The Zinc Finger Transcription Factor, MOK2, Negatively Modulates Expression of the Interphotoreceptor Retinoid-binding Protein Gene, IRBP *

The human and murine MOK2 orthologue genes encode Krüppel/TFIIIA-related zinc finger proteins, which are factors able to recognize both DNA and RNA through their zinc finger motifs. MOK2 proteins have been shown to bind to the same 18-base pair (bp)-specific sequence in duplex DNA. This MOK2-binding site was found within introns 7 and 2 of human PAX3 and interphotoreceptor retinoid-binding protein (IRBP) genes, respectively. As these two genes are expressed in the brain asMOK2, we have suggested that PAX3 andIRBP genes are two potentially important target genes for the MOK2 protein. In this study, we focused our attention onIRBP as a potential MOK2 target gene. Sequence comparison and binding studies of the 18-bp MOK2-binding sites present in intron 2 of human, bovine, and mouse IRBP genes show that the 3′-half sequence is the essential core element for MOK2 binding. Very interestingly, 8-bp of this core sequence are found in a reverse orientation, in the IRBP promoter. We demonstrate that MOK2 can bind to the 8-bp sequence present in the IRBP promoter and repress its transcription when transiently overexpressed in retinoblastoma Weri-RB1 cells. In the IRBP promoter, it appears that the TAAAGGCT MOK2-binding site overlaps with the photoreceptor-specific CRX-binding element. We suggest that MOK2 represses transcription by competing with the cone-rod homeobox protein (CRX) for DNA binding, thereby decreasing transcriptional activation by CRX. Furthermore, we show that Mok2 expression in the developing mouse and in the adult retina seems to be concordant with IRBP expression.

The human and murine MOK2 orthologue genes, which are preferentially expressed in brain and testis tissues, encode two different Krü ppel/TFIIIA-related zinc finger proteins. The human and murine genes have been localized to band q13.2-q13.3 of chromosome 19 and chromosome 6, respectively (1,2). The human hsMOK2 protein shows substantial differences with the murine MOK2 protein. The mouse MOK2 protein contains seven tandem zinc finger motifs with only five additional amino acids at its COOH-terminal end (3). The seven fingers motifs are highly similar to one another but are distinct from those of other zinc finger proteins. The structural feature of murine MOK2 protein is also found at the end of human hsMOK2 protein. Furthermore, the human protein contains three additional zinc finger motifs in tandem with the others and a nonfinger acidic domain of 173 amino acids at the NH 2terminal end (2). We have previously shown that human MOK2 RNA maturation results in three mRNAs with different 5Јuntranslated exons. One of these three mRNAs encodes a smaller MOK2 protein (hsMOK2⌬) containing 10 zinc finger motifs and a small NH 2 -acidic domain made up of 76 amino acids. We have shown that the human and murine MOK2 proteins are able to recognize both DNA and RNA through their zinc finger motifs (4). Electron microscopy and specific RNA homopolymer binding activity showed clearly that the murine and human MOK2 proteins are RNA-binding proteins that associate mainly with nuclear RNP components, including nucleoli and extranucleolar structures. Murine and human MOK2 proteins have been shown to bind the same 18-base pair (bp) 1 -specific sequence in duplex DNA (4). This 18-bp-specific sequence has been identified by two approaches, randomized oligonucleotide and whole genome PCR techniques. The 18-bp MOK2-binding site occurs in an intron of two different human genes. It is interesting to note that these two potential target genes for MOK2 protein function in the brain, where the MOK2 gene is preferentially expressed. The first one is the human PAX3 gene, a transcription factor expressed during brain development (5,6). Interestingly, in this gene, the MOK2-binding site occurs in the last intron (7), in which disruption is associated with the translocation in human alveolar rhabdomyosarcomas (8). The second potential target gene encodes the human interphotoreceptor retinoid-binding protein (IRBP), which is expressed exclusively in retinal photoreceptor cells and in a subgroup of pinealocytes. IRBP is thought to be involved in the visual cycle of vertebrate retina (9 -11). In developing mice, IRBP is first expressed at the birth of the photoreceptors, suggesting that it may also be involved in photoreceptor differentiation (12). The importance of IRBP in normal photoreceptor development has been demonstrated by the recent generation of mice with a targeted disruption of the IRBP gene (13). In the absence of the Irbp gene, there is a slowly progressive degeneration of retinal photoreceptors. The MOK2-binding site is located in intron 2 of human and bovine IRBP genes. It has been suggested that the highly conserved introns 2 and 3 of the IRBP gene might contain important regulatory elements for IRBP gene expression (14).
