Pax-6 and (cid:97) B-crystallin/Small Heat Shock Protein Gene Regulation in the Murine Lens INTERACTION WITH THE LENS-SPECIFIC REGIONS, LSR1 AND LSR2*

We have demonstrated previously that a transgene comprising the (cid:50) 164/ (cid:49) 44 fragment of the murine (cid:97) B-crystallin gene fused to the bacterial chloramphenicol acetyltransferase ( cat ) gene is lens-specific in transgenic mice. The (cid:50) 147 to (cid:50) 118 sequence was identified as a lens-specific regulatory region and is called here LSR1 for lens-specific region 1. In the present experiments, a (cid:50) 115/ (cid:49) 44- cat transgene was also lens-specific in transgenic mice, although the average activity was 30 times lower than that derived from the (cid:50) 164/ (cid:49) 44- cat transgene. The (cid:50) 115/ (cid:49) 44 (cid:97) B-crystallin fragment contains a highly conserved region ( (cid:50) 78 to (cid:50) 46) termed here LSR2. A (cid:50) 68/ (cid:49) 44- cat transgene, in which LSR2 is truncated, was inactive in transgenic mice. DNase I footprinting indicated that LSR1 and LSR2 bind partially purified nuclear proteins from either (cid:97) TN4-1 lens cells or the mouse lens as well as the purified paired domain of Pax-6. Site-specific mutation of LSR1 eliminated both Pax-6 binding and promoter activity of the (cid:50) 164/ (cid:49) 44- cat transgene in transgenic mice. Finally antibody/electro- phoretic mobility shift assays and cotransfection experiments

The crystallins comprise approximately 90% of the water-soluble proteins of the transparent eye lens and contribute to its optical properties (1,2). The ␣B-crystallin/small heat shock protein gene is expressed abundantly in lens and also in various other tissues, including skeletal muscle, heart, and to a lesser extent, lung (3,4). Previously we demonstrated that the differential constitutive expression of the murine ␣B-crystallin gene is under transcriptional control (4,5) and that the sequences between Ϫ426 and Ϫ259 of the murine ␣B-crystallin promoter functions as a muscle-preferred enhancer (5). Transgenic mice containing an ␣B-crystallin promoter/bacterial chloramphenicol acetyltransferase (cat) reporter transgene established that the sequences between Ϫ426 and Ϫ164 (which includes the enhancer) are required for expression in heart and skeletal muscle, while sequences downstream of Ϫ164 are sufficient to direct lens-specific gene expression (6). The Ϫ147/Ϫ118 sequence con-tains a putative lens-specific regulatory region called LSR (renamed here LSR1) (6). In the present investigation we have performed transient transfection and transgenic mouse experiments to identify another regulatory element at positions Ϫ78/ Ϫ46, called here LSR2, which also contains lens-specific promoter activity. Thus, the regulatory elements required for lens specificity of the diversely expressed mouse ␣B-crystallin gene are located downstream from the enhancer required for expression in skeletal muscle and heart (5).
Recent experiments have indicated that Pax-6, a member of paired-domain (PD) 1 family of transcription factors, is an essential factor for eye development (7)(8)(9)(10)(11). The Pax-6 gene also encodes an alternatively spliced variant (Pax-6-5a) (9,12). This variant has a 14-amino acid insertion in N-terminal half of the PD changing its recognition specificity. During eye development, the expression pattern of Pax-6 indicates its direct role in the formation of lens, retina, and cornea (9). Mutations in Pax-6 are associated with distinct eye defects, including small eye in mouse and rat (13), aniridia in humans (8, 14 -16), Peter's anomaly in humans (17) and eyeless in Drosophila (18). We have demonstrated that Pax-6 is involved in lens-specific expression of chicken (19) and mouse (20) ␣A, chicken ␦1 (21) and guinea pig (22) crystallin genes (23). In the present study, DNase I footprinting, antibody/electrophoretic mobility shift assay (EMSA), site-directed mutagenesis, and transient cotransfection experiments provide evidence that Pax-6 interacts at LSR1 and LSR2 to activate the ␣B-crystallin promoter in the lens.
