Mafs, Prox1, and Pax6 Can Regulate Chicken (cid:1) B1-Crystallin Gene Expression*

During lens fiber cell differentiation, the regulation of crystallin gene expression is coupled with dramatic morphological changes. Here we report that Mafs, Prox1, and Pax6, which are essential transcription factors for normal lens development, bind to three functionally important cis elements, PL1, PL2, and OL2, in the chicken (cid:1) B1-crystallin promoter and may cooperatively direct the transcription of this lens fiber cell preferred gene. Gel shift assays demonstrated that Mafs bind to the MARE-like sequences in the PL1 and PL2 elements, whereas Prox1, a sequence-specific DNA-bind-ing protein like its Drosophila homolog Prospero, interacts with the OL2 element. Furthermore, Pax6, a known repressor of the chicken (cid:1) B1-crystallin promoter, binds to all three of these cis elements. In transfection assays, Mafs and Prox1 activated the chicken (cid:1) B1-crystallin promoter; however, their transactivation ability was repressed when co-transfected with Pax6. Taken together with the known spatiotemporal expression patterns of Mafs, Prox1, and Pax6 in the developing lens, we propose that Pax6 occupies and represses the chicken (cid:1) B1-crystallin promoter in lens epithelial cells, and is displaced by Prox1

The lens is a transparent tissue composed of cuboidal epithelial cells on its anterior surface and fiber cells that comprise the remainder of the lens. Vertebrate lens development initiates when a portion of lens-competent cephalic ectoderm responds to an inductive signal from the optic neuroepithelium. The induced ectoderm, termed the lens placode, thickens, invaginates, and pinches off to form the lens vesicle. Cells in the anterior portion of the lens vesicle develop into lens epithelium, whereas cells in the posterior portion leave the cell cycle, elongate and differentiate into primary lens fibers. After the lens forms, its growth is mediated by lens epithelial proliferation and the differentiation of lens epithelial cells at the lens equator into lens fiber cells (1,2). This differentiation event is marked by dramatic changes in cell shape, the down-regulation of epithelial markers such as vimentin, the initiation of aqua-porin0/MIP and beaded filament expression, and the accumulation of crystallin proteins (3).
A number of transcription factors are essential for either the initiation of lens development or the differentiation of lens fiber cells from epithelial cells (4,5). Mafs are a group of basic leucine zipper transcription factors that bind to a common recognition element (TPA-MARE 1 or CRE-MARE) and regulate target gene expression (6). c-Maf expression is first detected in the lens placode and is highly up-regulated during fiber cell differentiation (7)(8)(9). In c-Maf knockout mice, the lens fiber cells fail to elongate and crystallin gene expression is severely repressed (8 -10). L-Maf is expressed in the embryonic chicken lens and is sufficient to reprogram retinal-pigmented epithelial cells into lens cells. In gel shift assays, L-Maf binds to MARElike elements in a number of crystallin promoters and can activate the expression of constructs consisting of these MAREs coupled with heterologous promoters (11). MafB is transcribed in the lens epithelium (8) and can transactivate the chicken ␣A-crystallin promoter in transfection assays (12).
Prox1, a divergent homeodomain protein, is expressed initially in the early lens placode and is up-regulated during fiber cell differentiation (13)(14)(15). Prox1 null lenses do not undergo lens fiber cell elongation, although the expression of several fiber cell-specific genes can be detected by RT-PCR (14). In co-transfection assays, Prox1 activates several ␥-crystallin promoters (16). However, it has not been established whether Prox1 is a DNA binding transcription factor like its Drosophila homolog, Prospero (17,18).
Pax6 is a paired domain/homeodomain transcription factor hypothesized to be a master control gene of visual system development because it can induce ectopic eyes in both vertebrates and Drosophila (19 -23). In vertebrates, Pax6 expression initiates in the head ectoderm and is essential for the development of the lens placode. Once the lens forms, Pax6 becomes restricted to the epithelium, which continues to express this transcription factor throughout life (24,25). Tissue recombination experiments demonstrated that lens epithelial cells lacking one functional copy of the Pax6 gene preferentially undergo fiber cell differentiation, suggesting that Pax6 maintains the epithelial cell phenotype in the mature lens (26). Pax6 is a positive transcriptional regulator of the mouse ␣B (27,28), chicken and mouse ␣A (29,30), and chicken ␦1-crystallin (31) genes. Furthermore, Pax6 cooperates with Mafs to activate -crystallin (32,33) and glucagon expression (34). However, Pax6 directly represses the transcription of the fiber cell preferred crystallin genes, chicken ␤B1 (25) and mouse ␥E,␥F (35).
