Because the catenins of p120-catenin subfamily play a key role in cell adhesion but have not been investigated in the adult lens, the localization of p120-catenin, Arvcf and δ-catenin were examined. Immunolabeling histological cryosections of juvenile mouse lenses (P7) revealed that p120-catenin is strongly localized to the cell membranes of the youngest elongating lens fiber cells and the lens epithelium (Figures 1A–C). However, its expression appears greatly reduced in lens fiber cells located immediately adjacent to the outermost lens fiber cells (Figure 1D) and is almost undetectable among deeper fiber cells (Figure 1E). In contrast, δ-catenin is barely detectable in the lens epithelium and superficial lens fiber cells (Figures 1F–H). δ-catenin is strongly localized to the tricellular junctions of cortical lens fiber cells (Figures 1I,J). Relatively strong Arvcf signal is detected in the bicellular junctions of both lens epithelial cells and lens fiber cells (Figures 1K–O). Developmental expression of Arvcf is also detected as early as E11.5 throughout the lens vesicle cell membranes (Figure 1Q) and continues throughout development. Strong localization along the length of primary fiber cell lateral membranes at E12.5 (Figure 1R) and secondary fiber cells at E17.5 is also observed (Figure 1S).

Given the strong expression of Arvcf in lens fiber cells it was hypothesized that it may be important for lens development and/or transparency. To test this, a mouse line harboring a targeted allele of Arvcf (Arvcf ^{tm1e(EUCOMM)Wtsi} ) was acquired from a commercial repository and analyzed. The allele was designed such that it disrupts endogenous RNA splicing by the insertion of a lacZ/neomycin cassette that is flanked by a splice acceptor and termination sequence 5’ of the first coding exon common to all known transcripts (Supplementary Figure S1A). Homozygous mice with two copies of the allele were generated and are usually morphologically indistinguishable from heterozygous or wild-type littermates but are born at a reduced Mendelian frequency (60/340 or 17.6% vs. the Mendelian 25%). The prevalence of heterozygous (189/340 or 55.6 vs. 50%) and wild-type (91/340 or 26.8 vs. 25%) mice are both slightly increased from expectations. The exception to a normal appearance was that some homozygous mice (10/60 or 16.7%) were noticeably runted at frequencies greater than heterozygous (6/189 or 3.2%) and wild-type mice (2/91 or 2.2%). To determine if homozygous animals were dying embryonically due to major developmental disruptions, embryos were collected from timed matings at E15.5. However, all embryos observed appeared morphologically normal, were present in Mendelian ratios, and their eyes and lenses appeared to have relatively normal morphology (Supplementary Figures S1C–F). It is yet unclear why these animals are underrepresented at birth but it could be a result of matriphagy before pups are observed.