Linear DNA fragments were injected into one pronucleus of a single celled mouse embryo (FVB/N strain) (24). The embryos were obtained from superovulated FVB/N females. Injected embryos were transferred into FVB/N females made pseudopregnant by mating to vasectomized FVB/N males. Transgenic mice were created by the National Eye Institute Centralized Transgenic Facility.
Analysis of Transgenic Mice-DNA was isolated (25) from tails of founder Fo mice and analyzed by Southern blot and polymerase chain reaction analyses for the presence of the transgene. We used 5Ј oligodeoxynucleotide primers (oligodeoxynucleotide 10660, 5Ј-CCCTGATCA-CAAGTCTCCATGAACT-3Ј, for ␣B115-cat and oligodeoxynucleotide 10661, 5Ј-ACCCCTGACCTCACCATTCCAGAAG-3Ј, for ␣B68-cat, and oligodeoxynucleotide 10662, 5Ј-TCTCTTTTCTTAGCTCAGTGTCTAG-3Ј, for ␣BLSR1(Mu-9760)-cat, which were specific for the murine ␣B-crystallin promoter, and a 3Ј oligodeoxynucleotide primer (oligodeoxynucleotide 7576, 5Ј-CGGTCTGGTTATAGGTACATTGAGC-3Ј) which was specific for the cat gene. Fo mice containing the transgene were mated to nontransgenic FVB/N mice to obtain F1 offspring, and sibling matings were used to establish homozygous mouse lines. The transgene copy number for each mouse was estimated by hybridization intensity in a slot blot analysis of the genomic DNA relative to the standard samples representing 0 -50 copies of the transgene, using the FIG. 1. Structure of the murine ␣Bcrystallin-cat chimeric transgenes. Approximately 2.7-kbp NdeI/PstI fragments isolated from p11-3, p61-7, and p9760 (6) containing a Ϫ115/ϩ44, a Ϫ68/ ϩ44, and a mutated Ϫ164/ϩ44 fragment (6) of ␣B-crystallin gene, respectively, linked to the bacterial cat gene were used as transgenes in transgenic mice. Approximately 60 bp of pBR sequences are present 5Ј to the ␣B fragments in the transgene and approximately 1,400 bp of simian virus 40 (SV40) sequences, including the small intron and poly(A) addition signal, were present 3Ј to the cat gene. The site-specific mutation in ␣BLSR1(Mu-9760)-CAT is shown in Fig. 7A. ECL 3Ј-Oligo Labeling Detection System (RPN 2130/2131, Amersham) and the 1.6-kbp NdeI/BamHI fragment from pSVO-CAT as the labeled probe. Blots were exposed at room temperature for 1 min.
Tissue Extraction and CAT Assays-Hemizygous mice between 1 and 3 months of age were sacrificed by CO 2 asphyxiation and the lenses, lung, heart, liver, kidney, spleen, brain (cerebrum), and skeletal muscle (from thigh) were homogenized with 0.25 M Tris-HCl (pH 7.8) in Duall glass homogenizers (Knotes, 0020). The lenses were homogenized in polypropylene Eppendorf microcentrifuge tubes, centrifuged for 10 min at 4°C, and the supernatant fraction heated at 65°C for 15 min followed by centrifugation for 10 min at 4°C. Protein concentrations were determined in the final supernatant fraction with the Bio-Rad protein assay kit according to the manufacturer's instructions using bovine serum albumin for the standards. Tissue extracts containing 0.5-20 g of total protein were analyzed for CAT activity by the biphasic assay (26). The level of CAT activity was taken as an indirect measure of promoter strength.
Nucleic Acid Isolation-Plasmid DNA was prepared by the alkaline lysis method (27) followed by ultracentrifugation banding in CsClethidium bromide. For transfection experiments plasmid DNA was isolated and purified using the QIAGEN plasmid kit (QIAGEN Inc., Chatsworth, CA). Mouse tail DNA was isolated by modification of the procedure of Hogan et al. (25).
