Mutually regulated expression of Pax6 and Six3 and its implications for the Pax6 haploinsufficient lens phenotype

Six3 Expression in the Lens Requires Pax6 Function in a Dosage-Dependent Manner.

We examined the regulatory hierarchy of genes involved in lens development by comparing the expression of eye-specific transcription or nuclear factors in the eyes of E12.5 wild-type and Pax6 heterozygous embryos. Reduced Six3 expression was found in lenses of Pax6^lacZ embryos (Fig. 1 A′, B′, and G), whereas expression levels of Eya1 and 2, FoxE3, Prox1, and Sox1 (Fig. 1 C–F′ and data not shown) were the same in wild-type and Pax6^lacZ embryos. In contrast, Six3 expression in the retina of Pax6^lacZ embryos was unchanged (Fig. 1 A′, B′, and G). The lens-specific reduction of Six3 expression in Pax6^lacZ embryos indicates a dosage dependence of Six3 expression on Pax6 levels in the lens, whereas Pax6 function is not essential for Six3 expression in the retina (23).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Dosage-dependent reduction of Six3 expression in lenses of Pax6^lacZ embryos. Reduced Six3 expression is found in lenses of E12.5 Pax6^lacZ embryos (B′), compared with controls (A′). Six3 expression in the retina (ret) is unchanged. Expression of Sox1 (C–D′) and Prox1 (E–F′) is unchanged in wild-type (WT) and Pax6^lacZ embryos. (A, B, E, and F) Bright field, hematoxylin. (A′, B′, E′, and F′) Dark field. (C–D′) Fluorescence microscopy. ae, anterior epithelial cells; lf, lens fiber cells; ret, retina; tz, transition zone; PI, propidium iodide. (Bar = 100 μm.) (G) Gray-scale measurement of Six3 expression. For each embryo, measurements were done in the lens (AE and LF) and retina (Ret) as described in Materials and Methods. Values were calculated in E12.5 Pax6^lacZ (gray) and wild-type (black) embryos. Six3 expression was reduced in lenses of Pax6^lacZ embryos, whereas expression in the retina was unchanged. Results are the average ± SE of measurements from four separate embryos. Statistical difference: **, P < 0.01; *, P < 0.05; ns, not significant.

Lens-Specific Expression of Six3 Rescues the Pax6 Haploinsufficient Lens Phenotype.

Because of its specificity, we speculated that this dosage-dependent reduction of Six3 expression might contribute to the Pax6 haploinsufficient lens phenotype. We therefore attempted to rescue this lens defect by increasing Six3 gene dosage, using a transgene (αASix3, Fig. 2A) linking a 0.4-kbp murine αA-crystallin gene promoter to a 1.4-kbp mouse Six3 cDNA (9). With a transgene-specific riboprobe, embryonic expression was initially detected at E12.0 (Fig. 2B). Expression of the transgene in lens fiber cells overlapped that of the endogenous Six3 gene from E12.0 to E13.5 and was maintained in the differentiated fiber cells (Fig. 2 B–E). We found that transgene expression levels were lower in αASix3/Pax6^lacZ embryos than in αASix3/Pax6^+/+ embryos (not shown), this difference being likely caused by the presence of a Pax6-responsive site in the αA-crystallin promoter sequences used in the transgene construct (25).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Structure and expression of the αASix3 transgene. (A) Diagram of the DNA construct used for the generation of αASix3 transgenic mice. The αASix3 transgene was generated by inserting the full-length coding sequences of the murine Six3 cDNA into polylinker restriction sites of plasmid pACP3. The arrow represents the site of transcription initiation. A 0.6-kbp fragment derived from the Six3 cDNA was used as an in situ hybridization probe (probe A). A transgene-specific riboprobe (probe B) was derived from the SV40 polyadenylation sequences. The location of primers used for PCR genotyping (S5 and S3) is shown. AATAA, SV40 polyadenylation (pAn) site; ATG: translation initiation site. (B–E) Expression of the transgene. Using riboprobe B, transgene expression was detected in αASix3 transgenic embryos from E12.0 (B) and was restricted to lens fiber cells at all stages examined (D, E13.5). Expression of the transgene is compared with Six3 expression in age-matched controls (C and E). (Dark field; bar = 100 μm.)