Here we have focused our attention on IRBP as a potential MOK2 target gene. Sequence comparison and binding studies of the 18-bp MOK2-binding sites present in intron 2 of human, bovine, and mouse IRBP genes show that the 3Ј-half sequence is the essential core element for MOK2 binding. Very interestingly, 8 bp of this core sequence are found in a reverse orientation, in the IRBP promoter. The results presented here demonstrate that MOK2 can bind to the 8-bp sequence present in the IRBP promoter and repress its transcription when transiently overexpressed in retinoblastoma Weri-RB1 cells. Furthermore, we show that Mok2 expression in the developing mouse organism and adult retina seems to be concordant with IRBP expression.

MATERIALS AND METHODS
Plasmid Constructions-All plasmids generated for this study were confirmed by DNA sequencing. The recombinant GST-hsMOK2 and CBD-hsMOK2 fusion proteins were obtained by inserting the bluntended XbaI-EcorI fragment (1850 bp) into EcoRI/Klenow fragmenttreated pGEX-3X (Amersham Pharmacia Biotech) and BamHI/Klenow fragment-treated pET35 (Novagen), respectively. The pET35 vector contains the cellulose-binding domain tag (CBD). The XbaI-EcorI fragment, which contains the entire coding sequence of hsMOK2, was obtained from pBhsMOK2. The pBhsMOK2 plasmid and the eukaryotic expression CMV-hsMOK2 vector were obtained by inserting the bluntended AcsI fragment from cDNA1 (2) into the SmaI site of pBluescript KS ϩ (Stratagene) and NotI/Klenow fragment-treated pCMV, respectively. The plasmids encoding recombinant maltose-binding (MBP)-MOK2 fusion proteins and the eukaryotic expression vector CMV-MOK2 were described previously (4). In this article, the eukaryotic expression vector referred to as pCMV-hsMOK2, which encodes an isoform containing a smaller NH 2 -acidic domain, was renamed CMV-hsMOK2⌬ (4).
The plasmid phsIRBPCAT contains a 332-bp fragment from the human IRBP gene (from Ϫ291 to ϩ41 relative to the transcription start site) inserted upstream of the CAT reporter gene (10). The fragment of the human IRBP gene promoter was obtained from genomic DNA by PCR using a 5Ј primer containing an added StuI site and a 3Ј primer containing an added Tth111I site. The product was digested by StuI-Tth111I and cloned into the corresponding sites of the PSV2CAT vector. In the phsIRBPCAT vector, the SV40 promoter was replaced by the human IRBP promoter.
A 276-bp fragment from intron 2 of the bovine IRBP gene (nt 7130 -7406 (15)) was obtained from Bos taurus genomic DNA by PCR and cloned into EcoRV pZero-2 (Invitrogen). Murine intron 2 of the Irbp gene was obtained from C3H mouse genomic DNA by PCR amplification with a human 5Ј primer that localizes to exon 2 (nt 5045-5065) and a 3Ј reverse human primer that localizes to exon 3 (nt 7017-7037 (14)). The fragment (ϳ2000 bp, EMBL/AJ294749) was cloned into the SmaI site of the pBluescript KS ϩ vector and sequenced.
