Ординатура / Офтальмология / Английские материалы / Eye, Retina, and Visual System of the Mouse_Chalupa, Williams_2008

collaborate to focus a clear image onto the retina. In terrestrial vertebrates, the lens is responsible for accommodation for near or far vision. This is accomplished by muscular activity outside the lens either to move the lens forward and backward or to change the shape of the lens within the eye (reviewed in Robinson and Lovicu, 2004). Perhaps more important than the functional role of the lens in the adult animal is the instructive role the lens plays in the overall development of other structures in the eye. The lens is the major determinant of the overall size of the vertebrate eye, and a small lens (microphakia) always results in a small eye (microphthalmia). In addition to dictating ocular size, the lens is absolutely required for the development of the anterior chamber and the neural crest–derived structures of the anterior segment, including the corneal stroma, corneal endothelium, iris stroma, and the trabecular meshwork (Beebe and Coats, 2000). The precise molecular signals sent by the lens to these developing structures are unknown.

History of lens research: The rise of experimental embryology

Although interest in the lens and its development predates the twentieth century (reviewed in Robinson and Lovicu, 2004), lens research has remained at the forefront of developmental biology since the classic experiments of Hans Spemann (1901). In these studies, Spemann demonstrated

that the OV was required for the subsequent formation of the lens from the overlying PLE in the frog, Rana fusca. This finding was the first description of the concept of embryonic induction. In addition to revealing a fundamental concept of developmental biology, these experiments were also the first to introduce microsurgery as an embryological tool (reviewed in Okada, 2000). Spemann’s discovery of embryonic lens induction spawned numerous subsequent investigations of lens development by Spemann and others, some of which demonstrated that the OV was not universally required for vertebrate lens induction (King, 1905; Lewis, 1904; Mencl, 1903; Spemann, 1907). These species-specific differences in the importance of the OV in the process of lens development have been addressed in more recently developed models, which have demonstrated that the OV is simply the last in a stepwise series of embryonic tissues responsible for inducing vertebrate lens formation (reviewed in Fisher and Grainger, 2004).

Nonmurine model systems dominated lens research throughout much of the twentieth century. Spemann’s initial discovery of lens induction in frogs, coupled with the accessibility and rapid development of tadpoles, fostered the popularity of this model system for embryonic studies. The discovery of lens regeneration in select species of salamanders and fishes (reviewed in Okada, 2000, 2004) also led to productive investigations of lens development in these species. The developing chick also provided an excellent

model for experimental studies on lens development. Chick embryos have large eyes and, like amphibian embryos, are amenable to surgical manipulation during development. This property was exploited by Alfred and Jane Coulombre in classic lens reversal experiments which revealed that the polarity of the lens (with the lens epithelium normally facing the cornea) is dependent on extrinsic factors within the eye (Coulombre and Coulombre, 1963). In contrast to chicks and tadpoles, mammalian embryos develop within a largely inaccessible uterine environment, making experimental manipulations more difficult. These difficulties are compounded in the mouse, as the developing eye is extremely small.

History of lens research: Enter the mighty mouse

Rodents have the advantage of being easy and inexpensive to breed in captivity, with short generation times and multiple animals per litter. Less than a decade after Hans Spemann’s initial experiments on lens induction in frogs, early mouse geneticists began creating the first inbred mice (reviewed in Morse, 1978). These animals, largely descendants of mice domesticated centuries ago by Asian and European mouse fanciers, were initially created to investigate cancer (Paigen, 2003a) but have subsequently served as keystone species for investigating all aspects of mammalian biology. Inbred mice, according to the 1989 International Committee on Standardized Nomenclature for Mice, are the result of at least 20 generations of brother-sister mating (Silver, 1995). This level of inbreeding theoretically results in genetic homozygosity for 98.7% of the genome (Green, 1981). In essence, this means that all animals in a given inbred strain are basically genetically identical. All of the common, commercially available inbred strains have been inbred many more than 20 generations, dictating that any heterozygosity within an inbred strain is the result of mutations arising since the initial creation of the strain. From the beginning, it was the power of mouse genetics that provided the most potent experimental benefit to this unassuming rodent.

Although the mouse lens did not start out as the most popular developmental system for investigations of lens development, it was not entirely ignored even in these early years. In 1938, Herman Chase obtained a strain of anophthalmic (having no eyes) mice from Clarence Cook Little (founder of the Jackson Laboratory in Bar Harbor, Maine). Chase continued inbreeding these mice, forming the strain now known as ZRDCT. Chase published a number of papers on these mice, the first of which described the ocular morphology of developing mutant embryos (Chase and Chase, 1941). The key morphological feature of these mice was an OV that was smaller than normal and that failed, in most cases, to make good contact with the overlying PLE.

About 90% of mice in the ZRDCT strain are anophthalmic, and the remainder are microphthalmic (having very small eyes). Chase also investigated the genetics of the ZRDCT strain by making crosses with other inbred mouse strains. These crosses confirmed that the anophthalmic trait was recessive, as all F1 progeny from crosses involving one ZRDCT parent had normal eyes. Subsequent crossing suggested that the anophthalmic trait was not the result of a single mutant gene but, more likely, the combination of a major genetic determinant, designated ey1, and a small number of additional modifying genetic characters (Chase, 1942). In later experiments, Chase hypothesized that most of the ZRDCT anophthalmic phenotype was the result of homozygosity for ey1 and a major modifier that he designated ey2 (Chase, 1944). Therefore, anophthalmia in the ZRDCT mice represented an early example of a complex genetic trait in mammals.

