutations occurring in the NLS sequence could disrupt the nuclear transportation of Crx protein and thus cause retinal degenerations. 

Materials and Methods

Generation of Crx-GFP Fusion Protein Expression Constructs

Wild-type Crx cDNA was released from a mouse Crx clone (a generous gift from Constance Cepko; GenBank accession number U77615) by HindIII/BglII digestion, gel purified, and cloned at the HindIII/BglII sites to the 5′ coding region of a GFP expression vector (pTH40). 3′ truncated or mutated Crx sequences were amplified from the wild-type Crx with PCR using pfu DNA polymerase (Stratagene, La Jolla, CA). Crx mutants harboring mutations at codon 41 (R41W), 80 (E80A), 88 (K88T), 90 (R90W), and 98 (R98L) were generated using site-directed mutagenesis based on the two-step PCR. ¹⁴ The PCR primers were the following: Ps (sense strand), 5′-agacccaagctggctagcgt-3′; antisense strand primers for Crx deletion constructs: D164, 5′-actagatctgactccaaatggacacggtg-3′; D106, 5′-actagatctgctgttgctgtttctgct-3′; D98, 5′-actagatctgtcgctgctgtctgcatt-3′; D87, 5′-actagatctggaaccagacctggaccctg-3′; D78, 5′-actagatctgcagattgatcttaagagcaac-3′; D43, 5′-actagatctgccgctcccgccgctgcttc-3′; D35, 5′-actagatctgggcacttgagtatgggaca-3′. Primer pairs for Crx missense mutant constructs were the following: R41W, 5′-aagcagcggtgggagcgga-3′ and 5′-tccgctcccaccgctgctt-3′; E80A, 5′-aatctgcctgcgtccagggt -3′ and 5′-accctggacgcaggcagatt-3′; K88T, 5′-aggtctggttcacgaatcgtag-3′ and 5′- ctacgattcgtgaaccagacct-3′; R90W, 5′-ggttcaagaattggagggcgaaat-3′ and 5′-atttcgccctccaattcttgaacc-3′; R98L, 5′-cagacagcagctacagcagca-3′ and 5′-tgctgctgtagctgctgtctg-3′. PCR was performed on a RoboCycler (Stratagene) with the cycle profiles: 1 cycle of 5 minutes at 94°C, 50 seconds at 52°C, and 1 minute at 72°C; 30 cycles of 1 minute at 92°C, 50 seconds at 52°C, and 1 minute at 72°C, and 1 cycle of 10 minutes at 72°C. Amplified Crx fragments were digested with HindIII/BglII, gel purified, and cloned to the GFP expression vector. Plasmid DNA was prepared with QIAprep spin miniprep Kit (Qiagen, Valencia, CA) and purified using Wizard Maxiprep DNA purification kit (Promega, Madison, WI). Correct fusion of each construct was confirmed by DNA sequencing. 

Cell Culture and Transfection

Human HEK293 T cells ¹⁵ were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin. Cultured cells were transfected with TransFast transfection reagent (Promega) according to the manufacturer’s protocol. Briefly, HEK 293T cells grown in 60-mm culture dishes were replated onto 35-mm microwell culture dishes (MatTek, Ashland, MA) at a density of 2 × 10⁵ cells/dish the day before the transfection. For transfection, 5 μg of each DNA construct was mixed with 2 ml of culture medium, and then 15 μl TransFast reagent at a 1:1 charge ratio of lipid:DNA was added to each DNA/medium mixture and mixed by vortexing. After incubating at room temperature for 15 minutes, the medium in each culture dish was replaced with the DNA/medium/lipid mixture and incubated in the 37°C CO₂ incubator for 2 hours. Then, 1 ml complete DMEM was added. Each fusion construct was tested at least three times. 

