Development- and Differentiation-Dependent Reorganization of Intermediate Filaments in Fiber Cells

Abstract

purpose. To define the remodeling of lens fiber cell intermediate filaments (IF) that occurs with both development and differentiation.

methods. Prenatal and postnatal mice were probed for the IF proteins phakosin, filensin, and vimentin, using light microscope immunocytochemical methodology.

results. The pattern of vimentin accumulation in elongating fiber cells changed with development. Early in development vimentin first emerged predominantly as focal accumulations in the basal region of both epithelial and primary fiber cells. A light diffuse cytoplasmic staining was also noted. Later in embryonic development, and through maturity, vimentin in fiber cells was predominantly associated with the plasma membrane with no anterior–posterior polarity. Phakosin and filensin were first detected in the very latest stages of primary fiber elongation and continued to accumulate well after cells had completed elongation. Initially, these proteins accumulated in the anterior half of the fiber cells and were cytoplasmic in distribution. After P13, the pattern of initial distribution in differentiating fiber cells changed to a predominantly plasma membrane localization. Neither beaded filament protein showed focal basal accumulations. In mature lenses, all three proteins ultimately disappeared from the nuclear fiber cells.

conclusions. Beaded filament protein accumulation lags significantly behind both primary and secondary fiber cell elongation, suggesting a functional role subsequent to elongation. The subcellular distribution of vimentin and the beaded filament proteins showed marked differences within the cell, with differentiation, and with development. The differences in time of initial synthesis and in distribution of these IF proteins may bear on hypotheses about the role of IFs in fiber cell elongation and in structural–functional polarity of the fiber cell.

Intermediate filaments (IFs) are cytoskeletal structures found in essentially all cells of higher organisms. In humans, the large IF gene family consists of several dozen distinct but related members, organized into five types or classes on the basis of several features. ^{1 2 3} IF protein expression is highly regulated, with each tissue type expressing only one, or a small number of IF proteins at any given time. All epithelia, for example, assemble IFs from heteropolymeric IF dimers composed of a type I and a type II IF protein (cytokeratins). ⁴ However, which pair of cytokeratins is expressed is both tissue and differentiation specific. Cells of the corneal epithelium, for example, express the K3/K12 pair of cytokeratins, whereas basal cells of the epidermal epithelium express K5/K14. As cells of a stratified epithelium differentiate, there is further modulation of IF gene expression. In epidermis, the expression of K5/K14 in the basal cells switches to K1/K10 as the cells differentiate and move to suprabasal positions. ⁴ Cells of mesenchymal origin, in contrast, generally assemble homopolymeric filaments from the type III class of IF proteins, which includes vimentin, desmin, and glial fibrillary acidic protein (GFAP). The lens expresses at least four successive generations of IF proteins during its development (discussed later). Each successive change in IF presumably meets a specific functional need of the lens, but their roles remain unknown. 

The functions ascribed to IFs are many. ⁴ However, these functions are often deduced only by inference. Epidermis represents one of the most intensively studied model systems of IF function. A wide array of data, including human genetic disorders, transgenic models, and in vitro assembly studies, indicate that IFs are critical to the stabilization of the differentiated cell phenotype. Mutations of epidermal cytokeratins lead to cell fragility and skin-blistering phenotypes. ^{5 6 7 8 9 10} Transgenic mice that express a dominant negative neuronal IF protein have difficulty in maintaining axonal diameter, resulting in nervous system disorders. ^{11 12 13} IFs have also been shown to interact with a variety of cellular structures, including cell junctions, organelles, and other cytoskeletal structures. ⁴  

The ocular lens of vertebrates is derived from surface ectoderm. This developmental process begins with a thickening of the ectoderm to form a lens placode. The placode invaginates to form the lens pit, which deepens and pinches off from the ectoderm to form the lens vesicle—a hollow, epithelial-walled sphere. The primary lens fibers are formed as the posterior epithelial cells of the lens vesicle elongate anteriorly, eliminating the lumen of the lens vesicle as they reach the epithelium. Secondary lens fibers are formed as epithelial cells at the lens equator undergo differentiation and elongate, ultimately contributing a new layer of fiber cells to the surface of the lens fiber mass. The lens grows in diameter as successive generations of secondary fibers differentiate and are added as new layers to the lens fiber mass. ^{14 15 16}  

