Aberrant Lens Fiber Differentiation in Anterior Subcapsular Cataract Formation: A Process Dependent on Reduced Levels of Pax6

Abstract

purpose. TGFβ can induce development in lenses of opaque subcapsular fibrotic plaques that have many features of human subcapsular cataracts. To understand further the events associated with the onset and progression of TGFβ-induced cataract, several different models for anterior subcapsular cataract (ASC) were used and characterized.

methods. Anterior subcapsular plaques were induced in rat lenses cultured with TGFβ and in transgenic mice overexpressing TGFβ in the lens. ASC was also examined in lenses of mice haploinsufficient for Pax6, as well as in human biopsy specimens. Immunofluorescence and in situ hybridization labeling were used to examine changes in patterns of gene expression associated with cataract formation in these models.

results. Examination of TGFβ-induced cataract in transgenic mice established that the subcapsular plaques are composed of a heterogenous cell population: a population of myofibroblastic cells as well as a population of lens-fiber–like cells. Further support for phenotypic change comes from the observation that the cells in these plaques no longer expressed lens epithelial markers, such as Pax6 and Connexin43. Subsequent examination of human biopsy specimens of ASC, as well as lenses from Pax6-deficient mice, showed that the anterior subcapsular plaques in both cases were also composed of a heterogenous population of cells. In contrast, anterior subcapsular plaques that developed in vitro in response to TGFβ did not have this same cellular heterogeneity, as no fiber-like cells were present.

conclusions. These findings suggest that in vivo, during TGFβ-induced cataract formation, some lens epithelial cells transform into myofibroblastic cells, whereas others differentiate into fiber cells. As this pathologic change is accompanied by altered expression of genes characteristic of the normal lens epithelial cell phenotype and as lenses from Pax6-deficient mice exhibit development of anterior subcapsular plaques closely resembling those induced by TGFβ in transgenic mice, the authors propose that a reduction in Pax6 levels may be essential for this pathologic process to progress. Furthermore, it is clear from these in vitro studies that TGFβ alone cannot reproduce the same morphologic and molecular changes associated with ASC formation in vivo, indicating that additional molecule(s) in the eye are important in this process.

The transparency of the vertebrate ocular lens is primarily attributed to its highly organized structure. ¹ The distinct polarity and growth of the lens is maintained throughout life as epithelial cells proliferate and subsequently elongate and differentiate into fiber cells. These two cellular processes are thought to be tightly regulated by growth factors found in the ocular media. To date, several mitogens have been identified for lens epithelial cells, including members of the fibroblast growth factor (FGF), insulin-like growth factor (IGF)/insulin, platelet-derived growth factor (PDGF), and epidermal growth factor (EGF) families (see Ref. ² for review). However, to date, FGFs are the only growth factors that have been shown to induce lens fiber differentiation in mammals ² (Zhao H, et al. IOVS 2003;44:ARVO E-Abstract 954). More recent studies have shown that members of the transforming growth factor (TGF)-β superfamily are also essential for lens fiber differentiation and maturation. ^{3 4 5}  

In contrast to a normal role for TGFβ in terminal lens fiber differentiation, there is an increasing body of data demonstrating that this molecule plays a role in lens disease—more specifically, it induces aberrant lens epithelial cell behavior, leading to the formation of cataract. ^{2 6 7 8} Earlier studies have reported that lens epithelial cells can be induced to differentiate into fibroblast-like cells, resulting in the formation of anterior subcapsular cataracts (ASCs) ^{9 10 11} ; however, it has only recently been shown that this pathologic change can be attributed to the influence of TGFβ. 

Using rat lens epithelial explants or whole rat lens culture models, studies from our laboratory have shown that TGFβ can induce lens epithelial cells to undergo an aberrant differentiation pathway, including the formation of spindle-shaped cells, accompanied by wrinkling of the underlying lens capsule, aberrant accumulation of extracellular matrix (ECM), and cell death by apoptosis. ^{12 13 14} These spindle-shaped cells express α-smooth muscle actin (α-SMA), ¹³ a cytoskeletal protein not normally found in the lens, but is characteristic of myofibroblasts. All the morphologic and molecular changes identified in these in vitro models have typically been found in different forms of human cataract, including ASC and posterior capsular opacification (PCO). ^{6 15 16 17}  

Studies by Srinivasan et al. ¹⁸ have shown that ASC can also be induced by expression of a self-activating form of human TGFβ1 in the lens of transgenic mice. The subcapsular plaques in these mice are not only similar in morphology to those induced in rat lenses by TGFβ in vitro, but demonstrate a range of molecular changes, including the expression of α-SMA, desmin, collagen types I and III, fibronectin, and tenascin. ⁸ As all the features described for TGFβ-induced subcapsular cataracts are not typical of lens epithelial cells, but are more typical of myofibroblasts/fibroblasts, we proposed that this pathologic process, characterized by the differentiation of lens epithelial cells along a mesenchymal pathway, involves a TGFβ-induced epithelial–mesenchymal transition (EMT). ⁸  