During lens development, the morphological transition between epithelial cells and fiber cells coincides with specific, regulated changes in crystallin gene expression. For example, ␤B1-crystallin, which accounts for 8.5% of the total protein of the newborn mouse lens (36), is not transcribed in lens epithelial cells, but its expression begins concurrently with the initial elongation of fiber cells (37,38). Thus, ␤B1-crystallin is a specific lens fiber cell marker whose transcription is likely to be controlled by molecular mechanisms that overlap those of fiber cell differentiation. Furthermore, these mechanisms are evolutionarily conserved because the chicken ␤B1-crystallin promoter is fully functional in transgenic mice (38).

EXPERIMENTAL PROCEDURES
RT-PCR Analysis of c-Maf Isoform Expression in Lens-Total lens RNA was harvested from 5-day-old mice using the SV Total RNA Isolation Kit (Promega, Madison, WI). RT-PCR was performed using the Superscript RT-PCR kit (Invitrogen, Carlsbad, CA). The common 5Ј primer for both c-Maf isoforms is GCAGGTAGACCACCTCAAGC, the 3Ј primer for the c-Maf long form is GCAAACTGCAAGAGGGTCTC, and the 3Ј primer for the c-Maf short form is AAAAGAGCCATCAC-CACCAC. The expected sizes of RT-PCR products are 249 bp for the c-Maf long form and 300 bp for the c-Maf short form. After reverse transcription, the products were PCR amplified for 35 cycles of 94°C/15 s, 55°C/30 s, and 72°C/30 s. The RT-PCR products were electrophoresed in 8% PAGE, isolated from the gel using Ultrafree-DA DNA purification columns (Millipore, Billerica, MA), cloned into pCR 2.1-TOPO (Invitrogen), and verified by DNA sequencing.
Chicken MafB, an intronless gene, was cloned from genomic DNA by PCR using the 5Ј primer AAGGAAGCGAGAGGCGAAGCGGA and 3Ј primer ATTAAGCACTCACATGAACTC. The PCR product was subcloned into PCR-Script Amp SK(ϩ) (Stratagene) and subjected to sequencing. The eukaryotic MafB expression vector was prepared by subcloning the BamHI-NotI fragment of PCR-Script-MafB, which contains the full coding region of MafB into pcDNA3.1 Zeo(ϩ) (Invitrogen).
Two EST clones, respectively comprising the 5Ј (GenBank TM accession number 2257419) and 3Ј (GenBank TM accession number 2988788) ends of the mouse c-Maf long form cDNA separated by a NotI site, were purchased from Invitrogen. The NotI-EcoRI fragment of the 3Ј mouse c-Maf EST was inserted into pVL1392 (BD Pharmingen, San Diego, CA) to create an XbaI site at its 3Ј end. The eukaryotic expression vector for the c-Maf long form was then constructed by sequentially subcloning the EcoRI-NotI fragment of the 5Ј EST and NotI-XbaI fragment of pVL1392-c-Maf into pcDNA 3.1 Zeo(ϩ) (Invitrogen). The c-Maf short form expression vector was generated by inserting the XhoI fragment from pCR-2.1 TOPO-c-Maf short form (see the RT-PCR step) into the XhoI site of the long form expression vector. All of the c-Maf expression vectors were confirmed by sequencing.
Cell Culture and Transfection-N/N 1003A cells (a rabbit lens epithelium derived cell line) (42) were transfected following the Lipo-fectAMINE Plus reagent protocol provided by Invitrogen. Generally 6 ϫ 10 5 cells were plated for each 60-mm dish at least 8 h before transfection, each dish received 2.5 g of promoter/chloramphenicol acetyltransferase (CAT) plasmid, 0.5 g of pCMV/␤GAL, and 0.5 g of various expression plasmids. Cells were harvested 48 h after transfection and cellular extracts were prepared by multiple cycles of freeze/thaw. The extracts were assayed for CAT and ␤-galactosidase activity as previously described (39). All co-transfection experiments were performed at least twice in triplicate and analyzed statistically by Student's t test.
