THE ANATOMICAL RECORD PART A 273A:714 –730 (2003) Lens Structure in MIP-Deﬁcient Mice KRISTIN J. AL-GHOUL,1–3 TYLER KIRK,2 ADAM J. KUSZAK,2 REBECCA K. ZOLTOSKI,4 ALAN SHIELS,5,6 AND JER R. KUSZAK2,3* 1 Department of Anatomy and Cell Biology, Rush-Presbyterian-St. Luke’s Medical Center, Chicago, Illinois 2 Department of Ophthalmology, Rush-Presbyterian-St. Luke’s Medical Center, Chicago, Illinois 3 Department of Pathology, Rush-Presbyterian-St. Luke’s Medical Center, Chicago, Illinois 4 Department of Basic and Health Sciences, Illinois College of Optometry, Chicago, Illinois 5 Department of Ophthalmology and Visual Science, Washington University School of Medicine, St. Louis, Missouri 6 Department of Genetics, Washington University School of Medicine, St. Louis, Missouri ABSTRACT In this study we used correlative light, scanning, and transmission (freeze-etch) electron microscopy to characterize lens structure in normal mice and compare it with that in mice deﬁcient in the major intrinsic protein (MIP) of ﬁber cells. Grossly, wild-type lenses were transparent and had typical Y sutures at all of the ages examined. These lenses had ﬁbers of uniform shape (hexagonal in cross section) arranged in ordered concentric growth shells and radial cell columns. In addition, these ﬁbers had normal opposite end curvature and lateral interdigitations regularly arrayed along their length. Ultrastructural evaluation of these ﬁbers revealed anterior and posterior end segments characterized by square array membrane on low-amplitude wavy ﬁber membrane. Approximately 13% of the equatorial or mid segments of these same ﬁbers were specialized as gap junctions (GJs). In contrast, heterozygote lenses, while initially transparent at birth, were translucent by 3 weeks of age, except for a peripheral transparent region that contained ﬁbers in the early stages of elongation. This degradation in clarity was correlated with abnormal ﬁber structure. Speciﬁcally, although the mid segment of these ﬁbers was essentially normal, their end segments lacked normal opposite end curvature, were larger than normal, and had a distinct non-hexagonal shape. As a result, these ﬁbers failed to form typical Y sutures. Furthermore, the nuclear ﬁbers of heterozygote lenses were even larger and lacked any semblance of an ordered packing arrangement. Grossly, homozygote lenses were opaque at all ages examined, except for a peripheral transparent region that contained ﬁbers in the early stages of elongation. All ﬁbers from homozygote lenses lacked opposite end curvature, and thus failed to form any sutures. Also, these ﬁbers were essentially devoid of interlocking devices, and only 7% of their mid segment was specialized as GJs. The results of this study suggest that MIP has essential roles in the establishment and maintenance of uniform ﬁber structure, and the organization of ﬁbers, and as such is essential for lens function. Anat Rec Part A 273A:714 –730, 2003. © 2003 Wiley-Liss, Inc. Key words: MIP; crystalline lens; knockout mice; sutures; gap junctions; structure; electron microscopy; freeze-etch; freeze-fracture The crystalline lens is a major optical component of the dynamic focusing process. During this process it develops and maintains an apparently simple, but in reality complex, tissue structure predicated on the formation and precise organization of specialized cells, the lens ﬁbers. All vertebrate lenses develop and grow as inverted, stratiﬁed epithelia. The ﬁbers are long, crescent-like cells arranged end to end in growth shells (GSs). As such, as more ﬁbers are formed throughout life, the outer, younger GSs constitute the lens cortex, while the inner, older GSs make up the lens nucleus. However, while this basic lens architecture is characteristic of all vertebrate lenses, differences in ﬁber anatomy signiﬁcantly affect the optical quality of the lens (see Fig. 1). © 2003 WILEY-LISS, INC. Grant sponsor: NIH NEI; Grant numbers: EY06642; EY11411; Grant sponsor: Louise C. Norton Trust; Grant sponsor: Doctor Bernard and Jennie M. Nelson Fund. *Correspondence to: J.R. Kuszak, Ph.D., Departments of Ophthalmology and Pathology, Rush-Presbyterian-St. Luke’s Medical Center, 1653 West Congress Parkway, Chicago, IL 60612. Fax: (312) 942-2371. E-mail: firstname.lastname@example.org Received 11 February 2003; Accepted 28 March 2003 DOI 10.1002/ar.a.10080 LENS STRUCTURE IN MIP-DEFICIENT MICE Fig. 1. Whole ﬁxed lenses as seen through a surgical dissecting microscope (A–C) and 3D-CADs (D–L) of lenses with umbilical sutures (left column), line sutures (center column), and “Y” sutures (right column). A posterior view of an avian lens with branchless, umbilical sutures (A and D) has no discernable surface features. All ﬁbers are sequentially overlaid in register onto existing GSs of ﬁbers, resulting in RCCs that extend from the center to the periphery of the lens (G). These ﬁbers are straight meridians that extend from pole to pole (J). Lenses with line sutures (e.g., rabbit and frog lenses) feature two branches oriented 180° apart to form a vertical line suture anteriorly and a horizontal line suture posteriorly (B and E). As new ﬁbers are overlaid in register onto existing GSs, suture branches are also overlaid in register. Thus, two continuous, triangular suture planes are formed that extend from the center of the lens to its periphery (H). Within GSs, ﬁbers are either straight or s-shaped 715 (K). The position of four straight ﬁbers effectively deﬁnes the orientation of the four suture branches. Between straight ﬁbers, the ends of sshaped ﬁbers are aligned as longitudinal arc lengths (suture branches). The ends of ﬁbers from two adjacent quadrants abut and overlap to form suture branches (E, H, and K). An anterior view of a Y-suture lens shows that there are three anterior suture branches oriented 120° apart that reach conﬂuence at the pole (C and F). The position of six straight ﬁbers effectively deﬁnes the orientation of the six suture branches. Between straight ﬁbers, the ends of s-shaped ﬁbers are aligned as longitudinal arc lengths (suture branches). The ends of ﬁbers from two adjacent sextants abut and overlap to form six suture branches. All of the ﬁbers are overlaid in register onto existing GSs, and hence the six suture branches also become overlaid in register to form six triangular suture planes that extend from the center to the lens periphery (F, I, and L). 716 AL-GHOUL ET AL. For example, avian lens ﬁbers are meridians, tapering at the ends as they extend from pole to pole (Fig. 1, left column). In contrast, ﬁbers of all other vertebrate lenses are not meridians. These ﬁbers have ends that ﬂare and curve away from the poles in opposite directions. As a result, the end-to-end arrangement of these ﬁbers produces “lens suture branches.” In line sutures (e.g., rabbit and frog lenses), two anterior branches are oriented at 180° to each other to form a vertical “line” suture, while opposite end curvature results in two posterior branches forming a horizontal line suture (Fig. 1, center column). Most other vertebrate lenses (e.g., in mice, rats, pigs, cats, dogs, bovines, and primates at birth) have “Y” sutures, three anterior branches oriented at 120° to each other to form a Y suture pattern, while opposite end curvature results in three posterior branches forming an inverted Y suture (Fig. 1, right column). In primate lenses, sutural complexity increases as ﬁbers are formed during speciﬁc periods of life. A Y suture is formed during fetal development, a six-branch “simple star” evolves during infancy, a nine-branch “star” is laid down throughout adolescence, and a 12-branch “complex star” is produced during the adult years. The precise