The ZRDCT mice are highlighted here for a number of reasons. First, these studies were among the first, if not the first, published works exploiting both the embryology and genetics of mice to study eye development. Second, these mice demonstrate that although the PLE undergoes inductive influences prior to encountering the OV (reviewed in Fisher and Grainger, 2004), physical contact between the OV and the surface ectoderm is indeed important for normal development of the mouse lens and eye. Third, the ZRDCT strain still exists nearly 70 years after Clarence Little first noticed their anophthalmic ancestors, and these mice continue to inspire scientific investigation. The ey1 mutation was identified by positional cloning as a point mutation in an alternative start codon for the Rx/rax homeobox transcription factor (Tucker et al., 2001), but the identity of the ey2 modifier gene(s) remains unknown. Rx/rax is expressed in the OV but is not expressed in the PLE or the lens, so the effects of the ey1 mutation on lens development are indirect. Mice homozygous for null (loss of function) mutations in Rx/rax fail to form OVs and are also anophthalmic (Mathers et al., 1997), and homozygous mutations in the human Rx/ rax orthologue are associated with human anophthalmia (Voronina et al., 2004). In many ways, the ZRDCT mice studies pointed the way toward the power of mouse genetics to elucidate problems of eye development.

There were other, nongenetic, studies of mouse eye development that were notable in these early years. As early as 1948, Ida Mann showed that mouse lens epithelial cells were capable of elongating in vitro (Mann, 1948) preceding similar findings by others in chicks (Philpott and Coulombre, 1965) and amphibians (McDevitt and Yamada, 1969). Muthukkaruppan performed tissue recombination experiments using E9 mouse embryos to explore the presumed requirement of the OV to induce formation of the mammalian lens (Muthukkaruppan, 1965). He found that both a lens and a retina could form in vitro from explanted mouse

OV with surrounding mesenchyme and PLE. In these experiments, lenses (as judged by the morphological appearance of elongated fiber cells) were not able to form from isolated PLE, from isolated OV, or from PLE cultured in association with various sources of embryonic mesenchyme. Muthukkaruppan was able to achieve lens formation from the PLE in association with optic cup tissue from E10 embryos or from E13 neural retina, but not from E13 spinal cord, demonstrating that lens-inductive influence was not present in all embryonic neural tissue. Furthermore, the lens-inductive influence of the optic cup or neural retina was able to pass through a Millipore filter (Muthukkaruppan, 1965).

Yujiro Yamamoto used lenses from 6- to 8-day-old mouse pups to perform lens transplantation experiments into the eyes of adult mice (Yamamoto, 1976). Yamamoto found that when a normal lens was placed into a normal adult eye with the normal orientation (epithelial cells facing the cornea), lens growth continued and the lens remained transparent for more than 5 months. As in experiments performed more than a decade earlier in chicks (Coulombre and Coulombre, 1963), mouse lenses transplanted into mouse eyes in reverse orientation (epithelial cells facing the retina) underwent a reorientation. In other words, the lens epithelium facing the retina elongated into fiber cells, and a new monolayer of epithelium formed over the original posterior surface of the lens facing the cornea. Yamamoto found that the reformation of the lens epithelium on the posterior surface of the reversed lens (facing the cornea) did not depend on the recipient mouse having an intact neural retina, but that the neural retina was necessary for the differentiation of fibers of the original lens epithelium in the reversed donor lens or for the continued lens growth in the correctly oriented donor lens (Yamamoto, 1976).

Perhaps the first morphological descriptions of normal mouse eye and lens development were provided by Paul Leonhard Kessler in 1877, and these were followed by Rugh in 1968 and beautiful photographic views of mouse eye development by Pei and Rhodin in 1970, and more recently by Kaufmann in 1992. Other morphological processes such as fiber cell denucleation (Kuwabara and Imaizumi, 1974; Vrensen et al., 1991) have been covered in the mouse lens using transmission electron microscopy, and excellent scanning electron micrographs of mouse eye and lens development by Kathleen Sulik can be found online (www.med.unc. edu/embryo_images).

One essential component of lens development and growth is lens cell proliferation. Cell proliferation in the lens is confined to the lens epithelium, and some of the finest early characterization of the normal mouse lens epithelium was carried out by Nancy Rafferty using adult CF1 strain outbred (nongenetically identical) mice (Rafferty, 1972; Rafferty and Rafferty, 1981). Rafferty used 3H-thymidine labeling to

characterize cell cycle kinetics in the mouse lens epithelium. These studies estimated that the mouse lens epithelium consisted of 44,000 cells that could be subdivided based on position and rate of proliferation. This estimate included approximately 3,600 cells of the most peripheral cells, where nuclei line up into what are called meridional rows. These cells are currently characterized as very young fiber cells. Rafferty also observed a slowly cycling subpopulation of approximately 5,000 cells in the periphery of the lens epithelium whose divisions would ultimately fuel the production of secondary fiber cells at the rate of 207 cells per day (Rafferty and Rafferty, 1981). This zone is now referred to at the germinative zone (figure 22.1G). An adjacent zone approximately seven cell diameters thick and consisting of approximately 9,000 epithelial cells that were no longer incorporating 3H-thymidine but had not yet begun to line up in meridional rows was also denoted by the Raffertys. This zone of postmitotic epithelial cells is now referred to as the transitional zone (see figure 22.1G), as these are epithelial cells in transition to becoming secondary fiber cells. Although these studies were not the first to recognize that different subpopulations of lens epithelial cells with different proliferative potentials exist in vertebrate lenses, they were benchmarks in the specific characterization of the mouse lens epithelium and served as a framework by which later molecular studies of mouse lens cell proliferation could proceed.