Epifluorescence Microscopy

Twelve to 48 hours posttransfection, living cells were counterstained with SYTO 59 (Molecular Probes, Eugene, OR) at a final concentration of 5 μm for 30 minutes and examined with an epifluorescence microscope using a 40× lens. GFP green fluorescence was detected with standard fluorescein optics (FITC HQ filter set; Chroma, Brattelboro, VT), whereas the SYTO 59 red fluorescence was detected with a Texas red filter set (Chroma). The GFP-tagged proteins were quite bright and easily imaged, but a counterstain was necessary to clearly identify the boundaries of the nuclei and cells. A variety of SYTO dyes were tested (SYTO Red Fluorescent Stain Sampler Kit; Molecular Probes), and SYTO 59 was identified as a live-cell counterstain that was quite bright in Texas red optics and that did not produce a signal within FITC optics. A cooled CCD camera (Princeton Instruments, Trenton, NJ) was used to acquire images of the transfected cells. The GFP fluorescence signal of the transfected cells was defined as the signal that exceeded the background, or dark noise, seen in parts of the images that contained no cells. Images were acquired and analyzed with IPLab software (Scanalytics Inc., Fairfax, VA). 

Scanning Laser Confocal Microscopy

Similar procedures described previously ¹⁶ were used. Briefly, transfected living cells with or without the SYTO 59 counterstaining were viewed on a BioRad MRC 600 (Hercules, CA) scanning laser confocal microscope 12 to 48 hours posttransfection. For quantification of GFP fluorescence intensity, the 488-nm laser line was used to examine the transfected cells without the SYTO dye counterstaining, and the emission was collected with a 510-nm bandpass filter. For determining the subcellular localization of the Crx fusion proteins, the SYTO dye–counterstained cells were examined in the same way, except the use of dual excitations with both FITC and rhodamine filters. Images were collected with a 100 × 1.4-na or 60 × 1.4-na objective lens and 10% laser. All images were acquired using a Kalman average of three scans and analyzed with IPLab and IGOR Pro software (WaveMetrics, Lake Oswego, OR). Image manipulations involved converting the digital images of GFP and SYTO 59 into 24-bit color images in which the GFP was represented as green and the SYTO59 red. In addition, the files were merged to create overlays for comparison. None of the values in the image files were manipulated except for where scale bars and boxes were added. 

Results

Subcellular Localization of the Wild-Type Crx Fusion Protein in Cultured Cells

GFP has been widely used as a protein tag to follow the intracellular trafficking of fusion proteins, including transcription factors, in living cells and to map nuclear localization signals. ^{17 18 19} GFP expressed in mammalian cells drifts freely throughout the cytoplasm and nucleus. ²⁰ This appears due to the low molecular weight of GFP, which enables it to diffuse passively into the nucleus. Cells transfected with the GFP expression vector exhibited fluorescence in both the cytoplasm and the nucleus, whereas cells transfected with the wild-type Crx fused to GFP (construct Wild, Fig. 1 ) demonstrated GFP fluorescence restricted within the nuclei (Fig. 2-1) . 

Subcellular Localization of Mutant Crx Fusion Proteins: Identification of a Novel NLS

The nuclear trafficking of transcription factors usually requires the presence of a positively charged nuclear localization sequence. ²¹ A Prosite database search revealed that two basic regions lying in the residues 36 to 43 and residues 116 to 122 of mouse Crx protein could be the putative nuclear localization signals. Based on these locations, truncated Crx-GFP fusion constructs (Fig. 1) were created to map the NLS that targets the Crx protein to the nucleus. 

Construct D164 exhibited nuclear localization of the Crx fusion protein (Fig. 2-1) , indicating that the NLS of Crx protein lies within the first 164 amino acids. Construct D106 also produced an exclusive nuclear localization of the fusion protein (Fig. 2-1) . This construct was truncated such that the C-terminal 193 amino acids of Crx were missing, which included one of the putative nuclear localization signals (residues 116–122). The strong nuclear signal produced by this construct indicates that residues 116 to 122 do not appear to contain an essential NLS. Two other constructs, D35 and D43 (Fig. 1) , were then constructed to test whether the putative proximal NLS (residues 36–43) would serve as a NLS. As shown in Figure 2-1 , the D35 construct with the putative proximal NLS deleted resulted in the fusion protein localized throughout the transfected cells. Surprisingly, the same pattern of localization was also observed in cells transfected with the construct D43 (Fig. 2-1) , carrying the putative proximal NLS, which demonstrates that this NLS is not essential, and that the real NLS is located somewhere between the residue 43 and residue 106. 