Both the rat lens culture and transgenic mouse models of TGFβ-induced cataract are similar, in that they reproduce many of the molecular and morphologic changes characteristic of human ASC and PCO; however, there are some pronounced differences. As we report for the first time in this study, the subcapsular plaques of transgenic mice are composed of a heterogenous population of cells, unlike the TGFβ-induced subcapsular plaques induced in whole rat lens cultures, which are primarily composed of α-SMA–reactive myofibroblastic cells. This is not surprising, considering that in our in vitro model, lenses are supplemented with TGFβ only, whereas in the transgenic mouse model, the TGFβ-induced pathologic differentiation pathway of epithelial cells may be influenced by several other factors in situ. One would predict that the latter case would be more typical of the pathologic processes involved in formation of human cataract. As a result of this, in the present study, we extended our analysis of these cataract models to identify and characterize the non-α-SMA–reactive cells that compose the subcapsular plaques of our transgenic mice. Elucidating the formation of these TGFβ-induced plaques in vivo will no doubt provide a better understanding of the molecular and cellular basis of cataract formation in humans. 

Methods

All experimental procedures with animals conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the University of Sydney Animal Ethics Committee. All research with human tissues adhered to the tenets of the Declaration of Helsinki. 

Mouse Models

Transgenic mice from families OVE853 and OVE918, which have been described, ^{8 18} were used in the present study. These transgenic lines express a secreted, self-activating form of human TGFβ1, from the lens-fiber–specific murine αA-crystallin promoter. ¹⁸ Eyes of comparable age were also collected from mice heterozygous for Pax6 mutations. The Small Eye alleles used in the present study were Pax6 ^{Sey-Neu 19} and Pax6 ^Sey-Dey. ²⁰ Both lines carrying the mutation were maintained on an FVB/N background. 

Induction of Cataractous Plaques by TGFβ In Vitro

Lenses were obtained from postnatal day (P)21 Wistar rats and cultured in serum-free medium 199 with or without TGFβ2 (Genzyme, Cambridge, MA), at a final concentration of 750 pg to 1 ng/mL, as described previously. ²¹ Lenses were cultured for up to 6 days, with renewal of medium every 2 days, but without further addition of TGFβ. By the end of the culture period, lenses cultured with TGFβ had distinct opacities. Whole lenses were fixed and processed for immunofluorescence, as described later. 

Human Cataractous Tissues

Human ASC specimens (circular sections of the anterior capsule) were obtained during cataract surgery from Japanese patients, with a mean age of 64 years. ⁷ Specimens were obtained at the Wakayama Medical College Hospital, Wakayama, Japan, or were supplied by the IOL Implant Data System Committee of the Japanese Society of Cataract and Refractive Surgery. Once informed consent was obtained, immediately after removal, ASC specimens were fixed in 10% formalin and further processed for routine histologic examination. Six-micrometer sections of plaques were immunolabeled for α-SMA or β-crystallin and stained with periodic acid-Schiff (PAS) reagent, as described later. 

Histology

Ocular tissues for histology were collected from transgenic mice at postnatal day 21 and from Pax6 ^Sey mice between postnatal days 21 to 40. Whole eyes were fixed overnight in 10% neutral-buffered formalin, dehydrated, embedded in paraffin, and processed for routine histology. For histochemical analysis, 6-μm-thick sections were stained with either hematoxylin and eosin or PAS reagent. 

For semithin sections, whole eyes from mice were fixed in 4% paraformaldehyde and 0.1 M phosphate-buffered 1% glutaraldehyde. After 3 days’ fixation, lenses were dissected and postfixed in 0.1 M phosphate-buffered 1% glutaraldehyde (24 hours), rinsed in 0.1 M phosphate buffer (24 hours), and fixed again in phosphate-buffered 1% osmium tetroxide (2 hours). After an overnight rinse in 0.1 M phosphate buffer, tissues were dehydrated and embedded in Spurr’s epoxy resin. For light microscopy, semithin sections (900 nm) were stained with toluidine blue. 

