Ectopic Pax6 Expression Disturbs Lens Fiber Cell Differentiation

Abstract

purpose. Pax6 is a transcription factor necessary for the specification and subsequent formation of the ocular lens. It is expressed in all lens cells at early stages of development. After lens formation, Pax6 expression is maintained in the lens epithelium, whereas its level abruptly decreases in differentiated fiber cells. This study is to test the hypothesis that normal fiber cell differentiation would be perturbed by sustained Pax6 expression.

methods. Transgenic mice expressing the canonical form of mouse Pax6 were created under the control of a modified mouse αA-crystallin promoter. The phenotypic changes in the transgenic lens were analyzed by light and electron microscopy. The effect of ectopic Pax6 expression on the lens fiber cells was investigated by in situ hybridization, immunohistochemical staining, real-time reverse transcriptase–polymerase chain reaction (RT-PCR), and two-dimensional (2-D) gel electrophoresis.

results. Transgenic mice from seven different lines all had cataracts with severity that correlated with the transgene expression level in lens fiber cells. In severely affected lines, a lumen was present between the apical surfaces of the epithelial and fiber cells, suggesting that secondary fiber cell elongation is incomplete. Electron microscopy analysis showed that the ball-and-socket interdigitations between neighboring fiber cells were underdeveloped or attenuated in the transgenic lens. Most interesting, elevated levels of Pax6 in fiber cells reduced the protein levels of transcription factor cMaf, which is known to be essential in fiber cell differentiation. Furthermore, the total amount of lens proteins was 60% less than normal in the Pax6 transgenic lens. Among the crystallins examined, the relative ratio of intact βB1-crystallin protein to total lens protein was significantly reduced. Real-time reverse transcriptase PCR showed that the ratio of βB1-crystallin transcript levels to total mRNA levels were reduced by 87%.

conclusions. The data demonstrate that high levels of Pax6 expression disrupt normal fiber cell differentiation and maturation.

The ocular lens has long been used as a model to study molecular mechanisms during development. ¹ Morphologic lens development begins with the head ectoderm responding to inductive signals from the underlying optic vesicle to form the lens placode. ² The lens placode subsequently invaginates to form the lens pit, which detaches from the surface ectoderm to form the lens vesicle. The lens vesicle develops into the lens when the anterior cells remain proliferative and form the lens epithelium, and the posterior cells exit from the cell cycle, elongate into lens fiber cells, and fill up the lumen of the lens vesicle. After the basic structure is established, the lens continues to grow throughout life through differentiation of the lens epithelial cells into fiber cells. Around the lens equator, the lens epithelial cells divide at the anterior germinative zone and begin to differentiate into the fiber cells at the transitional zone. Fiber cell differentiation is characterized by withdrawal from the cell cycle, cell elongation, and degradation of intracellular organelles including the nucleus. ³ Lens differentiation is also characterized by expression of fiber cell-specific markers, such as β- and γ-crystallins, as well as the intermediate filament proteins CP49 and filensin. 

Through molecular genetic approaches, several transcription factors critical for lens development have been identified. ^{4 5} Among them is Pax6, a paired/homeodomain–containing transcription factor that is a key regulator of lens determination and morphogenesis. ^{6 7 8} Pax6 expression is essential for lens formation, ^{9 10} and ectopic expression of Pax6 is sufficient to induce lenses in the absence of neural tissues. ¹¹ Pax6 is normally expressed in the head ectoderm overlying the optic vesicle and is continuously present in the lens placode, pit, and vesicle as development progresses. ¹² After lens formation, Pax6 expression becomes progressively restricted to the anterior lens epithelium and eventually disappears from the lens fiber cells. The functional significance of Pax6 downregulation in lens fiber cells is not well understood. Studies have indicated that Pax6 plays a dual role in lens crystallin expression. It activates several genes expressed in adult lens epithelial cells including αA-, αB-, δ-, and ζ-crystallin, ^{13 14 15 16} whereas it represses the expression of fiber cell specific βB1-crystallin. ¹⁷ These results imply that Pax6 is important for maintaining the lens epithelial phenotype and that loss of Pax6 is necessary for normal fiber cell differentiation. 

Normal lens development not only depends on the temporal and spatial expression of Pax6, but also on the gene dosage. ^{18 19 20 21 22} Furthermore, there are two Pax6 splice variants in the lens, Pax6 and Pax6(5a). ^{23 24} They have different but overlapping DNA-binding specificities ²³ and functions. ²⁵ Both isoforms are present in equal amounts in human lenses, ²⁶ but Pax6 is the predominant form in mouse and bovine lenses. ^{16 27} We have demonstrated that ectopic expression of Pax6(5a) in lens fiber cells results in cataracts associated with the upregulation of α5-integrin, β1-integrin, paxillin, and p120 Src substrate expression. ²⁸ Because Pax6, not Pax6(5a), is the predominant form of Pax6 in the mouse lens, we generated transgenic mice that overexpress the canonical Pax6 in the lens fiber cells to test our hypothesis that loss of Pax6 activity is essential for normal fiber cell differentiation. In this study, we investigated the phenotypic and molecular changes in lenses with ectopic Pax6 expression in lens fiber cells. We demonstrated for the first time that elevated levels of Pax6 in lens fiber cells reduce the protein levels of cMaf transcription factor and inhibit fiber cell terminal differentiation and maturation. 

