Diverse gap junctions modulate distinct mechanisms for fiber cell formation during lens development and cataractogenesis

Cataracts, defined as lens opacities, are the leading cause of blindness in the world. To date, cataract surgery remains the only way to cure them and there is no non-surgical method to delay or prevent cataractogenesis. Although decades of studies have significantly improved our understanding of risk factors for developing cataracts, as well as biochemical and morphological changes that occur during cataract formation, the mechanisms for how the lens establishes and maintains its transparency remain largely unknown. The lens is formed through sequential events including the differentiation and elongation of posterior lens vesicle cells to form primary fibers that fill the lumen of lens vesicle, followed by the differentiation and elongation of anterior lens epithelial cells to form secondary fiber cells that lie on top of primary fibers at the lens periphery. Cell elongation is a hallmark of lens fiber differentiation during lens development(Bassnett, 2004; Piatigorsky, 1981). A mechanistic understanding of fiber cell elongation in vivo has relied on speculations based on in vitro studies of cultured lens epithelial cells. Moreover, mechanisms that differentially regulate lens primary and secondary fiber cell formation are unknown.

Lens fiber cells are coupled by intercellular gap junction channels formed by α3 (connexin 46) and α8 (connexin 50) connexin subunits to maintain the homeostasis required for lens transparency(Goodenough, 1992; Mathias et al., 1997). Mutations in the α3 (Gja3) and α8 (Gja8)connexin genes are one of the common causes for inherited cataracts in humans and mice. Connexin proteins have four transmembrane domains with three intracellular regions (the N terminus, a cytoplasmic loop and the C terminus)and two extracellular loops (E1 and E2)(Yeager and Nicholson, 2000). Six connexin subunits oligomerize to form one connexon (hemichannel). A gap junction channel is formed by the docking of extracellular loops of two opposing connexons in the plasma membrane. Hundreds of gap junction channels come together to form gap junctions that are morphologically defined as specialized punctate `plaques' of cell-to-cell contacts. These channels with small pores provide pathways for the direct exchange of small molecules between adjacent cells (Fleishman et al.,2004; Unger et al.,1999). Co-expression of two types of connexin subunits in cells will allow the formation of homomeric connexons (consisting of one type of subunit), heteromeric connexons (consisting of two types of subunits),homotypic channels (the docking of two identical connexons) and heterotypic channels (the docking of two different types of connexons)(Kumar and Gilula, 1996).

Studies of α3^-/- knockout mice have suggested thatα3 connexin is essential for maintaining lens transparency(Gong et al., 1997), while the analyses of α8^-/- knockout mice and knock-inα3(50KI46/50KI46) mice have revealed that α8 connexin is important for lens growth (Martinez-Wittinghan et al., 2003; Martinez-Wittinghan et al., 2004; Rong et al.,2002; White et al.,1998). Although α8^-/- lenses show reduced epithelial proliferation and delayed fiber cell maturation, the mechanism for how the loss of α8 connexin leads to smaller lenses is unknown(Rong et al., 2002; Sellitto et al., 2004). The interaction between α3 and α8 subunits has been suggested by their colocalization in the fiber cells of different vertebrate lenses and the biochemical isolation of heteromeric connexons from the lens(Gong et al., 1997; Jiang and Goodenough, 1996; Konig and Zampighi, 1995; Lo et al., 1996). The α3 and α8 connexin subunits are able to form heteromeric and heterotypic channels in paired Xenopus oocytes and cultured cells in vitro(Hopperstad et al., 2000; White et al., 1994). Thus,diverse gap junctions formed by α3 and α8 subunits need to be further investigated in vivo.

The roles of diverse gap junctions have never been elucidated during lens development due to a lack of an appropriate experimental model in vivo. We hypothesize that dominant cataracts are caused by altered intercellular communication mediated by diverse gap junction channels consisting of mutant and wild-type connexin subunits in the lens. In this work, we have found that the combination of mutant α8-S50P subunits (a mutation in the extracellular loop 1) and wild-type α8 subunits specifically inhibits the elongation of embryonic lens fiber cells, while the combination of mutantα8-S50P and wild-type α3 subunits disrupts the differentiation and elongation of postnatal lens fibers. This work reveals that diverse gap junctions mediate distinct mechanisms to control the formation of lens primary and secondary fiber cells, and this also explains why and how a variety of cataracts can result from perturbations of different types of gap junctions during lens development.