variation in the amount of ﬁber end ﬂare and opposite end curvature necessary to create the different suture types is well documented (Kuszak, 1995a). In addition, both the effect of different suture types on lens optical quality, and the negative effects of abnormal sutural anatomy as a consequence of both naturally occurring and experimentally induced pathology on lens function is calculable (Kuszak et al., 1991, 1994, 1999, 2000, 2002; Sivak et al., 1994; Al-Ghoul et al., 1998, 1999). However, the factors that direct the exact arrangement of ﬁbers into the form that follows function, and are compromised such that malformation leads to malfunction, are not known. A major consequence of ﬁber terminal differentiation is the transformation of selected cuboidal lens epithelial cells, ⬍5 m high, into the ribbon-like ﬁbers that range in length from 150 m to 2.5 mm in small lenses (e.g., in mice), and from 150 m to 15 mm in large lenses (e.g., in rabbit and bovine). The creation of these exceedingly long ﬁbers is characterized by the elaboration of speciﬁc cytoplasmic and membrane proteins: the crystallins and major intrinsic protein (MIP), respectively. MIP is the most abundant ﬁber surface marker (Gorin et al., 1984), and is generally considered to comprise at least 50% of ﬁber plasma membrane proteins (Alcala et al., 1975; Alcala and Maisel, 1985). Thus, the signiﬁcance of MIP in lens function has been the target of numerous biochemical, morphological, immunocytochemical, and physiological investigations. It was proposed that MIP is the ﬁber gap junction (GJ) protein (Goodenough, 1979; Kuszak et al., 1981). However, subsequent studies revealed that MIP was found both in association with ﬁber communicating junctions and distributed throughout the nonjunctional membrane (Bok et al., 1982; Fitzgerald et al., 1983; Paul and Goodenough, 1983). Subsequently, MIP was also localized to other ﬁber membrane specializations, including lateral interdigitations (balls and sockets, and ﬂaps and imprints) and regions of close membrane apposition (i.e., square array membrane) (Fitzgerald et al., 1985; Sas et al., 1985; Dunia et al., 1987; Costello et al., 1989; Gruijters, 1989; Zampighi et al., 1989). This suggests that MIP may also have a role in ﬁber adhesion and the regulation of extracellular space volume. Signiﬁcant sequence homology with several transmembrane channel proteins responsible for the transport of small molecules indicated that MIP may function as a water channel (Sandal and Marcker, 1988; Shiels et al., 1988; Baker and Saier, 1990; Rao et al., 1990). When MIP was reconstituted into unilamellar vesicles and bilayers, it demonstrated properties consistent with both adhesion and ﬂuid transport (Ehring et al., 1990; Michea et al., 1994). Subsequently, MIP was identiﬁed as a member of a sequence-related family of proteins, including the aquaporins and the aquaglyceroporins (Reizer et al., 1993; Froger et al., 1998); therefore, it is now often referred to as aquaporin 0. MIP exhibited water transport in Xenopus oocyte membranes, which was qualitatively similar to aquaporin 1, but was quantitatively of signiﬁcantly less magnitude (Chandy et al., 1997). In isolated lens ﬁber membranes, MIP displayed moderate but consistent water transport activity (Varadaraj et al., 1999). Recent structural studies reiterated the idea that MIP may have more than one functional role in the lens. MIP was consistently found by freeze-fracture immunolabeling to localize only to the periphery of ﬁber GJs, which indicates that it may be involved in the formation of these cell-to-cell communicating plaques (Dunia et al., 1998). In addition, puriﬁed, reconstituted MIP displayed surface tongue-and-groove contours consistent with a role for adhesion of apposing protein-membrane lattices (Fotiadis et al., 2000). In order to further elucidate the role of MIP in vertebrate lenses, we characterized lens structure in MIP-deﬁcient mice. We utilized correlative light microscopy (LM) to examine lens histology, scanning electron microscopy (SEM) to characterize ﬁber surface morphology, and transmission electron microscopy (freeze-etch) to assess the ultrastructure of ﬁber membrane in lenses from agematched wild-type, heterozygote, and homozygote mice. Our data indicate that a lack of MIP results in dysplastic ﬁbers, which adversely affect the exact ﬁber architecture necessary for lens function. MATERIALS AND METHODS Mice MIP-deﬁcient mice with a hybrid 129S5/C57BL/6J-albino genetic background were generated from a library of gene-trap embryo stem cells, as described previously (Shiels et al., 2001). Sequence analysis showed that the gene-trap vector had inserted into the ﬁrst exon of the MIP gene, resulting in an in-frame translation stop signal at codon 57. MIP transcript and protein levels in heterozygous mutants were reduced relative to wild-type levels, and no MIP transcripts or translation products were detected in homozygous mutants as determined by Northern and Western blot analyses, respectively. At 3 weeks of age (postnatal day [P] 21), lenses from homozygous mice displayed polymorphic opacities, whereas lenses from heterozygous mice were translucent and did not develop frank opacities until ⬃24 weeks of age. All animals were handled in compliance with institutional animal care and use guidelines, and with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. LENS STRUCTURE IN MIP-DEFICIENT MICE Lenses Lenses from heterozygote and homozygote MIP-deﬁcient mice, as well as from wild-type mice, were examined by LM and SEM at the following time points: birth (P1), 1 week old (P7), 2 weeks old (P14), and 3 weeks old (P21). For freeze-etch it was necessary to utilize older mice in order to accurately dissect the lenses. Only 10-, 18-, and 24-week-old mice were examined by the freeze-etch technique. For each of the above time points, n ⫽ 8 –12 mice. Animals were killed by intraperitoneal injection of sodium pentobarbital or by CO2 asphyxiation. The lenses were dissected from the enucleated eyes and then processed for LM, SEM, or freeze-etch. All lenses were ﬁxed in 2.5% gluteraldehyde in 0.07 M sodium cacodylate buffer at pH 7.2 at room temperature for 2–3 days, with fresh ﬁxative changes daily. After the lenses were washed overnight in 0.2 M sodium cacodylate buffer, axial lens dimensions were taken and the lenses were photographed under a surgical dissecting microscope (Zeiss, New York, NY). LM For histology, the lenses were postﬁxed in 1% aqueous osmium tetroxide at 4°C overnight, then washed in cacodylate buffer and dehydrated through a graded ethanol series to propylene oxide. Tissue was inﬁltrated and ﬂat embedded in epoxy resin. The embedded lenses were sectioned along the polar axis. Tissue sections 1–2 m thick were cut with a glass knife, stained with a 1:1 mixture of methylene blue and azure II, and photographed with an Olympus Vanox AHBS3 microscope (Olympus America Inc., Melville, NY) equipped with a 35-mm camera. Color slides were digitized using a Polaroid Sprint Scan 35 (Polaroid Corp., Bedford, MA) and processed using Adobe PhotoShop version 6 (Adobe Systems Inc., San Jose, CA) on a Pentium III PC platform. SEM To expose suture patterns at the poles and/or ﬁber surface morphology along the ﬁber length, the lenses were dissected to remove the elongating ﬁber layers, as previously described (Al-Ghoul et al., 1998). To enable a morphological examination of the cortical and nuclear ﬁbers, the lenses were split along the polar axis, thereby exposing radial cell columns (RCCs) and GSs. The tissue was postﬁxed overnight in 1% aqueous osmium tetroxide at 4°, washed in cacodylate buffer, and dehydrated through a graded