Cataracts were not observed in young adult mice, however, aged mice developed noticeable opacities (Figures 2A,B). To examine the progression of these opacities, lenses were extracted from mice at increasing ages and imaged with a dark background (Figures 2C–P). Lenses from control and homozygous mice were completely transparent at 2 months of age (Figures 2C,D) but by 5 months, opacities were observed that appeared similar to clinical observations of cortical spoking in cortical cataracts (Figures 2E,F; Supplementary Figures S2A,B, arrowheads). These opacities were not present in heterozygous or wild-type littermate control animals (Figures 2E,G,I,K,M,O), the outer most lens fiber cells (Figures 2F,H,J, asterisks), or initially within the lens nucleus. This phenotype was observed bilaterally in 100% of all Arvcf ^−/− animals older than 6 months (n = 11). Interestingly, the region of opacity at 5 months appeared more deeply at 6 and 8 months (Figures 2H,J). Assuming that the opaque regions remain opaque, this observation suggests that the newly differentiating cells added to the lens during growth remained transparent. However this expanding outer region eventually acquired cloudy areas (Figures 2J,L, arrows) and became opaque by 12 months of age (Figure 2N). The nucleus also appears translucent by 12 months and by 20 months the entire lens appears completely white, has a ruptured lens capsule, and appeared to lose volume (Figure 2P). Additionally, opacities in the anterior pole were observed that overlap with suture lines by 6 months that became more prominent at 8 months and sometimes appeared as a whorl pattern (Figures 2H,J,L, Supplementary Figure S2C). To determine if the refractive ability of the lens is disturbed with the loss of Arvcf, 6 month old lenses were extracted and a Helium-Neon laser was aimed at the lens parallel to the visual axis at varying equatorial intervals (Figures 2Q,R). A composite image compiled from several individual images allowed the observation that the focal length appears longer in Arvcf ^−/− lenses vs. Arvcf ^+/+ littermates (Figures 2Q,R, arrow). These focal length changes occur outside of the regions of opacity which tend to block the laser rather than refract its light path (Figure 2S). To quantify these changes, at least six lenses per genotype (Arvcf ^+/+ , Arvcf ^+/- , and Arvcf ^−/− ) from different animals were used in the laser assay. The equatorial position of laser penetration was grouped into four equidistant regions along the lens radius and the focal length measured among laser lines from each group (Figure 2T). The average focal length was found to be significantly different when comparing Arvcf ^+/+ and Arvcf ^−/− lenses only when the laser was aimed in the region between 50 and 75% of the lens radius (Figure 2U, Supplementary Figure S2F). To test whether Arvcf has any deficiencies in its mechanical properties, the dimensions of dissected lenses from control and mutant animals were measured following a mechanical load (Figures 22V–W). Upon successive placement of coverslips onto the anterior pole of the lens similar to previously published experiments (Cheng et al., 2016a), we observed that Arvcf deficient lenses from ∼60 day old animals surprisingly resisted deformation more than control animals. This was observed in both the axial and equatorial dimensions and was statistically significant when greater loads were placed upon the lenses (Figure 2X). Although the lenses appeared stiffer in the mutant animals, no significant changes in the ability of the lenses to return toward their original dimensions after the mechanical load was removed for 2 min (known as tissue resiliency) were observed (Figure 2Y). While not significantly different, the axial and equatorial dimensions of mutant lenses, on average, were closer to their preload values after removing the mechanical load (axial −3.6%; equatorial +1.2%) versus the control lenses (axial −4.8%; equatorial +2.1%). Together these data demonstrate that the loss in transparency and refractive changes is preceded by alterations to the biomechanical properties of the lens. The stiffening of tissue may be a reflection of cellular changes that occur before lens opacities in cortical cataracts.

To ascertain what cellular abnormalities may underlie these opacities, histological staining of 1 month old lenses was performed. Wheat germ agglutinin (WGA) labeling revealed that the outermost, superficial lens fiber cells of Arvcf ^−/− animals are similar in size and periodicity to cells from Arvcf ^+/+ animals and are arranged in a hexagonal lattice (Figures 3A,B). However, breaks along radial fiber cell junctions are often observed in Arvcf ^−/− lenses (Figure 3B, arrowheads). Deeper cortical and nuclear lens fiber cells showed significantly more disruptions to the hexagonal lattice and shape of cells (Figures 3C–F). Large spaces between cell membranes are often observed in Arvcf ^−/− but not Arvcf ^+/+ lenses that appear to be irregularly sized fiber cells. It is unclear whether these are due to fusion events of two or more cells or are abnormally swollen, but they do appear to be cellular and not intracellular space or histological artifacts as cytoplasmic αB-crystallin labeling is localized to these regions (Supplementary Figure S3A). Strikingly, these lenses already have signs of cellular disorganization at a time point months before cortical opacities are readily observed suggesting that these disruptions in cellular architecture are not initially a source of opacity.