Generation of ENU-mutagenized mice, mouse breeding and genome-wide linkage analysis were described previously (Du et al., 2004). The dominant L1 mutation, in the C57BL/6J(B6) strain background, was outcrossed with wild-type C3H/Hej mice to produce affected G1 hybrid mice that were further crossed with wild-type C3H/Hej mice to produce second generation (G2) mice. The G2 mice were phenotyped with a slit lamp and genotyped with genomic DNA for a genome-wide linkage analysis by using a total of 59 microsatellite markers(Du et al., 2004). Based on the chromosomal location, causative gene candidates were identified from the Mouse Genome Database at the National Center for Biotechnology Information (NCBI)website. PCR fragments of the α8 connexin gene were amplified from mutant genomic DNA isolated from homozygous mutant mice, as previously described (Chang et al., 2002),and followed by DNA sequencing analysis.

Mouse pupils were dilated by using an eye drop containing 1% phenylephrine and 1% atropine before lens clarity was examined using a slit lamp. The cataract was directly imaged in living animals by a slit lamp (Nikon-FS3)using a camera and Kodak elite2 200ASA color slide films. Fresh lenses,dissected from enucleated eyeballs of wild-type and mutant mice, were imaged under a Leica MZ16 dissecting scope using a digital camera.

L1 heterozygous α8^S50P/+ α3^+/+mice were bred with α8^-/- α3^-/- double knockout mice to produce the α8^S50P/- α3^+/-and α8^+/- α3^+/- mutant mice. Both L1 homozygous α8^S50P/S50P α3^+/+ and double mutant α8^S50P/S50P α3^-/- mice were generated from the intercross of α8^S50P/-α3^+/- mice. Double mutant α8^S50P/S50Pα3^-/- mice were bred with α3^-/- knockout mice to produce α8^S50P/+ α3^-/- mice. Previously described PCR methods were used to distinguish α3 or α8 knockout alleles from wild-type or mutant α8 alleles(Chang et al., 2002).

Enucleated eyeballs were fixed in a solution containing 2% glutaraldehyde and 2.5% formaldehyde in 0.1 M cacodylate buffer (pH 7.2) at room temperature for at least 5 days and were postfixed in 1% aqueous OsO₄, stained en bloc with 2% aqueous uranyl acetate and then dehydrated through graded acetone. Samples were embedded in Epon resin (Ted Pella, Redding, CA). Sections (1 μm) were collected on glass slides and stained with Toluidine Blue. Bright-field images were acquired via a Zeiss Axiovert 200 light microscope with a digital camera. Thin sections (80 nm) were cut with a diamond knife, stained with 5% uranyl acetate followed by Reynold's lead citrate before examination under a JEOL JEM-1200EX electron microscope (JEOL,Tokyo, Japan).

A previously described method was used to prepare lens frozen sections for immunohistochemical analysis with different antibodies(Gong et al., 1997). The following reagents were used for immunostaining embryonic lens frozen sections: a rabbit polyclonal antibody against the C-terminal region ofα8 connexin (generously provided by Dr M. J. Wolosin at Mount Sinai School of Medicine, New York), rhodamine phalloidin (Molecular Probe) for detecting F-actin and DAPI (Vector Laboratories) for labeling cell nuclei. Fluorescent images were collected under a Zeiss Axiovert 200 fluorescent microscope with an Axiocam camera or a Leica laser confocal microscope with a high resolution digital camera.

A male cataractous mouse was identified from ENU-mutagenized mice by slit-lamp examination. This male founder was bred with wild-type C57BL/6J (B6)female mice, and half of the offspring developed a phenotype identical to the male founder. This mutant line was named lens mutation 1 (L1) and was maintained in the B6 background (Fig. 1A). The L1 heterozygous mice developed whole cataracts with small eyes (about 30% of the size of wild-type eyes) at weaning age(Fig. 1A,B). Genomic DNA samples of 30 normal and 36 affected G2 mice were used for a genome-wide linkage analysis. The L1 mutation was mapped to mouse chromosome 3 with a Lod score of 6.8 at the linkage marker D3Mit98, which is in the vicinity of the Gja8 (α8 connexin) gene(Fig. 1C).