Differentiation of fiber cells from lens epithelial cells includes not only major structural changes, but also extensive changes in gene expression. These changes include modifications in IF protein expression. At the onset of placode formation and invagination, the cells of the lens anlage express cytokeratins. ¹⁷ Soon thereafter, vimentin and GFAP are detected in the lens. ^{18 19 20 21 22} Ultimately, vimentin is expressed throughout the epithelium and elongating fiber cells of the adult mammalian lens. It is noteworthy that a targeted deletion of the vimentin gene in mouse does not result in an obvious phenotypic alteration of the lens, whereas overexpression of vimentin results in cataract. ²³ This establishes that vimentin is not required to either attain or maintain the highly differentiated architecture of the fiber cells, but that correct regulation of vimentin expression is essential to optical clarity. 

A fourth type of IF, the beaded filament (BF) is also expressed in the lens. The BF is composed of two proteins, phakosin and filensin, which are highly divergent members of the IF family. The BF also includesα -crystallin. BF protein expression appears to be unique to the lens. ^{24 25 26 27 28 29 30 31} The BF proteins themselves are unique among the many IF proteins in several ways, most notably in that they assemble into a structure (the BF) that is distinct from the classic cytoplasmic IF. The role of the BF is unknown, but two recent reports have implicated point mutations in human phakosin as the cause of at least two forms of human autosomal dominant congenital cataract, indicating that normal BFs are critical to optical clarity. ^{32 33}  

To explore the potential role of the BF in the elongation of both primary and secondary fiber cells we sought to define the developmental appearance of both BF proteins in a mammalian model system and to contrast this with the developmental appearance of vimentin. In addition to defining the temporal relationship between BF protein appearance and cell elongation, this study defines the normal progression of BF expression in the mouse model, through both development and differentiation. This information will be essential in evaluation of the impact of mutations in human autosomal dominant cataract, particularly the explanation for late onset of this cataract, as well as phenotypic changes in genetically engineered mice. 

Methods

Mice

Pregnant CD1 mice were received from Charles River Laboratories (Wilmington, MA). The presence of a vaginal plug in the morning defined day 1 of gestation. Embryos were collected on days 12 through 18 of gestation. Eyes from postpartum animals were collected at 0, 1, 2, 7, and 13 days of age. After excision, tissues were placed immediately into cold 4% paraformaldehyde in 0.1 M phosphate buffer overnight at 4°C. Selected tissues were embedded in paraffin, using routine methods. ^{34 35}  

Serial sections were cut at a nominal setting of 6 μm, collected on Superfrost/plus glass slides (Fisher Scientific, Fairlawn, NJ), and air dried. Before immunostaining, sections were deparaffinized with xylene and rehydrated by passage through decreasing concentrations of ethanol to water. 

All procedures involving animals were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research (available at www.arvo.org/arvo/animalst.htm). 

Antigens

cDNA for murine phakosin was cloned into pT7-7 and expressed in Escherichia coli BL21 (DE3). Bovine filensin was cloned into pBAD TOPO TA (Invitrogen, San Diego, CA; www.invitrogen.com), and expressed in E. coli TOP10 One Shot (Invitrogen). Bacterial inclusion bodies were solubilized in 8 M urea and fractionated by a combination of gel filtration and ion-exchange chromatography. 

Antisera and Antibodies

Highly enriched and purified proteins (phakosin and filensin) were used to immunize New Zealand White rabbits. Mouse monoclonal antibodies (mAbs) against bovine vimentin (clone Vim 3B4) were purchased from Chemicon (Temecula, CA; www.chemicon.com). Biotinylated secondary antibodies and streptavidin-peroxidase were purchased from Zymed (South San Francisco, CA; www.zymed.com). 