The cellular disruptions and opacities are likely due to the absence of any detectable Arvcf expression in these animals. Both RT-PCR and western blotting assays were unable to detect expression of Arvcf RNA and protein from Arvcf homozygous mutant whole eyes, lenses, and brain tissue of 21 day old mice (Figures 3G,H, Supplementary Figures S3B–D). In addition, Arvcf was not detected following immunofluorescent labeling of histological sections of lenses from Arvcf ^−/− animals (Figures 3I–M). Arvcf antibody labeling is strong along both the shorter bicellular junctions, orientated along the equatorial radius (radial), and longer bicellular junctions, orientated parallel with the equator (circumferential), in Arvcf ^+/+ lens fiber cells (Figure 3K). Localization of detectable protein to both junctions is reduced in Arvcf ^+/- lens fiber cell junctions (Figure 3L) and completely absent from those of Arvcf ^−/− lens fiber cells (Figure 3M). Protein expression also appears absent from Arvcf ^−/− embryonic lenses during development (Supplementary Figures S1G–I). Measuring the immunofluorescent intensity along the radial and circumferential fiber cell junctions from Arvcf ^+/+ Arvcf ^+/- Arvcf ^−/− lenses revealed significant decreases in junction localization corresponding to gene dosage (Figure 3N).

Because of the potential for compensatory increased expression of other p120-catenin subfamily members in the absence of Arvcf, the expression of δ-catenin and p120-catenin were examined by immunofluroescent labeling mutant lenses. Within 4-week old mice, the localization of δ-catenin appears unaffected by the loss of Arvcf and maintains its tricellular localization pattern (Figures 4A,B). However, the localization of p120-catenin in both the radial and circumferential bicellular junctions appeared more intense (Figures 4C,D). This is not due to differential labeling between experiments and was observed in multiple lenses (Supplementary Figure S3E, n = 4). Upon measuring their intensities, the tricellular junction intensity of δ-catenin does not significantly vary between Arvcf ^+/+ vs. Arvcf ^−/− (Figure 4E) but a significant increase in junctional p120-catenin was observed (Figure 4F). These results suggest that compensatory, feedback mechanisms may exist to regulate p120-catenin expression and/or localization but not δ-catenin in response to loss of Arvcf function.

The p120-catenin subfamily of catenins are known to bind to the juxtamembrane domain of cadherins and thought to increase the stability of adherens junctions. Histological cryosections were prepared from 4-week old animals to visualize whether a reduction in membrane localization of identified proteins in lens fiber cells is observed in the absence of Arvcf (Figure 4). N-cadherin localization to radial and circumferential junctions appears reduced, particularly along the longer, circumferential junctions (Figures 4G,H). Quantification of immunofluorescent intensity confirmed this observation (Figure 4I). Interestingly, the tricellular localization of N-cadherin appears retained (Figure 4H), possibly because δ-catenin functions similarly to Arvcf and is still present in Arvcf ^−/− lenses. Likewise, β-catenin (Figures 4J–L) and α-N-catenin (Figures 4M–O) are also reduced in the membranes of Arvcf ^−/− lens fiber cells to a significant degree and appear retained in the tricellular junctions. This result suggests their association with the adherens junctional complex in lens fiber cells is at least partially dependent on Arvcf.