It is known that α8 connexin mutations cause cataracts in both humans and mice. Therefore, we performed DNA sequencing analysis using PCR fragments amplified from the genomic DNA of L1 homozygous mice. We found a missense mutation (T→C) of the Gja8 gene, resulting in the replacement of the serine residue at codon 50 by a proline residue (S50P) in the extracellular loop 1 (E1-loop) of the α8 connexin protein(Fig. 1D). Thus, the dominant cataracts in the L1 mouse line are caused by the α8-S50P mutation. The genotypes for L1 heterozygous and homozygous mice are labeled as α8^S50P/+ α3^+/+ andα8^S50P/S50P α3^+/+, respectively. The wild-type, homozygous α8 knockout and α8/α3 double knockout mice are labeled as α8^+/+ α3^+/+,α8^-/- α3^+/+ and α8^-/-α3^-/-.

We further confirmed that the α8^S50P/S50Pα3^+/+ mice developed small and ruptured cataractous lenses similar to the α8^S50P/+ α3^+/+ mice at weaning age (Fig. 2A,B). Histological data showed severely disorganized fiber cells, vacuole formation and posterior capsule rupture in both α8^S50P/+α3^+/+ and α8^S50P/S50P α3^+/+lenses (Fig. 2C,D). Thus,secondary fiber cell formation was severely disrupted in bothα8^S50P/+ α3^+/+ andα8^S50P/S50P α3^+/+ postnatal lenses.

As altered fiber cells were obviously observed in the neonatal lenses of both α8^S50P/+ α3^+/+ andα8^S50P/S50P α3^+/+ mice (data not shown), we carried out histological studies of their embryonic lenses. At 15.5 days post-conception (E15.5), both α8^+/+ α3^+/+and α8^-/- α3^+/+ embryonic lenses showed no space between the primary fibers and overlying anterior epithelium(Fig. 3A,B). Unexpectedly, a large cystic lumen between posterior primary fiber cells and anterior epithelium was observed in the E15.5 α8^S50P/+α3^+/+ embryonic lenses with 100% penetrance(Fig. 3C), but not in the E15.5α8^S50P/S50P α3^+/+ embryonic lenses(Fig. 3D). Histology data further verified that at earlier stages (E13.5), posterior primary fiber cells did not reach the anterior epithelium in the α8^S50P/+α3^+/+ lenses, while primary fiber cells in wild-type lenses elongated and obliterated the lumen of the lens vesicle at the same embryonic stage (data not shown).

Thus, although mature fiber cells are severely altered in both postnatal day 21 (P21) α8^S50P/+ α3^+/+ andα8^S50P/S50P α3^+/+ lenses(Fig. 2), the elongation of primary fiber cells is significantly perturbed only inα8^S50P/+ α3^+/+ embryonic lenses, not inα8^S50P/S50P α3^+/+ and α8^-/-α3^+/+ lenses during embryonic development. We hypothesize that α8-S50P is a gain-of-function mutant subunit, as neither a loss-of-function of α8 connexin in α8^-/-α3^+/+ lenses nor a normal function of α8 connexin in wild-type α8^+/+ α3^+/+ lenses inhibits primary fiber cell elongation. Moreover, normal elongation of primary fiber cells in α8^S50P/S50P α3^+/+ lenses suggests that an interaction between mutant α8-S50P subunits and endogenous wild-type α8 subunits is probably required for the suppression of primary fiber cell elongation in α8^S50P/+α3^+/+ embryonic lenses.

We hypothesize that the gain-of-function α8-S50P mutation alters the intercellular communication in lens fiber cells by interacting with endogenous wild-type α8 and/or α3 subunits to form mutant gap junctions. To test this hypothesis, we examined the presence of gap junctions inα8^S50P/+ α3^+/+ embryonic lenses. A representative immunostaining image showed typical punctate fluorescent spots of α8 connexin in an E15.5 wild-type lens frozen section detected by an anti-α8 antibody (Fig. 3E). Similar fluorescent signals were also detected in posterior fiber cells of E15.5 α8^S50P/+ α3^+/+ lens sections (Fig. 3F,G). Transmission electron microscope (TEM) analysis further confirmed the presence of bona fide gap junctions in α8^S50P/+ α3^+/+lenses (Fig. 3H). Therefore, we investigated the subunit composition of mutant gap junctions that perturb the formation of primary and secondary fiber cells in α8-S50P mutant lenses by a genetic approach.