Western Blot Analysis

Antisera were characterized by Western blot analysis. Whole unfractionated lenses from 4-week-old mice were solubilized in sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer and resolved by SDS-PAGE, using 12.5% gels. Proteins were transferred to membranes (Immobilon P; Fisher Scientific) and blocked with 5% normal goat serum and 1% nonfat powdered milk in Tris-buffered saline with 0.1% Tween 20. Antisera were used at 1:4000 dilution for 4 hours, washed, and probed with alkaline phosphatase conjugated to goat anti-rabbit antibody (Cappel, Burlingame, CA) diluted 1:3000. To verify specificity, primary antisera were substituted with preimmune sera from the respective rabbits, or, for the mouse mAb, an irrelevant mAb of the same class. Identical Western blot analyses were performed with antisera preadsorbed with crude inclusion body preparations from bacterial lysates expressing murine phakosin and bovine filensin. 

Immunocytochemistry

Deparaffinized sections were exposed to 1.0 mg/ml pepsin (Sigma, St. Louis, MO) in 2.8% glacial acetic acid (pH 2.5) for 20 minutes at room temperature. Endogenous peroxidase activity was abolished by incubation in 3% H₂O₂ in methanol, for 15 minutes. Sections were blocked with 10% normal goat serum (polyclonal antibodies) or with reagents (HistoMouse; Zymed) for mAbs. Rabbit antisera to phakosin was used at 1:250 dilution, antiserum to filensin at 1:200 dilution, and the mAb to vimentin was used at 10μ g/ml for 120 minutes at room temperature. Negative controls consisted of preimmune rabbit sera, or, for vimentin labeling, the same class of mouse mAb (IgG_2a) at identical concentrations. Primary antibodies were localized with biotinylated secondary antibody (goat anti-rabbit or goat anti-mouse, as appropriate) then horseradish peroxidase conjugated to streptavidin. Color development was performed for 3 minutes with 0.8 mg/ml diaminobenzidine in 0.05 M tris buffer (pH 7.6) and 0.0008% H₂O₂. Tissues were counterstained with hematoxylin. 

Photography

Sections were coverslipped and photographed in bright-field conditions (E800 camera; Nikon, Tokyo, Japan). Images were recorded digitally and printed (PhotoShop image management software; Adobe, San Jose, CA; 932c printer, Hewlett Packard, Palo Alto, CA). 

Results

Western Blot Characterization of Antibodies

Antisera were raised against highly enriched recombinant murine phakosin (Fig. 1 , lane 3, serum IM26C-899) and recombinant bovine filensin (Fig. 1 , lane 5, serum URN0T-76). These were assessed for specificity by Western blot analysis. To ensure that all possible immunoreactivity was identified, whole unfractionated mouse lenses were solubilized in SDS-PAGE lysis buffer, resolved by SDS-PAGE, and stained with Coomassie blue (Fig. 1 , lanes 2 and 4). Filensin, vimentin, and phakosin, indicated by arrows in lane 2, represented a relatively small percentage of the total mouse lens protein. Bovine filensin (lane 5) differed from the murine homologue in having a repeat in the tail domain, giving it a higher mass and slower mobility on SDS-PAGE than the mouse homologue, indicated by the arrow in lane 2. Whole lens lysate, identical with that in lane 2 but at one-tenth the total amount, was used for characterization of the antisera by Western blot analysis. Antisera to phakosin (Fig. 1 , lane 6) and filensin (Fig. 1 , lane 7) both labeled multiple bands. This multiplicity of bands has been demonstrated in several previous reports and derives from the proteolytic breakdown products that accumulate in lens fiber cells as they age and lose their capacity for repair and renewal. The relationship between lower molecular weight bands and the parent molecule has been established by application of peptide-specific antisera and multiple mAbs. ^{20 24 25 36 37 38 39} Because these were newly generated antisera, the specificity was verified by creating identical blots and probing with antisera that had been preadsorbed with purified inclusion bodies from bacteria expressing the phakosin and filensin. Preadsorption of antisera with inclusion body preparations containing their cognate antigen eliminated labeling (not shown). Preadsoprtion of the phakosin antiserum with filensin inclusion bodies, and the filensin antiserum with phakosin inclusion bodies did not alter labeling, demonstrating that the preadsorption was specific for the expressed proteins and not a result of immunoreactivity to bacterial antigens. Further, this analysis established that this strain of mouse expressed full-length phakosin and filensin and did not harbor mutations that result in substantially truncated proteins, with possible impact on distribution within the lens. 

