Ординатура / Офтальмология / Английские материалы / Development of the Ocular Lens_Lovicu, Lee Robinson_2004

and Bassnett, 2000b). Fusion zones are frequently seen to occur between the anterior segments or posterior segments of fiber cells as they approach their sutural locations. The frequent cell-to-cell fusion zones near sutures are thought to constitute a means for fibers to change their direction or curvature to conform to the precise modeling of sutures required for proper lens function. Cell-to-cell fusion zones also provide large patent pathways for intercellular transport between fibers for substances too large to pass through fiber gap junctions.

4.5.2.2. Compaction of Aged Fibers in Different Developmental Regions

Because the lens grows throughout life, it is necessary to compact the fiber mass so that the lens does not outgrow its place within the eye. Compaction, although not evident in young humans or animals, is particularly evident in aging humans.

The age-related changes in fiber compaction within the different developmental regions are best appreciated by relating equatorial thin-section TEM images, where all fibers are cut in cross section at their midpoints (Fig. 4.30), to SEM analyses of lenses split along their anteroposterior axis to reveal the end-to-end arrangement of nuclear fibers in growth shells and radial cell columns (Fig. 4.31).

In thick and thin sections taken through the lens equatorial plane, the reduction in fiber width as a result of fiber compaction is most evident at the transition from the cortex to the adult nucleus (Figs. 4.30a and b). This view also reveals that fiber compaction occurs in the juvenile and fetal nuclei, although to a lesser extent in the fetal nucleus (Taylor et al., 1996; Al-Ghoul and Costello, 1997). Comparable thick and thin sections through the embryonic nucleus suggest that age-related compaction in the oldest region of lenses is insignificant (Figs. 4.30c and d).

However, significant age-related compaction in the fetal and embryonic nuclei is readily apparent in SEM images (Fig. 4.31; Al-Ghoul et al., 2001). Low-magnification overviews (Figs. 4.31a and b) show that a young human lens has a thicker embryonic nucleus and a less acute fetal nuclear fiber angle (more spherical shape) than an older lens. The highmagnification views demonstrate that the fibers from the older lens have accordionlike folds that decrease the overall thickness of the embryonic nucleus along the optic axis (Figs. 4.31c and d). The SEM images also reveal the complex pattern of furrowed membranes that correspond exactly to the undulating membranes visible in TEM thin sections or freezeetch replicas (Figs. 4.22 and 4.28). Finally, the knobs that appear on the fibers in the SEM images correspond to the collection of edge processes that are seen in TEM images (Figs. 4.22 and 4.30) and are exposed when adjacent fibers are split apart during the preparation of SEM specimens. An important conclusion is that a full appreciation of the complex topology of the fibers is only possible when both SEM and TEM techniques are employed.

4.6. Summary

The lens is a prime example of how an organ modifies its structure through terminal differentiation to accomplish its function. It produces fibers of defined shape and size and arranges these cells into radial cell columns and growth shells throughout a lifetime. It uses an abundance of specialized cytoplasmic crystallin proteins, cytoskeletal elements, intercellular contacts (gap junctions and square array membranes), and cell-to-cell fusion to produce a structure that will remain viable over many decades. The crystallin proteins

Figure 4.30. Transmission electron micrographs of human fibers. (a) Inner cortex showing smooth cytoplasm of five fibers in a radial cell column. Note the undulating membranes (arrowheads) and circular profiles (black arrows). It is likely, based on the marked intercellular projections (white arrows), that the circular profiles are derived from sections through the intercellular projections and do not represent vesicles or cellular breakdown products. (b) Adult nucleus revealing a very complex pattern of membranes. The increase in complexity can be accounted for by a significant compaction of fibers in which each fiber decreases its cross-sectional area by a factor of three. The interdigitations also become more complex, and more circular profiles appear within the fiber cytoplasm. These profiles are most likely derived from intercellular projections. (c) Embryonic nucleus from a young lens (22 years old). Note large irregular cell with undulating membranes (arrowheads) and cytoplasmic profiles (black arrows). (d) Embryonic nucleus from aged lens (67 years old). The fibers in panels c and d are similar because both have smooth cytoplasm and complex membrane interfaces. These fibers are influenced by compaction, which is more clearly demonstrated by SEM. (a) From Taylor et al., 1996, courtesy of ARVO. (b)–(d) From Al-Ghoul et al., 2001, courtesy of Academic Press, London, UK.