Proteins, Nuclear Extracts, Antiserum, EMSA, and DNase I Footprinting Assays-Nuclear extracts were prepared from ␣TN4-1 mouse lens cells (28), L929 fibroblasts (29), N/N1003A rabbit lens cells (30), and adult mouse lenses as described by Shapiro et al. (31). Complementary oligodeoxynucleotides were synthesized (model 380A synthesizer; Applied Biosystems) and annealed at 1:1 molar ratio as described previously (32). Double-stranded oligodeoxynucleotides were labeled on one strand, and EMSAs were performed as described previously (32). Wild type and mutated versions of an oligodeoxynucleotide containing the sequence Ϫ152/Ϫ121 (LSR1) or the wild type oligodeoxynucleotide containing the sequence Ϫ78/Ϫ28 (LSR2) of the ␣B-crystallin promoter were used for EMSA. For footprinting lens extract was partially purified on a heparin ultrogel A4R column and eluted with 400 mM KCl as described previously (33). Anti-quail-Pax-6 antibody was a gift from Dr. Simon Saule (34).
DNase I footprinting was performed by using the EcoRI-BamHI restriction fragment of pD96 (5) spanning positions Ϫ666 to ϩ76 of the ␣B-crystallin promoter. A polymerase chain reaction-generated frag-ment corresponding to the Ϫ190 to ϩ40 sequence of the ␣B-crystallin promoter was used for footprinting experiments with the paired domain of purified mouse Pax-6 (PD5 and its alternatively spliced form, PD5a) (12). The Pax-6 GST expression vectors were gifts of Dr. Richard Maas (Harvard Medical School, Boston, MA). End labeling, EMSA, and DNase I footprinting were performed as described previously (32).
Site-directed Mutagenesis-Mutations that were generated previously (6) within the Ϫ164/ϩ44 EcoRI/PstI fragment of the mouse ␣Bcrystallin gene obtained from pRD30A (5) were used for transfection experiments and EMSAs. In brief, site-specific mutations (Mu-9760 and Mu-9761) (6) were introduced by using a oligodeoxynucleotide-directed mutagenesis kit (Sculptor in vitro mutagenesis kit, Amersham). Mutagenic oligodeoxynucleotides contained the substitution sequence TCTAGA (XbaI site) and 20 bases on each side complementary to the ␣B-crystallin promoter sequence. The resulting mutated restriction fragments were subcloned into pRD30A at the unique BamHI site (5). All constructs were confirmed by sequencing the ligation junctions and mutated regions.
Cell Culture, Transfection, and Enzyme Assay-Mouse COP-8 fibroblasts (35) were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal calf serum and 50 g/ml of gentamicin in 10% CO 2 . The cells were propagated on 60-mm diameter  Footprinting is shown for the lower (antisense) strand of the EcoRI-BamHI fragment (Ϫ666 to ϩ76) of pRD28 (5). pRD28 was digested with BamHI and the lower strand was 5Ј end-labeled with [␥-32 P]dATP, using T4 polynucleotide kinase. Partially purified nuclear extracts were used from ␣TN4-1 lens cells (28) or adult mouse lenses, as indicated (see "Experimental Procedures" for the description of ␣H400 purification). The solid, hatched, and open boxes indicate the regions (LSR1, LSR2, and downstream sequence, respectively) protected from DNase I digestion after incubation with the ␣TN4-1 and lens nuclear extracts. plastic dishes. Seven g of wild type ␣B-promoter-cat plasmids (p65-7 and p11-3) (5) or mutated test plasmids (Mu-9760 and Mu-9761) (6), increasing amounts (0.1-3 g) of pKW10-Pax6 (which expresses the wild type PD, homeodomain, and transactivation domain of mouse Pax-6), and 2 g of internal control pCH110 which expresses ␤-galactosidase (Pharmacia Biotech Inc.) were cotransfected for 6 h as calcium phosphate precipitates as described previously (32). In the cotransfection experiments using p11-3, three mutant forms of Pax-6 cDNA expression vectors were used (p309C expresses the PD (1-140 amino acids), p310A expresses the PD and the homeodomain (1-286 amino acids), and p312A expresses the PD, homeodomain, and a part of transactivation domain (1-388 amino acids) of Pax-6). Cells were harvested and extracts were prepared 48 h after transfection. CAT activities were determined by the biphasic (26) assay, and ␤-galactosidase activities were determined as described previously (32). The transfection data represent the means of three separate experiments, with each experiment being conducted with duplicate plates.