Immunocytochemistry

Vimentin.

Immunocytochemical localization of vimentin at various times in development is shown in Figure 2 . At embryonic day (E)12 the lens vesicle lumen was still present, and primary fiber elongation was under way. Labeling (brown-black product) could be seen in both the epithelium and elongating primary fiber cells. Most of the labeling was present as well-defined, intensely labeled foci in the basal end of both epithelial and fiber cells. This has not been previously described. These foci were largest, and thus most obvious, in cells that were actively elongating (Fig. 2B , arrow). There was some suggestion of diffuse cytoplasmic labeling of the fiber cells, but it was very near the limits of detection and much lower than the levels seen in subsequent stages. At E14 (Fig. 2C) , vimentin reactivity was very evident at the periphery of the lens, because of the presence of these intensely labeled basal accumulations. At higher magnification, the nature of both the epithelial (Fig. 2D) and elongating secondary fiber cells (Fig. 2E) labeling could be seen. The size of the basal accumulations reached a maximum in cells that were actively elongating, then dissipated as cells matured (Fig. 2E , cells at lower right). Again, a diffuse, but now more obvious level of cytoplasmic labeling was present. 

The subcellular distribution of vimentin at this stage was dissimilar to that reported previously in mature lenses, ^{19 39} in two ways: Actively elongating cells showed the basal focal accumulations, and in the youngest animals, vimentin first appeared to be cytoplasmic in the fiber cells. This contrasted with previous reports that described vimentin reactivity as predominantly peripheral or membrane associated. Puzzled by this, we extended the study to include later embryonic and early postnatal time points. 

By E16 (Fig. 2F) , the pattern of fiber cell labeling changed dramatically, becoming clearly associated with the plasma membrane. Figure 2G is a higher magnification view of 2F, showing longitudinally cut fiber cells. Figure 2H is from a coronally sectioned lens, showing fiber cells in cross section, which more clearly reveals the peripheral localization of the labeling. 

In cross sections (Fig. 2H) this labeling often suggested a nonrandom distribution, as though concentrated at the corners of this roughly hexagonal cell, similar to the pattern described by Lo et al. ⁴⁰ for actin bundles. It is noteworthy that primary fiber elongation occurs with little or no vimentin at the plasma membrane, whereas secondary fiber elongation is characterized by pronounced vimentin presence at the membrane throughout the elongation process. 

In older lenses (Figs. 2I 2J) , the observed pattern of vimentin labeling was identical with that reported previously. ^{19 39} Vimentin was present in both epithelial cells and fiber cells but most intensely in the epithelium. Fiber cell vimentin distribution was principally membranous, at least in the youngest fiber cells (Fig. 2I) . Examination of Figure 2I shows that the predominant membrane labeling of the lens fiber cell faded gradually with maturation into cytoplasmic labeling, then disappeared entirely in the lens nuclear cells, evident at the right hand edge of Figure 2I and in a lower power overview in Figure 2J . A similar pattern can be seen in higher magnification views (Fig. 2K) of the P13 lens shown in Figure 2J . The final disappearance of vimentin immunoreactivity in the lens nucleus is consistent with previous reports that show that vimentin degradation occurs in cells that have lost the capacity for transcription and thus for protein replacement. This zone is evident for the absence of immunoreactivity in Figures 2I and 2J . 

Phakosin and Filensin

BF proteins colocalized throughout these studies, and their localization is therefore described together. 

At E12, the initial elongation of primary fibers was well under way (Fig. 3A ). However, at this stage, BF protein was either not present or was present at undetectable levels. BF protein immunoreactivity emerged by E14 (Fig. 3B) and was pronounced by E17 (Fig. 3C) . It is evident from Figures 3B and 3C that BF proteins were not detected until the elongation process was quite advanced. Also, the initial emergence of BF protein did not occur as discreet, intensely labeled focal accumulations at the basal end of the cell but instead occurred as diffuse, cytoplasmic labeling in the anterior half of the fiber cell (Figs. 3B 3B inset, 3C) , a pattern distinct from that seen for vimentin. 