ethanol series. After the tissue was dehydrated overnight in 100% ethanol, the ethanol was replaced through a graded ethanol/freon 113 series to pure freon 113. Specimens were critical-point dried in Freon 23 (Dupont, Wilmington, DE) in a Balzers CPD 020 (Balzers, Hudson, NH), secured on aluminum stubs with silver paste, sputter-coated with gold, and examined in a JEOL JSM 35c scanning electron microscope (JEOL USA, Peabody, MA) at 15 kV. Freeze-Etch As described earlier, all vertebrate lenses are stratiﬁed epithelia that develop and grow in an inverted vesicle conﬁguration. As a result, unlike other stratiﬁed epithelia, lenses retain every cell formed throughout life. Thus, in any lens there are terminally differentiating ﬁbers with a full complement of cytoplasmic organelles, mature ﬁbers 717 that have eliminated most cytoplasmic organelles, and, in long-lived species, ﬁbers with an age range that can span decades. In addition, a unique polar ﬁber morphology has been proposed to subserve lens physiology. Mathias et al. (1997) theorized that current circulates around and through the lens, inwardly at the poles and outwardly at the equator. The inward movement of ﬂuid is described as being along the polar intercellular clefts, or by structural deﬁnition the sutures, convecting glucose to the innermost ﬁbers, where it is used for anaerobic metabolism. The outward movement of ﬂuid is described as being intracellular, convecting waste products of metabolism out of the lens at the equator (presumably through GJs). Previous ultrastructural studies conﬁrmed that GJs are not uniformly distributed along the ﬁber length (Fitzgerald, 1986; Kuszak, 1995b). Fiber ends, or those portions of ﬁbers involved in sutures and thus the inward movement of ﬂuid, have few GJs. Fiber mid-portions, or those segments of ﬁbers not involved in sutures, are thought to be responsible for outward movement of ﬂuid, through their numerous GJs. In addition, the density of GJs conjoining the mid-segment of ﬁbers varies between species (Kuszak et al., 1985; Fitzgerald, 1986; Lo and Harding, 1986; Kuszak, 1995b). The mid-segments of chick, rat, frog, and human ﬁbers are reported to have 65%, 33%, 12%, and 5%, respectively, of their membrane specialized as GJs. Note that the lenses with the most GJs have the simplest suture system, while lenses with the fewest GJs have the most complex suture system. The above ﬁndings indicate that no one group of ﬁbers from any vertebrate lens, or one lens from any species, is completely representative of all ﬁbers or all lenses. When analyzing ﬁber ultrastructure, it is imperative that both the region of a ﬁber along its length and the age of the ﬁber be qualiﬁed. Therefore, we developed (and routinely employ) a dissection protocol that guarantees the unequivocal retrieval of anterior and posterior end segments of ﬁbers, as well as the mid-segment of the same ﬁbers, at progressive depths in any lens. The dissection protocol is shown in Figure 2. Brieﬂy, under a surgical dissecting microscope equipped with an ocular reticle, the axial dimensions and suture patterns of a lens are recorded prior to dissection in buffer. The Y sutures are apparent in a ﬁxed, normal mouse lens (Figs. 1C and 2A). With a pair of #5 electron microscopy (EM) forceps, the ﬁber ends that abut and overlap at two adjoining suture branches are separated. Then, by gently sliding the tips of the forceps beneath the group of ﬁbers that had been part of the two adjoining suture branches, the entire group of ﬁbers can be peeled off of the lens (Fig. 2A). The resulting crescent-shaped group of ﬁbers is shown in Figure 2B at the time of dissection, and in Figure 2D by SEM. Because of ﬁber opposite end curvature, the peel contains an intact suture branch, as conﬁrmed by SEM (Fig. 2D, arrow). Additional peels containing other suture branches are then retrieved. At this point, the equatorial diameter of the remaining lens mass is recorded. By subtracting this measure from that of the original, undissected lens, it can be ascertained whether the ﬁber peels contain superﬁcial cortical (peripheral 200 at the equator) and/or cortical ﬁbers (the next 500 – 800 at the equator). The peels are then cut into three pieces containing an intact anterior suture branch, mid-segment of ﬁbers, and posterior ends of ﬁbers as seen at the time of dissection (Fig. 2C) and conﬁrmed by SEM (Fig. 2E and F). 718 AL-GHOUL ET AL. Fig. 2. Dissection of mouse lenses for freeze-etch TEM. A–C: Photomicrographs taken through a surgical dissecting microscope. D–F: SEM micrographs of comparable specimens. To obtain specimens containing either a ﬁber end or mid-segment, groups of ﬁbers are carefully peeled away from the lens, beginning at one of the poles (A). The resulting crescent-shaped specimens (B and D) are then cut to isolate speciﬁcally the end segments from the mid-segment of ﬁbers (C and E). Because one set of ﬁber ends (either anterior or posterior ends of ﬁbers) are grasped by the EM forceps, they may be crushed and therefore are discarded. Progressive dissections to retrieve peels of deep cortical ﬁbers and, successively, nuclear ﬁbers can then be performed. If intact anterior suture branches were retrieved by the above procedure, then the contralateral lens is used for retrieval of posterior suture branches. Note that further dissection of groups of ﬁbers into distinct ⬍1 mm3 pieces at or near sutures (anterior and posterior segments), as well as off sutures (mid-segment of ﬁbers) guarantees that the distinct ultrastructure along ﬁber length can be studied without any overlap between adjacent segments. The above procedure has been performed successfully on mouse, rat, guinea pig, chicken, cat, dog, monkey, and human lenses. Anterior, posterior, and equatorial ﬁber segments of only mature cortical ﬁbers from 10-, 18-, and 24-week-old wild-type, heterozygote, and homozygote lenses were dissected from ﬁxed lenses and prepared for freeze-etch analysis by standard techniques (Kuszak et al., 1982). Brieﬂy, after primary ﬁxation, the tissue was cryoprotected by gradual inﬁltration of 25% glycerol prepared in cacodylate buffer. The specimens were mounted with appropriate orientation (broad faces mounted parallel to the frozen knife) on standard Balzers gold specimen carriers, and snap-frozen in supercooled liquid nitrogen. Fracture and LENS STRUCTURE IN MIP-DEFICIENT MICE replication were carried out at –115 C in a Balzers 301 freeze-etch unit according to standard techniques (Moor and Muhlethaler, 1963). A drop of 0.5% colloidon was placed on the replica to enhance its structural integrity. Replicas were retrieved onto formvar-coated, 100-mesh copper grids and examined at 2,500 –100,000⫻ magniﬁcation in a JEOL JEM 1200EX transmission electron microscope at 80 kV. Quantiﬁcation of GJ and non-GJ areas, as well as GJ size and percent distribution from the anterior and posterior ends and equatorial segments, was accomplished using PC Scion Image, Beta v. 4.0.2 (Scion Corp., Frederick, MD). Approximately 2,500 m2 of mature cortical ﬁber membrane were replicated in each of the three lens genotypes (wild-type, heterozygote, and homozygote). Of these amounts, 400 –500 m2 of membrane of each ﬁber region were randomly selected for statistical analysis. A KruskalWallis one-way analysis of variance (ANOVA) was used to test for signiﬁcant differences between groups. RESULTS Fiber Morphology (LM and SEM) At the lens equator, nascent ﬁbers rotate 90° about their axis as they begin elongating bidirectionally toward the poles (Duke-Elder and Wybar, 1961). This area, which is called the “bow” region because of the conﬁguration of the nuclei from the differentiating cells, was