118 Jer R. Kuszak and M. Joseph Costello

provide a continuous gradient of refraction between fibers. The extracellular space, a potential source of diffraction, is essentially eliminated by conjoining fibers via square array membranes and gap junctions. In addition, gap junctions and fusion zones provide intercellular pathways between the fibers, which become positioned farther away from their source of nutrition with the addition of each growth shell. But, ultimately, age-related changes in lens crystallins, cytoskeleton, and membrane render the lens incapable of preserving its necessary structure-function regimen indefinitely, leading to common lens pathologies, such as presbyopia and cataract.

120 Duncan, Cvekl, Kantorow, and Piatigorsky

(Graw, 1997) as well as αA/αB-crystallin double knockout mice (Brady and Wawrousek, 1997) argues for a direct role of at least some crystallins in developmental processes as well (Andley et al., 1998; Andley et al., 2000; Boyle and Takemoto, 2000).

The relative expression level of crystallin genes changes during development and results in quite large differences in crystallin composition in different portions of the lens. An example of this is the preponderance of δ-crystallin in the nucleus and β-crystallins in the cortex of the adult chicken lens (Ostrer and Piatigorsky, 1980; Thomson et al., 1978). This difference in crystallin composition results from changes in the relative amounts of crystallin gene transcription that occur in lens fibers born during the preand posthatching period (Hejtmancik et al., 1985). Since the efficiency of the vertebrate lens is dependent on a smooth refractive index gradient that corrects for spherical aberration (Land, 1988), it appears likely that differential transcriptional regulation of crystallin genes during development is important for lens function. In addition, since crystallins are often used as markers for lens differentiation, information on the control of their expression is essential to forward our understanding of the molecular mechanisms responsible for lens development.

Here we review the structure, function, and expression of crystallins during development.

5.2.Structure and Function of Crystallins

5.2.1.α-Crystallin

α-Crystallin, a member of the small heat-shock family of proteins, is an aggregate of two approximately 20 kDa polypeptides, αA and αB (de Jong et al., 1993; Groenen et al., 1994; Ingolia and Craig, 1982; Klemenz et al., 1991; Sax and Piatigorsky, 1994). These subunits most likely arose from a gene duplication event and share about 57% amino-acid identity. In humans, the αA-crystallin gene is localized on chromosome 21 and encodes a 173–amino acid polypeptide, and the αB gene is localized on chromosome 11 and encodes a 175–amino acid polypeptide. The two subunits form a soluble aggregate that has a molecular weight ranging from 300 kDa to over 1 thousand kDa, with an average of 800 kDa (Horwitz, 2003; Siezen et al., 1978; Veretout et al., 1989). When isolated from human, rat, and bovine lenses, the ratio of αA to αB is approximately 3:1 (Delcour and Papaconstantinou, 1974; Lampi et al., 2002; van Kamp et al., 1974).

Although the exact structure of α-crystallin has not been determined, it is known that α- crystallin consists of about 50% β-sheet and 10–15% α-helix (Horwitz et al., 1998; Koretz et al., 1998). Recently it has been shown by cryoelectron microscopy that aggregates of αB-crystallin contain approximately 32 subunits that form a globule with a hollow cavity (Haley et al., 1998; Horwitz et al., 1999; Smulders et al., 1998). The subunits making up this cavity freely exchange (van den Oetelaar et al., 1990) in a temperature-dependent manner (Bova et al., 1997).