Materials and Methods

Production of Transgenic Mice

The αA-crystallin/Pax6 cDNA construct (Fig. 1A) was made by cloning the full-length mouse Pax6 cDNA into the BclI/HindIII site of the pACP3 vector which contains the mouse αA-crystallin promoter (−342/+49), the SV40 small T-antigen intron, and the polyadenylation signal. ^{22 28} The Pax6 minigene (Fig. 1A) was released from the resultant plasmid by NotI digestion, purified by Glass Milk (PlantMedia, Dublin, OH), and microinjected into the pronuclear stage FVB/N embryos. Potential transgenic founder mice were screened by polymerase chain reaction (PCR), using tail genomic DNA and primers hybridizing to the SV40 region. ²⁹  

The modified αA-crystallin promoter/Pax6 cDNA construct (Fig. 1B) was made as follows: a BglII site was inserted into the αA-crystallin promoter (−368/−340 fused to −287/+45) between nucleotides (nt) −83 and −82 by overlapping PCR. Two copies of a Pax6 consensus binding site were inserted in opposite orientations at the BglII site and resulted in the following sequence (−83, 5′-ATCCA-TCACTCATGCGTGAAGATGGATCC ATCTTCACGCATGAGTGACTG GATCTAG-3′, −82). The forward orientation of the Pax6 consensus site is underscored, a single nucleotide deletion between A and T is denoted by a dash, and the BglII/BamHI restriction sites for cloning are italic. This αA-crystallin promoter with the additional Pax6 binding sites ³⁰ was then used and substituted for the αA-crystallin promoter in pCPV6 which contains a rabbit β-globin intron and human growth hormone polyadenylation sequences (hGH pA). ³¹ The mouse Pax6 cDNA, gift from Patrick Callaerts (University of Houston, TX), was inserted downstream of the β-globin intron and upstream of the hGH pA sequences. The resultant plasmid was digested with SacI to release the microinjection fragment (the Pax6 minigene). After purification with a gel extraction kit (Qiaex; Qiagen, Valencia, CA), the DNA fragment was microinjected into the pronuclei of FVB/N fertilized eggs. ³² Transgenic founder mice were screened by PCR using primers derived from the αA-crystallin promoter (Pr4, 5′-GCATTCCAGCTGCTGACGGT-3′) and the β-globin intron (β3, 5′-AAGGCATGAACATGGTTAGCAGAGG-3′). All experiments using animals were approved by the University of Missouri-Columbia institutional review board and conformed with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Histologic Analysis

Mouse embryo heads or postnatal eyes were removed and fixed overnight in 10% neutral buffered formalin. Tissues were rinsed briefly in PBS and transferred to 70% ethanol. After dehydration through a graded ethanol series and clarification in xylene, tissues were embedded in paraffin and sectioned serially at 5 μm thickness. Sections were stained with hematoxylin and eosin (H&E) for standard histology or used for in situ hybridization and immunohistochemistry. 

In Situ Hybridization

Bluescript-based (Stratagene, La Jolla, CA) plasmids containing either the mouse partial Pax6 or cMaf cDNA were gifts of Kathleen Mahon (Baylor College of Medicine, Houston, TX) and Brian Ring and Gregory Barsh (Stanford University, Sanford, CA), respectively. The plasmids were linearized by restriction enzyme digestion. Sense or antisense riboprobes were generated with an in vitro transcription kit (Promega, Madison, WI) with either T3 or T7 RNA polymerase. In situ hybridization was performed according to a procedure described previously. ³³ The hybridized slides were coated with emulsion (NTB-2; Eastman Kodak, Rochester, NY), dried, and exposed for 5 days at 4°C before development. The slides were counterstained briefly with hematoxylin and photographed by bright- and dark-field microscopy. 

Immunohistochemistry

For cMaf immunohistochemical staining (see Figs. 6B 7 ), formalin-fixed paraffin sections were dewaxed with xylene and rehydrated through a degraded ethanol series. After blocking in 5% normal horse serum for 1 hour, sections were incubated with anti-cMaf (SC-7866; Santa Cruz Biotechnology, Santa Cruz, CA) primary antibody (1:300 dilution) in PBS containing 2% serum at 4°C overnight. After they were washed with PBS, sections were incubated with rhodamine-conjugated secondary antibody (1:200 dilution) for 1 hour at room temperature. Fluorescent signals were visualized on a microscope (DMR; Leica, Deerfield, IL), and images were captured by a charge-coupled device (CCD) camera. For the Pax6-cMaf double-labeling experiment shown in Figure 7 , after the cMaf immunostaining, sections were boiled in a microwave in 10 mM sodium citrate buffer (pH 6.0) for 10 minutes to unmask the Pax6 antigen. The cMaf fluorescent signal was not affected by this treatment as monitored by fluorescence microscopy. After blocking in 5% normal horse serum for 1 hour, sections were incubated with an anti-Pax6 primary antibody (Biomeda, Foster City, CA) with 1:300 dilution in PBS with 2% serum at 4°C overnight. Control slides were incubated with 2% serum in PBS without the anti-Pax6 antibody. After washing with PBS, sections were incubated with Alexa Fluor 488–conjugated secondary antibody (Molecular Probes, Eugene, OR) with 1:200 dilution for 1 hour at room temperature. Slides were covered with mounting solution mixed with 4,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St. Louis, MO) to label cell nuclei. Fluorescence signals were visualized with the microscope (DMR; Leica), and images were captured by a CCD camera. 