Cells, Transient Transfections, and CAT Assays-For transient transfection assays, HeLa cells were plated at 10 6 cells on a 100-mm Petri dish for 24 h prior to the addition of the recombinant plasmids by the calcium phosphate method as described previously (16). For nuclear extracts, the HeLa cells were transfected with 15 g of MOK2 expression vectors. For CAT assays, Weri-RB1 retinoblastoma cells were transfected in 6-well plates with GenePORTER transfection reagent (Gene Therapy System) according to the manufacturer's recommendations. About 10 6 cells were cotransfected with 2 g of IRBP promoter-CAT reporter vector (phsIRBPCAT) and 2 g of MOK2 expression vector or parental pCMV plasmid. The cells were harvested 36 h later for the reporter gene assay using the freeze/thaw method (Promega). The activity of the resulting extracts was determined using the CAT assay protocol (Promega). Protein concentrations were determined by the Coomassie protein assay (Pierce). For data interpretation, CAT activity was normalized to the protein concentration of the extracts. All transfection experiments were repeated at least five times with different CsCl-DNA preparations.
DNA Probes-The plasmids containing human, bovine, or mouse intron 2 of IRBP genes (1.5 g) were cut once with the appropriate restriction enzymes, treated with intestine phosphatase alkaline, and end-labeled with T4 polynucleotide kinase. After a second digestion with appropriate restriction enzymes, each end-labeled fragment was purified on a 6% polyacrylamide gel. The double strand oligonucleotides were labeled with T4 polynucleotide kinase in the presence of [␥-32 P]ATP and purified on a 15% polyacrylamide gel. For DNA probes containing dITP, dUTP, or deaza-dATP, a 189 bp-fragment of intron 2 of human IRBP (nt 6660 -6849) was amplified by PCR using two primers, one of which was 5Ј end-labeled by treatment with T4 polynucleotide kinase in the presence of [␥-32 P]ATP. The PCR products were purified on a 6% polyacrylamide gel. The 92-bp fragment of the human IRBP promoter (Ϫ88 to ϩ4) was labeled by PCR using [␣-32 P]CTP and purified on a 6% polyacrylamide gel.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared from five 100-mm Petri dishes of transfected HeLa cells by a modified method of Dignam et al. (17). The crude nuclei were resuspended in 250 l of extraction buffer (20 mM HEPES, pH 7.9, 150 mM KCl, 0.05 mM ZnCl 2 , 0.1% Nonidet P-40, 0.5 mM dithiothreitol, 20% glycerol) with a protease inhibitor mixture without EDTA (Roche Molecular Biochemicals) and disrupted by sonication. After centrifugation at 15,000 ϫ g for 10 min, the nuclear extract was stored at Ϫ70°C. For bacteria expressing fusion proteins, the crude protein extracts were prepared as described previously (4). The standard DNA binding reaction (20 l) contained 20.000 cpm of 32 P-labeled probe, 2.5 g of crude MBP fusion protein or 10 g of nuclear extract, and 2 g of poly(dI-dC) in binding buffer (20 mM HEPES, pH 7.9, 100 mM KCl, 0.05 mM ZnCl 2 , 0.1% Nonidet P-40, 0.5 mM dithiothreitol, 20% glycerol). Complexes were analyzed by electrophoresis on a nondenaturing premigrated 4% polyacrylamide gel (acrylamide/bis ratio of 19:1) or 1% agarose gel in 0.5ϫ TB buffer (45 mM Tris borate, pH 8.3) at 4°C at 200 or 120 V, respectively. EDTA was omitted in all binding and electrophoresis buffers to avoid denaturing MOK2. All probes were gel-purified.
Antibody Preparation-The purified GST-hsMOK2 fusion protein was injected into a New Zealand White rabbit. Affinity-purified hsMOK2 antibodies were obtained by elution of immunoglobulins bound to the human CBD-hsMOK2 protein as described previously (4).