The relationship of mouse and human lenses

Mice, like humans, are mammals and therefore share a closer genetic relationship with humans than with other nonmammalian vertebrates such as amphibians or chickens. In fact, humans are genetically closer to mice than to cats, dogs, cows, pigs, or many other familiar nonprimate mammals (Reyes et al., 2004). Insofar as cataracts remain the number one cause of human blindness worldwide (Resnikoff et al., 2004), the study of lens development in an animal model closely related to humans has obvious potential medical benefit. That said, mice are not simply small, furry humans, and there are fundamental differences between the mouse and human lens. The mouse lens is much rounder and takes up a larger percentage of the eye volume than does the flatter human lens. The mouse and human lens also differ in the arrangement and type of lens sutures (reviewed in Kuszak and Costello, 2004). Lens sutures are formed when differentiating fiber cells lose contact with the basement membrane and associate with the ends of other differentiating fiber cells. This process explains why older, more mature fiber cells are found deep within the lens, while the youngest fiber cells remain in contact with the lens capsule at the surface. Mouse lenses have Y-suture lenses and primates have star suture lenses. Suture patterns, and the overall geometry of individual fiber cells, also dictate

the accommodating power of the lens, with human lenses having significant accommodation power, versus the negligible accommodating power of the mouse lens (Kuszak et al., 2006).

There are several biochemical differences between mouse and human lenses as well. Among these are differences in the composition of lens membrane lipids. Human lens cell membranes contain large amounts of dihydrosphingomyelin, a lipid virtually absent from mouse lens membranes, and mouse lens membranes have far more phosphatidylcholine than the membranes of human lenses (R. Truscott, pers. comm.). These differences in murine and human lenses are more apparent in their mature forms than during embryonic development, and there is strong reason to suspect that nearly all the genetic pathways dictating mouse and human lens development are similar.

Beware: Not all mice are created equal

The existence of inbred mice is one of the great strengths of working with the mouse system. Each inbred strain can be thought of as a pool of identical individuals from a genetic standpoint. At the same time, different inbred strains represent different pools of genetically identical individuals. It is prudent to be aware of the possible complications inherent to the interpretation of phenotypes arising in a particular strain of inbred mice. Other chapters in this book alert readers to the common Pde6brd1 mutation leading to rapid postnatal retina degeneration in many inbred mouse strains. Lurking in the eyes of several “normal” inbred mouse strains are mutations and polymorphisms that have an impact on lens development as well. Among the more recently identified of these is a null mutation in the Bfsp2 gene encoding the lens fiber cell–specific beaded filament protein, CP49. This Bfsp2 mutation is shared by several inbred strains, including 129X1/SvJ, 129S1/SvImJ, 129S4/SvJae, 129S2/ SvPas, C3H, CBA, 101 and FVB/N (Alizadeh et al., 2004; Sandilands et al., 2004; Simirskii et al., 2006). As a result, the lenses of these mice lack intermediate beaded filaments. Although the overall development of the lenses from Bfsp2 mutant mice appears normal, beaded filament loss leads to increased light scattering in the lens and alterations in other optical properties as well (reviewed in Perng and Quinlan, 2005). FVB/N and 129 are the most commonly used inbred strains for the production of transgenic and knockout mice, respectively. Therefore, it is not unlikely that mice derived from genetic engineering approaches carry Bfsp2 mutations.

Inbred mice derived from the C57 Black strains (most notably C57BL/6 and C57BL/10) are inherently prone to develop microphthalmia and/or anophthalmia. This was first noted in the experiments of Herman Chase, who used C57 Black mice as a control strain in comparisons with the

ZRDCT anophthalmic strain (Chase, 1942; Chase and Chase, 1941). The frequency of spontaneous microphthalmia or anophthalmia in C57BL mice has been reported to be as low as 4.4% (N = 2,200; Chase, 1942) to as high as 9.6% (N = 2,709; Kalter, 1968). For reasons that remain unclear, eye defects are more common in females than in males and, when unilateral, are more commonly seen in the right eye than in the left (Chase, 1942; Kalter, 1968). There is also evidence that environmental factors influence the frequency of eye defects in C57BL mice (Pierro and Spiggle, 1969), and C57BL/6 mice are particularly susceptible (compared to C3H, CF-1, and CD-1 strains) to ocular defects resulting from teratogenic insults such as alcohol (Cook et al., 1987). The apparently stochastic appearance of ocular defects within genetically identical C57BL mice suggests an inherent genetic deficiency of eye development universally present in the strain. Like the eye defects present in the ZRDCT strain (although seen at a much higher frequency), eye defects in C57BL mice are most likely a complex genetic trait consisting of multiple genetic loci.

There are two main hypotheses as to the mechanistic nature of C57BL ocular defects. The first derived from observations that C57BL OVs, like those of the ZRDCT strain, are smaller than normal and make less than optimal contact with the PLE, resulting in a smaller lens (Cook and Sulik, 1986; Harch et al., 1978). The second hypothesis is that the defects in C57BL mice are intrinsic to problems within the PLE or the lens cells derived from the PLE (LoCascio et al., 1987; Robinson et al., 1993). This hypothesis descends from the analysis of ocular tissues derived from chimeric mice made from the aggregation of preimplantation embryos from C57BL/6 mice and DBA/2 or A/J strain mice. Chimeras made with C57BL/6 embryos demonstrated a uniform and dramatic underrepresentation of C57BL/6 cells in the lens relative to the overall chimeric contribution of C57BL/6 cells to other ocular and nonocular tissues. Notably, the OV-derived chimeric retinas tended to have a C57BL/6 contribution reflective of the overall C57BL/6 chimeric contribution. Even those supporting the lens-intrin- sic hypothesis noted consistent morphological differences in the early stages of lens development between C57BL/6 and control (DBA/2 and A/J) embryos resulting in a universally smaller initial lens pit and vesicle (Robinson et al., 1993). The fact that most C57BL/6 mice do not exhibit frank postnatal microphthalmia suggests that size deficiencies present early in lens development can largely be compensated for, presumably by increased proliferation. It is interesting to note, however, that adult C57BL/6 mice were recently shown to exhibit a lens axial thickness significantly (P < 0.001) smaller than that of DBA/2 (Puk et al., 2006). Unfortunately, the molecular nature of C57BL/6 ocular deficiencies remains unknown. Perhaps with the application of current molecular genetic tools the answer will be

forthcoming. In any case, C57BL/6 is among the most commonly used inbred mouse strains, and it is frequently used as a “normal” control strain. It is also common practice to breed mutations created in 129 strain-derived embryonic stem (ES) cells onto a C57BL/6 genetic background. Therefore, it is important for those interested in ocular development to realize the C57BL/6 mice do carry a genetic predisposition to ocular abnormalities.