examined by LM of thick (1–2 m) sections. In all wild-type lenses, the transparent bow region was structurally consistent with normal ﬁber differentiation (Fig. 3A). The bow region of lenses from both heterozygote and homozygote MIP-deﬁcient lenses were comparable to the wild-type lenses at all ages examined (Fig. 3B and C). Further histological examination of MIP-deﬁcient lenses (translucent heterozygote lenses and opaque homozygote lenses) revealed structural abnormalities, especially at the ﬁber ends. The ﬁber ends were markedly larger than normal, and/or excessively disorganized as compared to ﬁber ends in wild-type lenses (Fig. 3D–I). These abnormalities became progressively more severe with time (data not shown). SEM was used to evaluate ﬁber surface morphology and to expand on LM observations. In wild-type lenses split along the anterior–posterior axis, ﬁbers were arranged in ordered RCCs and GSs throughout the cortical and nuclear regions (Fig. 4A). In addition, all cortical ﬁbers were of uniform shape (hexagonal) and size, and had typical lateral interdigitations regularly arrayed along their length (Figs. 4B and 5A). In contrast, the shape and arrangement of ﬁbers in the heterozygote lenses were less uniform (Fig. 4C). Speciﬁcally, some cortical ﬁbers were enlarged (Fig. 4C, black asterisks), thereby disrupting the ordered arrangement of RCCs and GSs in adjacent ﬁber layers. Furthermore, the cortical ﬁbers of heterozygotes had lateral interdigitations that were smaller and less regularly arrayed along the ﬁber length compared to the age-matched wild-type lenses (Figs. 4D and 5B). By 3 weeks after birth, anterior regions of the heterozygote lenses often displayed acellular material, which likely resulted from the breakdown of cortical ﬁber ends (Fig. 4C, white asterisk). SEM of homozygotes showed that although these lenses were grossly of comparable shape and size to both heterozygote and wild-type lenses, they had almost no semblance of normal ﬁber shape or arrangement (Fig. 4E). Rather, these lenses had amorphous ﬁbers that were only arranged in a vague conﬁguration of RCCs 719 or GSs (Fig. 4E and F). Nevertheless, very irregular lateral-edge processes were still present along the lengths of homozygote cortical ﬁbers (Figs. 4F and 5C) Because LM examination revealed early and marked structural compromise at the ﬁber ends, SEM was utilized to assess ﬁber-end morphology and sutural architecture. Wild-type lenses split along the polar axis revealed normal anterior and posterior suture planes (Fig. 4A, arrows; and Fig. 6A, asterisks) wherein all ﬁber ends exhibited the proper curvature to abut and overlap with opposing ends. Fiber peels exposed the suture branches within GSs in the deep cortex. The ends of mature wild-type ﬁbers were slightly ﬂared and irregularly shaped (Fig. 6B), as is typical of properly formed sutures (Kuszak et al., 1996). In contrast, the posterior ends of superﬁcial cortical ﬁbers from heterozygotes were curved away from the polar axis and toward the vitreous (Fig. 6C, arrows). Suture planes (Fig. 6C, asterisks) were excessively disorganized due to the presence of abnormally larger ﬁber ends. These markedly larger ends were especially obvious in ﬁber peels showing the suture branches formed by mature cortical ﬁbers in heterozygotes (Fig. 6D). The anterior ends of ﬁbers often appeared to be globular (data not shown), leading to complete cellular breakdown in this region (as mentioned above). Consequently, anterior suture planes were not present in most heterozygotes by 3 weeks of age (see Fig. 4C). In homozygote lenses split along the polar axis, there were no apparent suture planes at any age examined (Fig. 4E and 6E). Similarly, ﬁber peels failed to reveal any suture branches in these lenses (Fig. 6F). The complete lack of uniformity in ﬁber size, shape, opposite end curvature, and arrangement precluded suture formation in homozygote lenses. It is well established that ﬁber morphology in the nuclear region is less uniform than in the cortical region in vertebrate lenses (Willekens and Vrensen, 1982; Kuszak et al., 1983; Kuszak, 1995b; Taylor et al., 1996; Al-Ghoul and Costello, 1997; Shestopalov and Bassnett, 2000). The nuclear ﬁber morphology in wild-type lenses observed in the present study was consistent with the above-cited reports. Speciﬁcally, nuclear ﬁbers were wider than cortical ﬁbers, displayed variability in width along the ﬁber length (Fig. 7A, arrows), and had lateral interdigitations of variable size and shape (Fig. 7A, arrowheads). In comparison, nuclear ﬁbers in heterozygotes were markedly more irregular in shape and appeared to have roughened surface morphology, possibly due to membrane degeneration (Fig. 7B). As noted in cortical ﬁbers, nuclear ﬁbers in homozygotes were amorphous and had no regular arrangement (Fig. 7C). At higher magniﬁcation, it was revealed that these ﬁbers displayed a complete lack of interdigitations (Fig. 7D). Freeze-Etch Freeze-etch was used to examine the ultrastructure of cortical (fully elongated and mature) ﬁber membrane from wild-type, heterozygote, and homozygote mouse lenses at 10, 18, and 24 weeks of age. Thus, while we assessed the contribution of MIP to ﬁber development and lens growth by LM and SEM, freeze-etch was used only to examine cortical ﬁber membrane ultrastructure in adult lenses. All of the quantitative data (mean, median, and size range of GJs) from wild-type, heterozygote, and homozygote lens cortical ﬁbers are summarized in Table 1. 720 AL-GHOUL ET AL. Fig. 3. Light micrographs of thick (1–2 m) sections taken through the polar axis of mouse lenses. Left column, wild-type; center column, heterozygote; right column, homozygote. In the equatorial region of wild-type (A), and MIP-deﬁcient (B and C) lenses, morphology appeared normal at all ages examined. It was apparent that the initial differentiation and elongation of lens ﬁbers was not adversely affected by MIP deﬁ- ciency. However, by 1 week of age, MIP-deﬁcient lenses had abnormally larger ﬁbers that were not organized into proper RCCs and/or GSs. This was especially evident at the anterior (D–F) and posterior (G–I) ends. Panels A–C are at equivalent magniﬁcation, and panels D–I are at equivalent magniﬁcation. Fiber peels comprising approximately one-third of the lens equatorial diameter were removed from the periphery of lenses. This initial set of peels (approximately 250 m thick) contained superﬁcial cortical ﬁbers (both elongating and fully elongated and maturing), wherein the elaboration of MIP is an ongoing process. The second set of peels, containing only mature cortical ﬁbers, wherein the elaboration of MIP is complete, was taken for freeze-etch anal- ysis. These peels were further split into anterior, mid, and posterior segments (see Fig. 2). Freeze-etch replicas of all wild-type and heterozygote lenses simultaneously revealed the membrane ultrastructure of deﬁned BFs and narrow faces (NFs) on a number of ﬁbers aligned in deﬁned GSs and RCCs (Fig. 8A). In addition, replicas from all segments of wild-type ﬁbers conﬁrmed the uniformity in ﬁber shape (hexagonal) and typ- LENS STRUCTURE IN MIP-DEFICIENT MICE 721 Fig. 4. SEM micrographs of wild-type and MIP-deﬁcient mouse lenses split along the polar axis. Wild-type lenses (A and B) had transparent ﬁbers of uniform shape and size arranged in ordered RCCs and GSs. Normal anterior and posterior suture planes were present (arrows). By comparison, heterozygote lenses (C and D) had translucent ﬁbers that were disordered due to occasional abnormally large ﬁbers (black asterisks). Regions of apparent ﬁber breakdown were also noted (white asterisk). Homozygote lenses (E and F) had opaque ﬁbers that