The p120-catenin subfamily of catenins strengthen the cadherin complex through facilitating the clustering of cadherin monomers into structures called cadherin nanoclusters (Yap et al., 1998; Ishiyama et al., 2010; Yap et al., 2015). These nanoclusters have historically been referred to as spotty or punctate adherens junctions (pAJs) and are observed in lens fiber cells with electron microscopy (Lo, 1988; Niessen and Gottardi, 2008). Recent super-resolution, light microscopy studies have demonstrated that cadherin nanoclusters vary in size and range in diameter from approximately 50–60 nm at the smallest organizational size to as large as 1–2 microns (Yap et al., 2015). Utilizing high-resolution microscopy (Airyscan) to image the broader, circumferentially orientated junctions following immunolabeling of cortical lens fiber cells, demonstrated that Arvcf signal appears as puncta that are often in close association with N-cadherin puncta and appear approximately ∼150 nm in size (Figures 5A,B, white arrows). This supports a role for Arvcf in the stabilization of these nanocluster structures. To determine if these structures are affected by the loss of Arvcf, whole fiber cells dissected from the cortical region of the lens from 1 month old mice were imaged following immunofluorescent labeling with N-cadherin and β-catenin antibodies (Figures 5C–J). Upon focusing on a plane parallel to the bicellular junctions of control lens fiber cells, N-cadherin and β-catenin labeling appeared as discrete puncta along the entire lateral membrane surface (Figures 5C,G). The puncta appear reduced in intensity and are more diffuse in appearance when observed at the level of the cytoplasm where junctions are not present (Figures 5D,I), which may be indicative of endocytosed (N-cadherin) or adherens junction dissociated (β-catenin) protein. The contrast between the immunolabeling of junctional and cytoplasmic regions is also well defined in the z-stack projections (Figure C′ and D’). When imaging Arvcf-deficient lens fiber cells the difference between the bicellular junctional plane and cytoplasmic planes are difficult to discern from each other in either the x-y or z projection planes possibly because of destabilized cadherin leads to more cadherin endocytosis and accumulation of N-cadherin/β-catenin within the cytoplasm (Figures 5E,F,H,J). The size of the bicellular junctional puncta in Arvcf-deficient lens fiber cells also appear smaller (Figure 5E vs. Figure 5F). Quantifying the size and density of the junctional puncta of lens fiber cells from each phenotype revealed that while the density of individual puncta is not significantly different (Figures 5K,N), the area of individual N-cadherin and β-catenin puncta (Figures 5M,P) and the percentage of the total area occupied by either N-cadherin (Figure 5L) or β-catenin (Figure 5O) are significantly reduced. Utilizing the average area of puncta and assuming that the puncta are roughly circular, the calculated diameter of N-cadherin puncta are approximately ∼155 nm in diameter in control cells and are reduced to ∼124 nm in the absence of Arvcf. These data suggest that Arvcf is required to maintain the size of cadherin nanocluster complexes of lens fiber cell bicellular membranes and may be indicative of a reduction in cell adhesion.

In addition to their localization to bicellular junctions, high-resolution imaging also revealed that Arvcf and N-cadherin are strongly localized to the base of and along the lateral sides of the interlocking protrusions (Figures 6A,B). Interestingly, the observed signal appears largely absent from the broad sides of the interlocking protrusions (Figure 6B, asterisks) suggesting that cadherin-based junctional complexes are formed solely along the protrusion margins. Strikingly, the normally intense and continuous tricellular junctional signal of N-cadherin and β-catenin near the protrusion bases of cortical lens fiber cells (Figure 6C) appears in an intermittent pattern in the absence of Arvcf. (Figure 6D, arrowheads). Furthermore, N-cadherin and β-catenin are rarely observed at the margins of interlocking protrusions and no longer appear to delineate their margins (Figure 6C vs. Figure 6D). Because this observation could be due to the absence of protrusions, scanning electron microscopy was performed on 1 month-old lenses from control and Arvcf deficient mice. While the Arvcf mutant lenses have an abundance of interlocking protrusions that is similar to control lenses in the absence of Arvcf (Figure 6E vs. Figure 6F), they are often misshapen and elongated in several regions of varying depths (Figures 6G,I,K vs. Figures 6H,J,L and Supplementary Figures S4A,B). Furthermore, well-defined paddle structures are difficult to distinguish unlike in control lenses (Figures 6E,K,M, asterisks). Older, cataractous lenses display an exacerbated interlocking protrusion phenotype and possess extremely long interlocking protrusions that curl up within the spaces between fiber cells (Figures 6M–O vs. Figures 6N–P). Therefore, the inability of N-cadherin/β-catenin to define the protrusion boundaries is not due to their absence but instead due to their extreme mislocalization within them. To characterize this mislocalization, super-resolution images of individual protrusions immunofluorescently labeled with N-cadherin or β-catenin were collected and processed for comparison between genotypes (Figures 7A,B,G,H). Because individual protrusions from Arvcf deficient lens fiber cells are difficult to identify from N-cadherin or β-catenin localization alone (Figures 7B,H), co-labeling with Aquaporin-0 was utilized to locate them (Supplementary Figure S4C). The fluorescent intensity of every pixel from 10 to 16 aligned, morphologically similar individual interlocking protrusion images was quantified and compared with the pixels at the same position of other images (Supplementary Figure S4C). The average pixel intensity at each position was quantified and used to generate an “average image” for each genotype (Figures 7C,D,I,J). When the fold change at each pixel was determined and represented as a heat map image it indicated that changes were greatest along the borders of the protrusions (Figures 7E,K). Statistically significant differences were also represented as a heat map and found at the protrusion apices (Figures 7F,L) and along the lateral boarder (Figure 7L). Differences in fluorescent intensity are also visualized with more conventional line graphs (Figures 7M–P). Together these data demonstrate that N-cadherin and β-catenin require Arvcf to localize to the margins of interlocking protrusions.