Molecular biology of the lens: Developing the tools

The first half of this chapter described the historical context in which developmental studies on the mouse lens arose. It was only with the advent of molecular techniques and, most important, the ability to experimentally manipulate the mouse genome that the mouse truly became the keystone for mechanistic understanding of vertebrate lens development. It is appropriate here to briefly discuss lens crystallins. The literature on lens crystallins is rich and enormous, far greater than can be adequately discussed here.

Lens biochemists first isolated and characterized crystallin proteins from nonmurine sources, and resultant antibodies to these proteins gave lens biologists the tools to follow development in tissue sections. Despite their critical functional role in maintaining lens clarity, lens crystallins are likely not major regulators of lens development. Crystallins have, however, been used extensively as molecular markers of lens differentiation, and eventually any genetic pathway leading to lens differentiation will ultimately regulate crystallin gene expression. Pioneering work on the regulation of crystallin gene expression made the first transgenic studies on lens development possible. Crystallins are the most abundant water-soluble proteins in the lens, and their high concentrations are necessary to maintain lens transparency. The mammalian lens contains α-, β-, and γ-crystallins. In the mouse there are multiple genes encoding each crystallin subclass:

αA-, αB-, βA3/A1-, βA2-, βA3-, βB1-, βB2-, βB3-, γA-, γB-, γC-, γD-, γE-, γF-, and γS-crystallin (reviewed in Duncan et al., 2004). The developmental expression of the three different mammalian crystallin classes was first shown by John McAvoy with immunocytochemical studies in the rat (McAvoy, 1978). McAvoy showed that α-crystallins were expressed in lens epithelial cells and lens fiber cells, while β- and γ-crystallins were restricted to lens fiber cells. McAvoy also demonstrated that β-crystallin expression precedes γ- crystallin expression in the fiber differentiation pathway. This expression pattern was later seen by others in the mouse. The first crystallin genes expressed during mouse development are the α-crystallins (Zwaan, 1983), with αBcrystallin initiating expression at the lens placode stage and αA-crystallin appearing at the lens pit stage (Haynes et al., 1996; Robinson and Overbeek, 1996). Both α-crystallin genes are expressed in lens epithelial cells and lens fiber cells,

but αA-crystallin is expressed at significantly higher levels in lens fiber cells. β- and γ-crystallins have been and continue to be used as markers of lens fiber cell differentiation. This is not inappropriate, as the expression of γ-crystallin transcripts in the rodent embryo does not initiate until after fiber cell differentiation commences (Goring et al., 1992; Van Leen et al., 1987), and antibodies to β-crystallins do not detect these proteins in the embryonic mouse lens epithelium (Carper et al., 1986). Recently, however, βB1and γScrystallin proteins were found to be reasonably abundant in normal postnatal rat lens epithelium, and transcripts for βA2- and γS-crystallins are induced in lens epithelial cells cultured in nondifferentiating media, demonstrating that these particular crystallins are not exclusive to lens fiber cells (Wang et al., 2004).

The mouse αA-crystallin gene was the first crystallin promoter to be cloned and analyzed in detail (King and Piatigorsky, 1983). These analyses began by fusing 5′ flanking regions of the αA-crystallin gene to a chloramphenicol acetyl transferase (CAT) reporter gene. Two methods were initially used to evaluate the αA/CAT constructs. The first was to transfect the αA/CAT construct into primary explants of chick lens epithelia and then to analyze the explant for expression of the CAT reporter gene. The constructs containing 364 nucleotides upstream and 46 nucleotides downstream of the transcription initiation site (−364/+45) of the mouse were capable of driving CAT expression in explanted chick lens epithelia but were not capable of promoting CAT expression in chick fibroblasts. In contrast, a shorter, −87/+45 construct was not capable of eliciting CAT expression in either chick lens epithelia or fibroblasts (Chepelinsky et al., 1985). These experiments first demonstrated that sequences between −87 and −364 of the mouse αA-crystallin gene contained sufficient information to drive the expression of a heterologous gene (CAT) in a tissue-specific way, and that chicken regulatory proteins were able to interpret the instructions encoded in mouse noncoding DNA. Similar previous experiments by Kondoh and colleagues demonstrated that a genomic clone for δ-crystallin (a crystallin not normally expressed in the mammalian lens), containing 2 kb of upstream noncoding sequence, was able to direct δ-crystallin synthesis following microinjection into cultured mouse lens epithelial cells but not into nonlens cells (Kondoh et al., 1983). These experiments showed that the regulatory sequences directing gene expression to the lens exhibit amazing functional conservation despite millions of years of evolution separating chickens from mice.

Lens research enters the transgenic era

The second method used to investigate the regulatory activity of the αA-crystallin gene 5′-flanking region was to inject the αA/CAT constructs into pronuclear stage mouse zygotes

to create transgenic mice. The −364/+45 αA/CAT construct exhibited lens-specific CAT expression in the resultant transgenic mice, and expression was detected as early as E12.5 in the mouse embryo (Overbeek et al., 1985). At the time, this was very new technology. The first transgenic mice ever made were published in 1980 and 1981 (reviewed in Paigen, 2003b). The αA/CAT transgenic mice published in 1985 were the first transgenic mice produced in which transgene expression was directed specifically to the visual system (Overbeek et al., 1985). For the first time in the history of lens research it was possible to ectopically express virtually any gene in the developing lens.