Lenses from E14.5 αASix3/Pax6^lacZ embryos were larger than those of age-matched Pax6^lacZ embryos, but remained smaller than wild-type controls (Fig. 3 A–C). This phenotypic rescue was found in Pax6^lacZ mice hemizygous for the αASix3 transgene and was fully penetrant in the FVB/N background. The increased lens size was maintained in 3 week-old αASix3/Pax6^lacZ mice (Fig. 3 I–K), although at this later stage lens morphology was abnormal and included the presence of vacuoles and ballooned fiber cells (Fig. 3 K and O). To explain this phenotypic rescue, we first examined the effects of Six3 expression on lens fiber cell proliferation, differentiation, and death. We studied cell proliferation in embryos by using the BrdUrd incorporation assay, which detects cells entering the S phase. BrdUrd-incorporating lens fiber cells were present in E12.5 (Fig. 3S) and E13.5 αASix3/Pax6^lacZ embryos, but not in the lens fiber cell compartment of controls (Fig. 3 Q and R). We next examined the effects of Six3 expression on lens differentiation and cell death. At stage E14.5, the expression of the fiber cell differentiation markers αA-, β- and γ-crystallins, MIP-26, filensin, and CP-49 (Fig. 3 U–W and data not shown), as well as the number of TUNEL assay-positive fiber cells (Fig. 3 Y–Z′), were the same in fiber cells of αASix3/Pax6^lacZ and control embryos. Altogether, these results indicate that the improved lens size in αASix3/Pax6^lacZ embryos resulted, in part, from fiber cell proliferation. Importantly, differentiation and survival of fiber cells in αASix3/Pax6^lacZ embryos were normal, leading to overall lens growth.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Rescue of the Pax6 haploinsufficient lens phenotype in αASix3/Pax6^lacZ transgenic mice. (A–P) Effects of Six3 expression on lens morphology. Transverse sections from eyes of E14.5 (A--H) and 3-week-old (I–P) mice. Note the increased lens size in αASix3/Pax6^lacZ mice (C and K) compared with age-matched Pax6^lacZ mice (B and J), and the microphakia in αASix3/Pax6^+/+ transgenic mice (D and L). The boxed areas in A–D and I–L are enlarged in E–H and M–P, respectively. Note the presence of vacuoles (arrowhead, H) in fiber cells of E14.5 αASix3/Pax6^+/+ transgenic embryos. Large vacuoles (arrowheads, K and L) and disorganized fiber cell arrangement are present in lenses of 3-week-old αASix3/Pax6^lacZ and αASix3/Pax6^+/+ transgenic mice (O and P). Posterior rupture of the lens capsule is seen in the latter (arrow, P). (Q–T) Lens fiber cell proliferation in embryos harboring an αASix3 transgene. BrdUrd immunochemistry in E12.5 embryos demonstrating the presence of proliferating cells in the lens fiber cell compartment and transition zone (arrow) of αASix3/Pax6^lacZ (S) and αASix3/Pax6^+/+ transgenic embryos (T), but not in controls (Q and R). (U–X) Normal lens maturation in αASix3/Pax6^lacZ transgenic embryos. The differentiation marker β-crystallin is normally expressed in fiber cells of E14.5 αASix3/Pax6^lacZ transgenic embryos (W) [compare with wild-type (U) and Pax6^lacZ (V) controls]. The same marker is expressed heterogeneously in fiber cells of a E14.5 αASix3/Pax6^+/+ transgenic embryo (X). (Y–Z′′) TUNEL assay. The lens fiber cell compartment of E13.5 αASix3/Pax6^+/+ transgenic embryos contains rare nuclei that were positive for the TUNEL assay (Z′′). At the same stage, numerous TUNEL-positive nuclei are found in the degenerating hyaloid vessels of transgenic and control embryos (arrow). [Bar = 100 μm, except E–H (50 μm) and I–L (500 μm).]

Six3 Overexpression Causes Defective Lens Maturation and Fiber Cell Death.