Polymerase chain reaction analyses of tail DNA were performed from the transgenic mice containing ␣B115-CAT, ␣B68-CAT, and ␣BLSR1(Mu-9760)-CAT constructs, to establish the presence of the transgene. Five of the 21 Fo mice carried the ␣B115-cat transgene in their genome, 3 out of 29 Fo mice contained the ␣B68-cat transgene, and 5 out of 37 Fo mice contained the ␣BLSR1(Mu-9760)-cat transgene. Three lines carrying one integration site (all had multiple copy numbers) of each construct were chosen for further analysis. The transgene data are summarized in Table I.
Expression of the ␣B-crystallin Promoter-cat Fusion Genes in Transgenic Mice: Evidence for a Proximal Lens-specific Regulatory Region-To investigate the possibility that there are sequences downstream of LSR1 (6) that can confer lens-specific ␣B-crystallin promoter activity, we analyzed the ␣B115-CAT and ␣B68-CAT transgenic mice generated above. The transgenic mice were sacrificed and numerous tissues (heart, lung, liver, spleen, kidney, skeletal muscle, brain, and lens) were assayed for CAT activity. The transgenic data are summarized in Table I. Average CAT activities for ␣B115-CAT, ␣B68-CAT, and ␣BLSR1(Mu-9760)-CAT in various tissues are shown in Fig. 2A. All the mice from multiple independent lines (different founders) containing the ␣B115-cat transgene expressed CAT activity exclusively in the lenses at significantly higher level than the background CAT activity observed in the lenses of nontransgenic mice (see Table I and Fig. 2A). In contrast to the results obtained with the ␣B115-cat transgene, the mice containing the ␣B68-cat transgene had no CAT activity in the lens or any other tissue examined (Table I and Fig. 2A) except for mouse #55 in which CAT activity in the lenses was above the background. The low lens activity of the ␣B68-cat transgene in mouse #55 may be due to its different integration site, which may have reconstructed an adequate binding site for a positive transcription factor. Unexpectedly, ␣BLSR1(Mu-9760)-cat transgene containing mice did not exhibit CAT activity in any of the tissues analyzed, including the lens (Table I and Fig. 2A). With LSR2 still intact in the ␣BLSR1(Mu-9760)-CAT construct, a low CAT activity was expected in the lenses of the transgenic mice. Several possibilities to explain this observation are discussed below. Fig. 2B depicts the average CAT activities in lenses of all the transgenic mice for the present and earlier experiments carrying the ␣B426-cat (6), ␣B164-cat (6), ␣B115-cat, and ␣B68-cat transgenes. The transgenic mice containing the ␣B426-cat transgene (which includes the muscle-preferred enhancer) had an average of seven times more CAT activity in the lens than the transgenic mice containing the ␣B164-cat transgene (which lacks the enhancer). For the ␣B115-cat transgene (which lacks both enhancer and LSR1) there was an additional 30-fold decrease in average CAT activity in the lens. Finally there was no CAT activity in the lenses of the transgenic mice harboring the ␣B68-cat transgene. These data indicate that the sequences downstream of Ϫ115 contribute to the lens expression of the ␣B-crystallin gene. Sequences downstream of Ϫ115 contain a motif highly conserved in the human, mouse, and duck ␣Bcrystallin promoter, previously called Block 2 (37), which we rename here LSR2 for lens-specific region 2. Taken together, these experiments demonstrate that the minimum lens-specific promoter for the ␣B-crystallin gene is sequence Ϫ115 to ϩ44 and that the Ϫ164/Ϫ115 sequence enhances activity of the ␣B promoter in a lens-specific fashion.