displayed a complete lack of uniformity with respect to size, shape, and arrangement. Speciﬁcally, these ﬁbers were not arranged in any discernible RCCs and GSs, and had not formed any sutures. It also appeared that cortical ﬁber interdigitations in MIP-deﬁcient lenses (D and F) were less regular in size and shape as compared to those from wild-type lenses (B). Panels A, C, and E are at equivalent magniﬁcation; white boxes indicate the enlarged regions shown at medium magniﬁcation in B, D, and F. Panels B, D, and F are at equivalent magniﬁcation. ical, lateral edge processes arrayed along the ﬁber length (Fig. 8A, arrowheads). The ability to simultaneously replicate identiﬁable BFs and NFs correlated well with the SEM results (Figs. 4 – 6). The translucent, heterozygote ﬁbers still had a relatively hexagonal cross-sectional shape. In contrast, freeze-etch replicas of the opaque homozygote cortical ﬁbers never simultaneously revealed deﬁned BFs and NFs in any ﬁbers, and the ﬁbers were never aligned in either deﬁned GSs and/or RCCs (Fig. 8B). The inability to simultaneously replicate identiﬁable BFs and NFs on cortical ﬁbers from homozygote lenses also correlated well with the SEM results (Figs. 4 – 6). The homozygote ﬁbers always had a non-hexagonal, abnormally large, polygonal, cross-sectional shape. All segments of wild-type ﬁbers had a very high non-GJ particle density, reﬂective of a full complement of MIP having been produced and inserted into the ﬁber membrane. However, square array membrane (orthogonal arrays of MIP) was only observed at high magniﬁcation on low-amplitude membrane of anterior and posterior end segments from wild-type lenses (Fig. 9A). In addition, these ﬁber segments were essentially devoid of GJs (⬎1%). 722 AL-GHOUL ET AL. In contrast, 13.18% of the mid-segment of wild-type ﬁbers were specialized as GJs. Quantiﬁcation of individual GJ plaque areas revealed that wild-type GJ size ranged from 0.004 to 1.574 m2. Freeze-etch replicas of all segments of ﬁbers from heterozygote lenses conﬁrmed an alteration in ﬁber uniformity (shape and size), and a less than normal array of lateral-edge processes initially found by SEM examination (data not shown). All regions of these ﬁbers had reduced intramembrane particle (IMP) density, indicative of the decreased production and/or lack of insertion of MIP into ﬁber membrane (Figs. 9B and 10). Square array membrane was not observed on any segment of ﬁbers from heterozygote lenses (Figs. 9B and 10). Examination of the mid-segment of heterozygote cortical ﬁber membrane revealed that the percentage (14.5%) of this membrane specialized as GJs, was essentially identical to that quantiﬁed in the comparable region of wild-type lenses (13.15%). However, the mean, median, and range of GJ size between these two groups was signiﬁcantly different (P ⱕ 0.001). Freeze-etch replicas from all segments of homozygote lens ﬁbers conﬁrmed the lack of any semblance of ﬁber uniformity in shape and size. In addition, there were reduced numbers and more irregularly arrayed lateral-edge processes, as shown previously (Fig. 8B). Furthermore, as expected, a paucity of non-GJ particles was observed, conﬁrming that MIP production had been “knocked out” (Fig. 11). It was also noted that, as in the heterozygote ﬁbers, square arrays were not observed on any segment of the homozygote ﬁbers. In addition, only 6.81% of the midsegment of these ﬁbers was specialized as GJs, literally half of that quantiﬁed in both wild-type and heterozygote lenses. Furthermore, whereas the heterozygote and homozygote cortical ﬁbers had the same mean GJ size, the median and range of GJ size between these two groups was signiﬁcantly different (P ⱕ 0.001). DISCUSSION Fig. 5. Higher-magniﬁcation SEM micrographs of the mid-segment of cortical ﬁbers. In wild-type lenses, lateral interdigitations formed at the angle of two narrow faces were of consistent shape, size, and spacing (A). High magniﬁcation conﬁrmed that alterations to the lateral interdigitations on cortical ﬁbers were present in MIP-deﬁcient lenses. Speciﬁcally, interdigitations on both heterozygote (B) and homozygote lens ﬁbers (C) were smaller than on ﬁbers from wild-type ﬁbers (A). Additionally, in homozygotes these membrane specializations were less numerous and lacked the orderly arrangement seen in both wild-type and heterozygote lenses. Panels A–C are at equivalent magniﬁcation. This study details the morphology of lenses from MIPdeﬁcient mice. An examination of heterozygote lenses revealed speciﬁc structural changes as a result of reduced synthesis of MIP in mature lens ﬁbers that was considerably more ampliﬁed in homozygote lenses. At birth, the heterozygote lenses appeared unaffected. However, by P7, these lenses were translucent and both the anterior and posterior ends of mature cortical ﬁbers were abnormally large, with a very nonuniform shape. In addition, these ﬁbers had atypical lateral interdigitations. These lateral-edge processes were smaller and less regularly arrayed along the ﬁber length than in age-matched wild-type lenses. By P21, it was apparent that the atypical structure of heterozygote cortical ﬁber ends precluded the formation of normal Y sutures. The posterior ends of these ﬁbers were consistently seen to be aberrantly curved away from the polar axis. Nuclear ﬁbers from these lenses lacked lateral interdigitations by 4 weeks of age. In a related study (Shiels et al., 2001), it was shown that the structural abnormalities in heterozygote lenses were sufﬁcient to effect a signiﬁcant compromise in lens function (sharpness of focus). In the same study it was also found that the structural abnormalities in homozygote lenses were so severe that the lenses were rendered completely opaque, and thus were completely nonfunctional. It is clear from this report that a deﬁciency in MIP does not prevent lens ﬁber differentiation and elongation. LENS STRUCTURE IN MIP-DEFICIENT MICE 723 Fig. 6. SEM micrographs of ﬁber-end and sutural architecture. Lenses split along the polar axis (left column) revealed RCCs and GSs in the cortex as well as the suture planes (white asterisks). Crescentshaped ﬁber “peels” (right column) exposed the BFs of cortical ﬁbers within one GS, as well as the corresponding suture branch. In wild-type lenses (A and B), ﬁbers always featured proper end curvature to form sutures. Fiber ends were slightly ﬂared and irregular, forming continuous suture planes and branches. However, in heterozygotes (C and D), the posterior ﬁber ends were aberrantly curved away from the polar axis and toward the vitreous (C, arrows). These cortical ﬁber ends were abnormally large, resulting in excessive disorder at sutures. Examination of homozygote lenses (E and F) failed to reveal any suture branches or planes. The lack of suture architecture in homozygote lenses was due to the complete absence of ﬁber uniformity in size, shape, opposite end curvature, and/or arrangement in proper RCCs and GSs. These ﬁndings are consistent with results in two MIP mouse mutants, the Fraser cataract mice (CATFr) and lense opacity (LOP) models (Novak et al., 1999; Shiels et al., 2000). A transposon-induced splicing error in the CATFr mouse results in the addition of a long terminal repeat sequence, and MIP remains in intracellular membranes. In the LOP mouse model, a missense mutation inhibits targeting of MIP to the ﬁber membrane, and it remains in the cytoplasm. In both of the mutants, MIP was shown to have been produced but not inserted into ﬁber plasma membrane. Comparable structural alterations in mature ﬁbers were seen in both of these animal models. Mature ﬁbers in both the cortex and nucleus of MIP-deﬁcient lenses were unable to