Examination of the bow region of an E16 lens (Fig. 3D) suggested that the labeling of the fiber cells was cytoplasmic and not membrane associated, as had been reported in studies on adult lens. Examination of these cells at higher magnification (Fig. 3E) and in cross section (Fig. 3F) shows unequivocally that the labeling was cytoplasmic. At no point was membrane-associated labeling seen at this stage of development. However, sometime between postnatal day (P)7 and P13 BF immunoreactivity ceased to first emerge as cytoplasmic labeling in fiber cells and instead first emerged as membrane-associated labeling. This membrane-associated pattern of labeling was clearly demonstrated in coronal sections of the adult lens (superficial fiber cells in Fig. 3H ), as well as in the P13 lens (superficial cells of Fig. 3I , and at higher magnification in Fig. 3J ). Note that the pattern of BF protein labeling in mature lens was similar to that reported for vimentin: label was predominantly associated with the plasma membrane, but there was a gradual accumulation of label in the cytoplasm before a relatively abrupt loss of signal occurred in the fiber cells that had lost their nuclei (3G, and right edge of 3H). 

In a coronal section of a P13 lens (Fig. 3I) , the transition between these two labeling patterns could be seen in the same lens: simultaneous presence of both membrane-associated labeling in recently formed fiber cells (superficial cells in Fig. 3I , and at higher magnification in 3J) and cytoplasmic labeling in deeper, older regions (3I, and higher magnification in 3K). In the mature lens (Fig. 3H) the pattern showed membrane labeling, with a gradual accumulation of cytoplasmic labeling, then the signal disappeared. At no point did the pattern in the mature lens resemble that seen in Figure 3F or in the deeper regions of Figure 3I , where there was intense cytoplasmic label and no membrane label. 

The loss of fiber cell nuclei coincides with a loss of immunoreactivity for BF proteins in the fiber cells (Fig. 3G and right edge of 3H). This presumably derived from a combination of a loss of transcriptional activity, combined with proteolysis of BF proteins, as has been described for bovine lenses. ³⁸ However, to assure that the absence of immunoreactivity in the central lens was not a function of a gradient of tissue fixation, we froze unfixed lenses and hemisected them in a cryostat. The hemisected lens was then immersed in fixative. This freshly exposed surface was thus uniformly fixed across its diameter, ensuring that both peripheral and central lens received identical exposure to fixative. This lens was then processed into paraffin, sectioned, and probed with antibodies. The pattern of antibody labeling was identical with that achieved by immersion fixation of intact lenses, confirming that the absence of immunoreactivity in the central lens was a function of physiologic antigen degradation. 

Discussion

The lens is remarkable in using multiple iterations of IF proteins during its development, expressing type I and II cytokeratins, ¹⁷ type III IF proteins (GFAP and vimentin), ^{19 21 22} and the lens-specific BF proteins. In the present study, we focused on IF proteins expressed in fiber cells, specifically with respect to the remarkable elongation shown by these cells as they differentiate and the role that BF proteins may play in this process. 

Immunocytochemical studies of mature lenses report that vimentin is found in the lens epithelium and maturing fiber cells. ^{19 39 41 42} These studies have shown that vimentin immunoreactivity is predominantly at the plasma membrane of nucleated fiber cells but also that it disappears from older cells that have lost their nuclei. ³⁹ The precise role of vimentin in the biology of the lens is not known, but deletion of the vimentin gene results in no reported changes in lens architecture or clarity, whereas cataract develops in transgenic mice that overexpress vimentin. ^{23 43} The localization we report on mature lenses is consistent with that reported in the literature. The first significant emergence of vimentin occurred in the form of intensely labeling basal foci, most evident in the elongating primary fibers. The pattern of this accumulation in cells at the very first stage of elongation suggests that it represents some form of cytoplasmic repository to be deployed for the elongation process. However, as noted, targeted deletion of the vimentin gene did not block this process, establishing that vimentin was not required for the elongation process. The ultrastructural basis of this focal accumulation is being investigated. 