DNase I Footprinting with Lens Nuclear Extract and Pax-6 PD-We next performed DNase I footprinting experiments in order to localize lens nuclear protein binding sites in the ␣Bpromoter. A DNase I footprint obtained with a partially puri- fied nuclear extract from the ␣TN4-1 lens cells (28) or adult mouse lenses is shown in Fig. 3. The results showed that the sequences between positions Ϫ147 and Ϫ118 (LSR1), Ϫ72 and Ϫ51 (LSR2) and Ϫ30 and Ϫ7 of the ␣B-crystallin 5Ј-flanking region were protected from DNase I digestion (Fig. 3). Since we noted sequence similarities between the footprinted regions and binding sites for Pax-6 (23), we examined the possibility that the paired domain of Pax-6 (PD5) can bind to the DNase I protected regions. A DNase I protection assay of the Ϫ190/ϩ40 fragment revealed that the regions covering LSR1 (Ϫ160/ Ϫ131), LSR2 (Ϫ97/Ϫ27), and a sequence in exon 1 (ϩ2/ϩ29) on the upper (sense) strand were footprinted by PD5 (Fig. 4). Positions Ϫ164 to Ϫ120 (LSR1), Ϫ72 to Ϫ28 (LSR2) and ϩ2 to ϩ27 (in exon 1) of the lower (antisense) strand were also protected from DNase I digestion by PD5. There were some differences in the footprinted regions obtained with PD5 and its alternatively spliced form, PD5a (PD of Pax-6 that contains 14 extra amino acids) (12). PD5a did not create a DNase I footprint on the upper (sense) strand of LSR2. On the lower (antisense) strand, a more extended footprint covering the sequence between Ϫ106 and Ϫ28 was obtained for PD5a, consistent with more than one binding site for this form of Pax-6. In a control test, GST-fused Prox-1 (recombinant protein) did not create a footprint in either the LSR1 or the LSR2 regions (data not shown). The Pax-6 footprinted sequences for upper and lower strands and the surrounding nucleotides are shown in Fig. 5A. Regions that resemble the consensus binding sequence for the paired-domain of Pax-6 (PD5) and PD5a are shown in Fig. 5B; the uppercase nucleotides conform with the Pax-6 consensus binding site and the lowercase nucleotides deviate from the consensus binding site.
EMSA Using Anti-Pax-6 Antibody and Competitor-Oligodeoxynucleotides-In order to investigate the possibility that the lens nuclear proteins that bind to LSR1 of the ␣B-crystallin promoter have antigenic similarity to Pax-6, EMSAs were per-formed using the Ϫ152/Ϫ121 oligodeoxynucleotide and partially purified nuclear extracts prepared from ␣TN4-1 cells and adult mouse lenses. A complex was formed between the probe and the lens nuclear proteins (Fig. 6A, lane 2, arrow), which was strongly reduced in the presence of anti-Pax-6 antibody (Fig. 6A, lane 3). A gel shift complex was also formed with the LSR1 (Ϫ152/Ϫ121) oligodeoxynucleotide when using a nuclear extract from ␣TN4-1 lens cells (Fig. 6B, lane 1), which was diminished in the presence of anti-Pax-6 antibody (data not shown). Complex formation using the ␣TN4-1 nuclear extract was abolished by competition with self-oligodeoxynucleotide (Fig. 6B, lane 2). In addition, oligodeoxynucleotide 9718/9719 containing the consensus PD-binding site for Pax-6 (38) competed for complex formation (Fig. 6B, lane 3). Oligodeoxynucleotides H2B2.2 (containing a Pax-5 binding site that has been shown to interact with Pax-6 (38)), 9922/9923 (comprising the chicken ␣A-crystallin Ϫ60/Ϫ27 sequence that has a Pax-6 binding site) (19) and 9708/9709 (comprising the chicken ␣A-crystallin Ϫ177/Ϫ130 sequence) all competed for complex formation although to a lesser extent (Fig. 6B, lanes 4 -6,  respectively). These data suggest that the lens nuclear protein, which binds to ␣B-crystallin promoter via LSR1, is Pax-6 or an antigenically related protein.