In contrast to the improved lens size in αASix3/Pax6^lacZ transgenic mice, αASix3/Pax6^+/+ transgenic mice were microphakic (Fig. 3 D and L). The severity of the microphakia was greater in embryos homozygous for the αASix3 transgene (not shown). The latter effect was found in either the Pax6^+/+ or Pax6^lacZ background, indicating that Six3 gene dosage was the primary factor affecting the lens phenotype. αASix3/Pax6^lacZ transgenic embryos were microphakic from stage E14.5 onward (Fig. 3D). In sections done on adult eyes, the lenses had a disorganized fiber cell arrangement accompanied by posterior rupture of their content into the vitreous chamber (Fig. 3 L and P).

BrdUrd incorporation assay revealed lens fiber cell proliferation in αASix3/Pax6^+/+ transgenic embryos, an effect similar to that observed in αASix3/Pax6^lacZ embryos (Fig. 3 S and T). We examined the effect of increased Six3 expression on lens fiber cell maturation and survival. From E14.5 onward, a heterogeneous distribution of lens differentiation markers was observed in lenses of αASix3/Pax6^+/+ embryos, when compared with controls (Fig. 3 U and X). Lastly, from stage E14.5 onward, vacuoles were present in fiber cells of αASix3/Pax6^+/+ embryos, but not in age-matched controls (Fig. 3 E and H), indicating a process of cell death. However, fiber cells were rarely stained with TUNEL (Fig. 3Z′′). Cell death thus resulted either from necrosis or from a process differing from the common forms of apoptosis. These results thus suggest that, in addition to defective maturation, Six3 overexpression causes fiber cell death.

Mutually Regulated Expression of Pax6 and Six3 in the Lens.

Members of the vertebrate Pax and Six gene families participate in gene feedback loops regulating eye, kidney, ear, and muscle formation (reviewed in ref. 26). To further address the effects of Six3 expression in the developing lens, we examined the regulatory relationship between Pax6 and Six3. In lenses of transgenic embryos, expression of Six3 up-regulated that of Pax6 (Fig. 4 A–F). Reciprocally, up-regulated Six3 expression was found in fiber cells of αA-crystallin (modified)/Pax6 (αAmPax6) transgenic embryos (Fig. 4 G–L). Expression of other eye-specific factors (Eya1, Eya2, FoxE3, Prox1, and Sox1) was unchanged in lenses of transgenic embryos and controls (not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Mutually regulated expression of Pax6 and Six3 in the developing lens. (A–L) Mutual activation of Six3 and Pax6 expression in transgenic embryos. Fiber cell nuclei from E14.5 αASix3 transgenic embryos are immunoreactive for the Pax6 protein (B, E, and F) whereas only faint staining is detected in the fiber cell compartment of wild-type controls (A, C, and D). Reciprocally, Six3-immunoreactive fiber cells are present in E14.5 αAmPax6 transgenic embryos (H, K, and L), but not in controls (G, I, and J). [Bar = 100 μm (A, B, G, and H).] (M and N) Cell transfection assays. COS7 cells were transiently transfected with a Pax6-luciferase reporter and increasing amounts of a Six3 expression vector (M), or with a Six3-luciferase reporter and a Pax6 expression vector (N). In both M and N, a dosage-dependent activation of the reporter is seen after transfection of increasing amounts of the expression construct. Values are expressed as the fold increase in luciferase activity in the presence of the expression plasmid, compared with the level of activity with reporter plasmid alone.