The second IF network to emerge in fiber cells is composed of two IF proteins, phakosin and filensin, ^{28 29 44 45 46} which, together with α-crystallin, ⁴⁷ assemble into the BF, a form of IF unique to the lens fiber cell. ^{24 26 27 48} Western and Northern blot analyses and immunocytochemical data all indicate that the BF proteins are expressed only in the lens, and then only in the lens fiber cell (see, however, Ireland et al. ⁴⁹ ). 

The observations that BF proteins are found in the lenses of all vertebrate orders thus far examined, but only in the fiber cells, ³⁶ suggests that the BF plays a role critical to lens function. This hypothesis is supported by the recent demonstration that point mutations in the human phakosin gene have been implicated as the cause of at least two forms of human autosomal dominant congenital cataract. ^{32 33} However, the function of the BF remains unknown. 

To pursue further the role of the BF, we sought to determine whether there is a temporal relationship between BF protein accumulation and the processes of primary and secondary fiber cell elongation. We have shown that BF protein is either not present or is present at very low levels during both primary and secondary fiber cell elongation. It is also clear that BF immunoreactivity does not reach a maximum until after most of the fiber cell elongation is completed. If the principle function of the BF is in the elongation process, it would seem unlikely that their expression would be so low during this process, then continue to accumulate after the process was completed. Although circumstantial, the present evidence is consistent with the hypothesis that BFs, similar to other IFs, are not required for achievement of the highly differentiated phenotype, but may be required for the stabilization of that phenotype. This is consistent with the late onset of the cataract seen in the autosomal dominant congenital cataract patients ascribed to point mutations in phakosin. However, any hypothesis that proposes a stabilization function for the BF network would have to consider the polarized emergence of the BF proteins: Does the anterior half of the embryonic fiber cell have a more pressing need for stabilization early in development, or is the BF stabilization not of functional significance until well after birth, when the adult pattern of distribution is seen? Such asymmetry must also be considered for other hypothesized BF functions, such as in organelle regulation. 

It was also clear that the subcellular distribution and patterns of accumulation of the BF proteins were sharply different from that of vimentin and that this pattern changed markedly with development of the lens. In embryonic fiber cells BF proteins were absent from the epithelium, first emerged in the anterior end of the fiber cell well after elongation has started, and were cytoplasmic in distribution. Vimentin initially appeared in both the epithelium and fiber cell, as focal accumulations at the posterior end of the cell, before elongation began. At no point during embryonic and early postnatal development was BF immunoreactivity seen to be predominantly membrane associated, as reported for the adult lens. Vimentin, in contrast, is predominantly membrane associated throughout much of development, with the possible exception of a very light, diffuse cytoplasmic staining until approximately day E14. BF proteins underwent a dramatic change in the method of accumulation approximately 1 to 2 weeks after birth, whereas vimentin showed no such change. All these differences between vimentin and the BF proteins suggest that the cell biology of synthesis and assembly of these two networks is sharply different and that careful examination of the basis for these differences may yield insights to the distinct functions of these two networks. 

The change in BF proteins from cytoplasmic distribution to membrane-associated distribution also suggests a developmental switch in BF-associated proteins that may govern such a change and mediate linkage to the plasma membrane. 

Fig. 3I is a low-magnification view of a P13 lens. This image reveals that the newly emergent fiber cells near the surface (short arrow) showed the membrane-associated localization of BF proteins, whereas cells deeper in the lens (long arrow) showed cytoplasmic distribution. Between is a layer that is distinct from both in having a more diffuse and disordered cytoplasmic labeling. We hypothesize that this pattern represents a snapshot in time when cells that have used both modes of BF protein assembly are still present. That is, the deeper cells (longer arrow) were formed in the embryonic and early postnatal period, when assembly was purely cytoplasmic and never membranous. Because this is a younger lens, these deeper cells are still viable, and the BF proteins have not been lost to proteolysis. Because these deeper cells will eventually lose viability and because this mode of synthesis–assembly ceases at approximately the second postnatal week, they are not evident in mature lenses, as seen in Figure 3H . 