Western Blot Analysis-The polyclonal MafB antibody was made in rabbit against mouse MafB peptide QPLQSFDGFRSAHH (amino acids 119 -131) and affinity purified. For Western blotting, lenses were dissected from adult mice. Epithelium and fiber cells were separated and homogenized in RIPA buffer (150 mM NaCl, 10 mM Tris, pH 7.2, 0.1% SDS, 1.0% Triton X-100, 1% sodium deoxycholate, 5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 2 g/ml pepstatin A, and 100 M sodium orthovanadate) to obtain water-soluble protein. Cell extracts of ␣TN4-1 cells (a mouse lens-derived cell line) (43) and N/N 1003A cells were prepared by harvesting the cultured cells and homogenizing them in RIPA buffer. 20 g of each protein preparation was subjected to electrophoresis through 12% SDS-polyacrylamide gels and electroblotted onto a nitrocellulose membrane (Invitrogen). The membranes were blocked overnight in 10% powdered milk in phosphatebuffered saline with 0.1% Tween 20 (PBS-T) and incubated with a 1:500 dilution of the antibody raised against the MafB for 1 h. The blots were washed 3 times with PBS-T, incubated with horseradish peroxidaseconjugated goat anti-rabbit IgG (Upstate USA, Charlottesville, VA) for 1 h, washed as above, incubated with LumiGlo chemiluminescent reagents following the manufacturer's instructions (Upstate USA), and exposed to BioMax MR film (Kodak, Rochester, NY).
Recombinant Protein Expression-The coding sequence for the bZip DNA binding region of MafB (from amino acid 195 to the carboxyl terminus) was synthesized by PCR with the 5Ј primer GGGGCTC-GAGAGGACCGGTTCTCGGATGAC and 3Ј primer GGGGGAGATCT-TAAGCACTCACATGAA. The PCR product was subcloned into pCR 2.1-TOPO and confirmed by sequencing. The MafB expression vector was generated by subcloning the EcoRI fragment into pET 41(c) (Novagen, Madison, WI). The recombinant MafB expression construct was subjected to sequencing to confirm the correct reading frame. The GST-Prox1 expression vectors were previously described (44). The GST-Pax6 expression vector was a gift from Drs. Jonathan Epstein and Richard Mass (Howard Hughes Medical Institute, Boston, MA).
Electrophoretic Mobility Shift Assay (EMSA)-All EMSAs were performed in a total volume of 12.5 l containing 6 l of A100, 250 ng of poly(dA-dT) or poly(dI-dC), 2.5 g of bovine serum albumin, 2 g of GST fusion protein, and 20,000 Cerenkov counts of double-stranded DNA probe as described previously (25). Nonradioactive competitors were added at 50-fold molar excess. After a 15-min preincubation at room temperature, reactions were electrophoresed through 5% polyacrylamide gels, using 0.5ϫ Tris borate-EDTA as the buffer, at 4°C. The oligonucleotides used for EMSA (sense strands only shown) were: PL1 (mut PL1) and Ϫ126/Ϫ46 (mut PL2) were described in Duncan et al. (25). TDA-Pros and mPbs were described in Hassan et al. (18).

PL1 and PL2 Elements
Are Essential for Chicken ␤B1-Crystallin Promoter Activity-The full-length chicken ␤B1-crystallin promoter (Ϫ432/ϩ30) can be separated into two portions: Ϫ126/ϩ30 is the minimal promoter in transfections (39) and contains all of the signals necessary for lens fiber cell-specific gene expression (38); whereas Ϫ432/Ϫ126 directs crystallin level gene expression (25,45) (Fig. 1A). Three distinct functional cis elements have been identified in the minimal promoter by DNase I footprinting and transfection analyses (39). The PL1 and PL2 elements are similar to each other and to the TPA-MARE sequence for Maf transcription factors (11). In contrast, the OL2 element is similar to the consensus DNA recognition sequence for the Drosophila protein, Prospero (18) (Figs. 1A and 6A).