Developmentally, α-crystallin is the first crystallin synthesized in the mammalian lens, where it is first detected in the lens placode (Haynes et al., 1996; Robinson and Overbeek, 1996; Zwaan, 1983). In chickens, α-crystallin appears after δ-crystallin in the lens vesicle (Ikeda and Zwaan, 1966, 1967). In the adult lens, α-crystallin transcription is mainly confined to the fiber cells (Sax and Piatigorsky, 1994), but significant amounts of both αA and αB mRNA can be detected in the lens epithelium (Robinson and Overbeek, 1996). In the mature mammalian lens, α-crystallin accounts for about 20–30% of water-soluble protein (Lampi et al., 1998; Lampi et al., 2002; Mehta and Lerman, 1972; Ueda et al., 2002).

α-Crystallin was first thought to be restricted to the lens, where it was assumed to have a completely refractive function. This view of α-crystallin changed when it was discovered by two independent groups that αB-crystallin was expressed in numerous tissues in addition to the lens (Bhat and Nagineni, 1989; Dubin et al., 1989), most notably in the heart, where it makes up as much as 3–5% of total cell protein (Benjamin et al., 1997). Although αA- crystallin expression is highly lens preferred, it can be detected in trace amounts in the spleen, thymus (Kato et al., 1991), and retina (Deretic et al., 1994). α-Crystallins are notably expressed in many other ocular tissues, including the Muller¨ ’s cells of the retina (Deretic et al., 1994; Lewis et al., 1988; Moscona et al., 1985), the retinal photoreceptors (Deretic et al., 1994), the retinal pigmented epithelium (Nishikawa et al., 1994; Robinson and Overbeek, 1996), the corneal endothelium (Flugel et al., 1993), the ciliary muscle, and the trabecular meshwork (Siegner et al., 1996).

The α-crystallins have been implicated in numerous nonrefractive pathways, including those for stress response, phosphorylation, and cell protection (Jakob et al., 1993). In 1992, Joseph Horwitz found that α-crystallin is a molecular chaperone that binds unfolded or denatured proteins, thereby suppressing nonspecific irreversible protein aggregation (Horwitz, 1992). This discovery was later confirmed by Jakob et al. (1993), who demonstrated that, in addition to αB-crystallins, other small heat-shock proteins were capable of acting as molecular chaperones. Since this pioneering work, researchers have found evidence indicating that the binding of nonnative proteins to α-crystallin and other small heat-shock proteins creates a reservoir of unfolded proteins that can later interact with other chaperones to restore the unfolded proteins to a native state in an ATP-dependent process (Ehrnsperger et al., 1997; Lee et al., 1997; Wang and Spector, 2000). Consistent with the chaperone function of α-crystallin, it has been shown that both α-crystallin expression (Klemenz et al., 1991) and phosphorylation (Wang et al., 2000) are induced by physiological stress. Further, αB-crystallin is involved in numerous diseases outside of the eye (Head and Goldman, 2000; Welsh and Gaestel, 1998).

α-Crystallins have properties consistent with other nonrefractive functions. They participate in both cAMP-dependent (Chiesa et al., 1987; Spector et al., 1985; Voorter et al., 1986) and non-cAMP-dependent phosphorylation events (Kantorow et al., 1995; Kantorow and Piatigorsky, 1994), where they may be involved in signal transduction pathways. They also can bind cytoskeletal elements (Bloemendal et al., 1984; Del Vecchio et al., 1984; FitzGerald and Graham, 1991; Head and Goldman, 2000; Wisniewski and Goldman, 1998) and are associated with membrane binding (Cobb and Petrash, 2000; Ramaekers et al., 1980). α-Crystallins can also translocate to the nucleus (Bhat et al., 1999), where they may play a role in controlling lens cell differentiation (Boyle and Takemoto, 2000). α-Crystallins also exhibit antiapoptotic activity, with αA-crystallin having a greater protective effect than αB-crystallin (Andley et al., 2000).