The αA/CAT transgenic mice opened a virtual floodgate to lens transgenic experiments. Many of these were designed to directly test various promoter constructs for their lens activity and specificity in vivo. Transgenic mice were soon being made with the murine −759/+45 γF-crystallin promoter, in which transgene expression was directed specifically to central lens fiber cells (Goring et al., 1987), and a δ-crystallin genomic clone containing a 2.5 kb promoter that directed δ-crystallin expression to both the lens and the brain (Kondoh et al., 1987). Transgenic studies proved to be the gold standard for functional activity of lens promoter regions in vivo. Over the years, several different promoter constructs have directed transgene expression to the mouse lens with varying degrees of lens specificity. Among these are various crystallin promoters from a variety of species (reviewed in Duncan et al., 2004), as well as the chicken vimentin (Capetanaki et al., 1989) and human keratin 14 (Nguyen et al., 2002) promoters and elements of the murine Pax6 promoter (Ashery-Padan et al., 2000; Williams et al., 1998). These promoters have proven to be valuable tools for lens developmental biologists.

A common misconception, even among lens biologists, is that the overexpression or ectopic expression of any gene in the lens will cause a cataract. While the expression of many genes in the lens does lead to cataract, this is not universally true. In fact, a number of different transgenes have been expressed ubiquitously, as well as in the lens specifically, without causing lens cataracts. The first lens transgenic mice produced with the αA/CAT construct did not develop cataracts (Overbeek et al., 1985).

Other lens transgenic experiments were designed to ectopically express ectopic proteins in the lens to alter normal lens development. Lenses are not prone to the development of tumors in any vertebrate species, but transgenic mice expressing the SV40 T-antigen readily formed lens tumors (Mahon et al., 1987). These lens tumor–bearing mice spawned numerous productive investigations of cell cycle regulation and apoptosis in which the transgenic mouse lens was used as a model system (reviewed in Griep and Zhang, 2004). Other early ectopic lens expression experiments directed toxin genes to the lens to achieve lens-specific cell

ablation (Breitman et al., 1987; Landel et al., 1988). Specific loss of lens cells in these mice uniformly led to microphthalmia, demonstrating that eye size in mice is directly dependent on lens size during development.

In addition to the ectopic expression of viral proteins, reporter genes, and toxin genes, the lens was used to express a wide variety of growth factors. Some of these growth factors were expressed endogenously by the lens and others were not normally expressed by the lens at all. Often these experiments were designed to influence either the development of the lens or the development of tissues in the eye surrounding the lens. Growth factors and cytokines expressed in the lens of transgenic mice include transforming growth factor-alpha (TGF-α; Decsi et al., 1994; Reneker et al., 1995, 2000), epidermal growth factor (EGF; Reneker et al., 2000), TGF-β (Flugel-Koch et al., 2002; Srinivasan et al., 1998; Zhao and Overbeek, 2001), bone morphogenic protein 7 (BMP7; Hung et al., 2002), vascular endothelial growth factor (VEGF; Ash and Overbeek, 2000), insulinlike growth factor-I (IGF-I; Shirke et al., 2001), plateletderived growth factor (PDGF; Reneker and Overbeek, 1996), neurotrophin-3 (NT3; Robinson, 2008), insulin (Reneker et al., 2004), optineurin (Kroeber et al., 2006), interleukin-1β (IL-1β; Vinores et al., 2003), leukocyte inhibitory factor (LIF; Graham et al., 2005), and several different fetal growth factors (reviewed in Robinson, 2006). In addition, a number of different growth factor receptor genes (wild-type, constitutively active and dominant negative) have been expressed in the lens. These have contributed significantly to our understanding of the regulation of lens development, particularly in the regulation of lens fiber differentiation (reviewed in Lovicu and McAvoy, 2005). Although transgenic-mediated overexpression provided insights into lens development, they were limited. It was often difficult to determine if lens phenotypes in transgenic mice were the result of a fundamental interference with a specific normal developmental process or if overexpression of proteins in the lens resulted in nonspecific, ectopic effects.

Customizing the mouse genome to study lens development

The advent of homologous gene targeting in mouse ES cells quickly made it possible to delete or alter genes in the mouse genome at will. This was the technology that firmly cemented the mouse at the forefront of understanding lens development. Today, the mouse remains the only vertebrate in which targeted germline genetic alterations are routine. Genes in which null mutations are introduced by gene targeting are often referred to as knockout mice. Not all targeted mutations are knockouts, however. Targeted mutations can lead to hypomorphic (partial function) or dominant negative alleles. Sometimes gene targeting

is used to replace the coding sequence of one gene with another to create a “knock-in” mouse. Targeted mutations in many different mouse genes, sometimes unexpectedly, led to problems with eye development. The opposite also proved true.

Amazingly, lens development in mice with null mutations in genes encoding major lens proteins such as αAcrystallin (Brady et al., 1997), αB-crystallin (Brady et al., 2001), CP49 (Alizadeh et al., 2002), or filensin (Alizadeh et al., 2003) was largely normal, at least in the sense that the lenses were formed and were initially free from cataract.