maintain the long, ribbon-like shape that was achieved during elongation. In addition, the concentric arrangement of ﬁbers into highly 724 AL-GHOUL ET AL. Fig. 7. SEM micrographs of nuclear ﬁber morphology. Transparent nuclear ﬁbers in wild-type lenses (A) were wider than cortical ﬁbers, displayed variability in width along ﬁber length (arrows), and had variably-sized interdigitations between cells (arrowheads). By comparison, translucent heterozygote nuclear ﬁbers (B) were more irregular in shape and had roughened surface morphology. Opaque nuclear ﬁbers in ho- mozygotes (C) were amorphous and had no regular arrangement. In addition, higher magniﬁcation revealed that these ﬁbers completely lacked interdigitations between cells (D). Panels A–C are at equivalent magniﬁcation. White box in panel C indicates the region shown at higher magniﬁcation in panel D. TABLE 1. Quantitative data derived from freeze-etch replication of wild type, heterozygote and homozygote mip knockout mice cortical ﬁbers showing the number, mean and median size and size range of gap junctions Number of GJs analyzed Mean area/GJ ⫾ s.e.m. (m2) Median area of a GJ (m2) Range of GJ size (min. -max.; m2) Total GJ area (m2) Total membrane area (m2) % membrane specialized as GJ Wild type Heterozygote Homozygote 699 0.087 ⫾ 0.006 0.036 0.004–1.574 60.580 460.62 13.15% 572 0.102 ⫾ 0.006* 0.048 0.003–0.832* 58.343 402.28 14.50% 341 0.102 ⫾ 0.015** 0.017 0.002–2.410*** 34.857 512.073 6.81% Statistical analysis was only used to compare mean GJ size and range in GJ size. These are the variables that were derived from measurements made on the micrographs. The other values presented in the table (median size of GJs and % of membrane specialized as GJ) are total values calculated from average values. Since there is only one value per variable, statistical analysis of these values is not possible (Jaeger, 1993). P ⬍ 0.001 for the following comparisons: *Wild type vs. heterozygote; **wild type vs. homozygote; ***heterozygote vs. homozygote. ordered GSs was progressively degraded as ﬁbers aged. Results from the present study were consistent with observations in both CATFr and LOP mouse lenses (Shiels et al., 2000). The similarities among these three models indicate that while MIP is not essential for ﬁber formation, it is important in the establishment and maintenance of both individual ﬁber shape and overall ﬁber arrangement. One explanation for the observed changes in ﬁber structure is that MIP may be involved in the maintenance of ﬁber–ﬁber interactions that regulate morphology. A clue to the mechanism by which MIP could regulate lens ﬁber structure may be seen in the morphology of ﬁber interdigitations shown in this study. Qualitatively, MIP-deﬁcient cortical ﬁbers appeared to have fewer lateral interdigitations with altered size and shapes as compared to normal control lenses. MIP-deﬁcient nuclear ﬁbers were even more compromised, with sparse or no lateral interdigitations, as a function of development and growth. Fiber lateral interdigitations, such as balls and sockets, ﬂaps, and imprints (Kuszak et al., 1996), and both high- and LENS STRUCTURE IN MIP-DEFICIENT MICE Fig. 8. TEM of the mid-segment of freeze-etch replicated cortical ﬁbers from wild-type (A) and homozygote (B) lenses. The characteristic cross-sectional shape (hexagonal) of a number of wild-type ﬁbers (F1– F3) is readily apparent. Large expanses of both BFs and NFs of these ﬁbers aligned in GSs and RCCs are apparent. In contrast, while the 725 mid-segment of freeze-etch-replicated homozygote cortical ﬁber membrane revealed large expanses of membrane, the markedly irregular (non-hexagonal) shape of these ﬁbers precluded positive identiﬁcation of either deﬁned BFs or NFs. 726 AL-GHOUL ET AL. Fig. 9. TEM of freeze-etch-replicated, cortical ﬁber-end segments from wild-type (A) and heterozygote (B) lenses. Both the anterior and posterior ﬁber-end segments of ﬁbers from all three groups were characterized by a paucity of GJs (⬍1% of these ﬁber regions were special- ized as GJs). While square array membrane, or orthogonal arrays, comprised of MIP, were readily observed in wild-type lenses, they were never observed in either heterozygote or homozygote lenses. Panels A and B are at equivalent magnifcation. low-amplitude wavy membranes (which are equivalent to polygonal domains of furrowed membrane seen by SEM (Taylor et al., 1996; Kuszak and Costello, 2003)), keep ﬁbers in close apposition, minimizing extracellular space and thereby contributing to transparency by reducing large-particle scatter. Lateral interdigitations may also serve to maintain the relative positions of ﬁbers in RCCs and concentric GSs during accommodation or dynamic LENS STRUCTURE IN MIP-DEFICIENT MICE Fig. 10. TEM of the mid-segment of freeze-etch-replicated cortical ﬁbers from a heterozygote lens. Although the total percentage (14.5%) of this ﬁber region specialized as GJs was the same as that quantiﬁed in the comparable region of wild-type lenses (13.18%), the median and range in size of the heterozygote GJs were signiﬁcantly different (P ⱕ 0.001) from that of the wild-type. focusing. Therefore, the absence of properly formed and arrayed ﬁber lateral interdigitations would likely lead to a degradation in lens optical quality. The initial structural alteration that occurred at mature ﬁber ends in the MIP knockout, LOP, and CATFr (Novak et al., 1999; Shiels et al., 2000) mouse lenses may have been the result of swelling. The unique vulnerability of ﬁber ends to cellular swelling is consistent with the fact that square array membrane, composed of MIP (Costello et al., 1989; Chandy et al., 1997), was shown in this study to occur predominantly at the anterior and posterior end segments of ﬁbers, while GJs were found to predominantly conjoin the mid-segment of ﬁbers. The fact that the size range of ﬁber GJs was signiﬁcantly altered in both heterozygote and homozygote lenses supports the proposition that MIP also has a role in establishing and maintaining ﬁber–ﬁber communication in the lens (Dunia et al., 1998). The polar distribution of ﬁber membrane specializations shown in this study may also relate to the prevailing opinion that an internal circulatory system is an important component of lens physiology (Mathias et al., 1997). However, a quantitative comparison of the polar distribution and the number of ﬁber membrane specializations in vertebrate lenses remains to be conducted. Indeed, while the predominance of GJs conjoining the middle segment of 727 Fig. 11. TEM of the mid-segment of freeze-etch-replicated cortical ﬁbers from a homozygote lens. The total percentage (6.81%) of this ﬁber region specialized as GJ was literally half of that quantiﬁed in either wild-type or heterozygote lenses. However, whereas the mean GJ size of homozygote GJs was the same as that quantiﬁed for heterozgote GJs, the median and range in size of the GJs were signiﬁcantly different between these two groups (P ⱕ 0.001), as well as between either of these two groups and the GJs quantiﬁed from wild-type lenses. Note the paucity of nonjunctional IMPs in this replica as compared to that seen in heterozygote lenses (Fig. 10), owing to the lack of MIP production and insertion into ﬁber membrane in these KO lenses. ﬁbers has been shown to be a consistent lens structural feature in all vertebrates (Fitzgerald, 1986; Kuszak et al., 1996), the amount of this ﬁber region specialized as GJs varies considerably between species (Kuszak et al., 1985; Lo and Harding, 1986; Kuszak et al., 1996). Freeze-etch studies have shown that GJs constitute ⬃60%, 30%, 15%, and 5% of middle-segment ﬁber plasma membrane in chick, rat, frog, and human