The development of age-related cortical cataracts and loss of cadherin-associated proteins in fiber cell membranes observed in the absence of Arvcf is a significant finding because it demonstrates that this type of cataract can be caused by a reduction in adherens junction mediated adhesion. This information has not yet come to light likely because loss of function analyses in mice focused on proteins with adhesive function have severe lens defects or die prematurely. Conditional ablation of N-cadherin, the dominant cadherin in lens fiber cells, disrupts secondary lens fiber cell elongation leading to a severely disorganized lens (Pontoriero et al., 2009; Logan CM. et al., 2017; Logan C. M. et al., 2017). Lens-specific removal of β-catenin similarly causes significant disruptions to lens fiber cells during development (Cain et al., 2008). It is possible that lens specific removal of other genes encoding cadherin-complex proteins expressed in the lens fibers such as α-N-catenin, p120-catenin, δ-catenin and/or afadin may also cause age-related cortical cataracts however these experiments have yet to be reported (Jun et al., 2012; Lachke et al., 2012; Lang et al., 2014; Maddala and Rao, 2017). Aquaporin-0 and Connexin 50 deficient mice also have lens fiber cell adhesive function, but the cataracts that occur in these models are nuclear, present at an early age, and are likely related to ion and water transport functions of the proteins (White et al., 1998; Shiels et al., 2001; Kumari et al., 2013; Hu et al., 2017; Gu et al., 2019; Varadaraj and Kumari, 2019). Thus, the Arvcf mutant model is ideal for determining what occurs downstream of cell adhesion failure to cause opacities and will be the subject of future investigation.

Arvcf does not appear to be a required component of adherens junctions during secondary lens fiber cell differentiation at the embryonic or even at young adult stages due to the absence of an embryonic phenotype and transparency of the lens within superficial fiber cells. This could be due to genetic compensation from p120-catenin, which is the paralog with the greatest sequence similarity to Arvcf and is expressed in primary and secondary lens fiber cells during development (data not shown) (Pieters et al., 2012). Because p120-catenin and Arvcf function similarly, p120-catenin is likely sufficient for N-cadherin retention within adherens junctions during development. The upregulation of p120-catenin in fiber cells in the absence of Arvcf may also suggest why opacity occurs in a specific region (∼100–150 microns from the surface). The outermost cells where p120-catenin is highest may have sufficient stabilization of the cadherin complex to maintain adhesive contact and transparency. Because the p120-catenin upregulation appears restricted to only the cortical region of the lens, it does not explain why the central lens remains transparent. The development of opacities may be related to age-related changes in lens fiber cell mechanics where interfaces between the stiffer, more central regions and comparatively softer lens cortex may be subject natural breaks or slippage between regions as has been previously proposed (Fisher, 1973; Pau, 2006; Michael et al., 2008). The hydrostatic pressure gradient between central and more peripheral lens fiber cells may also contribute to mechanical environment differences and influence where opacity occurs (Gao et al., 2011).