Targeted disruption of the mouse αA-crystallin gene induces cataract (Brady et al., 1997), demonstrating its importance in lens function. The lens fiber cells of αA-crystallin null mice exhibit dense inclusion bodies consisting mainly of αB-crystallin, suggesting that high concentrations of αB-crystallin may be unstable in the absence of αA-crystallin in vivo and that αA-crystallin may have a solubilizing effect on αB-crystallin (Brady et al., 1997). In contrast, targeted disruption of the αB-crystallin gene and its linked relative the HSB2 gene (Iwaki et al., 1997) result in mice with relatively normal lenses. However, mice homozygous for targeted deletions of the αA-crystallin, αB-crystallin, and HSB2 genes have severe disruptions in lens morphology (Brady and Wawrousek, 1997). Studies of lens epithelial cells from the αA-crystallin knockout and αB-crystallin knockout mouse indicate

122 Duncan, Cvekl, Kantorow, and Piatigorsky

that α-crystallin subunits also play major roles in regulating lens epithelial cell division and chromosomal stability (Andley et al., 1998; Andley et al., 2000).

A point mutation in the human αA-crystallin gene is linked to cataract (Litt et al., 1998), and a point mutation in the human αB-crystallin gene is linked to desmin-related myopathy (Vicart et al., 1998). These point mutations disrupt chaperone function (Kumar et al., 1999; Perng et al., 1999; Shroff et al., 2000), suggesting that α-crystallin chaperone function is important for the maintenance of both lens and nonlens tissues. Other diseases associated with αB-crystallin include Alexander’s disease (Iwaki et al., 1989), Creutzfeldt-Jacob disease (Iwaki et al., 1992; Renkawek et al., 1992), Huntington’s disease (Iwaki et al., 1992), multiple sclerosis, and a multitude of other disorders (Head and Goldman, 2000; Iwaki et al., 1992; Van Noort et al., 1998). The study of α-crystallin structure and function both in lens and nonlens tissues is one of the fastest growing and most exciting areas of crystallin research.

5.2.2. β/γ-Crystallins

Native β-crystallin is found in solutions of water-soluble vertebrate lens proteins as a polydisperse mixture of hetero-octomers (βHigh) and heterotetramers (βLow), along with lower amounts of dimers (Bindels et al., 1981). These β-crystallin macromolecules are formed from the seven known subunits (βA1, βA2, βA3, βA4, βB1, βB2, βB3), which range in size from 22 to 35 kDa, depending on the species (Asselbergs et al., 1979; Lampi et al., 1998; Ueda et al., 2002; Wistow et al., 1991). Cloning of the cDNAs revealed that these seven polypeptides are encoded by six genes, with the βA1and βA3-crystallin subunits formed from alternate translational initiation from a common mRNA (Peterson and Piatigorsky, 1986). Orthologs of all six β-crystallin genes have been cloned from a number of vertebrates, including human, mouse, rat, cow, and chicken (Duncan et al., 1996a; Graw et al., 1999; Lampi et al., 1997; Lampi et al., 2002; Quax-Jeuken et al., 1984; vanRens et al., 1991). Since direct orthologs of several of these genes have also been reported in frogs and fish (Lu et al., 1996b; Wistow, 1995), it has been proposed that all six β-crystallin genes are found in all vertebrates.

In mammals, native γ -crystallin is found as a mixture of 22to 25-kDa monomers consisting of the members γ S (previously known as βS), γ A, γ B, γ C, γ D, γ E, and γ F (Lubsen et al., 1988). In mice and rats, all seven of these genes are functional, while in humans, only the γ A-, γ C-, γ D-, and γ S-crystallin genes contribute significantly to the lens proteins (Meakin et al., 1987; Zarina et al., 1992). Although birds generally lack

γ-crystallins, it has been controversial whether bird lenses have γ S-crystallin (Reton et al., 1984; vanRens et al., 1991). However, proteomic analysis of the chicken lens did not reveal

γS-crystallin as a major lens protein (Wilmarth et al., under review). Proteins with sequence similarity to the mammalian γ -crystallins have been shown to be major contributors to the water-soluble proteins of fish, shark, reptile, and amphibian lenses (Chang et al., 1991; Chiou et al., 1987; Chuang et al., 1997; Lu et al., 1996a; Pan et al., 1994).