For cMaf and Sox2 immunofluorescence labeling (see Fig. 8 ), fresh eyes were embedded in tissue freezing medium (Triangle Biomedical, Durham, NC), and 16-μm frozen sections were prepared. A more detailed protocol has been published. ³⁴ Tissue sections were fixed in 1:1 acetone/methanol at −20°C for 20 minutes and then air dried. After blocking in 1% bovine serum albumin in PBS for 1 hour, sections were incubated with a primary antibody against cMaf or Sox2 (Chemicon International, Temecula, CA) in PBS for 1 hour at room temperature. After washing with PBS, sections were incubated with secondary antibody conjugated with AlexaFluor 568 (Molecular Probes) at a 1:50 dilution for 1 hour. Cell nuclei in Figures 8A and 8B were counterstained with nuclear stain (1:1000 dilution, Syto13; Molecular Probes). Immunofluorescence signals were visualized on a confocal microscope (510 LSM; Carl Zeiss Meditec, Oberkochen, Germany). 

Transmission Electron Microscopy

For ultrastructural analysis, 4-day-old lenses were dissected from the eye and fixed in 2% paraformaldehyde and 2% glutaraldehyde in 0.13 M Na-cacodylate and 0.13 mM CaCl₂ (pH 7.4) for at least 2 hours at room temperature. Samples were washed with cacodylate buffer, postfixed in 1% osmium tetroxide in 0.1 M cacodylate buffer for 1 hour, dehydrated through a graded ethanol series, and embedded in epoxy resin. Thin sagittal sections were cut through at least one third of the lens and stained with uranyl acetate and lead citrate. Sections were examined and photographed with a transmission electron microscope (1200EX; JEOL Tokyo, Japan). 

Two-Dimensional Gel Electrophoresis

One-week-old (P7) lenses were isolated from wild-type or LR2 transgenic mice and frozen immediately in dry ice. Water-soluble lens proteins were prepared and protein concentration determined using the bicinchoninic acid (BCA) method. Lens proteins were first separated by isoelectric focusing on an immobilized-pH gradient gel strip (pH 5–9), and then by size on a 12% polyacrylamide gel. The gel was stained with protein dye (Sypro Red; Molecular Probes), and the identities of the protein spots were compared with standardized proteome maps of the FVB/N mouse lens. ³⁵ The relative amount of each polypeptide was quantitated on scanned gels (Delta2D software, ver. 3.1; Decodon, Greifswald, Germany). 

Quantitation of βB1-Crystallin Transcript Levels

Real-time RT-PCR analysis of βB1-crystallin transcript levels was completed as described previously. ³⁶ Briefly, total RNA was isolated from lenses using a commercial system (SV Total RNA Isolation System; Promega, Madison, WI). The RNA was reverse transcribed into cDNA using mixed reverse transcriptases (Omniscript and Sensiscript; Qiagen) and a reverse primer specific for the mouse βB1-crystallin gene (5′-CCT GCT CGA AGA CGA TCA GCC-3′). Then, the reverse transcriptase reaction was used for real-time PCR, by using a kit (QuantiTect SYBR Green PCR Kit; Qiagen) and the mouse βB1-crystallin forward primer (5′-CAG GGC CTG ATG GCA AGG GAA-3′) in a thermocycler (iCycler; Bio-Rad Laboratories, Hercules, CA). 

Results

Generation of Transgenic Mice that Overexpress Pax6 in Lens Fiber Cells

In the normal lens, Pax6 expression was downregulated during fiber cell differentiation (Fig. 2 , and see Fig. 7A , WT). To test the importance of this downregulation, we constructed a minigene using the mouse αA-crystallin promoter (−342/+49) to drive the mouse Pax6 cDNA (Fig. 1A) . This promoter has been used in numerous studies to target transgene expression specifically in the lens fiber cells. ^{22 28 29 31 37} We generated three independent founders, but none of them showed any lens defects. RT-PCR analysis indicated that the transgene was not expressed (data not shown). To enhance the αA-crystallin promoter activity through a potential feedback mechanism, we used a modified αA-crystallin promoter containing two additional copies of the Pax6 consensus binding site (Fig. 1B) . ³⁰ Using this modified promoter, we generated seven transgenic founders designated as OVE1076 to -1078 and LR1 to -4. In all of them, cataracts developed with different degrees of severity. Seven independent lines were established and designated as OVE1076 to -1078 and LR1 to -4. 

We analyzed transgene expression patterns by in situ hybridization (Fig. 2) and immunohistochemistry (e.g., Figs. 7A 7D 7G ) in embryonic day (E)18 eyes from lines OVE1076 to -1078. In the normal lens, Pax6 is highly expressed in the lens epithelial cells, and its expression is downregulated in the lens fiber cells (Fig. 2 , WT). In transgenic lenses, Pax6 mRNA was detected in the lens epithelium at qualitatively normal levels with additional ectopic expression in the fiber cells. Among the three lines tested in parallel (OVE1076 to -1078), OVE1078 had the highest level of transgene expression, whereas OVE1077 had the lowest. We estimate, based on the severity of the lens defects, that LR2, a line used for many of our analyses, expressed the transgene at intermediate levels between OVE1078 and -1076. Immunohistochemical staining using an anti-Pax6 antibody suggested that the amount of Pax6 protein correlates with the level of Pax6 mRNA in the different lines (see Figs. 7A 7D 7G ). Notably, the severity of the lenticular defects in different transgenic lines correlated with the Pax6 transgene expression levels. The severest cataracts developed in mice from lines OVE1078 and LR2, and they had smaller than normal lenses with concomitant microphthalmia. Mice from OVE1076 had obvious cataracts and only a slight reduction in lens size, whereas OVE1077 mice exhibited very mild cataracts that were only transiently detectable at weaning ∼P21 (Table 1) . 