Immunofluorescence

For immunofluorescence, 6-μm-thick paraffin sections of lenses (cultured tissue) or eyes (animal models) were hydrated and incubated for 30 minutes in 3% normal goat serum to reduce non-specific staining. Sections were then incubated overnight at 4°C with primary antibodies specific for either α-SMA, α-crystallin, β-crystallin, filensin, or fibronectin. Crystallin antibodies were specific for crystallin classes (i.e., total α- or total β-crystallins). All primary antibodies, with the exception of anti-α-SMA, were diluted 1:100 with PBS supplemented with 3% normal goat serum. After a brief rinse in PBS, bound primary antibody was visualized with a fluorescein-isothiocyanate (FITC)– or a Cy3-conjugated secondary anti-rabbit antibody, diluted 1:50 (Silenus, Hawthorn, Victoria, Australia). For α-SMA labeling, a monoclonal α-SMA antibody (clone 1A4; Sigma-Aldrich, Castle Hill, Australia) conjugated to Cy3 (diluted 1:200 in PBS) was used for direct labeling. All sections were counterstained with 1 μg/mL bisbenzimide (Hoechst dye; Calbiochem, La Jolla, CA) to label cell nuclei. Sections were then rinsed with PBS, mounted, and examined by fluorescence microscopy. 

In Situ Hybridization

The expression patterns of mRNA transcripts for Pax6, Connexin 43 (Cx43), αB1-crystallin, βB2-crystallin, and major intrinsic protein (MIP), in lenses of transgenic mice, were examined by in situ hybridization, using (³⁵S)UTP-labeled riboprobes, as previously described. ²² For Pax6 and βB2-crystallin, sense and antisense transcripts were synthesized using T7 and T3 RNA polymerases (Promega, Sydney, Australia), respectively. The antisense riboprobe for Pax6 was generated from a 300-nucleotide cDNA derived from the full-length mouse cDNA, ²³ whereas for βB2-crystallin, it was generated from a 591-bp cDNA template. Riboprobes for Cx43 and MIP were generated from 360- and 556-bp templates, using T7 and T3 polymerases to generate antisense and sense transcripts, respectively. For αB1-crystallin, Sp6 and T7 polymerases were used to synthesize antisense and sense transcripts, respectively. Hybridizations were performed on 6-μm-thick sections of lenses collected from transgenic mice and processed as just described. Hybridized sections on slides coated with photographic emulsion were exposed for up to 7 days before developing and counterstaining with Harris hematoxylin. 

Photography

Sections were photographed using a microscope (Dialux 20; Leitz, Wetzlar, Germany) equipped with normal, dark-field, and epi-illumination. Immunofluorescence micrographs were captured with 400 ASA film (T-Max; Eastman Kodak, Rochester, NY) pushed to 1600 ASA during processing, and bright-field micrographs were photographed on 64 ASA film (Ektachrome; Eastman Kodak), processed according to the manufacturer’s instructions. In situ hybridization slides were viewed by dark-field illumination and photographed using 400 ASA film (T-Max; Eastman Kodak) processed according to manufacturer’s instructions. Alternatively, some bright-field and dark-field micrographs were captured digitally with a digital camera (model DC100; Leica Microscopy Systems Ltd., Heerbrugg, Switzerland). 

Results

Transgenic Mice Overexpressing TGFβ

Our earlier studies of transgenic mice overexpressing TGFβ in the lens, have indicated that the anterior subcapsular plaques are composed of a heterogenous population of cells. ⁸ To characterize the cellular composition of these plaques, we first set out to determine the distribution of the myofibroblastic cells that have been reported to be present. 

Distribution of Myofibroblastic Cells.

Using serial sections of subcapsular plaques, we showed a differential distribution of α-SMA reactivity in the plaques (Fig. 1) . Cells immunoreactive for α-SMA formed a radial band that encircled the plaque. This was most evident when comparing peripheral sections (Fig. 1A) with midsagittal sections (Fig. 1B) through the same plaque. This distribution pattern clearly indicated that the TGFβ-induced subcapsular plaques are composed of a heterogenous population of cells: a population of myofibroblastic cells (reactive for α-SMA) and a population of cells nonreactive for α-SMA. 

Semithin sections demonstrated differential staining of the different cell types present in TGFβ-induced subcapsular plaques of our transgenic mice. A prominent mass of dark-labeled cells (Fig. 2 , asterisk) was flanked by lighter staining elongate cells (Fig. 2 , arrows). Consistent with our identification of myofibroblastic cells in these plaques, the darker labeled cells (attributed to their greater organelle content) corresponded to the region of cells that were reactive for α-SMA (Fig. 1B) . Interspersed among these cells were small pockets of darker stained material (Fig. 2 , arrowheads) which we have previously reported to be aberrant ECM deposition. ⁸ The elongate morphology of the lighter staining cells was typical of differentiating lens fiber cells. 

Localization of Lens Fiber Cell Markers.