lenses, respectively. Comparable quantitative analyses of the amount of ﬁber plasma membrane specialized as square array membrane between species have not been performed to date. Such freeze-etch studies will be inherently more difﬁcult to conduct. Aggregates of MIP as square array membrane are generally smaller and less circumscribed than ﬁber GJs, since they do not exist as raised oval and circular plaques on P-faces of replicas, or as similarly shaped depressions on E-face exposures. However, in bovine lenses, vast expanses of ﬁber plasma membrane are specialized as square array membrane (Costello et al., 1989; Zampighi et al., 1989). Qualitatively, the amount of ﬁber membrane specialized as square array membrane appears to be at 728 AL-GHOUL ET AL. least as variable between species as are the GJs. Thus, if both the square array membrane and the GJs are elements of a lens internal circulatory system, then variations in their polarized distribution and relative amounts in different species would inﬂuence the effectiveness of ﬂuid transport in each lens type as a function of development, growth, age, and pathology. Finally, the results reported herein may appear to be inconsistent with those from a recent study by Zampighi et al. (2002), but actually they are not. In the study by Zampighi et al. (2002), freeze-fracture labeling was used to assess the spatial arrangement and interactions of MIP with other proteins in the plasma membrane of the most nascent ﬁbers. It was concluded that 1) MIP was arranged in microdomains that extended along the long axis of nascent ﬁbers, and 2) the density of MIP varied along the long axis of nascent ﬁbers, being least at the apical end and greatest at the mid-segment. Furthermore, while the MIP microdomains of nascent ﬁbers contained densely packed MIPs, they were not arranged in characteristic orthogonal arrays or square array membrane (Costello et al., 1989). As described above, crystalline lenses consist of ﬁbers in different states of terminal differentiation, maturation, and aging. The nascent ﬁbers replicated by Zampighi et al. (2002) were probably only elongated to approximately 5% of their eventual mature length (2 mm). As such, they contained only a small portion of their future full complement of the major plasma membrane proteins (MIP, membrane protein 19kd [MP19], and GJ connexins 46 and 50) and underlying cytoskeletal network. Furthermore, such nascent ﬁbers have not even begun the process of developing the extensive array of ﬁber–ﬁber lateral interdigitations and sutures necessary to establish and maintain the exact arrangement of ﬁbers as RCCs and GSs required for lens function (i.e., dynamic focusing) (Kuszak et al., 1996). Therefore, the nascent ﬁbers examined by Zampighi et al. (2002) were different from the mature cortical ﬁbers investigated in the current study. It is also important to note that since both studies involved young animals (maximum age 6 months), the effects of aging on the contribution of MIP to ﬁber ultrastructure must still be addressed. This is an important consideration in lens physiology because the lens retains every ﬁber formed throughout life. Fig. 12. Color-coded 3D-CADs summarizing diagrammatically the contributions of MIP to lens structure. Epithelial membrane is depicted as white, MIP is yellow, and GJs are blue. A: During normal differentiation of epithelial cells into ﬁbers, MIP is inserted into ﬁber plasma membrane. The mid-segment of ﬁbers contains both GJs (blue) and MIP (yellow), and is therefore depicted as green (blue ⫹ yellow ⫽ green). Mature cortical ﬁbers have a uniform shape and size, resulting in ordered GSs, RCCs, and sutures. B: In heterozygote lenses, the end segments are depicted as a lighter yellow, while the mid-segment are depicted as more blue. Both color changes are indicative of the reduction in MIP synthesis and/or insertion into the membrane. The cortical ﬁbers of heterozygotes are less uniform in shape and size, especially at ﬁber ends, resulting in less ordered GSs, RCCs, and sutures. C: During ﬁber differentiation in homozygote lenses, MIP is not produced, and therefore the end segments of these ﬁbers are depicted as white. Consistent with the lack of MIP resulting in essentially a 50% reduction in the percentage of the mid-segment of these ﬁbers being specialized as GJs, this region of homozygote ﬁbers is depicted as light blue. These cortical ﬁbers lack any semblance of uniformity in shape and size along their entire length, resulting in malformation of GSs, RCCs, and sutures. LENS STRUCTURE IN MIP-DEFICIENT MICE In summary, the contributions of MIP to lens structure are shown diagrammatically by color-coded, three-dimensional, computer-assisted drawings (3D CADs) presented in Figure 12. As ﬁber terminal differentiation occurs, the principal membrane protein produced and inserted is MIP (represented as yellow). Lens epithelial membrane, which lacks MIP, is depicted as white. The mid-segment of ﬁbers are depicted as green, consistent with the presence of both GJs (represented as blue) and MIP (blue ⫹ yellow ⫽ green). Mature ﬁbers are of a uniform shape and size, resulting in ordered GSs, RCCs, and sutures. By comparison, as ﬁber terminal differentiation occurs in heterozygote lenses, elongating ﬁber membrane is depicted as light yellow at the ends, and more blue at the mid-segment, consistent with a reduction in MIP production and/or insertion. The mature ﬁbers of heterozygote lenses are of less uniform shape and size, particularly at the ﬁber ends, resulting in less ordered GSs, RCCs, and sutures. Finally, in homozygote lenses, ﬁber membrane is depicted as white at the ﬁber ends, and light blue at the mid-segment, consistent with essentially a complete lack of MIP. The mature ﬁbers of homozygote lenses are of nonuniform shape and size along their entire length, resulting in malformation of GSs, RCCs, and sutures. In conclusion, even a 50% loss of MIP is sufﬁcient to signiﬁcantly compromise lens function (dynamic focusing) and eventually lead to cataract, because MIP is necessary to establish and maintain ﬁber shape and (by extrapolation) lens sutures, and ﬁber lateral interdigitations. It is also essential in establishing and determining the size and distribution of ﬁber GJs (predominantly at a ﬁber’s mid-segment), and establishing square array membrane (predominantly at ﬁber-end segments). ACKNOWLEDGMENTS This work was supported in part by NIH NEI grants EY06642 (to J.R.K.) and EY11411 (to A.S.). The technical assistance of Mr. Layne Novak and Kurt L. Peterson is gratefully acknowledged. LITERATURE CITED Alcala J, Lieska N, Maisel H. 1975. Protein composition of bovine lens cortical ﬁber cell membranes. Exp Eye Res 21:581–595. Alcala J, Maisel H. 1985. Biochemistry of lens plasma membrane and cytoskeleton. In: Maisel H, editor. The ocular lens: structure, function and pathology. New York: Marcel Dekker, Inc. p 169 –222. Al-Ghoul KJ, Costello MJ. 1997. Light microscopic variation of ﬁber cell size, shape and ordering in the equatorial plane of bovine and human lenses. Mol Vis 3:2. Al-Ghoul KJ, Novak LA, Kuszak JR. 1998. The structure of posterior subcapsular cataracts (PSCs) in Royal College of Surgeons (RCS) rats. Exp Eye Res 67:163–177. Al-Ghoul KJ, Novak LA, Peterson KL, Kuszak JR. 1999. Internalization of posterior subcapsular cataracts (PSCs) in Royal College of Surgeons (RCS) rats. I. Mol Vis 5:6. Baker ME, Saier Jr MH. 1990. A common ancestor for bovine lens ﬁber major intrinsic protein, soybean nodulin-26 protein, and E. coli glycerol facilitator. Cell 60:185–186. Bok D, Dockstader J, Horwitz J. 1982. Immunocytochemical localization of the lens main intrinsic polypeptide (MIP26) in communicating junctions. J Cell Biol 92:213–220. Chandy G, Zampighi GA, Kreman M, Hall JE. 1997. Comparison of the water transporting properties of MIP and AQP1. J Membr Biol 159:29 –39. 