Lens Fiber Defects in Pax6 Transgenic Mice

We could not detect any lenticular defects in OVE1077 mice by histology under light microscopy. Representative lens histology for other transgenic lines is shown in Figure 3 . In the normal lens, fiber cell differentiation and maturation is associated with denucleation. As a result, the mature fibers at the lens core do not contain DNA and therefore were free of hematoxylin staining (Fig. 3A) . Lenses from all transgenic lines (except OVE1077) retained hematoxylin-positive materials in the central fiber cells (white arrows in Fig. 3F ), suggesting that denucleation is impaired in the transgenic lens. In all but one line (OVE1077), vacuoles were present in the cortical fiber cells, further suggesting that fiber cell differentiation is abnormal in the transgenic lens. In the severely affected lines OVE1078 and LR2, fiber cell elongation was incomplete, and fiber cells were disorganized. The apical surface (Fig. 3C , arrows) of the secondary fiber cells failed to contact the anterior lens epithelium. To investigate further the lens defects in OVE1078, a developmental analysis was performed on lenses isolated from earlier developmental stages (Fig. 4) . At E15, the architecture of the transgenic lens appeared normal with no obvious defects. At E18, abnormalities in lens fiber cells became apparent. The transgenic lens exhibited vacuoles, fiber cell elongation was incomplete (Figs. 4D 4F , arrows), and denucleation was inhibited. These data suggest that high levels of Pax6 in lens fiber cells interfere with normal fiber differentiation and maturation. 

The structure of the lens fiber cells was further examined by transmission electron microscopy (Fig. 5) . In the normal P4 lens, fiber cells showed ball-and-socket junctions that formed lateral interdigitations with the neighboring cells (Figs. 5A 5B , arrows). In contrast, the fiber cells in the OVE1078 transgenic lens either completely lacked such structures (Fig. 5C) or had significantly fewer interdigitated joints than did the normal lens (Figs. 5D 5E) . Similar defects were also observed in transgenic mice the expressed only the paired and homeodomains of Pax6, ²² suggesting that the inhibitory effect is mediated through these two DNA-binding domains. 

Reduction of cMaf but Not Sox2 Protein Levels in Pax6 Transgenic Lenses

Pax6 is known to interact with other transcription factors to control lens and eye development. ^{8 38} To investigate the inhibitory effect of Pax6 on fiber cell differentiation, we examined the expression of some transcription factors, including cMaf and Sox2, which are known to be important for normal fiber cell differentiation. Figure 6 shows the cMaf mRNA (Fig. 6A) and protein (Fig. 6B) localization in E14 and E18 lenses from OVE1078 mice. In the normal lens, cMaf mRNA levels were upregulated during fiber cell differentiation (Figs. 6A , WT), and a similar expression pattern was seen in the OVE1078 transgenic lens (Fig. 6A , OVE1078). Quantitative RT-PCR also confirmed that there was no significant difference in cMaf mRNA levels between wild-type and OVE1078 transgenic lenses (data not shown). Similarly, we detected high levels of cMaf protein in the fiber nuclei of the wild-type lens by immunohistochemistry (Fig. 6B , WT). However, very little cMaf immunoreactivity was detected in the fiber cells of the OVE1078 transgenic lens at either age (Fig. 6B , OVE1078). Further, cMaf protein levels were also significantly reduced in the LR2 transgenic lens (see Fig. 8B ) suggesting that ectopic Pax6 expression in lens fiber cells reduces cMaf protein levels but does not affect cMaf gene expression. 

To confirm that the reduction of cMaf protein in the transgenic lens results from ectopic expression of Pax6 in the fiber cells, we compared cMaf protein levels with Pax6 levels in lenses from OVE1076, -1077, and -1078 mice (Fig. 7 ; data not shown for OVE1077). In all lines, Pax6 protein levels did not appear altered in the lens epithelium compared with wild type. In the low-expressing OVE1077 mice, little ectopic Pax6 expression was detectable in lens fibers, and cMaf protein levels were similar to those in the wild-type lens (data not shown). In the intermediate-expressing line OVE1076, some fiber cell nuclei had high Pax6 levels, whereas other fibers expressed Pax6 at levels similar to or lower than the lens epithelium. Notably, double labeling for Pax6 and cMaf proteins revealed that nuclei containing high levels of Pax6 protein had low levels of cMaf (Figs. 7D 7E 7F , representative nuclei, arrows), whereas cMaf protein levels were close to normal in cells with low levels of Pax6 (representative nuclei; arrowheads). In the high-expressing OVE1078 line, a significant number of fiber cell nuclei stained positively for Pax6 protein, and few to no cMaf-positive nuclei were detected (Figs. 7G 7H 7I) . This suggests that Pax6 negatively regulates cMaf protein levels. 