The most superficial cells in Figure 3I , in contrast, differentiated after the switch in assembly modes and clearly showed the membrane-associated assembly. The transition zone between where the membrane labeling was gradually lost, and the reactivity becomes cytoplasmic, probably represents the gradual transition of membrane labeling to cytoplasmic labeling that has been documented in mature lenses for both vimentin and the BF. This appears to be the endstage fate of IFs in aging fiber cells. Sandilands et al., ³⁹ for example, reported a shift in BF protein localization from membranous to cytoplasmic when comparing young to old fiber cells in mature lenses. This shift from membrane to cytoplasm that occurs as a cell differentiates and matures could have several possible explanations. It may reasonably represent nucleation of filament assembly at the plasma membrane in younger cells, with subsequent extension into the cytoplasm as the cell matures and filaments grow in length. Alternatively, this shift could also represent the early stages of proteolytic release of antigenic fragments. Filaments that may be functional at the plasma membrane undergo proteolysis as the cell ages, potentially releasing fragments from whatever constraint bound them to the plasma membrane. The documented degradation of vimentin and BF proteins that occurred in older cells of the lens nucleus is consistent with the latter scenario. This pattern was similar for both vimentin and the BF proteins (compare Figs. 2J and 3G ). More work is required to verify this and is in progress. 

We are confident, however, that the patterns that we have described are not the result of variations in the plane of section of fiber cells in the different layers. Membrane labeling, for example, can appear cytoplasmic if the membrane is parallel to the plane of section. The divergence of fiber cell paths as the cells approach the sutures can add further confusion, creating the artifactual suggestion of zonation in labeling patterns as the cells present in different orientations. We explored multiple lenses for each time point, examined both sagittal and coronal sections, and followed the pattern changes through serial sections to rule out plane of section as an explanation for the differences we report. 

Ireland et al. ⁴⁹ report that the chick homologues to the BF proteins can be detected slightly earlier in development than reported here for the mouse. Whether this is due to the relative levels of sensitivity between the two localization methods, to the much more rapid maturation that occurs in chick lens, or to the differences between chick and mouse lens structure is not clear. Notably, chick lenses are operational at hatching, whereas mouse lenses are not used until after the eyelids open, some 2 weeks after birth, or about the time that the shift in BF patterns of accumulation change. This suggests that functional maturation of the lens is incomplete even in the postnatal mouse, an observation that may also provide clues to the functions of these two IF networks. Chick lens also express a splice variant of the CP49, referred to as CP49ins. ⁵⁰ Whether this is relevant to the differences in expression patterns is not clear. 

Whatever the ultimate explanation, there is clearly a difference in BF protein cell biology between embryonic and postnatal fiber cells. This suggests that the fine tuning of IF function is not achieved solely by switches in gene expression, but also by modulation in processing of a given IF network. These data also suggest developmental regulation of proteins that mediate BF linkage to the plasma membrane or to other subcellular structures, as well. 

The evidence in the present study suggests that the BFs probably do not have a central role in the elongation process, because their accumulation lags substantially behind the elongation process and does not reach a maximum until the elongation process is completed. Further, these data also show pronounced differences in when, where, and how these two IF networks are assembled in the fiber cell. These differences may provide keys to identifying the separate functions of these two networks. Subsequent studies will explore the asymmetry in distribution to determine whether this yields further clues to the function of both networks in the biology of the fiber cell. 

 Supported by Grants RO1EY08747 (PGF) and P30EY12576 (University of California, Davis) from the National Eye Institute.

 Submitted for publication August 7, 2000; revised November 22, 2000; accepted November 30, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

SDS-PAGE and Western blot characterization of antisera. Lanes 1 through 5: Coomassie blue stained SDS-PAGE gel showing in lane 1, recombinant vimentin; lane 2, total mouse lens lysate; lane 3, recombinant mouse phakosin; lane 4, total mouse lens lysate; lane 5, recombinant bovine filensin. Lanes 6 and 7: Western blot analysis showing in lane 6, antiserum to phakosin; and lane 7 antiserum to filensin.

Figure 2.