To further define the region containing the NLS required for nuclear trafficking of the Crx protein, additional deleted Crx fusion constructs (D98, D87, and D78 in Fig. 1 ) were generated. As shown in Figure 2-1 , both D87 and D78 constructs extending to amino acids 87 and 78 resulted in the fusion proteins being localized evenly throughout the cells. In contrast, the majority of cells transfected with the construct D98, which carries the Crx N-terminal 98 residues, including the intact homeodomain, exhibited strong nuclear-predominant fluorescence. These data indicate that there is a NLS located between residues 88 and 98 in the C-terminal end of the Crx homeodomain. However, the D98 construct did not produce exclusive nuclear localization in every cell, and comparing this partial effect with the one produced by the D106 construct (Fig. 3) suggests that the glutamine-rich domain between residues 99 and 106 is also required for the complete nuclear transport of the Crx protein. To confirm that the residues 88 to 98 of the Crx homeodomain is the minimal functional NLS, two mutations were placed at residues 88 and 98 (K88T and R98L, Fig. 4 ) of Crx in the D164 construct. Cells transfected with these constructs exhibited a heterogeneous pattern of fusion protein localization, where cytoplasmic and nuclear localization was detected in 42% to 60% transfected cells and exclusive nuclear localization in the remaining cells (Fig. 2-2) . This indicates that the lysine at position 88 and the arginine at position 98 are important for complete nuclear transport of Crx protein and that the residues 88 to 98 in the homeodomain is the minimal functional NLS of the Crx protein. 

Effect of the Disease-Causing Crx Mutations in the Homeodomain on the Nuclear Trafficking of Crx Protein

It has been reported that mutations in several human genes disrupt the nuclear import of the mutant proteins they produced. ^{22 23 24} To determine whether mutations in Crx had analogous effects, we analyzed the effect of the Crx mutations (Fig. 4) that cause inherited retinal diseases on the nuclear trafficking of the Crx protein. The R41W and E80A mutations within the Crx homeodomain (residues 39–98) but outside the NLS sequence had no effect on the nuclear localization of the Crx fusion proteins (Fig. 2-2) . However, cells transfected with the R90W construct revealed a heterogeneous localization pattern. The majority of the transfected cells (51–69%) showed a nucleus-predominant whole-cell localization of the fusion protein (Figs. 2-2 and 3C) , whereas a small portion of the cells exhibited only nuclear localization. As shown in Figure 3C , the fluorescence intensity of the cytoplasm is about half that of the nucleus. Interestingly, such a heterogeneous localization pattern was also observed in cells transfected or transformed with the mutant NLSs of SV40 large T antigen ²⁵ and the nucleoplasmin. ²⁶ The heterogeneous localization may reflect a deficiency in the rate of nuclear accumulation, ²⁵ which may allow the visualization of intermediate stages in the nuclear accumulation of Crx protein. 

Discussion

Crx is a homeodomain protein expressed specifically in developing and mature photoreceptor cells and required for photoreceptor differentiation and maintenance. ^{4 6} The mouse Crx protein is 96% identical with that of bovine and human, with 100% conservation of their homeodomains. ⁵ Like other transcription factors, Crx protein is synthesized in the photoreceptor cytoplasm and has to be transported to the photoreceptor nucleus to carry out its function. Nuclear localization signals that target nuclear proteins to the nucleus not only have an important function mediating the nuclear transport of the proteins, but also play a role in the pathogenesis of some genetic diseases. ^{22 23 27}  

Previous studies revealed that Crx can bind to and activate several photoreceptor genes, supporting the idea that Crx functions as a transcription factor. ^{4 5 7 8} Although Crx mRNA has been localized to the photoreceptor cells, ^{4 5 6} the subcellular localization of the Crx protein is unknown but is presumed to be in the nucleus. The data presented here demonstrated that the Crx protein is indeed a nuclear protein, which provides a direct evidence supporting the proposal of Crx’s function as a transcription factor. 