We then investigated whether Pax6 also inhibits the expression of other transcription factors such as Sox2 in the LR2 transgenic lens (Fig. 8) . In contrast to the situation with cMaf (Figs. 8A 8B) , Sox2 protein levels were unaffected in the LR2 transgenic lens (Fig. 8D) . We also examined the Sox2 mRNA and protein levels in E18 wild-type and OVE1078 lenses by in situ hybridization and immunostaining and again detected no differences in either expression pattern or levels (data not shown). 

Reduction of βB1-Crystallin Expression Levels in Transgenic Lenses

Pax6, cMaf, and Sox proteins all have the ability to regulate the expression of crystallins, the major structural proteins of the lens. ^{8 39 40 41 42 43} Because the expression levels of Pax6 and cMaf are altered in the transgenic lens, particularly in the high-expressing lines OVE1078 and LR2, we decided to analyze the crystallin protein profile in the P7 wild-type and LR2 transgenic lens by 2D gel electrophoresis (Fig. 9) . In the LR2 transgenic lens, the total amount of protein was approximately 60% less than that in the wild-type lens, which is consistent with the smaller lens (microphthalmia) phenotype. In the transgenic lens, the relative ratios of intact αA-, αB-, βA2-, βA4-, γA-, γB/C-, γD-, γE/F-, and γS-crystallins to the total amount of protein were close to normal, whereas the levels of intact βB2- and βB3-crystallin were slightly reduced. In addition, the levels of intact βΒ1-crystallin were significantly reduced (approximately 40% to 50% less than normal) although approximately 40% of this decrease was due to the proteolytic processing of βB1-crystallin. Notably, the transgenic lens also contain elevated levels of the proteolytic products of αA-crystallin (Fig. 9 , αAΔ) and βA3-crystallin (Fig. 9 , βA3-11). These crystallin proteolytic products are usually seen only at appreciable levels in adult normal lenses. ³⁵  

Because βB1-crystallin protein levels were decreased in the transgenic lens, we quantitated βB1-crystallin transcript levels in P7 LR2 transgenic lenses by real-time RT-PCR. Notably, βB1-crystallin levels in the LR2 transgenic lens were reduced by more than 80% (87% ± 13% P = 0.003) compared with wild-type lenses. 

Discussion

At early stages of lens development, Pax6 is expressed in head ectoderm overlying the optic vesicle and then in the primitive lens as it progresses through the lens placode, cup, and vesicle stage. ¹² Previous studies have established that Pax6 expression is necessary and, in some cases, sufficient for lens precursor cell specification and lens induction. ^{6 11} As lens development proceeds, Pax6 expression is downregulated in differentiated fiber cells and becomes restricted in the lens epithelium. We have shown that Pax6 negatively regulates βB1-crystallin expression. ¹⁷ To test our hypothesis that normal fiber cell differentiation and maturation requires downregulation of Pax6, we generated seven transgenic lines that overexpress Pax6 in the fiber cells. Lens defects, with different degrees of severity, developed in all the transgenic lines. In the severely affected lines (OVE1078 and LR2), elongation of the secondary fiber cells was incomplete, leaving a lumen beneath the anterior lens epithelium. Electron microscopy analysis showed that the specialized membrane structures—the ball-and-socket interdigitations among the neighboring fiber cells—were underdeveloped or attenuated in the transgenic lenses. Similar defects were also observed in transgenic lenses that expressed only the paired domain and the homeodomain of Pax6. ²² Most interesting was the finding that elevated levels of Pax6 in the lens fiber cells reduced the protein levels of cMaf, which is an important transcription factor in fiber cell differentiation. Our study implies that normal lens development depends on the proper levels of Pax6 in each lens cell type (epithelial and fiber), and loss of Pax6 expression is essential for fiber cell differentiation and maturation. 

cMaf belongs to the family of basic-leucine zipper (bZip) transcription factors. ⁴⁴ Loss of cMaf results in defects in lens fiber cell differentiation. ^{39 40 43} During early lens induction and specification, Pax6 is essential for activating and maintaining the expression of cMaf in the lens. ^{45 46} After the lens is formed, cMaf is upregulated in fibers, whereas Pax6 levels decline. The functional significance of this reciprocal expression profile is unclear. In our transgenic study, we demonstrate that higher levels of Pax6 in the lens fiber cells sharply reduces the amount of cMaf protein. Therefore, although Pax6 activates the basal expression of cMaf during early stages of normal lens development ⁴⁵ and Pax6 can directly activate the cMaf promoter in cotransfections, ⁴⁶ we propose that downregulation of Pax6 activity in lens fiber cells at later stages of lens development is essential for upregulating the cMaf protein levels during fiber cell differentiation. Using the OVE1078 mice we generated, Goudreau et al. ⁴⁷ showed that ectopic expression of Pax6 can activate Six3 expression. Taken together, our data imply that during late stages of lens development, Pax6 and Six3 expression mutually regulate each other, whereas cMaf protein levels are negatively regulated by Pax6. In our study, Sox2 levels were not affected by Pax6. Because the lens expresses three Sox genes (Sox1, -2, and -3) with overlapping but different expression patterns, ⁴⁸ the effect of Pax6 overexpression on Sox1 and -3 should be investigated. 