The contribution of PL1 and PL2 to the activity of the chicken ␤B1-crystallin promoter was tested by generating Ϫ432/ϩ30 reporter constructs with either PL1, PL2 or both deleted. Transfection assays performed in lens-derived N/N 1003A cells demonstrated that deletion of either one or both of the PL1 and PL2 elements reduced ␤B1-crystallin promoter activity by at least 70% (Fig. 1B). Similar results were also obtained in the non-lens cell line CHO (data not shown).
MafB and c-Maf Isoforms Are Expressed in Lens-Because TPA-MARE consensus sites can be recognized by multiple Maf family members, we investigated Maf expression in the adult lens. Previous in situ hybridization studies showed that MafB mRNA was present in the embryonic lens (8,46). Expression of MafB in the adult lens was determined by Western immunoblotting of extracts obtained from microdissected adult lens fiber cells, lens epithelial cells, and extracts of two lens-derived cell lines, ␣TN4-1 (43) and N/N 1003A (42). MafB protein was detected in microdissected lens epithelial cells as well as in the established lens epithelial cell lines, but not in the fiber cells ( Fig. 2A).
c-Maf protein is expressed at moderate levels in the lens epithelium, and its expression increases sharply with fiber cell differentiation (9). However, the mouse c-Maf gene is alternatively spliced, producing a long form with 10 extra amino acids at the carboxyl terminus compared with the short from (47). RT-PCR analysis detected the transcripts of both isoforms in 5-day-old mouse lenses (Fig. 2B), although this qualitative RT-PCR analysis suggests that the short form predominates over the long form. (sequence is shown in bold) are cis elements known to be important for ␤B1crystallin promoter function (38,39). B, transfection analysis of the chicken ␤B1crystallin promoter in lens-derived N/N 1003A cells. Loss of either or both the PL1 and PL2 element significantly decreases promoter activity (p ϭ 0.003 for p⌬PL1, p ϭ 0.003 for p⌬PL2, p ϭ 0.002 for p⌬PL1/ PL2). CAT activity is expressed relative to the activity of pCAT-Basic, which was arbitrarily set at 1. p432, the wild type chicken ␤B1-crystallin promoter (Ϫ432/ϩ30) linked to the cat gene; p⌬PL1, p432 without the PL1 element; p⌬PL2, p432 without the PL2 element; p⌬PL1/PL2, p432 without both the PL1 and the PL2 elements.  (43); N/N 1003A, an established rabbit lens epithelial cell line (42). Note that MafB was detected in adult lens epithelial cells and two lens-derived epithelial cell lines, ␣TN4 -1 and N/N 1003A, but was not detected in lens fiber cells from adult mice. B, expression of c-Maf splice forms in the lens. Total mRNA was isolated from 5-day-old mouse lenses and used as the template for RT-PCR analysis with isoform-specific 3Ј primers. c-MafL, c-Maf long form; c-MafS, c-Maf short form; RT-, reactions without reverse transcriptase. Both splice forms are expressed by the lens but the short form appears to predominate. tively (Fig. 3, A, lane 3, and B, lanes 3 and 4).
Transfection assays also demonstrated that MafB-mediated transactivation is dependent on both PL1 and PL2 because deletion of either PL1 or PL2 diminished MafB-mediated transactivation in N/N 1003A cells (Fig. 3C). Similar results were also obtained in the non-lens cell line CHO (data not shown).
Maf Specifically Binds to the PL1 and PL2 Elements of the Chicken ␤B1-Crystallin Promoter-Because Mafs can transactivate the chicken ␤B1-crystallin promoter and this transactivation is dependent on the MARE-like PL1 and PL2 elements, we tested whether these elements can bind Maf proteins in gel shift assays (Fig. 4). When there was no nonradiolabeled PL2/OL2 competitor present, MafB⅐PL2/OL2 complexes can be visualized by autoradiography. The formation of a labeled complex was significantly reduced by competition with nonradioactive self-oligonucleotide PL2/OL2, PL1, Ϫ126/Ϫ46, or Ϫ126/Ϫ46 (mut PL1), but not with PL1 mut1 and PL2 mut1, suggesting that MafB can specifically bind to PL1 and PL2 elements of the chicken ␤B1-crystallin promoter. The PL2 element appears to be more important for MafB binding because Ϫ126/Ϫ46 (mut PL2), which has an intact copy of PL1, competed less efficiently than Ϫ126/Ϫ46 (mut PL1), which has an intact copy of PL2.