To determine whether some of the cells comprising the plaques demonstrate any molecular features of lens fiber cells, we studied the expression of several different lens (and lens fiber-specific) genes, including α- and β-crystallin, filensin, and major intrinsic protein (MIP). Consistent with our histologic identification of fiber-like cells, we observed strong immunoreactivity for α- and β-crystallin within the subcapsular plaques (Fig. 3) . Strongest immunoreactivity for α- and β-crystallin was found in cells closest to the lens capsule, with cells deeper in the plaque demonstrating weaker labeling (Fig. 3C , arrow). This pattern of labeling was supported by the expression of αΒ1- and βΒ2-crytallin transcripts in the plaques, with cells deeper in the plaque demonstrating weak to no mRNA expression (Figs. 3E 3F , arrowheads). With antibodies specific for another fiber-differentiation marker, filensin, we noted a pattern similar to that of β-crystallin, with strongest immunoreactivity in cells positioned closest to the lens capsule of the plaques (Fig. 3D) . MIP transcripts showed this same expression pattern, with cells in more posterior regions of the plaques demonstrating no MIP mRNA expression (Figs. 3G 3H , arrows). These distinct cellular regions of plaques that did not express MIP corresponded with the regions that were reactive for α-SMA (Fig. 1) . 

Loss of Lens Epithelial Markers.

It is currently not clear whether some of the cells composing the plaques of our animal models remain as lens epithelial cells. To address this, we examined the expression of some molecular markers characteristic of lens epithelial cells, including Pax6 and Cx43. Pax6 is normally expressed in epithelial layers of the eye, including the corneal epithelium (Figs. 4A 4B , arrowheads) and the lens epithelium (Figs. 4E 4F , arrow). In lenses of transgenic mice, this was still the case, especially where the epithelium remained as a monolayer and continued to be associated with the overlying lens capsule (Figs. 4A 4B , arrow). However, there was a marked change in expression of Pax6 transcripts in cells comprising the deeper parts of the TGFβ-induced subcapsular plaques. As cells of the plaque were displaced from the lens capsule, there was a reduction in Pax6 expression, with little to no Pax6 mRNA detected throughout the body of the developing plaque (Figs. 4C 4D , asterisk). Similar to Pax6, Cx43 is normally expressed in epithelial layers of the eye, including the corneal epithelium (Fig. 5 , arrowheads) and throughout the lens epithelium (Fig. 5C , arrows). Furthermore, compared with Pax6 mRNA expression in transgenic mice, a similar result was found when we examined the expression of Cx43. Lens epithelial cells of transgenic mice expressed Cx43 until the point when they began to change in morphology and contribute to the TGFβ-induced subcapsular plaque (Fig. 5B , arrows). However, this loss of Cx43 expression was more abrupt than that of Pax6, with no transcripts for Cx43 detected in any of the cells that composed the plaque (Fig. 5B , asterisk). The loss of Pax6 and Cx43 lends further support to the notion that the cells that largely contribute to the subcapsular plaques lose their epithelial identity. Notably, the loss of Pax6 and Cx43 are also key markers in the process of normal lens fiber differentiation. 

Lenses Cultured with TGFβ

Earlier studies in vitro from our laboratory, examining the effect of TGFβ on whole lenses and lens epithelial explants, did not report any morphologic changes characteristic of the normal fiber differentiation process. To investigate whether the in vivo induction of the fiber-specific markers in subcapsular plaques observed in transgenic mice is the result of the direct influence of TGFβ, we assayed for fiber differentiation (the expression of β-crystallin) in subcapsular plaques derived from whole rat lenses cultured for up to 6 days with only TGFβ. As previously reported, ¹⁴ plaques from cultured lenses are predominantly composed of a multilayered mass of cells, most of which express α-SMA. Although β-crystallin reactivity was observed in fiber cells of whole lenses cultured with or without TGFβ (Fig. 6) , no reactivity for β-crystallin was detected in the subcapsular plaques induced by TGFβ (Figs. 6A 6B) , similar to the epithelial cells of control lenses (Figs. 6C 6D) . As this finding indicates that TGFβ may not necessarily be a direct inducer of lens fiber differentiation in the subcapsular plaques of transgenic mice, we examined whether secondary pathway(s) may be involved in this fiber cell differentiation. 

Pax6 Mutant Mice

There is compelling evidence from numerous studies that Pax6 is essential for eye and lens development, not only in early lens differentiation but also in the regulation of crystallin genes involved in the lens differentiation process. ^{24 25} As our present findings indicate that levels of this transcription factor were reduced in cells that compose the TGFβ-induced subcapsular plaques, we questioned whether this was sufficient to destabilize the lens epithelial cells, rendering them susceptible to following alternate differentiation pathway(s). To address this, we adopted the Pax6 ^Sey mouse model to characterize further the lens phenotype attributed to the presence of cataracts, previously reported in these mice. ¹⁹ For this study, we examined lenses from Pax6 ^Sey mice carrying two different alleles of Small Eye, including Pax6 ^Sey-Neu (an ethylnitrosourea-induced mutation in a splice acceptor, leading to a truncated Pax6 protein ¹⁹ ) and Pax6 ^Sey-Dey (a spontaneous mutant resulting from a large deletion of the loci for Pax6 and Wilms’ tumor ¹⁹ ). Eyes from both of these heterozygous lines of Pax6 ^Sey mice showed similar morphologic and molecular changes, as described next. 