 

Immunocytochemical localizations of vimentin. E, embryonic; P, postnatal. (A) Lens vesicle day E12. Dark brown-black label was evident at the periphery of the lens, in the basal regions of both the epithelium and primary fiber cells, but was most intense in the elongating primary fibers. (B) Higher magnification view of a region in (A). The reaction product was largely basal (arrow), but some indication of reaction could be seen along the length of the elongating fiber cells. (C) Day E14. Primary fiber elongation was completed, and secondary fiber elongation was in progress. Labeling was seen at the periphery of the lens. Light, diffuse cytoplasmic labeling of fiber cells was present, though difficult to display, but appeared as a slight brown cast. (D) Higher magnification view of lens epithelium from (C), showing the nature of the epithelial labeling. (E) Higher magnification view of (C), showing the nature of the labeling in the basal region of elongating fiber cells and how it dissipated in cells that were further progressed in the elongation process. Tracks of label were evident extending anteriorly along the fiber cells, but the pattern was not membrane associated. (F) Day E16. The intense nature of epithelial labeling was evident, and the emergence of fiber cell membrane labeling was well-established throughout the lens. (G) Higher magnification view of (F), showing fiber cells cut in longitudinal section and demonstrating the predominantly membrane-associated nature of the labeling at this stage. (H) Cross section of an E16 lens, showing the membrane-associated nature of the labeling, but also suggesting that the label was not uniform around the periphery of the fiber cell. The tendency for label to be more pronounced at the point of contact between three cells was evident in several places. (I) Mature lens (approximately 6 weeks after birth). The epithelium could be seen to be heavily labeled, whereas fiber cell labeling was principally membrane associated, at least in younger cells. In deeper fiber cells, there was a gradual accumulation of signal in the cell cytoplasm, resulting in both membrane and cytoplasmic labeling until the signal finally disappeared at about the same time that fiber cell nuclei disappeared. (J) Low-magnification view of P13 lens. Labeling of the lens epithelium was intense (left edge of the lens). The cells that were predominantly membrane labeled created a zone of slightly lighter apparent labeling. Loss of label in the deep cortex and nucleus was evident. (K) Higher magnification view of cortex of P13 lens, showing labeling pattern similar to that of adult.

Figure 3.

 

Immunocytochemical localization of BF proteins. (A) Lens vesicle, day E12. No indication of reaction product was seen. (B) E14 lens: Antibodies to filensin showed emergence of reactivity at the anterior end of primary and secondary fibers. Inset: Higher magnification view of the anterior ends of these cells, displaying the vesiculation that occurred during processing. (C) E17 lens: Signal intensity was greatly increased, showing anterior polarity and absence of label until elongation had progressed substantially. (D) Higher magnification view of bow region of an E16 lens, oriented with anterior to the left. The delayed onset of BF protein accumulation was evident, and a cytoplasmic distribution of signal was suggested. (E) Higher magnification view of longitudinally cut fiber cells from E16 lens. Label is clearly cytoplasmic. Compare with membrane labeling seen for vimentin, Figure 2G . (F) Cross-sectioned fiber cells from an E16 lens, showing dense cytoplasmic labeling, with an absence of stain concentrated at the cell periphery. Contrast with comparable vimentin labeling in Figure 2H . (G) Low-magnification overview of a day P13 lens. The bow region appeared negative, but at higher magnification would show membrane-associated labeling in the elongating cells. This pattern was very similar to that seen for vimentin (Fig. 2J) , except that the epithelium and early stages of elongating cells were unlabeled. (H) Adult lens, showing that BF protein label in older animals emerged first at the plasma membrane and then accumulated in the cytoplasm as the fiber cell ages. (I) P13 lens near the anterior pole. The membrane labeling of the younger cells was seen near the surface (shorter arrow), whereas the densely cytoplasmic labeling was seen in deeper cells (longer arrow). (J) Higher magnification view of superficial cortex from P13 lens, showing membrane-associated labeling. The plane of section shifts from mostly perpendicular to the long axis of fiber cells near the surface to distinctly oblique. (K) Higher magnification view of deeper cortex from day P13 lens, showing cytoplasmic labeling of fiber cells in cross section and the absence of a distinct membrane-labeling pattern.