A Novel NLS in the Crx Protein

The NLS is responsible for targeting proteins to the nucleus. ² Although sequence analysis indicates residues 36 to 43 and 116 to 122 could be the NLSs of Crx protein, our data show that they are not functionally essential and that the nuclear trafficking of Crx protein is mediated by a minimal NLS located at residues 88 to 98 in the C terminus of the Crx homeodomain. This novel NLS bears no homology to either the monopartite NLS of the SV40 large T antigen ²⁵ or the bipartite NLS of nucleoplasmin. ²⁶ However, this is not surprising because many NLS sequences are more complex ²⁸ and do not fit these two classical NLS models. ^{26 29} Contrary to conventional views, even neutral and acidic amino acids can play crucial roles in NLSs. ³⁰  

Could the function of Crx be regulated at the level of posttranslational nuclear trafficking? Crx is critical for photoreceptor differentiation and morphogenesis. ⁴ Cell fate determination is achieved through interactions between both extrinsic and intrinsic factors. One mechanism by which extracellular signals might control pattern is by directing the graded nuclear localization of homeodomain proteins such as extradenticle. ³¹ It is not clear by what extracellular signals and through which pathways of signal transduction, the Crx gene expression is activated during photoreceptor development, nor whether and how the regulatory function of Crx protein on other photoreceptor genes is itself regulated. Crx has a high percentage of serine and threonine residues in the C-terminal half, ⁶ which are potential phosphorylation sites. Perhaps the Crx function could be modulated at the posttranslational level through a regulated nuclear transport by phosphorylation or dephosphorylation. 

What is the functional significance of the conserved domains in the Crx protein? The homeodomain is functionally the most characterized region of the Crx protein, whereas the functional significance of other Crx domains have not been characterized. ⁴ Crx and other Otx/otd family members share the conserved polyglutamine domain of residues 99 to 105. ⁵ Our results indicated that this domain appears to be required for complete nuclear transport of the Crx protein. The lysine residue at position nine of the third recognition helix of Crx is highly conserved among the OTX/Otd homeodomain proteins family. ^{4 5} This lysine residue seems to be important for the nuclear transport of these transcription factors. Our results reveal that the Crx homeodomain also plays a critical role in the nuclear trafficking of Crx protein, in addition to its function in DNA recognition and binding. 

Disruption of Nuclear Transport of Crx Protein and the Molecular Pathogenesis of Leber Congenital Amaurosis

Impaired nuclear transport of mutant proteins has been identified as a contributing factor in the molecular pathology of inherited human diseases such as Bloom syndrome ²² and Werner’s syndrome. ²³ Crx mutations are associated with several forms of retinal degenerations with a wide range of phenotypes. ^{6 10 11 12 13} The molecular pathogenesis of how mutant Crx lead to retinal degeneration has yet to be elucidated. It is also poorly understood why similar mutations of the Crx gene cause different clinical phenotypes. ⁶ The R41W and E80A mutations causing cone–rod dystrophy ^{6 10 12} reside in the homeodomain but outside the NLS of Crx, and our data showed that they did not affect the nuclear localization of the Crx protein. The Crx R90W mutation was recently found to cause Leber congenital amaurosis. ¹³ The mutant protein decreased but did not abolish its binding to rhodopsin promoter in vitro. In transient transfection, this mutant alone was unable to transactivate the rhodopsin promoter, although it did not appear to abolish its transactivating activity in the presence of cotransfected transcription factor NRL. ¹³ We found that this mutation impairs, but does not prevent, the nuclear transport of the Crx protein. It seems possible that the defective nuclear trafficking of the mutant protein results in less amount of the Crx protein available in the cell nucleus for transactivating its targeting genes. This may, at least in part, explain why the mutant protein totally loses transactivating function without cotransfection of NRL, but still partially retains its DNA binding activity. Thus, the defective nuclear transport of Crx protein may contribute to the molecular mechanism of Leber congenital amaurosis. 