The lens complementation system

Perhaps the most challenging aspects of using gene targeting approaches to study lens development is that many genes important for lens development are also essential for the development of other organ systems. In this case, lens phenotypes in a particular knockout mouse may be secondary to developmental deficiencies in another tissue. Even worse, many homozygous knockout mice die prior to the lens placode stage, complicating the use of these knockouts to study lens development. One way to address this issue is with the use of chimeras, in which wild-type cells can often compensate for the early lethality that may be present in an uncompensated homozygous knockout. Nanette Liegéois and colleagues devised a lens complementation system using embryos homozygous for the mouse aphakia (ak) mutation for precisely this purpose (Liegéois et al., 1996). The ak mutation consists of deletions in the promoter region of the Pitx3 transcription factor, and ak homozygotes are unable to form lenses (reviewed in Graw and Loster, 2003). Injection of wild-type ES cells into homozygous ak blastocysts rescues lens development in the chimeras, and the resulting lenses are entirely derived from the ES cell component (Liegéois et al., 1996). The lens complementation strategy was recently used to determine if Fgfr1 played an essential role in mouse lens development. Fgfr1 homozygous knockouts die shortly after gastrulation, well before morphological signs of lens development (Deng et al., 1994; Yamaguchi et al., 1994). Chimeras made from Fgfr1 null ES cells and ak/ak embryos exhibited normally formed lenses entirely lacking Fgfr1, demonstrating that Fgfr1 was not required for normal lens development (Zhao et al., 2006). The use of the aphakia lens complementation strategy has at least two complications. First, lack of lens development in a chimera might be indirect and the result of problems with mutant ES cell migration or differentiation prior to lens induction. Second, lens development does initiate in ak mutants, and early lens structures such as the lens placode, lens pit, and lens vesicle, are likely to consist of a mixture of cells rather than strictly being derived from the ES cell component.

Conditional gene alterations using Cre recombinase

The manipulation of gene expression, both temporally and in a tissue-specific manner, is the most powerful way to exploit genetics to gain developmental insight. Fortunately, lens biologists have several tools at their disposal to do this, and new and better tools are doubtless forthcoming. Today, the most widely used system for temporal control of genome manipulation in the mouse is the Cre-loxP system, derived from P1 bacteriophage (reviewed in Branda and Dymecki, 2004). In essence, genomic DNA flanked by directly oriented 34 bp loxP sites can be deleted, in the presence of Cre recombinase, in vivo. Gene targeting constructs can be designed to insert loxP sites on each end of a gene, or within introns surrounding essential exons within that gene to preserve gene function in the targeted allele. The loxP-flanked gene or exons can then be deleted at will by controlling the temporal and spatial expression of Cre recombinase. The mouse lens was the first tissue in which tissue-specific DNA deletion using Cre recombinase was demonstrated in vivo (Lakso et al., 1992).

There are several different transgenic strains with patterns of Cre expression potentially useful for lens developmental studies. Rx-Cre transgenic mice, where Cre expression is under the regulatory control of a 4 kb medaka fish Rx3 promoter fragment, express Cre in the entire OV prior to lens placode formation, making these mice useful for evaluating the roles of OV-expressed genes on the subsequent induction and differentiation of the lens (Swindell et al., 2006). Within the lens, several different transgenic lines have been used to delete loxP-flanked genes (figure 22.2). The earliest acting of these are directed by genomic sequences from the murine Pax6 gene. The Le-Cre transgenic line exploits a 6.5 kb genomic fragment that includes the P0 promoter from the mouse Pax6 gene to drive Cre expression (Ashery-Padan et al., 2000). In Le-Cre mice, Cre expression initiates in the Pax6-expressing head ectoderm, including all the cells of the PLE at 9.0. Conditional (loxP-flanked) gene deletion in these mice takes place in virtually all of the ocular tissues derived from the surface ectoderm, including the lens, corneal epithelium, conjunctiva epithelium, and portions of the eyelid. The Le-Cre mice also experience conditional gene deletion in cells of the developing endocrine pancreas and in the olfactory epithelium (see figure 22.1). The Pax6(Lens)-Cre line is similar to Le-Cre except that it only includes the 340 bp Pax6 ectoderm enhancer fused to the P0 promoter and exhibits a more eye-specific expression pattern (Yoshimoto et al., 2005). MLR10 transgenic mice express Cre under the control of a murine αA-crystallin promoter modified by the insertion of a Pax6 consensus binding site (Zhao et al., 2004). In MLR10 mice, Cre expression initiates at the lens pit/vesicle stage (E10.5) and effectively deletes many loxP-flanked sequences in the majority of lens epithe-

Figure 22.2 Cre expression patterns in three different transgenic strains: Le-Cre (A–C), MLR10 (D–F ), and MLR39 (G–I ). Cre expression in whole-mount (A and B ) or tissue sections (C–I ) is indicated by blue (B and D–I ) or purple (A and C) staining following histochemical detection of a Cre-activated reporter allele. Arrows in

lial cells and lens fiber cells. The lens is the only site of ocular Cre expression in MLR10 mice. MLR39 transgenic mice express Cre from the αA-crystallin promoter, initiating in the lens fiber cells at approximately E12.5. Although Cre expression can also be seen in the developing retina pigment epithelium, within the lens, Cre expression in MLR39 mice is largely fiber cell specific (Zhao et al., 2004). Although the endogenous αA-crystallin gene is expressed in both lens epithelial cells and lens fiber cells, the −364/+45 αA-crystallin promoter (most commonly used in transgenic mice) efficiently directs gene expression only to lens fiber cells. It is now clear that the αA-crystallin gene contains several regulatory elements missing in the −364/+45 αA-crystallin promoter (Yang et al., 2006).

The different Cre-expressing mouse strains allow genes to be deleted in the lens at distinct developmental time points. This makes the evaluation of gene function at multiple stages of lens development and differentiation possible (see figure 22.2). For studies of lens induction, it may be valuable to delete genes in the presumptive lens prior to the initiation of Pax6 expression in the PLE. One potentially useful mouse strain for this purpose is the AP2α-Cre strain (Macatee et al., 2003), where Cre expression initiates in the head surface ectoderm at the 10 somite stage (E8.5). Mouse strains also exist in which Cre expression can be induced in a global fashion following induction with a drug such as tamoxifen (Hayashi and McMahon, 2002). These tools, and better ones sure to follow, permit exquisite manipulation of specific gene

A and B indicate Cre expression in the developing pancreas of LeCre mice. Developmental time points are indicated. See color plate 12. (Adapted from Ashery-Padan et al., 2000, and Zhao et al., 2004.)

expression during lens development. With any conditional gene deletion strategy it is important to determine precisely when and where Cre is expressed and to assess the efficiency with which the loxP-flanked DNA of interest is deleted within the particular Cre driver strain chosen. This information is essential to interpreting the results of conditional gene deletion experiments.