729 Costello MJ, McIntosh TJ, Robertson JD. 1989. Distribution of GJs and square array junctions in the mammalian lens. Invest Ophthalmol Vis Sci 30:975–989. Duke-Elder S, Wybar KC. 1961. The refractive media. In: Duke-Elder S, editor. Systems of ophthalmology. Vol. II. St. Louis: C.V. Mosby Co. p 309 –324. Dunia I, Manenti S, Rousselet A, Benedetti EL. 1987. Electron microscopic observations of reconstituted proteoliposomes with the puriﬁed major intrinsic membrane protein of eye lens ﬁbers. J Cell Biol 105:1679 –1689. Dunia I, Recouvreur M, Nicolas P, Kumar N, Bloemendal H, Benedetti EL. 1998. Assembly of connexins and MP26 in lens ﬁber plasma membranes studied by SDS-fracture immunolabeling. J Cell Sci 111:2109 –2120. Ehring GR, Zampighi GA, Horwitz J, Bok D, Hall JE. 1990. J Gen Physiol 96:631– 664. Fitzgerald PG, Bok D, Horwitz J. 1983. Immunocytochemical localization of the main intrinsic polypeptide (MIP) in ultrathin frozen sections of rat lens. J Cell Biol 97:1491–1499. Fitzgerald PG, Bok D, Horwitz J. 1985. The distribution of the main intrinsic membrane polypeptide in ocular lens. Curr Eye Res 4:1203–1218. Fitzgerald PG. 1986. The main intrinsic polypeptide and intercellular communication in the ocular lens. In: Hilfer RS, Shefﬁeld JB, editors. Development of order in the visual system. New York: Springer-Verlag. p 61–96. Fotiadis D, Hasler L, Muller DJ, Stahlberg H, Kistler J, Engel A. 2000. Surface tongue-and-groove contours on lens MIP faciitate cell-to-cell adherence. J Mol Biol 300:779 –789. Froger A, Tallur B, Thomas D, Delamarche C. 1998. Prediction of functional residues in water channels and related proteins. Protein Sci 7:1458 –1468. Goodenough DA. 1979. Lens gap junctions: a structural hypothesis for non-regulated low-resistance intercellular pathways. Invest Ophthalmol Vis Sci 11:1104 –1122. Gorin MB, Yancey SB, Cline J, Revel JP, Horwitz J. 1984. The major intrinsic protein (MIP) of the bovine lens ﬁber membrane: characterization and structure based on cDNA cloning. Cell 39:49 –59. Gruijters WT. 1989. A non-connexon protein (MIP) is involved in eye lens gap-junction formation. J Cell Sci 93(Pt 3):509 –513. Jaeger RM. 1993. Statistics: a spectator sport. Newbury Park, CA: Sage Publications, Inc. Kuszak JR, Alcala J, Maisel H. 1981. Biochemical and structural features of chick lens gap junctions. Exp Eye Res 33:157–166. Kuszak JR, Rae JL, Pauli BU, Weinstein RS. 1982. Rotary replication of lens gap junctions. J Ultrastruct Res 81:249 –256. Kuszak JR, Macsai MS, Rae JL. 1983. Stereo scanning electron microscopy of the crystalline lens. Scan Electron Microsc III:1415– 1426. Kuszak JR, Shek YH, Carney KC, Rae JL. 1985. A correlative freezeetch and electrophysiological study of communicating junctions in crystalline lenses. Curr Eye Res 4:1145–1153. Kuszak JR, Sivak JG, Weerheim JA. 1991. Lens optical quality is a direct function of lens sutural architecture. Invest Ophthalmol Vis Sci 32:2119 –2129. Kuszak JR, Peterson KL, Herbert KL, Sivak JG. 1994. The interrelationship of lens anatomy and optical quality. II. Primate lenses. Exp Eye Res 59:521–535. Kuszak JR. 1995a. Development of sutures in the lens. Prog Retin Eye Res 14:567–591. Kuszak JR. 1995b. The ultrastructure of lens epithelial and ﬁber cells in the crystalline lens. In: Jeon KW, editor. International review of cytology: a survey of cell biology. San Diego: Academic Press. p 305–350. Kuszak JR, Peterson KL, Brown HG. 1996. Electron microscopic observations of the crystalline lens. J Electron Microsc Technol 33:441– 479. Kuszak JR, Al-Ghoul KJ, Novak LA, Peterson KL, Herbert KL, Sivak JG. 1999. The internalization of posterior subcapsular cataracts (PSCs) in Royal College of Surgeons (RCS) rats. II. The interrelationship of optical quality and structure as a function of age. Mol Vis 5:7. 730 AL-GHOUL ET AL. Kuszak JR, Sivak JG, Herbert KL, Scheib S, Garner W, Graff G. 2000. The relationship between rabbit lens optical quality and sutural anatomy after vitrectomy. Exp Eye Res 71:267–282. Kuszak JR, Sivak JG, Moran KL, Scheib SA, Garner WH, Ke TL, Hellberg MR, Graf G. 2002. Suppression of post-vitrectomy lens changes in the rabbit by novel benzopyranyl esters and amides. Exp Eye Res 75:459 – 473. Kuszak JR, Costello MJ. 2003. The structure of vertebrate lenses. In: Lovicu FJ, Robinson ML, editors. Development of the ocular lens. Cambridge, UK: Cambridge University Press (in press). Lo WK, Harding CV. 1986. Structure and distribution of gap junctions in lens epithelium and ﬁber cells. Cell Tissue Res 244:253– 263. Mathias RT, Rae JL, Baldo GJ. 1997. Physiological properties of the normal lens. Physiol Rev 77:21– 49. Michea LF, de la Fuente M, Lagos N. 1994. Lens major intrinsic protein (MIP) promotes adhesion when reconstituted into large unilamellar liposomes. Biochemistry 33:7663–7669. Moor H, Muhlethaler K. 1963. Fine structure in frozen yeast cells. J Cell Biol 17:609. Novak LA, Al-Ghoul KJ, Kuszak JR, Collis HJ, Bassnett S, Shiels A. 1999. The structure of lens opacity (LOP) mouse lenses. Invest Ophthalmol Vis Sci 40:S883. Paul DL, Goodenough DA. 1983. Preparation, characterization, and localization of antisera against bovine MP26, an integral protein from lens ﬁber plasma membrane. J Cell Biol 96:625– 632. Rao Y, Jan LY, Jan YN. 1990. Similarity of the product of the Drosophila neurogenic gene big brain to transmembrane channel proteins. Nature 345:163–167. Reizer J, Reizer A, Saier Jr MH. 1993. The MIP family of integral membrane channel proteins: sequence comparisons, evolutionary relationships, reconstructed pathway of evolution, and proposed functional differentiation of the two repeated halves of the proteins. Crit Rev Biochem Mol Biol 28:235–257. Sandal NN, Marcker KA. 1988. Soybean nodulin 26 is homologous to the major intrinsic protein of the bovine lens ﬁber membrane. Nucleic Acids Res 16:9347. Sas DF, Sas MJ, Johnson RG. 1985. Lens junctional protein: analyzing MP26 with monoclonal antibodies. Curr Eye Res 4:1171–1182. Shestopalov VI, Bassnett S. 2000. Three dimensional organization of primary lens ﬁber cells. Invest Ophthalmol Vis Sci 41:1–5. Shiels A, Kent NA, McHale M, Bangham JA. 1988. Homology of MIP26 to Nod26. Nucleic Acids Res 16:9348. Shiels A, Mackay D, Bassnett S, Al Ghoul K, Kuszak J. 2000. Disruption of lens ﬁber cell architecture in mice expressing a chimeric AQP0-LTR protein. FASEB J 14:2207–2212. Shiels A, Bassnett S, Varadaraj K, Mathias R, Al-Ghoul K, Kuszak J, Donoviel D, Lillenberg S, Friedrich G, Zambrowicz B. 2001. Optical dysfunction of the crystalline lens in aquaporin-0 deﬁcient mice. Physiol Genomics 7:179 –186. Sivak JG, Herbert KL, Peterson KL, Kuszak JR. 1994. The interrelationship of lens anatomy and optical quality. I. Non-primate lenses. Exp Eye Res 59:505–520. Taylor VI, Al-Ghoul KJ, Lane CW, Davis VA, Kuszak JR, Costello MJ. 1996. Morphology of the normal human lens. Invest Ophthalmol Vis Sci 37:1396 –1410. Varadaraj K, Kushmerick C, Baldo GJ, Bassnett S, Shiels A, Mathias RT. 1999. The role of MIP in lens ﬁber cell membrane transport. J Membr Biol 170:191–203. Willekens B, Vrensen G. 1982. The three-dimensional organization of lens ﬁbers in the rhesus monkey. Arch Clin Exp Ophthalmol 219: 112–120. Zampighi GA, Hall JE, Ehring GR, Simon SA. 1989. The structural organization and protein composition of lens ﬁber junctions. J Cell Biol 108:2255–2275. Zampighi GA, Eskandari S, Hall JE, Zampighi L, Kreman M. 2002. Micro-domains of AQP0 in lens equatorial ﬁbers. Exp Eye Res 75:505–519.