The use of super-resolution fluorescent microscopy has also enabled this study to probe the consequences of Arvcf deficiency on pAJ organization in the junctions juvenile lens fiber cells. With this technique we observed that the N-cadherin/Arvcf-residing adherens junctions are organized in clusters that are on the order of ∼150 nm, making them smaller than and distinct from larger microclusters found in places such as the apical adherens junctional complex (Yap et al., 2015). Because Airyscan imaging has a lower end resolution capability of ∼140 nm, it is possible that these structures are a concentration of smaller order nanoclusters which can be as small as 50–60 nm in size (Quang et al., 2013; Huff, 2015; Wu et al., 2015). Thus, the smaller structures we observe could be due to fewer nanocluster subunits that fail to congregate together or that they are simply smaller whole nanocluster subunits that consist of fewer N-cadherin/β-catenin molecules. Nanoclusters of cadherin molecules can form in the absence of adhesion and it remains possible that some of these structures we observe are non-adherent (Wu et al., 2015). However transmission electron microscopy (TEM) has revealed that the spotty/pAJ of lens fiber cells from several species appear to be approximately ∼100–200 nm in size suggesting that these structures are indeed adherent (Lo, 1988; Lo et al., 1997). These data also support the possibility that the smaller nanoclusters we observe are likely due to smaller adhesion sites rather than fewer smaller nanoclusters. While the absence of Arvcf appears to diminish the size of these nanoclusters it is important to note that they are still present. Given that the p120-catenin can stabilize cadherins by binding to the juxtamembrane domain and masking the endocytosis signal, it is unclear why these cadherin nanoclusters persist in the absence of homologous Arvcf (Xiao et al., 2005; Miyashita and Ozawa, 2007). It is possible that Arvcf is necessary for some but not all N-cadherin retention and that additional proteins that have previously been implicated in protecting N-cadherin from endocytosis, such as β-catenin or NMDAR, can serve this role (Tai et al., 2007; Chen and Tai, 2017).

Another consequence of Arvcf deficiency is the disruption of N-cadherin and β-catenin localization within interlocking protrusions of lens fiber cells. Along with distortions of their shape and the abundance of Arvcf protein normally observed suggest that Arvcf stabilization of adherens junctions is integral to their function. These protrusions likely prevent fiber cells from sliding past each other via their interlocking nature but also likely provide stability through the creation of a large amount of surface area for adherens junction protein complexes to reside as well. Our observations of N-cadherin within the lateral domains of interlocking protrusions stands in contrast to previously published data where it appeared absent (Cheng et al., 2016b). However, we observe a similar absence when individual lens fiber cells have been pulled from their neighbors and imaged individually (Supplementary Figure S4) and it is likely that the cadherin complexes can be pulled out of the membrane when dissected in this manner. While this occurs most of the time, we have observed paired N-cadherin localization along the protrusion membrane on occasion when fiber cells become separated from each other (Supplementary Figure S4). Furthermore, pAJ-like structures have also been observed via TEM between interlocking protrusions and the pocket of neighboring cells supporting the presence of N-cadherin (and Arvcf) containing adherens junctional complexes in protrusion membranes (Biswas et al., 2010; Biswas et al., 2016). Adherens junction complex formation within these structures may also be important for stabilizing the membrane to allow optimal function for aquaporin-mediated water transport. Because water/ion movements are extremely important for lens physiological mechanisms that protect it from oxidative damage and preserve transparency it is possible that Arvcf deficiency may result in cortical cataracts via a disruption to these protective mechanisms.

The lack of adhesion complexes within interlocking protrusions may also be the reason for the disrupted organization and shape of deeper cortical lens fiber cells. The cellular separations we observed along the radial junctions may be easily explained as a direct result of reduced adhesion and therefore could be more prone to separation. The radial regions may be particularly sensitive to this because it is where the interlocking protrusions are concentrated. The abnormal elongation of interlocking protrusions could also be a direct effect of reduced adhesion. These lens structures have been morphologically and molecularly compared to dendritic spines of neurons (Frederikse et al., 2012) Reduced N-cadherin or α-N-catenin function cause spine elongation due to reduced adhesion (Togashi et al., 2002; Abe et al., 2004; Xie et al., 2008). Therefore loss of Arvcf in lens fiber cells may mimic what occurs in the absence of adhesion in neurons and cause remodeling of the protrusion structures. Part of this mechanisms may include the ability of members of the p120-catenin subfamily to regulate Rho-GTPases and actin dynamics in many contexts including during dendritic spine morphogenesis (Elia et al., 2006; Arikkath et al., 2009). Elongation of interlocking protrusions in lens fiber cells could therefore be due to an increase in actin polymerization that is dependent or independent of reduced adhesion. Interestingly, these elongation phenotypes, as well as the presence of swollen fiber cells are similar to what is observed in Aquaporin-0 knock out mice (Lo et al., 2014). The shared phenotypes could be due to Aquaporin-0’s adhesive or water transport functions and determining whether Aquaporin-0 function is related to Arvcf is a worthy avenue of investigation.