In order to determine the subunit composition of mutant gap junction channels that mediate the inhibition of primary fiber cell elongation inα8^S50P/+ α3^+/+ embryonic lenses, we replaced wild-type α3 and α8 alleles with the null alleles. By breeding theα8^S50P/+ α3^+/+ mutant mice with theα3^-/- knockout and α8^-/- knockout mice, we have generated two different compound mutant mice: α8^S50P/+α3^-/- mice that lack wild-type α3 connexin andα8^S50P/- α3^+/+ mice that lack wild-typeα8 connexin. Histological data showed a cystic lumen only inα8^S50P/+ α3^-/- embryonic lenses(Fig. 4A), but not inα8^S50P/- α3^+/+ lenses(Fig. 4B). Gap junctions were also detected in the fiber cells of these compound mutant embryonic lenses by immunohistochemistry and TEM (data not shown). These results suggest thatα8-S50P subunits interact with endogenous wild-type α8 subunits to inhibit primary fiber cell elongation, while the presence of endogenous wild-type α3 subunits does not affect primary fiber formation. Thus, it is possible that α8-S50P and wild-type α8 subunits form mutant gap junction channels that modulate a unique mechanism essential for the elongation of lens primary fiber cells.

In order to understand why both α8^S50P/+α3^+/+ and α8^S50P/S50P α3^+/+mice developed whole cataracts at weaning age(Fig. 2), we further examined the postnatal lens phenotypes of α8^S50P/+α3^-/- and α8^S50P/- α3^+/+mice. Surprisingly, the α8^S50P/+ α3^-/- mice developed a nuclear cataract rather than a whole cataract. A lens from a P21α8^S50P/+ α3^-/- mouse revealed a nuclear cataract with a transparent cortex (Fig. 4E). Histological data showed degenerated nuclear fibers but normal peripheral cortical fibers in the P14 α8^S50P/+α3^-/- lens (Fig. 4C). By contrast, α8^S50P/- α3^+/+mice developed microphthalmia with whole cataracts similar to theα8^S50P/+ α3^+/+ andα8^S50P/S50P α3^+/+ mice. A lens from a P21α8^S50P/- α3^+/+ mouse was small and ruptured with a cataract (Fig. 4F). Disrupted secondary fiber cells, enlarged vacuole-like extracellular spaces and posterior capsule rupture were observed in the P14α8^S50P/- α3^+/+ lens(Fig. 4D).

These data confirm that the combination of endogenous wild-type α3 and mutant α8-S50P subunits disrupts the formation of postnatal lens secondary fibers, while the presence of endogenous wild-type α8 connexin does not affect secondary fiber formation. These results also provide a molecular explanation for why the L1 heterozygous(α8^S50P/+ α3^+/+) and L1 homozygous(α8^S50P/S50P α3^+/+) mice developed similar phenotypes, such as ruptured lenses and microphthalmia, at the weaning age. Thus, it is possible that α8-S50P and wild-type α3 subunits form mutant gap junction channels that modulate a different mechanism for regulating proper formation of secondary fibers in postnatal lenses.

We have also generated α8^S50P/- α3^-/-mutant mice that lack both endogenous wild-type α8 and α3 connexins and compared their lens phenotypes with those of double knockoutα8^-/- α3^-/- mice. Histological data revealed that both α8^S50P/- α3^-/- andα8^-/- α3^-/- embryonic lenses had normal elongation of primary fiber cells, and both mutant mice developed large nuclear cataracts at the age of three weeks(Fig. 5A). A histological section of a P14 α8^S50P/- α3^-/- lens displayed normal secondary fiber cells in lens periphery but degenerated inner mature fiber cells (Fig. 5B). Lens phenotypes of different types of α8-S50P mutant mice are summarized in Table 1. In summary, these data suggest that mutant α8-S50P subunits probably have no function in vivo and that the interactions between these mutant subunits and endogenous wild-type α8 or α3 connexins perturb primary fiber cell elongation or secondary fiber cell formation, respectively.