In the early 1980s, sequence analysis of the β- and γ -crystallins made it apparent that they have a common evolutionary ancestor (Driessen et al., 1981). The common structural feature of the β/γ -crystallins is a distinctive type of antiparallel β-sheet called the Greek key motif (Bax et al., 1990; Blundell et al., 1981). It has been proposed that the ancestor of the Greek key–containing proteins comprised a single motif that duplicated into two subtypes (A and B). A β/γ -crystallin domain formed when A- and B-type Greek key motifs fused. The complete β/γ -crystallin proteins were created when two β/γ -crystallin domains were linked

together with an interdomain spacer (Wistow, 1990). In γ -crystallin, the two β/γ -crystallin domains interact intramolecularly, resulting in the monomeric native structure (Blundell et al., 1981). The two β/γ -crystallin domains of βB2-crystallin interact with their opposite number on another βB2-crystallin molecule (Bax et al., 1990). This results in the ability of βB2-crystallin to form stable dimers. In contrast, the β/γ -crystallin domains of βB1-crystallin interact intramolecularly, leaving several hydrophobic patches exposed to mediate dimerization (Van Montfort et al., 2003). While β-crystallin dimers further oligomerize into βHighand βLow-crystallin, the structure of these polydisperse molecules is not well characterized.

In all vertebrate lenses examined, the N- and C-terminal extensions of the β-crystallins are cleaved during the course of normal aging and development (David et al., 1993; Lampi et al., 1997; Ueda et al., 2002). This proteolysis is accelerated during the formation of cataract in rodent lenses treated with agents that disrupt calcium homeostasis (David et al., 1994). The mechanism responsible for the posttranslational proteolytic processing of human β-crystallins is still under investigation, but proteases of the calpain family are likely to be involved in rodents (David et al., 1994). While the function of crystallin proteolysis during lens maturation is still unknown, the observation that the processed

β-crystallins tend to be found in the insoluble fraction of the lens nuclear proteins (Lampi et al., 1998; Ueda et al., 2002) suggests that the cleavage of the hydrophilic arms aids in the exclusion of water from the lens nucleus during lens maturation.

Unlike the α-crystallins, no nonrefractive physiological functions have been directly ascribed to the β/γ -crystallins. However, it is likely that nonrefractive functions do exist. First, the β/γ -crystallins are evolutionarily related to a number of proteins involved in stress response, including protein S, a spore coat protein of the colonial prokaryote Myxococcus xanthus, and spherulin 3a, an encystment-specific protein of the slime mold Physarum polycephalum (Bagby et al., 1994; Wistow, 1990). More recently, sequence similarity has been noted for the EP37s, a group of proteins expressed in the integument and digestive system of newts (Ogawa et al., 1997), and AIM1, a gene often disrupted in malignant melanoma of humans (Ray et al., 1997). Second, like protein S,

β-crystallin can apparently bind calcium (Sharma et al., 1989), and several papers have reported that some β-crystallins are phosphorylated (Kantorow et al., 1997; Kleiman et al., 1988; Voorter et al., 1989). Further, it has recently been reported that β-crystallin is expressed in the neuronal cell line N1E-115 and that this protein translocates from the perinuclear zone to the cytoplasm in response to heat stress and cold shock (Coop et al., 1998).

The cataract in eye lens obsolescence (Elo) mice results from a frameshift mutation that destroys the fourth Greek key of γ E-crystallin (Cartier et al., 1992). Developmental analysis of this cataract has shown that the earliest detectable morphological alteration is an impairment of primary fiber elongation at 12.5 days post coitum (dpc). Subsequently, a high proportion of these defective fibers undergo apoptosis or necrosis, which leads to a severe disruption in lens morphology (Oda et al., 1980). Further, both the Cat2nop mutation (caused by insertion or deletion of nucleotides from γ B-crystallin) and the Cat2ns mutation (caused by the deletion of the 3 end of γ E-crystallin) have defects in secondary fiber elongation and nuclear degradation (Graw, 1997). Finally, both the expressed sequence tag (EST) effort of the human genome project and traditional approaches have found that βB2-,

βB3-, βA3/A1-, and βA2-crystallin are expressed outside of the lens in mammals and birds (Head et al., 1991; Head et al., 1995; Magabo et al., 2000) and that some γ -crystallins are expressed outside of the lens in amphibians (Smolich et al., 1994) and mice (Jones et al., 1999; Sinha et al., 1998). In fact, some β- and γ -crystallins have been identified

124 Duncan, Cvekl, Kantorow, and Piatigorsky

as major components of the vertebrate tooth proteome (Thyagarajan and Kulkarni, 2002). In aggregate, these observations strongly suggest that β/γ -crystallins have nonrefractive functions that remain to be identified.