Prox1 Activates the Chicken ␤B1-Crystallin Promoter-The vertebrate homolog of Drosophila Prospero, Prox1, is crucial for lens fiber cell elongation (14) and is able to activate ␥-crystallin promoters in transfection assays (16). To test whether Prox1 is also capable of activating the chicken ␤B1-crystallin promoter, co-transfection assays were performed with a Prox1 expression vector in N/N 1003A cells. The relative cat activity was increased 7.0 Ϯ 0.3-fold when the Prox1 expression vector was co-transfected (Fig. 3A, lane 4).
Prox1 Is a DNA-binding Protein-Whereas Prox1 can activate crystallin gene expression, its ability to function as a DNA binding transcription factor has not been demonstrated. However, the COOH terminus (amino acids 572 to 736) of chicken Prox1 consisting of its homeo-and Prospero domains is 78% similar to Drosophila Prospero (amino acid 1241 to 1403) (13), a sequence-specific DNA-binding protein and transcription fac- . C, the wild type reporter, p432 was significantly transactivated by MafB (p ϭ 0.00007). However, MafB could not activate the reporters lacking either the PL1 or PL2 elements (p ϭ 0.36 for p⌬PL1, p ϭ 0.78 for p⌬PL2). The relative CAT activity of the MafB expression vector and p⌬PL1/PL2 co-transfected cells was even lower than that of p432 (p ϭ 0.002). CAT activity is expressed relative to the activity of p432, which was arbitrarily set at 1.

FIG. 4. MafB specifically binds to the chicken ␤B1-crystallin
promoter. An oligonucleotide consisting of the PL2/OL2 element of ␤B1-crystallin was radiolabeled and used for EMSAs with the MafB/ GST fusion protein. MafB bound to the PL2/OL2 oligonucleotide and was competed by a 50-fold molar excess of nonradioactive oligonucleotides corresponding to self, PL1, 126/Ϫ46, and Ϫ126/Ϫ46 (mut PL1). By contrast, little to no competition was observed with a 50-fold molar excess of PL2 mut1 (M7) and PL1 mut1 (M6A), each of which contains a functional mutation in the respective MARE (38,39). Both mutations can significantly decrease chicken ␤B1-crystallin promoter activity in both transfection assays and transgenic mice (38,39).
tor (18). The COOH-terminal 236 amino acids (residues 1171-1406) of Prospero binds to a consensus sequence containing C(A/t)(c/t)NNC(T/c), where N is any nucleotide (18). To investigate whether Prox1 is a DNA-binding protein like Prospero, EMSA was performed using recombinant Prox1-GST fusion proteins and either TDA-Pros, the Prospero consensus binding site determined by a target detection assay, or mPbs, which contains multiple copies of the Prospero consensus binding site (18). The Prospero domain of Prox1 (Pros) formed a complex with mPbs (Fig. 5B), and the homeo domain/Prospero domain (HD ϩ Pros) of Prox1 forms a complex with both TDA-Pros (Fig.  5A) and mPbs (Fig. 5B). All DNA-protein complexes were significantly reduced by competition with nonradioactive self-oligonucleotide. As expected, the negative control GST-MafB fusion protein did not bind to mPbs (Fig. 5B).
Prox1 Binds to the OL2 Element of the Chicken ␤B1-Crystallin Promoter-Because the OL2 element of the chicken ␤B1crystallin promoter is similar to the Prospero consensus binding sequence (Fig. 6A), we tested the ability of Prox1 to bind the PL2/OL2 region. EMSA revealed that Prox1 can interact with the radiolabeled PL2/OL2 element in the absence of nonradioactive competitor (Fig. 6B). We designed a series of PL2/OL2 mutations, PL2 mut1 (the M7 mutation of the PL2/MARE that significantly reduces chicken ␤B1-crystallin promoter activity (38,39)), PL2 mut2 (mutation of the 5Ј end of the PL2/MARE), OL2 mut1 (mutation of the 5Ј end of OL2), OL2 mut2 (mutation of the 3Ј end of OL2) (Fig. 6A) to test the binding specificity of Prox1 in gel shift assays. The formation of the Prox1⅐PL2⅐OL2 complex was completely eliminated by competition with the nonradioactive self-oligonucleotide PL2/OL2. The mutated oligonucleotides, PL2 mut1 (M7), PL2 mut2, and OL2 mut2, competed for HD ϩ Pros binding, suggesting that these sites were not important for Prox1 binding. However, OL2 mut1 and PL1 (which lacks a Prospero consensus site) competed less efficiently for HD ϩ Pros binding (Fig. 6B). These data suggest that the 5Ј end of the Prospero consensus element is critical for Prox1 binding.