Mouse lens induction

Perhaps no stage of mouse lens development has been more clearly elucidated from a molecular perspective than that of lens induction. Mouse lens induction is the subject of a recent review article by Richard Lang, who has contributed much to the current understanding of this subject (Lang, 2004). At the heart of lens induction in mice (and likely all vertebrates) is the transcription factor Pax6. Pax6 expression is absolutely required for eye formation in Drosophila as well as in vertebrates. Ectopic expression of Pax6 is also sufficient to induce the formation of ectopic eyes in both Drosophila and Xenopus, demonstrating that Pax6 exhibits a conserved and critical role in metazoan eye development. Mutations in Pax6 are responsible for the dominant small eye (Sey) series of allelic mutations in mice and aniridia (as well as other ocular anomalies) in humans. Pax6 expression in the mouse PLE can be divided into two distinct phases (figure 22.3). The first phase of Pax6 expression (Pax6pre-placode) covers a large region of the head surface ectoderm that subsequently (after the

Upstream Regulators

separation Proliferation

Lens Development

Figure 22.3 A model for the genetic pathways regulating lens induction. (Reprinted with permission from Lang, 2004.)

PLE comes in close contact with the OV) becomes restricted in the second phase (Pax6placode) to a region including and immediately surrounding the lens placode. The second phase of Pax6 expression is dependent on functional Pax6 protein being produced in the first phase, indicating that Pax6 regulates its own expression (Grindley et al., 1995). Although there are many regulatory elements controlling Pax6 gene expression, two have been analyzed in detail with respect to their ability to regulate the Pax6placode phase of Pax6 expression. These elements are the Pax6 ectoderm enhancer, upstream of the P0 promoter, and the SIMO element, approximately 140 kb downstream from Pax6 (reviewed in Lang, 2004).

Inhibition of Fgf receptor (Fgfr) signaling (either by a chemical Fgfr inhibitor or by expression of a dominant negative Fgfr transgene) reduces the level of Pax6 in the mouse lens placode. Lenses that subsequently form under Fgfr inhibition are abnormally small, with reduced rates of cell proliferation. Heterozygous loss of Bmp7 exacerbated the reduction of Pax6placode expression in combination with Fgfr inhibition, suggesting a genetic cooperation of the Bmp and Fgfr signaling pathways in regulating Pax6 expression during lens induction (Faber et al., 2001). Mice homozygous for null mutations in either Bmp7 or Bmp4 most often fail to induce a lens. Although Bmp4 is not expressed in the PLE, and its loss does not reduce Pax6placode expression, it is expressed in the OV and required for the upregulation of the transcription factor Sox2 (Furuta and Hogan, 1998). Bmp7 is expressed

in the PLE, and while there is phenotypic variability in lens defects resulting from Bmp7 loss (Dudley et al., 1995; Luo et al., 1995), Bmp7 deficiency often results in the loss of Pax- 6placode expression (Wawersik et al., 1999). Pax6placode expression is also regulated by Meis transcription factors, at least in part, through binding sites identified in the Pax6 ectoderm enhancer (Zhang et al., 2002).

Vertebrate lens development also depends on members of the Sox family of HMG box transcription factors. Sox2 expression in the mouse PLE follows that of Pax6 and is dependent on BMP4. Pax6 and Sox2 proteins physically interact to control δ-crystallin gene expression in the chick lens (Kamachi et al., 2001). Recent evidence also suggests that Sox2 may cooperate with Oct-1 in mice to activate the Pax6 promoter through binding sites in the ectoderm enhancer (Donner et al., 2007). Null mutations in Sox2 cause early embryonic lethality and mice with Sox2 specifically deleted in the PLE have not been reported (Avilion et al., 2003). Gene-targeting mutations have determined a specific requirement for Sox1 for fiber cell elongation and γ- crystallin expression (Nishiguchi et al., 1998).

The regulation of Pax6pre-placode expression is less clear. There is evidence suggesting that vertebrate sensory and neurogenic placodes arise from a common preplacodal region (PPR) near the edge of the neural plate that expresses transcription factors from the Six (from Drosophila, sine oculis), Eya (from Drosophila, eyes absent), and Dach (from Drosophila, dachshund) families (reviewed in Donner et al., 2006; Schlosser, 2006). Pax6 is subsequently expressed in the portion of the PPR fated to become the lens. Although the direct relationship of Six/Eye/Dach expression and Pax6pre-placode expression is not firmly established in mouse lens induction, there is experimental evidence that Six3 may be required. First, Six3 expression may precede Pax6 expression in the PLE, and deletion of loxP-flanked exons of Six3 at E8.5 reduces Pax6pre-placode expression. Furthermore, Six3 binds both the Pax6 ectoderm enhancer and the SIMO element in vitro and in vivo, transactivates a reporter construct consisting of the Pax6 ectoderm enhancer fused to luciferase in cultured cells, and expands the Pax6 expression domain of chick embryo head ectoderm following in ovo electroporation (Liu et al., 2006). Pax6placode expression is required for the upregulation of Six3 expression in the lens placode, and a Six3 transgene under the regulatory control of the αA-crystallin promoter upregulates Pax6 expression, leading to a rescue of Pax6 heterozygous deficiency in the lens (Goudreau et al., 2002). Evidence suggests that Pax6 and Six3 are each able to regulate the expression of the other, and this reciprocal regulatory relationship is also seen in eyeless and sin oculis, the Drosophila orthologues of these genes.