How Pax6 reduces cMaf protein level is still unclear. There are two possibilities: Pax6 either inhibits the synthesis of cMaf proteins or reduces its stability. Based on a recent publication, ⁴⁹ it is more likely that overexpression of Pax6 reduces cMaf protein stability in lens fiber cells. Previously Goudreau et al. ⁴⁷ found that active ERK levels are elevated in the fiber cells of the OVE1078 lens, associated with abnormal expression of PDGFα-R (platelet-derived growth factor receptor α). Further, Ochi et al. ⁴⁹ demonstrated that cMaf protein degradation is regulated by ERK phosphorylation. Thus, we speculate that the reduction of cMaf protein levels in the Pax6 transgenic lens, results in part from an increase in active ERK levels in the lens fiber cells, probably through abnormal expression and activation of PDGFα-R in these cells. Whether PDGFα-R expression is directly regulated by Pax6 remains to be investigated. 

The varied effects of ectopic Pax6 expression on lens fiber cell biology are consistent with previously studies showing that eye development and function is highly dependent on the precise regulation of Pax6 levels. Transgenic mice harboring a YAC consisting of a large portion of the known Pax6 locus exhibit malformation of the iris and ciliary body with occasional animals exhibiting severe microphthalmia with associated lens and retinal abnormalities. ⁵⁰ Although the lenses of most of the animals harboring the Pax6 YAC were normal, it is difficult to assess the relevance of this result to the present study, because the Pax6 YAC mice would presumably not express additional Pax6 in lens fiber cells since gene expression was controlled by the endogenous Pax6 promoter. Mice heterozygous for mutations in the Pax6 gene have defects in the iris ⁵¹ and cornea ^{52 53} similar to those in the human disease aniridia. Lenses from mice heterozygous for Pax6 mutations have anterior subcapsular cataracts ²² that appear to arise from localized epithelial-to-mesenchymal transitions. It is intriguing to note that, in newborn mice chimeric for wild-type Pax6 and heterozygous Pax6 loss-of-function cells, only the cells with wild-type Pax6 are found in the lens epithelium, because cells heterozygous for the Pax6 mutant gene preferentially undergo fiber cell differentiation between E12.5 and E16.5. ¹⁸ In light of the present data, it is possible that the cMaf levels are abnormally upregulated in the cells heterozygous for a Pax6 mutation, and this Pax6 haploinsufficiency forces these cells to differentiate preferentially into the fiber cells when exposed to the appropriate environment. Thus, Pax6 expression in the lens epithelium may both maintain the commitment to an epithelial phenotype and prevent premature differentiation into fiber cells during late-stage lens growth and development. 

Pax6 overexpression in lens fiber cells results in multiple defects in lens morphology and structure. Some of the defects are directly or indirectly related to the disruption of coordinated levels of Pax6 and cMaf proteins in the lens. (1) Secondary fiber cell elongation is not complete in the Pax6 transgenic lens, leaving a lumen beneath the anterior lens epithelium. This result is consistent with the observation that loss of cMaf results in the failure of fiber cell elongation in gene knockout mice. ^{39 40 43} (2) cMaf activity is essential for the expression of the major fiber cell proteins β- and γ-crystallin, and to some extent, it is also important for αA-crystallin expression. ^{39 40} In the severely affected Pax6 transgenic mice, the total amount of proteins was approximately 60% lower than normal. Because crystallins make up approximately 90% of the total lens protein, we assume the levels of crystallins are reduced to the similar extent. The decrease in total crystallin proteins could directly result from the low levels of cMaf in the Pax6 transgenic lens. Furthermore, elevated levels of Pax6 in the fiber cells may inhibit β-crystallin expression as shown previously in cotransfections. ¹⁷ Indeed, the relative ratios of βB1-crystallin mRNA and protein are significantly reduced in the Pax6 transgenic lens. This result coincides with the most recent finding by Cui et al. ⁵⁴ that Pax6 can block cMaf-mediated transactivation of the chicken βB1-crystallin promoter in cotransfection experiments. 

One notable feature of the present study is that obvious phenotypic changes were not seen in transgenic lenses until E18, although the modified αA-crystallin promoter used in this study are active by E11. ³⁰ This is similar to mice that overexpress Pax6(5a) in lens fibers that do not develop major lens abnormalities until after birth although the transgene is active by E12.5. ²⁸ Because it has been observed that appreciable Pax6 expression is found in all lens cells shortly after the lens forms, whereas it is lost from lens fibers later in development, ¹⁰ it is probable that the late onset of these phenotypes reflects a late susceptibility to Pax6 overexpression in lens fibers. This may be at least part of the mechanism controlling developmental changes in crystallin gene expression in lens fiber cells. ^{35 55} However, the onset of the morphologic alterations is also likely to be influenced by the observation that not all lens fiber cell nuclei in transgenic lenses contain the same amount of Pax6 protein, possibly due to a complex feedback loop between the promoter used in these studies and Pax6. Further, although Pax6 is a transcription factor whose effects should be cell autonomous, the need for lens fibers to form ball-and-socket joints and extensive gap-junctional communication with their neighbors can result in extensive abnormalities, even in lenses containing only a few mutant cells. ^{56 57}  