Subcapsular Plaques in Pax6 Mutant Mice.

Characterization of lenses from Pax6 ^Sey mice demonstrated many morphologic and molecular features analogous to those found in the TGFβ-induced subcapsular plaques. This includes both previously reported features ^{20 26} and the novel features described in the present study. Histologic examination of Pax6 ^Sey mice lenses demonstrated the presence of anterior subcapsular plaques, which predominantly appeared in two forms. In some cases, the anterior pole of the lens and the overlying corneal stroma were attached, creating what has been described as a “lens-corneal plug” ²⁶ (Fig. 7A) , where the developing lens and cornea failed to separate during ocular morphogenesis. ^{20 26} PAS staining revealed that the cells composing this plug were mostly contained by a thinner than normal lens capsule (Fig. 7A) . In other cases, the lens, although still closely associated with the overlying cornea, was not attached to the corneal stroma and developed an anterior subcapsular plaque (Fig. 7B) , similar to those observed in the transgenic mouse model. PAS staining of the Pax6 ^Sey subcapsular plaques also demonstrated a thinner than normal lens capsule, which in some places infiltrated the underlying plaque cell mass (Fig. 7B , arrows). PAS-reactive material was also evident within the subcapsular plaques (Fig. 7B) , indicative of aberrant ECM deposition. For all further characterization, both forms of Pax6 ^Sey plaques derived from either Pax6 ^Sey-Neu or Pax6 ^Sey-Dey mice showed similar immunolabeling patterns. 

Presence of Myofibroblastic Cells.

As shown in Figure 1 a characteristic feature of TGFβ-induced subcapsular plaques is the expression of α-SMA. In lenses of Pax6 ^Sey mice, we report for the first time, the presence of myofibroblastic cells (immunoreactive for α-SMA) within the subcapsular plaques (Fig. 7C) , notably associated with the aberrant deposition of PAS-reactive material (Fig. 7B) . When we looked for the expression of an ECM component previously reported in TGFβ-induced subcapsular plaques, ⁸ we found that the plaques of Pax6 ^Sey mice also expressed fibronectin (Fig. 7D) which was not normally found in the lens of postnatal wild-type mice (data not shown). 

Localization of Lens-Fiber–Specific Markers.

Based on these findings, it appeared that the subcapsular plaques from Pax6 ^Sey mice were composed of a population of myofibroblastic cells, similar to that previously reported in TGFβ-induced subcapsular plaques ⁸ and in some forms of human cataract. ^{9 10} To assay for fiber differentiation in plaques derived from Pax6 ^Sey mice, we examined for the expression of β-crystallin and filensin. In the anterior region of the Pax6 ^Sey lenses, where subcapsular plaques form, we observed immunoreactivity for both β-crystallin (Figs. 8A 8C) and filensin (Figs. 8B 8D) , indicative of aberrant fiber differentiation. In these same lenses of Pax6 ^Sey mice, these proteins were localized only in fiber cells and not in the epithelium (Figs. 8F 8G) . Overall, the Pax6 ^Sey lens phenotype was similar to that reported in TGFβ-induced subcapsular cataracts, ⁸ not only in morphology but also in its composition of myofibroblastic cells, aberrant deposition of ECM, and the presence of aberrant fiber-like cells. 

Human ASCs

As this is the first report of fiber-specific markers in our models of ASC, we were interested to determine whether fibrotic plaques from human ASCs also contain fiberlike cells. To investigate this, we localized β-crystallin and α-SMA using immunohistochemistry on human biopsy specimens of ASCs. Mature plaques were composed of ECM as shown by PAS staining (Fig. 9A) . Around this ECM containing the plaque were elongate cells immunoreactive for α-SMA (Figs. 9B 9C) . Flanking these myofibroblastic cells was a population of cells immunoreactive for β-crystallin (Figs. 9D 9E) . Most of these cells were devoid of nuclei (Fig. 9D) as expected for mature fiber cells. This is the first report of the presence of lens fiber cells in subcapsular plaques of human ASC. 