Intercellular gap junction channels formed by α1 (Cx43), α3(Cx46) and α8 (Cx50) subunits have been suggested to provide a sophisticated regulatory network to coordinate lens growth and to maintain lens transparency throughout life(Goodenough, 1992). Lens fiber cells are coupled by gap junction channels consisting of α3 and α8 connexin subunits. This work demonstrates that the combination ofα8-S50P and endogenous wild-type α8 subunits specifically inhibits the elongation of primary fiber cells in embryonic lenses, while the combination of α8-S50P and endogenous wild-type α3 subunits disrupts the proper formation of secondary fiber cells in postnatal lenses. These results suggest that gap junctions formed by α8-S50P and wild-typeα8 subunits alter a unique mechanism required for the elongation of lens primary fiber cells, while gap junctions formed by α8-S50P and wild-typeα3 subunits perturb a separate and distinct mechanism needed for proper formation of secondary fiber cells in postnatal lenses. Thus, this work provides the first in vivo evidence for a working model that diverse gap junction communications modulate different signaling mechanisms for primary fiber cell elongation or secondary fiber cell formation during lens development (Fig. 6).

This work provides an in vivo model to understand a fundamental mechanism for why and how diverse gap junction channels are used in almost all organs supported by the fact that two or more types of connexin subunits are commonly co-expressed in cells (Goodenough et al.,1996; Kumar and Gilula,1996). It is generally accepted that diverse gap junctions provide a broad spectrum of pathways to ensure the homeostasis needed for various cellular functions in vivo. However, studies to define the roles of these diverse gap junctions have previously been hindered by an inability to distinguish gap junction channels formed by different types of subunits from the channels formed by one type of subunits in vivo. Previous results from studies of loss-of-function knockout mice suggest that α3 connexin is essential for maintaining lens transparency, while α8 connexin is required for lens growth. Functional differences between channels consisting of mixed α8 and α3 subunits and channels consisting of eitherα8 or α3 subunits in vivo are unclear(Martinez-Wittinghan et al.,2003). Studies of the α8-G22R mutation (a mutation in the N-terminal domain) have demonstrated that severe lens phenotypes are partly caused by the mutant gap junction channels consisting of both mutantα8-G22R and endogenous wild-type α3 subunits(Chang et al., 2002). Theα8-G22R mutant probably acts as a dominant-negative inhibitor to perturb gap junction communication by oligomerizing with wild-type α8 and/orα3 subunits. This predication is also supported by the fact that the elongation of primary fiber cells is normal in both α8-G22R heterozygous and homozygous embryonic lenses (C.-h.X, D.C. and X.G., unpublished), and inα8^-/- α3^-/- lenses without gap junction channels. Therefore, based on the result that a combination of α8-S50P and wild-type α8 subunits inhibits the elongation of primary fiber cells, we propose that α8-S50P mutant proteins are gain-of-function subunits that perturb the intercellular gap junction communication or possibly hemichannels by interacting with wild-type α8 subunits. A gain-of-function approach has never been used to investigate the roles of intercellular gap junction communications in vivo. The downstream signaling mechanisms modulated by diverse intercellular gap junctions formed byα8-S50P and α8 or α3 connexins in the lens are unknown. Thus, the α8-S50P mutation provides a useful experimental model with which to further investigate the mechanisms that uniquely regulate the elongation of primary fibers and the formation of secondary fiber cells during lens development.