Additional evidence for the role of Fgfr signaling in early lens development comes from recent analysis of mice lacking

Ndst1. Ndst1 is a member of a family of enzymes responsible for the synthesis of heparan sulfate required for heparan sulfate proteoglycan (HSPG) formation. HSPGs have been shown to participate in several different signal transduction pathways during development (reviewed in Lin, 2004), including the Wnt, Bmp, and Fgf pathways. Mice with targeted deletions in Ndst1 have ocular defects ranging from microphthalmia to anophthalmia resulting from a small lens placode invagination and reduced proliferation in the resultant lens (Pan et al., 2006). Despite the ability of HSPGs to participate in several different signaling pathways, the early lens in Ndst1 knockout mice is specifically compromised in Fgfr signaling and exhibits dramatic reductions in the level of phosphorylated (active) Erk in the lens (Pan et al., 2006). Targeted mutations in Frs2α phosphorylation sites, specifically linking Fgfr signaling to Erk activation, also led to defective lens induction (Gotoh et al., 2004).

AP2α is widely expressed in the embryonic head ectoderm, and its expression is not dependent on Pax6. Targeted mutations in AP2α lead to numerous craniofacial malformations, including abnormal lens development. Furthermore, Pax6 expression in lens tissue present in AP2α null mice is reduced relative to that in wild-type lenses, suggesting that AP2α may play a role in Pax6 gene regulation (West-Mays et al., 1999).

Several genes regulating lens development appear to be downstream from Pax6 in the lens induction pathway. Another gene that is upregulated in lens placode is Mab21/1, named after the C. elegans gene mab-21 required for ray identity specification. Mice lacking Mab21/1 exhibit severe microphthalmia, at least in part because of lowered proliferation in cells comprising the lens placode and early lens vesicle. Mab21/1 deficiency in lens cells is cell autonomous, as Mab21/1 homozygous mutant cells are specifically underrepresented in the lenses of chimeric mice. Pax6 expression is not altered in Mab21/1-deficient embryos, but Mab21/1 expression is lost in the absence of Pax6 (Yamada et al., 2003). Although Mab21/1-deficient lens cells are able express transcripts for differentiation-specific crystallins, Foxe3 expression within the lens is dramatically reduced.

Foxe3 is a winged-helix transcription factor that is normally expressed in the PLE prior to placode formation, and Foxe3 expression becomes restricted to the lens epithelial cells as development progresses. A mutation in Foxe3 is responsible for the spontaneously occurring dysgenetic lens (dyl) allele in mice (Blixt et al., 2000). Foxe3-deficient lenses display numerous developmental anomalies. Primary fiber cell differentiation, as assessed by the expression of differen- tiation-specific crystallins, does take place in the absence of Foxe3, but proliferation in the lens epithelium is severely reduced (Blixt et al., 2000; Medina-Martinez et al., 2005). Epithelial cells in Foxe3 mutants also display characteristics consistent with premature differentiation and undergo high

rates of apoptosis (Blixt et al., 2000). Foxe3-deficient lens vesicles also fail to separate from the surface ectoderm, leading to a persistent lens stalk. In the eye, Foxe3 is not expressed in the absence of Pax6 (Brownell et al., 2000), and Foxe3 expression is reduced in Sey heterozygotes (Blixt et al., 2007). Foxe3 deficiency also leads to abnormal differentiation of the neural crest–derived structures of the anterior segment, including the corneal endothelium, iris, and trabecular meshwork. This is not particularly surprising, as many mutations interfering with early lens development affect these tissues (Beebe and Coats, 2000). More surprising is the finding of subtle abnormalities in these anterior segment tissues in Foxe3 heterozygous mutants, where morphological lens development appears normal (Blixt et al., 2007). Within the eye, Foxe3 is not expressed outside the lens. Therefore, Foxe3 may be involved in the signaling from the lens responsible for the organization of neural crest cells into ocular structures of the anterior segment.

Foxe3 expression is also lost in the absence of Sip1, a Smad-binding homeodomain transcription factor (Yoshimoto et al., 2005). Sip1 is normally expressed in the lens placode and after lens formation; Sip1 is expressed in the lens epithelium and at the equator where secondary fiber differentiation occurs. Sip1 acts cooperatively with Smad8 to activate the Foxe3 promoter in transfection assays, but the promoter element responsible for this activation does not confer lens specificity in transgenic mice (Yoshimoto et al., 2005). Conditional deletion of Sip1 in the lens placode leads to lens cell apoptosis and failure of the lens to separate from the surface ectoderm, features common to Sey heterozygotes and Foxe3-deficient mice. In addition, the loss of Sip1 inhibits lens fiber differentiation. Sip1-deficient fiber cells express low levels of β-crystallins, fail to fully elongate, and do not express γ-crystallins (Yoshimoto et al., 2005). Sip1 expression is not affected by Foxe3 loss, but the dependence of Sip1 expression in the lens placode on Pax6 is unknown.

In addition to factors that promote lens induction, there are also factors that appear to act to inhibit lens induction in the head ectoderm. Conditional deletion of β-catenin in the PLE leads to ectopic appearance of lentoid bodies expressing β-crystallins and Prox1 in the surface ectoderm located nasally relative to the normal lens (Smith et al., 2005). β-catenin is an essential component of the canonical Wnt signaling pathway, and analysis of canonical Wnt signaling reporter mice suggests that this pathway is not active in the ectoderm that will form the lens but is active in adjacent ectoderm that will normally form the conjunctival epithelium and nasal epidermis (Miller et al., 2006; Smith et al., 2005). Furthermore, ectopic activation of canonical Wnt signaling in the developing lens inhibits lens formation (Miller et al., 2006; Smith et al., 2005). This evidence suggests that the normal role for Wnt signaling in the early lens