Microarray analysis on P7 wild-type and LR2 transgenic lenses has revealed more than 500 differentially expressed genes. ⁵⁸ Because our present data suggest that Pax6 can regulate gene activity at the protein level as well as the transcription level, the full spectrum of Pax6 effects on the lens are likely to be extremely complex. Further, there are two Pax6 isoforms, Pax6 and Pax6(5a), that result from alternative splicing of the 5a exon. ²³ Both isoforms are found in human lens at equal levels, whereas Pax6 predominates in mouse and bovine lenses. The two Pax6 isoforms have overlapping but different DNA-binding specificities. ²⁴ For example, Pax6 and Pax6(5a) can both transactivate some Pax6 targets such as αB-crystallin ¹⁴ but only Pax6 is able to bind to the PISCES element of the glucagon promoter ⁵⁹ and to repress the chicken βB1-crystallin promoter. ¹⁷ Mice without the 5a exon undergo eye formation, unlike complete Pax6-null animals; however, they have iris hypoplasia. ²⁵ Transgenic mice that overexpress Pax6(5a) in lens fiber cells have been generated and exhibited lens defects phenotypically similar to those in the Pax6 mice in this study. ²⁸ However, the population of genes with expression levels that are altered by Pax6(5a) and Pax6 expression are quite different as assayed by microarray analysis. ^{58 60} Further, preliminary studies have shown that, unlike the Pax6(5a) mice, α5 and β1-integrin protein levels are not changed in Pax6 transgenic lenses (data not shown). In contrast to the Pax6 mice, the cMaf protein levels are unaffected in the Pax6(5a) transgenic lenses (data not shown). These observations support the hypothesis that Pax6 and Pax6(5a) control the expression of overlapping yet distinct sets of genes in vivo. 

In summary, we tested the hypothesis that ectopic expression of Pax6 in lens fiber cells would disrupt normal cell differentiation and maturation. We found that elevated levels of Pax6 in lens fiber cells inhibits fiber cell elongation, denucleation, and formation of specialized membrane structures. Most interesting was our demonstration, for the first time in vivo, that Pax6 negatively regulated the protein levels of cMaf and transcript levels of βB1-crystallin in the lens, implying that the role of Pax6 in late-stage lens development is to maintain the undifferentiated phenotype of the lens epithelium. 

 Supported by National Eye Institute Grants EY13146 (LWR), EY12221 (MKD), EY07755 (LLD), and EY12995 (MLR), core Grants EY10572 and EY14795, Departmental unrestricted grants from Research to Prevent Blindness (RPB) Inc. (LWR), and a Sigma XI-National Academy of Science graduate research award (WC).

 Submitted for publication February 12, 2004; revised May 27, 2004; accepted June 7, 2004.

 Disclosure: M.K. Duncan, None; L. Xie, None; L.L. David, None; M.L. Robinson, None; J.R. Taube, None; W. Cui, None; L.W. Reneker, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Diagrammatic representation of the Pax-6 transgenes. (A) The original transgenic construct using the mouse αA-crystallin promoter (−342/+49). SV40 viral sequences containing a small intron and polyadenylation signal (SV40 pA) were added to the 3′ end of the Pax6 cDNA. (B) The Pax6 transgene construct used to generate the transgenic mice described in this study. The mouse αA-crystallin promoter (−368/−340 fused to −287/+45) was modified to include two copies of a canonical Pax-6 consensus DNA-binding site (5′-CAT CTT CAC GCA TGA GTG ACT G-3′) inserted between positions −82 and −83. The mouse Pax6 cDNA is flanked by the rabbit β-globin intron at the 5′ end and human growth hormone polyadenylation signal (hGH pA) sequences at the 3′ end. All seven lines created with this construct expressed the transgene and exhibited cataracts. The primers used to genotype the transgenic mice, designated Pr4 and β3, are shown in (B).

Figure 2.

 

Pax-6 mRNA detected in E18 lenses by in situ hybridization. E18 embryos were isolated and processed for in situ hybridization with a mouse Pax-6 cDNA used as a probe. In the wild-type lens (WT), Pax-6 was highly expressed in the lens epithelium (epi), and its expression was downregulated at the transitional zone (t, arrows) during fiber cell (fib) differentiation. In transgenic lenses from line OVE1076 to -1078, Pax-6 transcripts can be detected not only in the epithelium but also in the fiber cells. Among the three lines tested, the order of transgene expression level is OVE1078>OVE1076>OVE1077, which correlates with the severity of the fiber cell defects in the transgenic lenses (see Fig. 3 ).

Table 1.

 

Summary of Findings in Pax6 Transgenic Lenses

Line	Expression Level (mRNA)	Expression Level (Protein)	Cataract (Severity)
OVE1076	++	++	++
OVE1077	+	+	+/−
OVE1078	++++	++++	++++
LR2	+++ (estimated)	ND	+++

 ND, not determined.

Figure 3.

 

Histology of 1-week-old (P7) lenses from different transgenic lines. (A) Normal lens morphology with an anterior epithelial layer (epi) and uniformly stained fiber mass (fib). The transitional zone is indicated by t. (B–F) The lens epithelial layer appeared normal in all the transgenic lines. Cataracts developed with different severities because of lens fiber cell defects. Vacuoles were present in the cortical fibers (except in the lens of line OVE1077). In the severely affected line OVE1078 (C), the secondary fiber cells failed to elongate properly and the apical surface was not in contact with the anterior epithelium (arrows: anterior margin of the secondary fiber cells). Fiber cell denucleation was impaired in all the transgenic lines (except OVE1077). Even in the mildly affected line LR3 (E, F), hematoxylin-staining-positive materials (F, arrows), probably chromosomal debris, can be seen in the central fiber cells. The boxed region in (E) is shown in a higher magnification in (F). (A) through (E) are of the same magnification. Scale bar, 100 μm.

Figure 4.