The OL2 Element Is Important for Chicken ␤B1-Crystallin Promoter Activity and Prox1-mediated Transactivation-Because recombinant Prox1 binds the chicken ␤B1-crystallin pro-moter via the 5Ј end of the OL2 element, we tested whether these nucleotides are important for ␤B1-crystallin promoter activity and Prox1-mediated transactivation. In N/N 1003A cells, the three-nucleotide mutation (OL2 mut1) reduced promoter activity by half over wild type. Prox1 activated the p432 reporter 5.8 Ϯ 0.9-fold, whereas it can only activate the pOL2mut1 reporter 3.6 Ϯ 0.4-fold, which is significantly less FIG. 5. Prox1 can interact with the Prospero consensus binding site. A, TDA-Pros, an oligonucleotide consisting of two copies of the Prospero consensus determined by target detection assay (18), was radiolabeled and used for EMSAs with the homeodomains and Prospero of Prox1 (HDϩPros). HD ϩ Pros bound to TDA-Pros and was competed by a 50-fold molar excess of nonradioactive oligonucleotides corresponding to TDA-Pros and mPbs, which consists of multiple Prospero consensus binding sites. B, mPbs was radiolabeled and used for EMSAs with the Prospero domain of Prox1 (Pros), HD ϩ Pros, and MafB/GST fusion proteins.
FIG. 6. Prox1 can bind to the OL2 element of the chicken ␤B1-crystallin promoter. A, the PL2 and OL2 elements of the chicken ␤B1-crystallin promoter (Ϫ93/Ϫ70). MARE and Prospero consensus sequences are aligned with the WT PL2/OL2 sequence to identify putative functional regions. Mutated oligonucleotides used in the EMSA shown in panel B are listed below with mutations shown in bold: PL2 mut1 is the M7 mutation in PL2 (39), OL2 mut1 is a mutation in the 5Ј of the OL2 element, PL2 mut2 is a mutation in the 5Ј of the PL2 element, OL2 mut2 is a mutation in the 3Ј of the OL2 element. B, EMSA analysis of Prox1 binding to the PL2/OL2 region of the chicken ␤B1crystallin promoter. An oligonucleotide consisting of wild type PL2/OL2 shown in panel A was radiolabeled and used for EMSAs with the Prox1 (HD ϩ Pros)/GST fusion protein. HD ϩ Pros of Prox1 bound to the wild type PL2/OL2 oligonucleotide and was competed by a 50-fold molar excess of nonradioactive oligonucleotides corresponding to self, PL2 mut1 (M7), PL2 mut2, and OL2 mut2. Less competition was observed with a 50-fold molar excess of either OL2 mut1 or PL1. C, N/N 1003A cells were transfected with p432, pOL2mut1, and a plasmid encoding Prox1. The activity of pOL2mut1 is 60% lower than p432 (p ϭ 0.009). Prox1 transactivated the p432 promoter 5.8 Ϯ 0.9-fold. In contrast, Prox1 could only activate pOL2mut1 3.6 Ϯ 0.4-fold, which was significantly lower than its ability to transactivate the p432 promoter (p ϭ 0.02). than wild type (p ϭ 0.02) (Fig. 6C). These data suggest that the Prospero consensus site found in the OL2 element contributes to Prox1-mediated transactivation of p432.
Prox1 and Mafs Cooperate to Activate the Chicken ␤B1-Crystallin Promoter-It has been shown that Prox1 and the c-Maf short form synergistically activate the mouse ␤B2-crystallin promoter in transfection assays (48). Here we investigate whether Prox1 and Mafs can cooperatively activate the ␤B1crystallin promoter in transfection assays. Co-transfection of Prox1 with MafB did not significantly increase MafB-mediated transactivation in either N/N 1003A (Fig. 3A, lane 6) or CHO cells (data not shown). In contrast, Prox1 and the c-Maf short form alone activated the p432 promoter 4.0 Ϯ 0.2-and 2.4 Ϯ 0.3-fold, respectively, co-transfection of both Prox1 and the c-Maf short form expression vectors increased promoter activity by 5.3 Ϯ 0.7-fold (Fig. 3B, lane 7). Similar results were also obtained when Prox1 and c-Maf long form expression vectors were co-transfected (Fig. 3B, lane 9).