NLSs are usually defined by systematic deletion analysis. ² Nevertheless, there can be multiple, redundant NLSs. Although deletions of the putative NLSs did not produce an effect on the nuclear localization of Crx fusion proteins in our study, we cannot formally eliminate the possibility that the putative NLSs are redundant or have an addictive effect on the NLS identified in this study. In addition, nuclear targeting sequences can be of different strengths. ²⁸ It will be of interest to test whether this minimal NLS of Crx protein is sufficient to target a nonnuclear protein to the nucleus. Recently, renal epithelial cells have been used successfully to identify a novel apical sorting signal of rod photoreceptor–specific rhodopsin. ³⁶ In the present study, we used human embryonic kidney cells to map the NLS of the Crx protein. Although these cells do not normally express Crx, it seems reasonable to expect that they can be used to map NLSs. This speculation is based on the idea that the basic mechanisms of nuclear transport have been highly conserved during evolution. ² Indeed, the nuclear localization signals as well as the nuclear transport machinery are generally conserved between cell types, as exemplified in nucleoplasmin. ³⁷ Nevertheless, variations of the efficiency of a NLS can be present among different cell lines, ³⁸ so it is possible that photoreceptor cells may use additional cell-specific molecular information to regulate the nuclear transport of the native Crx protein. Further studies in photoreceptor cells will be important to test the findings of the present study and to establish the intracellular localization of the native Crx protein in vivo. 

 Supported by Grant EY08362 from the National Institutes of Health, Bethesda, Maryland, and the Dolly Green Scholars award from the Research to Prevent Blindness, New York, New York.

 Submitted for publication February 3, 2000; revised April 25, 2000; accepted May 12, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Schematic diagrams showing Crx-GFP fusion constructs made by fusing wild-type (W) or deleted mouse Crx C terminus to the N terminus of GFP (pTH40) at the BglII site (The sequence of each construct is available at http://info.med.yale.edu/ophtha/thom/seq.html/.) The constructs for deleted Crx were named with a “D,” signifying deletion and a number signifying the C-terminal amino acid of Crx fused to GFP. The locations of the homeodomain, putative, and identified NLSs and other important regions of Crx are indicated. The numbers above each diagram signify the positions of the amino acid residues of Crx protein.

Figure 2.

 

Subcellular localization of GFP-tagged wild-type and truncated Crx proteins in cultured HEK293T cells with SYTO 59 counterstaining. The green (GFP signal), red (SYTO), and merged images (OVERLAY) from representative transfected cells are shown for each construct. The Crx fusion protein is localized exclusively within the nucleus in images from constructs Wild, D164, and D106, predominantly in the nucleus in D98, and homogeneously throughout the whole cells in D78, D43, and D35. The images in D98, D87, and D78 were collected with a confocal microscope, and the remaining from an epifluorescence microscope and CCD camera. Scale bars, 10 μm.

Figure 3.

 

Subcellular localization of mutant Crx fusion proteins in cultured HEK293T cells with SYTO 59 counterstaining. Representative confocal microscope images showing the GFP signal in green, the SYTO 59 signal in red, and the merged files in the right hand column (OVERLAY) were collected from cells transfected with the mutant fusion constructs (shown in the left column) that contain point missense mutations in the Crx coding sequence. The Crx fusion protein is localized exclusively within the nucleus in R41W and E80A and in both the cytoplasm and nucleus in K88T, R90W, and R98L. Scale bars, 10 μm.

Figure 4.

 

Representative confocal microscope images of fluorescent cells transfected with constructs D106 (A), D43 (B), and R90W (C), respectively, without SYTO 59 counterstaining and plots of the average pixel intensity at each point along the horizontal boxes outlined in white. Cells transfected with construct D106 has fluorescence restricted within the nuclei (A), and the fluorescence distribution is approximately homogeneous, except that the level of fluorescence in the nucleoli is only half of that in other areas of the nucleus. Cells transfected with construct D43 (B) exhibit a homogeneous distribution of fluorescence in both the nucleus and the cytoplasm, whereas the construct R90W-transfected cell (C) reveals a nuclear-predominant distribution of the GFP fluorescence in the whole cell. The fluorescence intensity of the cytoplasm is about half that of the nucleus. Scale bars, 10 μm.

Figure 5.

 

Mutant Crx fusion constructs and their partial amino acid sequences showing the locations of the missense mutations, which are either in or outside the NLS, but within the homeodomain, of the Crx protein. The numbers signify the positions of the amino acid residues. The subcellular localization patterns of these fusion proteins in the transfected cells are also indicated here.

The authors thank Constance L. Cepko and Takahisa Furukawa for kindly providing the cDNA encoding mouse Crx. We are also grateful to the IOVS reviewers for their careful, critical reading this manuscript and their valuable suggestions.