Mutations in β/γ -crystallins have been associated with a number of congenital autosomal dominant cataract phenotypes in both humans and rodents. In humans, a mutation that results in the loss of the 51 C-terminal amino acids of βB2-crystallin results in the cerulean blue dot (Litt et al., 1997; Vanita et al., 2001) and Coppock-like cataracts (Gill et al., 2000). Mutations in the βA3/A1-crystallin gene have been linked to autosomal dominant cataract in two different human families (Bateman et al., 2000; Burdon et al., 2004; Kannabiran et al., 1998; Qi et al., 2004). The variable zonular pulverulent cataract is caused by mutation of the γ C-crystallin gene (Ren et al., 2000). Finally, the aculeiform cataract has been associated with a mutation in the γ D-crystallin gene (Heon et al., 1999; Stephan et al., 1999).

In mice, the Philly cataract is caused by an in-frame deletion of 12 basepairs from the

βB2-crystallin gene that results in the deletion of four amino acids from the C-terminal Greek key motif (Chambers and Russell, 1991). Recently, it has been reported that mice homozygous for this mutation are subfertile, apparently due to defects in egg production and sperm production/function (Robinson et al., 2003). This may suggest a function for the

βB2-crystallin protein detected in the adult rat testis (Magabo et al., 2000). In addition, a number of independent autosomal dominant cataract phenotypes have been mapped to the γ -crystallin locus, and potentially causative mutations have been found in the γ A- (Klopp et al., 1998), γ B- (Graw, 1997), γ C- (Ren et al., 2000), γ D- (Smith et al., 2000), and γ E- crystallin (Klopp et al., 2001) genes. Finally, mutation of the γ S-crystallin gene leads to the mouse Opj cataract, which exhibits defects in cortical fiber structure (Wistow et al., 2000b). Interestingly, the cataracts associated with γ B-crystallin mutations have been attributed to the aberrant folding of the abnormal protein into amyloid-like structures which translocate into lens cell nuclei and disrupt their function (Sandilands et al., 2002).

5.2.3. Enzyme-Crystallins

Crystallins are very diverse proteins that are defined more for their prevalence in the lens than as members of any particular group of proteins. The first inkling of their diversity came when δ-crystallin was found to be a major protein in the chicken lens (Rabaey, 1962). Subsequently, it was found that δ-crystallin was confined to the lenses of almost all birds and reptiles (for a review, see Piatigorsky, 1984). Further studies showed that many species had specific crystallins in their lenses that were not found in the lenses of other species. Surprisingly, these so-called taxon-specific crystallins are similar or identical to metabolic enzymes. For example, ε-crystallin in ducks is lactate dehydrogenase B4, τ -crystallin in turtles is α-enolase, and δ-crystallin in chickens is argininosuccinate lyase (Wistow et al., 1987; Wistow and Piatigorsky, 1987). In many cases, the enzyme that is expressed at low levels outside of the lens and the enzyme-crystallin that is expressed at high concentration in the lens are encoded by the same gene; in other cases, one or more gene duplications have taken place, and the daughter gene (or genes) has specialized for lens expression, such as in the example of argininosuccinate lyase and δ-crystallin (Piatigorsky et al., 1988). The dual utilization of a gene as an enzyme for metabolism in many tissues and as a structural protein for refraction in the lens has been called “gene sharing” (Piatigorsky et al., 1988). The implications of gene sharing and the numerous examples of its use among the lens crystallins have been reviewed extensively and are beyond the scope of this chapter (for reviews, see de Jong et al., 1994; Piatigorsky, 1998; Wistow and Piatigorsky, 1988). Two of