Discussion

By use of different in vivo models of ASC, the present study reports for the first time that fibrotic plaques contribute to this form of cataract, and in both human disease and mouse models, the plaques are composed of a heterogenous cell population. The best characterized of these cell types are the myofibroblastic cells reactive for α-SMA. These cells have been observed in human ASC, ^{10 15 16 17 27} as well as in both in vitro ^{12 14 21} and in vivo ^{8 18} animal studies. The other predominant cell type that we report for the first time in the present study are lens-fiber–like cells. In both anterior subcapsular plaques derived from human biopsy specimens, and those that develop in lenses of our mouse models, we demonstrate the presence of β-crystallin–reactive cells. Semi-thin sections of plaques derived from transgenic mice show these cells to be morphologically similar to elongating lens fiber cells. Further support for this comes from the fact that these cells also express filensin and MIP, two other well established molecular markers for differentiating lens fiber cells. It is important to note that these fiberlike cells do not express α-SMA. Similarly, neighboring myofibroblastic cells, reactive for α-SMA, do not label for β-crystallin, filensin, or MIP, consistent with the presence of distinct cell populations in the subcapsular plaques. The presence of fiber cells is also consistent with the loss of lens epithelial cell markers (e.g., Cx43, Pax6)—additional characteristic features of the normal fiber differentiation process. Earlier studies characterizing the same lines of transgenic mice used in the present study did not report β-crystallin–reactive cells in the anterior subcapsular plaques. ¹⁸ One explanation for this is that these fiber cells are eventually lost from mature subcapsular plaques (Steven P, McAvoy JW, Lovicu FJ, unpublished data, 2002). 

Although TGFβ receptor signaling has recently been shown to be necessary for fiber cell maturation, ³ in our earlier lens explant studies ¹² and in our present in vitro studies, TGFβ did not induce any of the morphologic and molecular changes characteristic of early lens fiber cell differentiation. In all these in vitro studies, the epithelial cells that respond to TGFβ undergo an EMT, transforming into myofibroblastic cells. These cells did not express β-crystallin (Fig. 6) . This is in contrast to all the in vivo models of ASC we described in the present study, where a distinct population of lens-fiber–like cells is clearly evident. These findings indicate that other regulatory factors may play an important role in ASC formation in vivo. For example, it is highly likely that other growth factors (possibly derived from the ocular media) play a prominent role in ASC formation in vivo. By virtue of its well-established role in normal lens fiber differentiation in mammals, a strong candidate for fiber cell induction in anterior subcapsular plaques is FGF. Although FGF has not yet been shown to play a direct role in ASC formation, a role for FGF has been reported in other forms of human cataract. ²⁸ The fact that FGF is expressed in plaques of human ASC (Ishida I, et al. IOVS 2002;43:ARVO E-Abstract 4002), taken together with our report that fiber-like cells contribute to plaques of human ASC, indicates that FGF may play a role in ASC formation. Future studies will be designed to investigate this hypothesis further. 

One of the most striking features associated with the formation of the subcapsular plaques in the present study was the tightly regulated differential gene expression associated with the loss of the epithelial phenotype, to either myofibroblastic or fiber cells. In the transgenic model, the centrally located subcapsular plaques are surrounded by an intact monolayer of lens epithelial cells. As these epithelial cells change in morphology, they progressively contribute to the body of the plaque, in the process switching off epithelial-specific genes, such as Pax6 and Cx43, and switching on others, such as the fiber-specific β-crystallin and filensin genes. Hence, one of the advantages of this animal model is that the morphologic and molecular changes contributing to the development of cataract can be readily followed at any stage of plaque formation. 

As noted earlier, Pax6 is an epithelium-specific gene. Numerous studies have identified this transcription factor, a member of the paired domain family, as a key regulator of eye development (see Ref. ²⁴ for review). Much of this stems from the fact that cells of the eye, including the lens epithelium, are exceptionally sensitive to changes in Pax6 activity levels. ²⁹ Based on the characterization of the transgenic mice in the present study, we propose that the reduction of Pax6 expression in the lens epithelium of these mice may, at least in part, account for the disruption of the lens epithelial monolayer. 

There have been several studies examining ocular morphogenesis in heterozygous Pax6 ^Sey mice ^{20 26} ; however, to date there have been very few detailed histologic studies undertaken to examine postnatal lenses from these mice. In the present study, we report several novel features characterizing the cataracts that form in postnatal eyes of Pax6 ^Sey mice. Although some earlier studies have attributed the presence of cataract in Pax6 ^Sey mice to vacuolization of the lens, ²⁶ the results of the present study, consistent with earlier studies, ²⁰ clearly demonstrate the development of ASCs in these mice. Our identification of both myofibroblastic (α-SMA-reactive) and fiberlike (β-crystallin–reactive) cells in these anterior subcapsular plaques, are consistent with some of the earliest histologic reports by Theiler et al. ²⁰ that first characterized the eye phenotype of Pax6 ^Sey-Dey mice. In adult lenses of heterozygous Pax6 ^Sey-Dey mice, they report the development of a “peculiar cataract” resulting from the presence of a “patch of abnormally differentiated and oriented fibers” in the anterior of the lens. They note the absence of a continuous lens epithelium beneath the capsule of the lens, replaced by “darkly staining slender fibers” (some irregularly swollen). 