Northern Blot and S1 Nuclease Analysis-Total cellular RNAs were extracted from normal tissues of 1-month-old mice by the guanidinium thiocyanate procedure (18). Polyadenylated RNAs were prepared using oligo(dT) cellulose (Type III, Collaborative Research) columns (19). Five g of poly(A) ϩ RNAs from each tissue was used for Northern blot analysis, using nucleotides 1848 -2276 of the mouse Mok2 cDNA as probe (3). For S1 nuclease analysis, 10 g of poly(A) ϩ RNAs from mouse embryos were hybridized with a single strand of the mouse genomic fragment (nt Ϫ563 to ϩ51 (16)). 5Ј end-labeled with 32 P at nt ϩ51. 5Ј end-labeling and strand separation were carried out by standard techniques. The S1 nuclease-resistant products were resolved by electrophoresis on 10% polyacrylamide denaturing gels and detected by autoradiography.
In Situ Reverse Transcriptase-PCR Experimental Procedure-In situ reverse transcriptase-PCR was performed as described by Thaker (20). The eyes of 1-month-old mice were enucleated, immediately frozen in OCT (TissueTek, Sakura, Netherlands) and sectioned (10 m). Cryosections were collected on silane-coated glass slides. The sections were fixed for 1 h with a 10% Formalin/PBS for 1 h, washed three times in a 0.1% Triton X-100 in PBS, permeabilized for 10 min at Ϫ20°C in an ethanol/acetic acid solution, washed three times in PBS, and dehydrated in graded alcohols. mRNAs were reverse-transcribed for 1 h at 42°C with 200 units of Moloney murine leukemia virus (M-MLV) reverse transcriptase RNase H minus, using random hexamers (Promega). Sections were washed twice in PBS and dehydrated in graded alcohols. PCR reactions were carried out with the 5Ј primer (5Ј-TCTA-ACTGTCTCCACTTCCCA-3Ј) and the 3Ј primer (5Ј-AAGGCACATA-ATTTCAGAGGA-3Ј) located in the 3Ј-untranslated region of Mok2 mRNA at nucleotides 1848 -1869 and 2276 -2297 in the presence of Biotin16-dUTP (Roche Molecular Biochemicals) with a 1:19 ratio of dTTP. The amplification program was as follows: 94°C for 1 min, 54°C for 1 min, and 72°C for 1 min repeated for 40 cycles. The slides were subsequently washed three times in PBS and incubated for 1 h in 1/100 extravidin alkaline phosphatase conjugate (Sigma) in PBS solution. The signal was revealed using a nucleic acid detection kit (Roche Molecular Biochemicals) containing levimasol and a subsequent phosphatase alkaline-catalyzed color reaction with Xphosphate and nitro blue tetrazolium salt, which produces a precipitate. Control sections were done in the same way except that the reverse transcriptase reaction was omitted.

RESULTS
Conservation of the MOK2-binding Site in Intron 2 of the Murine IRBP Gene-The high homology found in introns 2 and 3 of the human and bovine IRBP genes suggested that these sequences might contain important regulatory elements for IRBP gene expression (14). One of these elements might be the MOK2-binding site, which is highly conserved between the intron 2 regions of human and bovine IRBP genes. In bovine intron 2, we observed a deletion of one nucleotide at position 7 in the 18-bp MOK2-binding site (Fig. 1A). To determine whether this sequence was also conserved in murine intron 2, a fragment (ϳ2000 bp) obtained from mouse genomic DNA by PCR using human-specific primers, which localized to exon 2 and 3, was sequenced. Intron 2 of the murine Irbp gene shows a similar high level of homology with intron 2 of human and bovine IRBP genes (62.3 and 58.2%, respectively). Sequence comparison of the 18-bp MOK2-binding sites of the human, bovine, and murine genes showed that the MOK2-binding site of mouse intron 2 was the most divergent. Four nucleotides present in the first 9 bp did not agree with the previously determined consensus sequence for MOK2 binding (4) or with the human or bovine MOK2-binding sites in intron 2 (Fig. 1A). Nevertheless, EMSAs showed that the truncated human MBP-hsMOK⌬ and mouse MBP-MOK2 fusion proteins were still able to interact with the murine MOK2-binding site (Fig. 1B). These results suggested that the nucleotides at positions 3, 5, 7, 8, and 9 within the 18-bp MOK2-binding site were not crucial for MOK2 protein binding.