Pax6 Represses MafB-, c-Maf-, and Prox1-mediated Transactivation-Pax6 and Mafs cooperate to activate -crystallin (33) and glucagon (34) promoter activity. To examine whether the chicken ␤B1-crystallin promoter is regulated by the same way, a Pax6 expression vector was co-transfected with MafB, Prox1, or c-Maf expression vectors into N/N 1003A cells. As previously shown (25), Pax6 represses the chicken ␤B1-crystallin promoter (Fig. 3A, lane 5). Moreover, when the Pax6 expression vector is co-transfected with MafB or c-Maf, transactivation was lower than in transfection tests with MafB or c-Maf alone (compare Fig. 3, A, lanes 3 and 7; B, lanes 3 and 8; and lanes 4  and 10). Furthermore, Prox1-mediated transactivation is decreased at least by half when co-transfected with Pax6 expression vector (compare Fig. 3A, lanes 4 and 8). Similar results were also obtained in the non-lens cell line CHO (data not shown).
Pax6 Can Bind to cis-Elements of the Chicken ␤B1-Crystallin Promoter-In previous studies, Pax6 was able to bind to an oligonucleotide consisting of a PL2 trimer (25). Here we demonstrate that Pax6 can bind a single copy of the PL2/OL2 element and this complex was reduced by competition with nonradioactive self-oligonucleotide (PL2/OL2), PL1, Ϫ126/Ϫ46 or Ϫ126/Ϫ46 (mut PL1) (which has an intact copy of PL2), but not with PL2 mut1 (Fig. 7), suggesting that Pax6 can specifi-cally bind to PL1 and PL2/OL2 elements of the chicken ␤B1crystallin promoter. The PL2 element appears to be more important for Pax6 binding because Ϫ126/Ϫ46 (mut PL2), which has an intact copy of PL1, competed less efficiently than Ϫ126/ Ϫ46 (mut PL1), which has an intact copy of PL2. DISCUSSION Crystallin genes are among the most transcriptionally active single copy genes and produce from 1 to 50% of total lens mRNA depending on gene, species, and developmental stage (49). Some crystallin promoters are active in all lens cells as well as some non-lens tissues (50,51). In contrast, ␤B1-crystallin transcription is highly lens fiber cell preferred (38,41). In this study, we demonstrate that Pax6 can repress Maf-mediated transactivation by binding to the same composite element of the chicken ␤B1-crystallin promoter. We also show that Prox1 activates the chicken ␤B1-crystallin promoter by specifically binding to sequences adjacent to the MARE-like PL2 composite element. Furthermore, this Prox1-mediated transactivation is inhibited by Pax6. Thus, the promoter activity depends on occupancy of these transcription factors on the promoter: if Pax6 concentrations are high (as seen in the lens epithelium) (19), ␤B1-crystallin gene expression is off; if Prox1 and Maf concentrations are high and Pax6 concentrations are low (as in lens fiber cells) (8,14,15), Pax6 is displaced from the promoter by c-Maf and Prox1, which stimulate ␤B1-crystallin gene expression (Fig. 8).
Pax6 is essential for lens specification and early development (52). At later stages, loss of Pax6 expression in lens fiber cells is essential for normal fiber cell differentiation and crystallin gene expression. 2 In transgenic mice overexpressing Pax6 in lens fiber cells, the lens is reduced in size and secondary fiber cell elongation is incomplete. Notably, ␤B1-crystallin transcription is repressed by 78% relative to total lens RNA, confirming the importance of Pax6 down-regulation for ␤B1-crystallin expression. 2 Previous studies showed that c-Maf is critical for lens crystallin gene expression. In c-Maf null mice, the expression of ␣-, ␤-, and ␥-crystallins is severely repressed or completely abolished (8 -10). In transfection assays, c-Maf was able to activate the ␣A (12), ␤B2 (48), ␦1 (12), and ␥ (53) crystallin promoters. Other members of the large Maf family, L-Maf/MafA, MafB, and NRL are detected in the developing lenses of vertebrates (8,46,54,55). All of them can activate crystallin promoters in co-transfection assays, even though there are quantitative, but not qualitative, differences in their transactivation capacities (12,33). Thus, transcriptional activation of a crystallin gene may result from various Maf members at different developmental stages.