As mentioned earlier, in our transgenic mouse lines, we have shown that TGFβ-induced ASC formation is accompanied by a reduction in Pax6 expression. Our characterization of Pax6 ^Sey mice, which also display ASC, leads us to propose that TGFβ in our transgenic mice may negatively regulate Pax6 expression in the lens. Although TGFβ has yet to be shown to regulate Pax6 transcription directly, other members of the TGFβ superfamily have been reported to influence Pax6 expression. For example, activin A negatively regulates Pax6 expression in the spinal cord, ³⁰ whereas BMP7 has been reported to regulate Pax6 expression positively in lens placode formation (see Ref. ² for review). In the chick, phosphorylated SMAD1 labeling (an indicator of BMP signaling) is strongest in the region of early differentiating fiber cells, ⁵ a region corresponding to Pax6 downregulation. Another indication that TGFβ may negatively regulate Pax6 comes from the observation that Pax6 ^Sey mice have many ocular features similar to those found in eyes of transgenic mice overexpressing TGFβ. For example, further to the lens phenotype reported in this study, the cornea of Pax6 ^Sey mice have defects similar to that of the transgenic lines. As previously shown by Srinivasan et al., ¹⁸ examining eyes of transgenic mice overexpressing TGFβ specifically in the lens, and more recently by Davis et al., ³¹ examining eyes from postnatal and adult heterozygous Pax6 ^Sey mice, the corneal epithelium in both animal models is thinner due to a reduction in the epithelial cell layers. Furthermore, the range of severity of the ocular defects in both these animal models is very similar. Heterozygous Pax6 ^Sey mice present ocular defects ranging from relatively normal eyes with corneal opacification and/or cataracts, to microphthalmia or anophthalmia. Depending on the levels of transgene (TGFβ) expressed in the different lines of the transgenic mice, eyes may also show corneal opacification with or without cataract (low level of TGFβ expression), and in more severe cases, microphthalmia (higher levels of TGFβ expression). ¹⁸  

Normal levels of expression of Pax6 are clearly a requisite for differentiation and maintenance of the lens. Not only is Pax6 required for the maintenance of its own transcription, ³² but it has been reported to be important for the regulation of several different genes, namely those of the lens crystallins. ²⁴ In the case of the β-crystallin genes (Cryb), however, Pax6 acts as a negative regulator, binding sites in its promoter resulting in the inhibition, rather than the activation of β-crystallin transcription. ²⁵ This is supported by the inverse spatial expression pattern of Pax6 and β-crystallin in the lens. β-Crystallin expression is upregulated with fiber cell differentiation as Pax6 mRNA expression is progressively lost. Based on this, we cannot entirely rule out that the reduced level of Pax6 demonstrated in cells composing the subcapsular plaques of our transgenic mice in the present study is sufficient to induce the upregulation of β-crystallin expression. This, however, would not explain why other fiber-specific markers, such as filensin, are expressed in subcapsular plaques of lenses of Pax6 ^Sey mice, and also why only a select population of anterior epithelial cells undergo this aberrant fiber differentiation in these mice. 

Overall, with our in vivo animal models, the present study provides important new insights into the formation of human ASC. Not only have we established that subcapsular plaques contributing to this form of cataract comprise a heterogenous population of cells, but for the first time, we have identified a population of these cells to be fiber-like cells. This is consistent with our identification of β-crystallin–reactive cells in biopsy specimens of human ASC. Based on our understanding of the role of growth factors in regulating lens cell behavior, we propose that growth factors such as FGF, in concert with TGFβ, may also play role in ASC formation. Furthermore, as lenses from Pax6 haploinsufficient mice display many ocular defects similar to those we report in ASC, combined with the fact that we observed a decrease in expression of Pax6 in subcapsular plaques of our transgenic mice, we propose that reduced levels of Pax6 may be a major contributing factor in the formation of ASC. 

³  Present address: Department of Anatomy, Christian-Albrecht University at Kiel, Keil, Germany.

 Supported by the Sydney Foundation for Medical Research and the National Health and Medical Research Council (NHMRC) of Australia (FJL).

 Submitted for publication November 4, 2003; revised December 16, 2003, and January 21, 2004; accepted January 24, 2004.

 Disclosure:. F.J. Lovicu, None; P. Steven, None; S. Saika, None; J.W. McAvoy, 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.

 

Immunofluorescent localization of α-SMA in sections of anterior subcapsular plaques in the lens of a P21 transgenic mouse (red reactivity). Sections were counterstained with Hoechst dye (blue reactivity). In these subcapsular plaques, α-SMA was detected in a distinct population of cells that formed a radial band that encircled the plaque. This was evident when sections through the same plaque were compared. Peripheral sections of the plaque show a reactive band of cells across the width of the plaque (A), whereas midsagittal sections have reactive cells on either side of the plaque (B). Scale bar, 100 μm.

Figure 2.