We performed EMSAs using a series of probes containing a modified nucleotide to determine which bases were crucial for MOK2 binding. The probes were obtained by PCR using dITP, deaza-dATP, or dUTP instead of dGTP, dATP, or dTTP, respec-tively. Fig. 2 shows that the human and murine MBP-MOK2 fusion proteins were only unable to bind to the probe containing deaza-adenines instead of adenines. This result suggested that the adenines located in the 18-bp DNA-binding site were essential for the binding of human and murine MOK2 proteins.
MOK2 Binds to the TAAAGGCT Sequence of the Human IRBP Promoter-Transgenic and other studies have identified regulatory regions in the promoter important for the expression of the IRBP protein (10,(21)(22)(23)(24)(25). The IRBP upstream region has been found to have a 156-bp sequence that is well conserved between human, bovine, and mouse (26). The short promoter fragment from Ϫ123 to ϩ 18 (relative to the transcription start site) was found to confer photoreceptor-specific expression in transgenic mice (22). In the 123-bp sequence, we found 8 bp corresponding to the 3Ј end of MOK2-binding consensus sequence. The TAAAGGCT sequence is present in a reverse orientation compared with that of intron 2 (Fig. 3A). To determine whether this sequence binds MOK2, we performed an EMSA. The nuclear extract from control HeLa cells (transfected with the pCMV vector alone) and cells overexpressing human or murine MOK2 were used for comparison. Nuclear extracts were mixed with a labeled 92-bp fragment from the human IRBP promoter (Ϫ88 to ϩ4) and then electrophoresed. The weak bands with different mobilities obtained with control HeLa cell extract correspond to endogenous binding proteins that were not affected by anti-hsMOK2 antibody (Fig. 3B, left  panel, lanes 1 and 2). A retarded complex was detected with nuclear extract from hsMOK2-overexpressing HeLa cells (lane 3). The anti-hsMOK2 antibody clearly supershifted the DNAprotein complex, showing that this complex contains human hsMOK2 protein (lanes 4 and 5). In lane 6, a faster migrating complex produced by murine MOK2 protein can be seen. The difference in the mobilities of murine and human MOK2 proteins-DNA complexes can be accounted for by the difference in the molecular mass of these two proteins (22.8 and 51.5 kDa, respectively). To test whether MOK2 interacts directly with the 8-bp partial MOK2-binding site, we used two different mutated oligonucleotides as probes. These two oligonucleotides contained 6-and 5-bp modifications of the potential MOK2-binding site and the previously described AP-4-binding element, respectively (21). The DNA-protein complex was abolished when we used the mutant MOK2 probe but not the mutant AP-4 probe (Fig. 3B, right panel). These data illustrate that MOK2 proteins are able to bind to the 8-bp partial MOK2-binding site present in the IRBP promoter.
MOK2 Represses the Human IRBP Promoter Activity-Having shown that human and murine MOK2 proteins bind the 8-bp partial MOK2-binding site in the IRBP promoter, we analyzed whether MOK2 proteins were able to repress or transactivate the IRBP promoter. The retinoblastoma cell line Weri-RB1, which expresses IRBP as well as other photoreceptor genes (27), was used for transient transfections to assay the effect of MOK2 on the IRBP promoter. Cells were transiently cotransfected with the reporter construct phsIRBPCAT, which contains the Ϫ291 to ϩ41 region of the IRBP promoter, and the human or murine MOK2 expression vectors or the corresponding empty plasmid (pCMV). In the presence of MOK2 protein, a significant reduction in transcription activity was observed (Fig. 4). The most important reduction, about 70%, was found with the human hsMOK2 protein. Both the hsMOK2⌬ isoform, which is truncated for its NH 2 -acidic domain, and the murine MOK2 protein repressed CAT activity by about 50%. No effect was seen on the SV40 early viral promoter (data not shown). This result directly demonstrates the ability of MOK2 to act as a transcriptional repressor of the IRBP promoter.