The spatiotemporal expression patterns of crystallin genes can be attributed to the placement of cis elements and assortment of a small set of developmentally essential transcription factors (56,57). Here we show that, for the chicken ␤B1-crystallin gene, the activators, Mafs bind to MARE-like sequences in both the PL1 and PL2 elements; the repressor, Pax6, binds to both of these elements. Notably, the distribution of Maf and Pax6 binding sites appears to be unique to the chicken ␤B1crystallin promoter and may be essential for its stringently regulated transcription (38). In crystallin promoters that function in both lens epithelial and fiber cells, such as mouse ␣A (9, 30), chicken ␣A (11,29), and chicken ␦1 (31, 58), Pax6 activates these genes by binding to cis elements that are separated, or at most, partially overlapping with the consensus MAREs. It is possible that Maf and Pax6 bind to the corresponding elements and cooperatively activate the transcription of these crystallin genes as they do in the guinea pig (Cavia porcellus) -crystallin (33) and glucagon (34) promoters. Taken together, it is likely that the relative orientation of MAREs and Pax6 sites in a promoter determine the structure of the Pax6⅐Maf⅐DNA complex formed that results in the observed divergent effects on transcriptional regulation. Prox1 is another transcription factor known to be critical for fiber cell differentiation (14). In this study, we determined that the COOH terminus of chicken Prox1 can specifically interact with both the Prospero consensus sequence (18) and a similar sequence located in the OL2 element that is adjacent to the PL2/MARE element of the chicken ␤B1-crystallin reporter. However, another group has reported that in vitro translated zebrafish Prox1 failed to bind either the Prospero consensus binding sequence or the mouse ␥-crystallin promoter in gel shift assays (59). These differences may reflect species difference in Prox1 function or may simply arise from alternative experimental design because both Prospero (18) and Prox1 are relatively weak DNA-binding proteins in gel shift assays. Recently, it has become apparent that Prox1 probably interacts with other proteins while performing its functions. Prox1 can bind to CBP/p300, a transcriptional cofactor, and work synergistically with Maf to activate the mouse ␤B2-crystallin promoter (48). It is possible that Maf and Prox1, respectively, bind to adjacent cis elements, PL2 and OL2, and cooperatively recruit CBP/p300 to the chicken ␤B1-crystallin promoter as they do for the mouse ␤B2-crystallin promoter.
Not only is chicken ␤B1-crystallin transcription regulated by finely arrayed cis acting elements in the promoter, but also by changes in the expression pattern of the corresponding transcription factors during development. In embryonic lenses, Pax6 protein is found throughout the lens with the highest level in the lens epithelium (19). Pax6 cooperates with Mafs to activate crystallin genes expressed in all lens cells, such as mouse ␣A and chicken ␦1-crystallins, whereas Pax6 inhibits the expression of fiber cell preferred ␤B1 (25) and ␥-crystallins (35). Furthermore, Prox1 (14,15) and c-Maf (9) levels are strongly up-regulated in the transitional zone where fiber cells begin to form. Pax6 levels decrease sharply in chicken lens fiber cell nuclei late in embryonic development (25), coincident with the down-regulation of ␦1-crystallin and up-regulation of ␤B1-crystallin expression (60), consistent with our hypothesis.
In conclusion, the regulated expression of chicken ␤B1-crystallin appears to be achieved at two levels: at the promoter level, cis elements are recognized by positive (Maf and Prox1) and negative (Pax6) transcription factors simultaneously; at the transcription factor level, the repressor, Pax6, is expressed at high levels in the lens epithelium but not in lens fiber cells. In contrast, the activators, Prox1 and c-Maf are strongly upregulated in lens fiber cells coincident with the initiation and up-regulation of chicken ␤B1-crystallin expression. Thus, the spatiotemporal expression of chicken ␤B1-crystallin could be harmoniously directed during lens differentiation.