A double-stranded radiolabeled LSR2 oligodeoxynucleotide comprising the Ϫ78/Ϫ28 sequence of the ␣B-crystallin promoter produced several complexes when incubated with N/N1003A and ␣TN4-1 nuclear extracts (Fig. 6C, lanes 2 and 9,  respectively). These complexes were almost abolished by self competition (Fig. 6C, lanes 3 and 8) and by cross-competition with the LSR1 oligodeoxynucleotide (Fig. 6C, lanes 4 and 7). The formation of several specific complexes with the LSR2 oligodeoxynucleotide is consistent with the presence of multiple Pax-6 binding sites within LSR1. Preincubation of N/N1003A and ␣TN4-1 nuclear extracts with anti-Pax-6 antibody significantly reduced the formation of the complexes (Fig. 6C, lanes 5   FIG. 5. A, summary of DNase I footprinting for upper (sense) and lower (antisense) strands. Footprints for PD5 are shown in the hatched boxes and footprints for PD5a are shown as open boxes. B, alignment of Pax-6 recognition sequences in mouse ␣B-crystallin promoter. Uppercase letters indicate matching of the nucleotides with the consensus binding site for Pax-6. and 10). In a control test, preincubation of N/N1003A lens nuclear extract with anti-E12 antibody (from Pharmingen, San Diego, CA) did not significantly affect complex formation (Fig.  6C, lane 6). GST fused prox-1 (recombinant protein) did not complex with either LSR1 or LSR2 oligodeoxynucleotides (data not shown), confirming the specificity for binding of Pax-6 to these regions.
Competition EMSAs-Competition EMSAs were performed to test whether the inability of Pax-6 to stimulate activity of the mutated ␣B promoters correlates with a corresponding decrease in the ability of the mutated promoters to bind lens nuclear factors. The wild type double-stranded LSR1 oligodeoxynucleotide (Ϫ152/Ϫ121) produced a complex when incubated with the ␣TN4-1 nuclear extract (Fig. 8A, lane 1, arrow), which was greatly reduced by self-competition (Fig. 8A, lane 2). The mutated LSR1 sequences did not compete to the same extent as the unlabeled wild type LSR1 oligodeoxynucleotide for complex formation with the wild type oligodeoxynucleotide (Fig. 8A, compare lanes 3 and 4 with lane 2). Complex formation was also reduced significantly by competition with LSR2 oligodeoxynucleotide (Ϫ78/Ϫ28) (Fig. 8B, lane 2). These results provide a positive correlation between protein-DNA complex formation and functional activity of LSR1 and LSR2 within the ␣B-crystallin promoter. DISCUSSION Our recent experiments have shown that ␣B-crystallin gene sequences between Ϫ164 and ϩ44 are sufficient to promote lens-specific expression of a heterologous reporter gene in transgenic mice (6). In addition, we have previously identified five distal cis-acting regulatory elements (␣BE-1, ␣BE-2, ␣BE-3, MRF, and ␣BE-4) located within the muscle preferred enhancer (Ϫ427/Ϫ259) of the ␣B-crystallin gene (5). ␣BE-1, ␣BE-2, and ␣BE-3 are utilized for ␣B-crystallin enhancer activity in lens cells and in other tissues (32); MRF is essential for enhancer activity in heart and skeletal muscle, and ␣BE-4 is utilized selectively for enhancer activity in heart (36). Activity of fragment Ϫ164 to ϩ44 in the lens depends at least on a regulatory region LSR1 (Ϫ147/Ϫ118), which is physically separated from the distal enhancer control elements required for expression in non-lenticular tissues.
The present transgenic mouse experiments demonstrate that even in the absence of LSR1 (for ␣B115-CAT construct) CAT activity still occurs specifically in lens, while further deletion to Ϫ68 of the ␣B promoter (for ␣B68-CAT construct) abolishes CAT expression in the lens. Thus, the sequence Ϫ115 to ϩ44 is sufficient to generate lens-specific activity of the ␣B-crystallin promoter. The overall levels of CAT activity were approximately 30-fold lower in the lenses of the ␣B115-CAT lines than in the lenses of the transgenic mice harboring the ␣B164-cat transgene (6), indicating that LSR1 increases ␣Bcrystallin promoter activity in the lens. In the present studies using DNase I footprinting, transient transfection, and transgenic mice we have identified a lens-specific regulatory region (LSR2) between positions Ϫ78 and Ϫ46 of the ␣B-crystallin promoter. LSR2 is located in a highly conserved region of the mouse ␣B-crystallin promoter (37), and truncation of this site eliminates promoter activity in transgenic mice. Mutations in LSR1 also strongly reduce or eliminate promoter activity in transfected lens cells (6) and transgenic mice. Taken together, these experiments indicate that LSR2 is sufficient for lens expression of ␣B-crystallin, but for high lens expression of the ␣B-crystallin gene, coupling between the enhancer elements (␣BE-1, ␣BE-2, ␣BE-3), and lens-specific cis-regulatory regions (LSR1 and LSR2) is required. This coupling may be facilitated by the ability of Pax proteins to bend DNA (40).