Mok2 Is Expressed in Mouse Embryos and in Photoreceptor Cells-It has been shown that IRBP protein and mRNA appear at early developmental stages (12,28,29). Therefore, we investigated Mok2 expression during mouse embryogenesis. In previous work, we found that MOK2 mRNAs are present at low levels in cells and tissues proficient for the expression of the murine or human MOK2 gene (2,3). Expression of the Mok2 gene during mouse embryogenesis was consequently determined by S1 nuclease, a more sensitive method than Northern blot analysis, as described under "Experimental Procedures." As shown in Fig. 5A, S1 nuclease products were detected at all developmental stages tested. The presence of Mok2 mRNA at embryonic day 9.5 shows that the Mok2 gene is expressed early during mouse embryogenesis.
We have previously shown that murine and human MOK2 genes are preferentially expressed in the brain (2,3). First, we investigated MOK2 expression in the eyes. Northern blot analyses showed that the murine Mok2 gene is faintly expressed in the eye of 1-month-old mice (Fig. 5B). To know whether Mok2 is expressed in the retina, we used the in situ polymerase chain reaction (PCR) technique, which provides a very powerful tool for the study of the in situ expression of rare genes. The results showed that Mok2 expression in the retina of 1-month-old mice was restricted to the outer nuclear layer, which corresponds to the photoreceptors cells of the retina (Fig. 6). However, we observed that not all of the cells were stained within the outer nuclear layer (ONL, green or white arrow in Fig. 6). The outer nuclear layer of the mouse retina is composed of two kinds of photoreceptors, rods and cones, but contains a majority of rods. One possible explanation could be that only one kind of photoreceptor was stained within the outer nuclear layer. DISCUSSION In our previous studies, we showed that human and murine MOK2 bind to an 18-bp sequence (A/G)CCTT(A/G)TCAG(A/ G)GCCTTTA in duplex DNA (4). This MOK2-binding site was found within introns 7 and 2 of human PAX3 and IRBP genes, respectively. As these two genes are expressed in the brain as MOK2, we suggested that PAX3 and IRBP genes are two potentially important target genes for the MOK2 protein. In this study, we focused our attention on IRBP as a potential MOK2 target gene. First, we compared the sequences of the MOK2binding site found in intron 2 of human, bovine, and murine IRBP genes. The sequence comparison showed that the 3Ј half-site of MOK2 is strictly conserved between human, bovine, and mouse. The most divergence was observed on the 5Ј-side of the MOK2-binding site of murine intron 2 compared with those of the corresponding regions of the bovine and human MOK2binding sites. EMSAs showed that MOK2 proteins could interact with the human, bovine, and murine MOK2-binding site, suggesting that nucleotides at positions 3, 5, 7, 8, and 9 were  1 and 2) and overexpressing hsMOK2 (lanes [3][4][5] or murine MOK2 (lane 6) HeLa cells. The indicated amount (in microliters) of the affinity-purified anti-hsMOK2 antibody was added to EMSA reactions containing the wild-type probe (WT) and 2.5 g of nuclear extracts. SS indicates the super-shifted bands. Right panel, EMSA with wild-type, mutant MOK2 (MOK2mut), or mutant AP-4 (AP4mut) probes and 2.5 g of nuclear extracts from normal or overexpressing hsMOK2 cells. Complexes were analyzed by electrophoresis on a nondenaturing 4% premigrated polyacrylamide gel. not crucial for MOK2 protein binding to DNA. Furthermore, missing base contact probing suggested that MOK2 interacts with the adenines of the MOK2-binding site. The conserved 3Ј half-site of the MOK2-binding site contains a triplet of adenines. These results suggest that the essential core element of the MOK2-binding site consists of the 9-bp sequence G(A/ G)GCCTTTA. Very interestingly, 8-bp of this core sequence are found in a reverse orientation in the IRBP promoter.