A 107-bp BglII–AccI DNA fragment, located 5 kbp upstream of the Pax6 (P0) transcription start site, has previously been shown to direct reporter gene activity in lenses of transgenic embryos (24). This fragment was examined by EMSA for the presence of Six3-binding sites. We observed the shift of electromobility of a 20-bp sequence (5′-CTCATTGCCCATTCAAATAC-3′) derived from the 5′ part of the enhancer (not shown). We examined whether this sequence has Six3-dependent transcriptional activity. In cell transfection assays, luciferase activity was increased after cotransfecting a reporter construct containing a multimer of this sequence with a Six3 expression vector (Fig. 4M). Additionally, overlapping phage clones containing 27 kbp of the Six3 gene were sequenced and evaluated for the presence of sequences related to the Pax6-paired domain DNA-binding sites (27). An oligonucleotide derived from one potential inverted site (5′-TGCAGTTGAGCGGAAAAG-3′), located 2 kbp upstream of the first coding exon, shifted the Pax6 protein in EMSA (not shown). Moreover, a luciferase reporter construct containing a multimerized oligonucleotide was activated after cotransfection with a Pax6 expression vector (Fig. 4N). The results of EMSAs and cell transfection assays, combined with those in transgenic embryos, clearly indicate that each gene constitutes a direct target for the counterpart transcription factor, and that Pax6 and Six3 expression is mutually activated during mouse lens development.

Up-Regulation of PDGFα-R and Cyclin D1 After Lens-Specific Expression of Six3 or Pax6 in Transgenic Mice.

We further evaluated the cell proliferation in αASix3/Pax6^lacZ and αASix3/Pax6^+/+ transgenic embryos by studying the expression of cell cycle regulatory proteins in fiber cells. We found increased levels of cyclin D1 (Fig. 5 A–D), whereas expression levels of Cdkn1c, Cdkn1b, Cdc2, and Cdk4 were unchanged (not shown). The up-regulated cyclin D1 activity was accompanied by an increase in the activated form of p44/42 mitogen-activated protein kinase (ERK), indicating a fiber cell response to an extracellular mitogenic signal (not shown). We further evaluated this signaling pathway by studying the lens expression levels of receptor tyrosine kinases that act upstream of ERK. Expression levels of PDGFα-R were up-regulated in fiber cells of αASix3 transgenic embryos, whereas expression levels of insulin-like growth factor I-R, and fibroblast growth factor types 1 and 2 were unchanged (Fig. 5 I–L and data not shown). Moreover, fiber cell proliferation, ERK activation, and up-regulated expression of PDGFα-R and cyclin D1 were found in lenses of αAmPax6 transgenic embryos (Fig. 5 E–H and M–N, and data not shown). Defective fiber cell maturation was also demonstrated in αAmPax6 transgenic embryos (Fig. 5 O and P), an effect similar to that observed after Six3 overexpression (see Fig. 3 U and X). Therefore, in gain-of-function studies, the effects of either transcription factor in the lens were similar, as a likely consequence of the cross-regulation between both genes. These results also argue that both transcription factors participate, along with FoxE3, Prox1, and Cdkn1c (18, 19, 28), in a gene network controlling cell proliferation in the lens. In this context, the mutual activation of Pax6 and Six3 may serve to limit the fluctuations of expression levels of either transcription factor (3). Stable expression of Pax6 and Six3 may be required for the coordinate expression of various components of the cyclin D1/PDGFα-R signaling pathway, thus facilitating the progression between different phases of the cell cycle.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

PDGFα-R and cyclin D1 up-regulation in transgenic embryos expressing Six3 or Pax6 in the lens. (A–H) Cyclin D1 immunochemistry. Increased numbers of cyclin D1-reactive nuclei are detected in the fiber cell compartment of E12.5 αASix3 (C and D) and E14.5 αAmPax6 (G and H) transgenic embryos, compared with controls (A, B, E, and F). The boxed areas in E and G are enlarged in F and H. (I–L) Up-regulation of PDGFα-R expression in fiber cells of αASix3 transgenic embryos. PDGFα-R activity is up-regulated in the lens fiber cell compartment of E12.5 αASix3 (K and L) transgenic embryos. PDGFα-R immunoreactivity is present in anterior epithelial cells, but not in fiber cells of controls (I and J). (M–P) Lens fiber cell proliferation and defective maturation in αAmPax6 transgenic embryos. Lens fiber cell proliferation is observed in E12.0 αAmPax6 embryos (N), whereas the same changes are absent in age-matched controls (M). A heterogeneous distribution of αA-crystallin is found in the lens of a E14.5 αAmPax6 transgenic embryo (P), compared with a control (O). WT, wild type. [Bar = 100 μm, except F and H (50 μm).]