Unexpectedly, the construct lacking LSR1 (␣B115-CAT) exhibited CAT activity in the lenses of the transgenic mice, while the construct containing a mutation in LSR1, but with an intact LSR2, was not active in the lens. Several possibilities can be considered to explain this apparent inconsistency. First, it is possible that a repressor binding site was created in the LSR1-mutated construct that resulted in the binding of a transcription inhibitor. Second, there may be a transcription inhib- The location of the mutations (TCTAGA, an XbaI site) are shown below the sequence. B, relative CAT levels in COP-8 cells cotransfected with pKW10-Pax-6 (wild type Pax-6 cDNA) and either p65-7 (wild type Ϫ164/ϩ44 ␣B-crystallin promoter fragment fused to cat gene) or the Ϫ164/ϩ44 promoter fragment containing the Mu-9760 or Mu-9761 mutations indicated in A. The CAT levels expressed are relative to that obtained by parallel cells cotransfected with pKW10-Pax-6 and pRD30A (the promoterless parent vector) (5). C, relative CAT levels in COP-8 cells cotransfected with pKW10-Pax-6 and p11-3 (wild type Ϫ115/ϩ44 ␣B-crystallin promoter fragment fused to the cat gene) or the truncated Pax-6 cDNA constructs (p309C, p310A, and p312A, see "Experimental Procedures"). CAT levels are relative to parallel tests using pRD30A. Cells were harvested 48 h after DNA removal; CAT activity was determined by the biphasic assay (28) and normalized with respect to the activity of ␤-galactosidase, which resulted from cotransfection of pCH110 (see "Experimental Procedures"). itor binding site present between the sequences Ϫ164 and Ϫ115, normally deactivated by LSR1, which was deleted in the smaller construct ␣B115-CAT. Third, in addition to Pax-6, LSR1 may bind to another factor that acts as an activator when Pax-6 is also bound to LSR1 and as a repressor in the absence of Pax-6, making LSR1 a composite regulatory element (41). Further experiments are required in order to distinguish among these or other possibilities.
The present data add to the possibility that the use of Pax-6 for crystallin transcription may be a conserved mechanism to regulate gene expression in the lens (23). Lens development is a multistep complex process that starts at gastrulation and the major inductive events that determine the ectoderm to become lens occur during neurulation (42). Pax-6 is expressed in the presumptive lens cells of the ectoderm overlying the outgrowing optic vesicle (9), prior to the beginning of mouse lens formation and expression of the crystallins. The expression of Pax-6 in lens continues during development well after ␣Bcrystallin expression commences (9,43). In addition to the ␣B-crystallin gene, Pax-6 has been shown to bind and activate the mouse and chicken ␣A (20,19), chicken ␦1 (21) and guinea pig (22) crystallin gene regulatory regions. The exact mechanism of action of Pax-6 in mouse ␣B-crystallin expression is still unknown. One cannot exclude the possibility that in addition to Pax-6 other factors may be used for lens-specific expression of ␣B-crystallin (via LSR1 and LSR2) such as SOX proteins (44 -46) and retinoic acid receptors (47)(48)(49). Indeed both SOX-2 (47) and Pax-6 (21) appear to interact to activate the ␦1-crystallin enhancer in the lens, and both SOX and retinoic acid receptors (50,51) are used for lens-specific activity of the murine ␥F-crystallin promoter. Our preliminary experiments suggest that retinoic acid receptors may also contribute to the regulation of the ␣B-crystallin promoter in the lens. Clearly further experiments are necessary to delineate the spectrum of transcription factors that are used for activating the ␣B-crystallin gene in the lens and to determine whether Pax-6 is used for expression of the multifunctional ␣B-crystallin/small heat shock protein gene in non-lens tissues or under conditions of physiological stress (52).