In addition, we demonstrated that MOK2 binds to the TA-AAGGCT sequence and represses transcription from reporter constructs carrying the Ϫ291 to ϩ41 regulatory sequence from the human IRBP promoter when transiently overexpressed in Weri-RB1 retinoblastoma cells. This repressive effect of MOK2 appeared to depend on the presence of a TAAAGGCT sequence; no effect was seen on viral promoters such as the SV40 early region promoter (data not shown). The IRBP upstream region has been found to have a 156-bp sequence that is well conserved between human, bovine, and mouse (26). The short promoter fragment from Ϫ123 to ϩ18 relative to the transcription start site was found to confer photoreceptor-specific expression in transgenic mice (22). Previous studies have identified several functional cis-acting elements in the 123-bp proximal promoter of the IRBP gene (21,23). These cis-acting elements included two photoreceptor-specific consensus sequence, Ret1/PCEI (AATTAG) and its reverse repeat, GAT-TAA. Ret1/PCEI has been identified as a functionally active cis element in several photoreceptor-specific genes (30,31) including IRBP (22,32,33). Ret1/PCEI is potentially one of the target sites of Rx, Rax, and Erx proteins of the paired-type homeobox family (34 -36). The reverse GATTAA sequence (CRXE) is recognized by a photoreceptor-specific pair-related homeobox protein, cone-rod homeobox protein (CRX), which acts as a transcriptional activator (37,38). It has been shown that CRX bound to and transactivated weakly from the Ret1/PCEI sequence. These observations have suggested that CRX is not likely to be the factor that most tightly binds to the Ret1/PCEI site (21,38). The TAAAGGCT MOK2-binding motif, which is conserved between human, bovine, and mouse, overlaps with the CRX-binding element (CRXE, Fig. 3). MOK2 may repress transcription mediated by CRX by competing for CRX binding to DNA, thereby decreasing activation. CRX has been shown to bind and regulate several photoreceptor-specific promoters such as the rhodopsin, IRBP, ␤-PDE, and arrestin promoters (37). Unlike the CRX-binding element, the 8-bp MOK2-binding site is found only in the photoreceptor-specific IRBP promoter. Our results show that the complete human hsMOK2 protein, which contains the NH 2 -acidic domain, represses transcription to a greater extent than the mouse MOK2 or human hsMOK⌬ truncated isoforms, which essentially contain only the zinc finger-binding domains. The higher repression could be due to the additional steric hindrance of the hsMOK2 and other positively acting factors.
Our results did not directly demonstrate MOK2 regulation of the IRBP promoter in vivo. However, in addition to the results of binding assays and transient transfections, we showed that Mok2 expression in the developing mouse and adult retina seems to be concordant with IRBP expression. First, we found that Mok2 is expressed in the photoreceptor cells of the mouse retina where IRBP is synthetized. Mok2 mRNA was not detected in all photoreceptor cells. The morphological similarity of rods and cones in the rodent retina does not allow one to discriminate between rods and cones, but it is known that the photoreceptor cells of the mouse retina are composed of a majority of rods. IRBP is synthetized by both rod and cone photoreceptors. Rod cells synthetize 4-fold higher amounts of IRBP (39). It is possible that the photoreceptor cells expressing Mok2 might correspond to cones cells that express lower levels of IRBP. Furthermore, like IRBP mRNA, Mok2 mRNAs are detected at early developmental stages. Mok2 mRNAs are found at embryonic day 9.5 before the beginning of IRBP expression. Previous studies have shown that Irbp expression begins on embryonic day 13 in the mouse, which is similar to the photoreceptor trans-acting factor CRX (12,37,38). Mok2 is not only expressed before the beginning of IRBP expression but at all of the developmental stages tested. Increasing evidence suggests that the regulation of many genes is the result of a fine balance between positive and negative regulatory proteins. The IRBP regulation could be, among other things, the result of a balance between the repressor MOK2 and the activator CRX. Unlike Crx, the Mok2 gene is not only a photoreceptor-specific gene. Mok2 is more highly expressed in the adult mouse brain than in the whole eye. The finding that Mok2 is expressed early in mouse embryonic development and in the adult brain suggests that Mok2 might play an important role in development and particularly in neuronal development.