 

Histology of OVE1078 lenses from mice of different ages. Lens histology was compared between wild-type (A, C, E) and OVE1078 transgenic (B, D, F) mice. (A, B) At E15, the transgenic lens appeared normal. The apparent size difference is artificial, because the section in (A) was from a more central region of the lens globe than the section in (B). (C, D) Fiber cell (Fib) defects were first seen in the E18 transgenic lens. The secondary fiber cells did not elongate completely (arrows, anterior margin of the fiber cells), leaving a lumen at the cortical region underneath the anterior epithelium (Epi) (D). (E, F) The defects in fiber cell elongation were still seen in the newborn transgenic lens (F, arrows). More vacuoles were present in the fiber mass. Compared with the wild-type lens (E), fiber cells were not well organized and aligned in the transgenic lens. Scale bar, 100 μm.

Figure 5.

 

Electron microscopy of the fiber cells at the cortical region of P4 lenses. (A, B) Normal lens fiber cells formed interdigitations between neighboring cells through ball-and-socket junctions (arrows). (C–E) The fiber cells in the transgenic lens have a significantly less interdigitated junction structure (D, E). In many areas, the ball-and-socket junctions were completely absent as shown (C). These defects were consistently seen in two lenses from different animals at P4 as well as from two additional animals at P7 (data not shown). Magnification: (A–C) ×7500; (D, E) ×15,000.

Figure 6.

 

cMaf mRNA and protein localization in E14 and E18 lenses. (A) cMaf mRNA was detected by in situ hybridization, with a mouse cMaf cDNA used as a probe. cMaf mRNA was detected in lens epithelial and fiber cells, and the expression level was upregulated in the differentiating fiber cells. The expression levels of cMaf were not significantly different between the nontransgenic (WT) and transgenic (OVE1078) lens at both ages. (B) Immunofluorescence labeling for cMaf protein in the lens. Rhodamine fluorescence appears white. In the wild-type lenses, cMaf proteins were found in lens epithelial and fiber cell nuclei, but the levels were much higher in the fiber cell nuclei (WT). Compared with the normal lens, significantly fewer nuclei stained positively for cMaf in the transgenic lens (OVE1078). Scale bar, 100 μm.

Figure 7.

 

Double labeling for Pax6 (A, D, G) and cMaf (B, E, H) proteins in wild-type lenses (WT, A–C) and transgenic lenses from line OVE1076 (D–F) or OVE1078 (G–I). Superimposed images are shown in (C, F, I). In the wild-type lenses, Pax6 protein (green) levels were reduced in fiber cells and were hardly detectable (A). Under the same labeling conditions, Pax6 was found in the fiber cell nuclei of transgenic lenses (D, G). The higher the expression level, the more Pax6-positive nuclei were detected among the fiber cell nuclei (compare OVE1076 with OVE1078, D–G). In the wild-type lenses, cMaf protein levels were upregulated during fiber cell differentiation. Many cMaf-positive nuclei (red) were found in the lens fiber cells of wild-type mice (B). In the transgenic lenses, the number of cMaf-positive nuclei correlated negatively with the expression levels of the Pax-6 transgene (E, H). For example, in OVE1076 mice (D–F), the nuclei containing high levels of Pax6 protein had low levels of cMaf (arrows, representative nuclei), whereas fiber cells with normal levels of cMaf proteins showed low levels of Pax6 (arrowheads, representative nuclei). In the high-expressing OVE1078 mice (G–I), a significant number of fiber cell nuclei stained positive for Pax6 protein and few to no cMaf-positive nuclei were detected. Scale bar, 100 μm.

Figure 8.

 

cMaf (A, B) and Sox2 (C, D) immunolocalization in P7 normal (WT) (A, C) and LR2 (B, D) transgenic lenses. As shown in transgenic OVE1078 mice (Fig. 3) , cMaf protein (red) was less abundant in the fiber cells of the LR2 transgenic lenses (B). Unlike cMaf protein, Sox2 protein levels (red) were unaffected in the LR2 transgenic lenses. Cell nuclei in (A) and (B) were counterstained with nuclear stain (green).

Figure 9.

 

Lens crystallin protein analysis by 2-D gel electrophoresis. Equal amounts of water-soluble proteins from P7 normal (A) and LR2 transgenic (B) lenses were used for analysis. The α-crystallin protein levels (αA and αB) were not significantly affected in the transgenic lens. However, the fiber cell crystallins, including different isoforms of β- and γ-crystallins, were reduced in the transgenic lens. In addition, transgenic lenses contained more proteolytic and modified crystallins (circled spots), which are usually seen only in normal aged lenses. ³⁵ These products include proteolytic products of αA-crystallin (αAΔ) (box), a modified form of αB-crystallin (αBm), a modified form of βB1-crystallin (βB1m), and βA3-crystallin with 11 amino acids removed from the N terminus (βA3-11). Other proteins on the gel included fatty acid binding protein (fab) and αAinsert-crystallin (αAi).

The authors thank Kirk Czymmek (University of Delaware Core Microscopy Facility), the Electron Microscopy Core Facility at the University of Missouri-Columbia, Phillip Wilmarth, Li Xu, and Shanyu Ho for technical support; Paul Overbeek (Baylor College of Medicine, Houston, TX) for generation of OVE1076 to -1078 transgenic mice and for a critique of the manuscript; the University of Missouri-Columbia Transgenic Animal Core, for generating the transgenic mice lines LR1- to -4, and Ales Cvekl (Albert Einstein College of Medicine, New York, NY) for valuable discussions.