To date, six α3 mutants (F32L, P59L, N63S, P187L, S380fs and N188T)and five α8 mutants (R23T, E48K, P88S, I247M and V64G) have been linked to dominant cataracts in humans (Bennett et al., 2004; Berry et al.,1999; Jiang et al.,2003; Li et al.,2004; Mackay et al.,1999; Polyakov et al.,2001; Rees et al.,2000; Shiels et al.,1998; Willoughby et al.,2003; Zheng et al.,2005). In addition, three α8 mutants (G22R, D47A and V64A)cause different dominant cataracts in mice(Chang et al., 2002; Graw et al., 2001; Steele et al., 1998). Previously reported mutations of α8 connexin are illustrated in a topological model of α8 protein (Fig. 7). These mutations cause a variety of dominant cataracts in humans and mice. A previous study of human α8-P88S mutant (located in the second transmembrane domain) suggests a hypothesis that the dominant cataract is caused by cellular defects triggered by mistrafficking and intracellular accumulation of toxic aggregates composed of mutant connexin proteins (Berthoud et al.,2003). However, this hypothesis cannot explain the cataracts caused by the α8-S50P mutation and the α8-G22R mutation. Data from our current study of the α8-S50P mutation and our previous study of theα8-G22R mutation reveal that different types of cataracts result from a perturbation of intercellular gap junction communication by the specific interactions between mutant subunits and endogenous wild-type subunits during lens development. Therefore, our current work demonstrates a new mechanistic explanation for why and how different connexin mutations lead to a variety of cataracts. Moreover, this new mechanism explains how mutations of other connexin isoforms cause different diseases in other organs, such as a recent finding that the Cx43-G60S mutation in extracellular loop 1 causes oculodentodigital dysplasia (Flenniken et al., 2005). The α1 connexin (Cx43 or Gja1) is predominantly expressed in lens epithelial cells. We have not characterized the properties of epithelial cells in different α8-S50P mutant mice. It would be an interesting area to explore in future studies.

Fiber cell elongation is a morphological hallmark of lens fiber cell differentiation. The mechanism for fiber cell elongation has never been investigated in vivo due to a lack of proper experimental systems. Current understanding of fiber cell elongation is based on the results obtained from elongation studies of cultured lens epithelial cells in vitro(Beebe et al., 1982; Parmelee and Beebe, 1988) and some descriptive in vivo information(Bassnett, 2005; Kuszak et al., 2004). Several transcriptional factors have been reported to be essential for the formation of lens primary fiber cells (Kim et al.,1999; Ogino and Yasuda,2000; Wigle et al.,1999). However, mechanistic differences between lens primary and secondary fiber formation are unknown. No previous results have ever demonstrated the involvement of α8 connexin or gap junction channels in the regulation of elongation or formation of lens primary fiber cells. This work provides molecular evidence for regulatory differences between lens primary and secondary fiber cells.

The α8-S50P mutant subunits alone are nonfunctional in the lens. The mutant subunits must rely on the presence of wild-type α8 subunits to inhibit the elongation of primary fiber cells. However, the combination ofα8-S50P and wild-type α8 subunits does not disrupt the formation of secondary fiber cells in postnatal lenses. As neitherα8^-/- knockout nor α8^-/-α3^-/- double knockout inhibits the elongation of lens primary fibers, a complete blockage of fiber-to-fiber coupling is not responsible for the inhibition of primary fiber cell elongation. Normal fiber-to-fiber coupling does not inhibit primary fiber cell elongation in wild-type lenses. We hypothesize that gap junction channels formed by α8-S50P and wild-type α8 subunits gain their mutant function by increasing channel permeability to facilitate the fiber-to-fiber transport or by transmitting a unique signal that inhibits the elongation of primary fiber cells. Future experiments will be needed to evaluate this hypothesis or an alternative hypothesis that the interaction of α8-S50P and wild-type α8 subunits perturbs a non-junction signaling event to inhibit the elongation of primary fiber cells.

The molecular basis for the proper formation of differentiating secondary fiber cells and the maturation of fiber cells remains largely unknown despite substantial descriptive information that have been published in the last few decades. This work demonstrates that α8-S50P must rely on the presence of endogenous wild-type α3 to disrupt the formation of secondary fiber cells in postnatal lenses. A previous study of the α8-G22R mutation also revealed the importance of the interaction between mutant α8 and wild-type α3 connexins in regulating the formation of secondary fiber cells (Chang et al., 2002). These results suggest that the intercellular gap junctions formed by α8 and α3 connexin subunits modulate the formation of secondary fiber cells in the lens periphery.

In summary, this work reveals that the α8-S50P mutation causes lens cataracts via distinct mechanisms, which are different from those causing cataracts in the α8^-/- knockout and α8-G22R mutant mice. More importantly, this study clearly demonstrates that connexin point mutations such as α8-S50P can be used as useful experimental models to investigate fundamental mechanisms during lens development and cataractogenesis in vivo.

We thank Alex Park and Eliza Lee for their assistance. We also thank other members of the Gong laboratory for critical reading this manuscript. This work was supported by the National Eye Institute grants EY13849 to X.G. and EY05314 to W.K.L.