IRBP gene expression is highly regulated. It is known that IRBP gene transcription can be modulated by light (40) and by agents such as cAMP (41), indicating that both activation and repression of IRBP activity are required for fine regulation of IRBP levels. Some factor seems to be responsible for regulating IRBP mRNA levels. For example, when developing or adult mice were deprived of normal light, there was a decrease in the IRBP mRNA level. Furthermore, it has been postulated that aberrant expression of IRBP may be implicated in certain genetically mediated retinal degenerations of the cat (42,43) and mouse (44). Abyssinian cats homozygous for a slowly progressive form of hereditary rod and cone degeneration show a 50% reduction in IRBP mRNA and protein as early as 4 weeks of age, well before the onset of significant changes in retinal structure. The repressor activity of MOK2 might play a role in the reduction of IRBP mRNA levels. It seems worth mentioning that the human hsMOK2 gene maps to chromosome 19q13.2-q13.3 near the disease locus for autosomal dominant cone-rod distrophy (CORDII, 19q13.3 (45,46)). The trans-acting factor CRX has been located to 19q13.3. In the human, several clinical phenotypes have been associated with CRX mutations, including cone-rod dystrophy, Leder congenital amaurosis, and retinitis pigmentosa (47)(48)(49)(50)(51). The locus for dominant retinitis pigmentosa also lies near the location of the CRX gene (RP11, 19q13.4 (52)).
The IRBP gene contains two MOK2-binding elements, a complete 18-bp MOK2-binding site located in intron 2 and the essential core MOK2-binding site (8 bp of conserved 3Ј-half site) located in the IRBP promoter. The results presented here demonstrate that MOK2 can bind to the 8-bp present in the IRBP promoter and repress transcription. Actually, we do not know the role of the 18-bp MOK2-binding site present in intron 2 of the IRBP gene. This site could allow MOK2 to repress transcription in another way by blocking transcriptional elongation. Interestingly, it has been suggested that a negative regulatory element affecting mRNA elongation might be involved in controlling IRBP gene expression during fetal retinal development (53). The arrangement of the two MOK2-binding sites, the conserved 3Ј-half site in the promoter and the 18-bp binding site in the intron, is reminiscent of that of another potential MOK2 target gene, PAX3. In this gene, the 18-bp MOK2-binding site is located in the last intron. A search for MOK2-binding sites in the proximal promoter region of human PAX3 reveals the presence of a TAAAAGGCT sequence that could bind to MOK2. Therefore, MOK2 might regulate the transcriptional activity of target genes at different levels. The six other potential genes isolated by whole genome technique are still unknown (4). We previously showed that MOK2 is also an RNA-binding protein associated mainly with nuclear RNP components (4). Numerous examples of multifunctional proteins that bind to both DNA and RNA have emerged (reviewed in Refs. 54 -56). The best known members in the zinc finger family are TFIIIA and WT1. MOK2 was shown here to be a transcriptional repressor, but, in other circumstances, it might also be an activator as has been shown for many other DNAbinding transcription factors. For example, WT1 has been shown to repress and activate transcription depending on the promoter and the physiological context (reviewed in Ref. 57). This hypothesis is supported by the fact that the human isoform hsMOK2 contains an NH 2 -acidic domain that has frequently been shown to act as an activation domain in many transcription factors.