 

Semithin midsagittal section of a region of an anterior subcapsular plaque in the lens of a P28 transgenic mouse (region shown in black in the schematic of the plaque, inset). Toluidine blue staining differentiated two main cell types: a ball of darker staining myofibroblastic cells ( ) underlying a layer of lighter staining lens-fiber–like cells (arrows). Note the presence of ECM-like material (arrowheads) within the ball of myofibroblastic cells. The absence of the lens capsule is an artifact of tissue processing. Scale bar, 50 μm.

Figure 3.

 

Representative midsagittal sections of anterior subcapsular plaques in lenses from P21 transgenic mice. Fluorescence microscopy was used to detect cell nuclei stained with Hoechst dye (A) or immunolabeling for α-crystallin (B), β-crystallin (C), or filensin (D). Bright-field (E–G) and dark-field (H) microscopy was used to detect silver grains representing the spatial expression of mRNA transcripts for αΒ1-crystallin (E), βΒ1-crystallin (F), and MIP (G, H). Note that the more posterior regions of the plaques have little or no expression of the respective genes (B–H, arrows, arrowheads). β-crystallin (F, arrowheads) was present in cell nuclei in some regions of the plaques not expressing MIP mRNA (G, H, arrows). Scale bar, 100 μm.

Figure 4.

 

(A, C, E) Bright- and (B, D, F) dark-field images of Pax6 mRNA expression in anterior regions of eyes from P14 transgenic (A–D) or wild-type (E, F) mice. Normal expression of Pax6 is shown in the corneal epithelium (A, B, arrowheads) and lens epithelium (A, B, E, F, arrows). Higher magnification of the plaque in boxed areas in (A) and (B) is shown in (C) and (D). The levels of Pax6 mRNA in anterior lens cells contributing to subcapsular plaque formation was markedly reduced (C, D, ). Scale bar: (A, B) 200 μm; (C–F) 100 μm.

Figure 5.

 

(A) Bright- and (B, C) dark-field images of Cx43 mRNA expression in anterior regions of eyes of P21 transgenic (A, B) and WT (C) mice. Normal expression of Cx43 was demonstrated in the corneal epithelium (A–C, arrowheads) and lens epithelium (A–C, arrows). In the central lens epithelium of transgenic mice, as the epithelial monolayer was disrupted, there was an abrupt loss of Cx43 transcripts (A, B). Cx43 mRNA was not expressed by any cells contributing to the subcapsular plaques of transgenic mice (B, ). Scale bar, 200 μm.

Figure 6.

 

Sections of whole rat lenses cultured for 6 days in the presence (A, B) or absence (C, D) of TGFβ. Sections were stained with either Hoechst dye (A, C) or immunofluorescently labeled for β-crystallin (B, D). β-Crystallin was detected in lens fibers of both treated and nontreated lenses. The multilayered cells of TGFβ-induced subcapsular plaques were not reactive for β-crystallin (A, B). Scale bar, 100 μm.

Figure 7.

 

Representative sections of anterior regions of eyes from heterozygous P21 Pax6 ^Sey mice, stained for PAS (A, B) or immunolabeled for α-SMA, counterstained with Hoechst dye (C), or immunolabeled for fibronectin (D). (A) Characteristic eye phenotype associated with heterozygous Pax6 ^Sey mice shows the corneal attachment to the underlying lens. (B) A different phenotype: the formation of anterior subcapsular plaques. PAS staining highlights the remodeling of the lens capsule (B, arrows) as well as aberrant ECM deposition within these plaques. Associated with the aberrant ECM in the subcapsular plaques is the presence of myofibroblastic cells, immunoreactive for α-SMA (C, red reactivity) and fibronectin (D). Scale bar, 100 μm.

Figure 8.

 

Representative sections of lenses from heterozygous P21 Pax6 ^Sey mice stained with Hoechst dye (A, B, E) or immunolabeled for β-crystallin (C, F, red reactivity) or filensin (D, G, green reactivity). β-Crystallin (F) and filensin (G) immunoreactivity is shown in differentiating fiber cells at the lens equator. More anteriorly, the anterior subcapsular plaques of these same lenses also contain (C) β-crystallin– and (D) filensin-reactive cells. le, lens epithelium; lf, lens fibers. Scale bar, 100 μm.

Figure 9.

 

Representative serial sections of a human anterior subcapsular plaque, stained with PAS (A), Hoechst dye (B, D), or immunolabeled for α-SMA (C, red reactivity) or β-crystallin (E, green reactivity). Human anterior subcapsular plaques, composed of ECM, demonstrated distinct populations of α-SMA– and β-crystallin–reactive cells. Scale bar, 100 μm.

The authors thank Jessica Boros and Louise van der Weyden for invaluable technical assistance, as well as those who provided generous gifts of materials: Mark Ireland for anti-filensin antibody, Nicolette Lubsen for cDNAs for αB1-crystallin and βB2-crystallin, Anna Chepelinsky for cDNAs for MIP and Cx43, and especially Paul Overbeek for providing the transgenic and Pax6 ^Sey mice.