Although adhesion is likely reduced between lens fiber cells of Arvcf deficient lenses, it is somewhat surprising that they are stiffer. It is reasonable to speculate that if fiber cells have a reduction in the protein complexes that adhere cells to one another then the tissue may be predisposed to deform more. However, the increased stiffness observed here in lenses from 60-day old mice demonstrates that Arvcf-dependent adhesion alone is not a major contributor to lens tissue stiffness at this age. Similar to our results, lenses from 2.5 months- to 4 month-old aquaporin-0 deficient mice have an increased capacity to resist deformation (Kumari et al., 2015; Gu et al., 2019). It is possible that the increase in stiffness for both models indicate a shared mechanism, especially given the observations from this study and others (Logan C. M. et al., 2017; Maddala and Rao, 2017). Alternatively, increased stiffness in Arvcf deficient lenses may also be related to the Rho-GTPase regulatory function of the p120-catenin subfamily (Fang et al., 2004; Anastasiadis, 2007). Loss of Arvcf may cause disrupt Rho-GTPase-dependent actin-cytoskeleton regulation and cause a stabilizing network of F-actin to polymerize. Intermediate filament and tropomodulin knock-out mice have lenses with decreased stiffness (Fudge et al., 2011; Gokhin et al., 2012) making it plausible that altering the actin cytoskeleton may result in tissue stiffening. Identifying the underlying mechanisms behind why the biomechanics of the lens is altered in this mouse model may ultimately facilitate an understanding of the etiology of age-related cortical cataracts.

Together, these data demonstrate that Arvcf is required to maintain transparency with age through the stabilization of the N-cadherin containing adherens junctions between lens fiber cells. By serving as a model of age-dependent cortical cataracts, further analysis of Arvcf-deficient mice has the potential to provide new information on the development of cortical cataracts with age. Furthermore, this study may indicate that the stability of the adherens junctional complex could be a biomarker of the aging process that predisposes it for cataract formation. It is speculated that loss/reduction of adherens junctions between fiber cells may induce opacities by disrupting ion and water transport, the architecture of the cytoskeleton, and/or the biomechanical properties of the lens.

The Arvcf targeted mouse line, Arvcf ^tm1e(EUCOMM) Wtsi, was purchased from the European Mouse Model Archive (EMMA ID 04548). The line was generated from the insertion of a lacZ/Neomycin cassette (L1L2_gt0) to generate an ES cell clone (EUCOMM EPD0102_1_D06) which targeted the Arvcf locus on chromosome 16 at position 18396535 between exons 3 and 4 of the longest Arvcf transcript (Supplementary Figure S1A). The 3′ loxP site was not preserved during targeting preventing this allele from having conditional potential. Genotyping was performed utilizing the cassette flanking primers 5′-GCT​GAC​CTA​ACC​ATG​GTT​ACG (for) and 5′-CAA​GAC​AAG​TCC​ATC​TGG​ACC (rev) and the targeting cassette residing primer 5′CAA​CGG​GTT​CTT​CTG​TTA​GTC​C (p3) to yield a 564 nucleotide and/or a 430 nucleotide band in a 1% agarose gel with PCR using Dream Taq (Thermofisher) and the following protocol: 95°C for 5 min (94°C for 30 s, 65°C for 45 s, 72°C for 45 s, 35cycles) 72°C for 60 s (Supplementary Figure S1B).

The refraction chamber was made from a clear acrylic box with a ¾” hole drilled in its side and a 1” disc of optical glass (Thorlabs: WG11050) glued to the outside. The mouse lenses were placed upon a pedestal made from a glass cloning cylinder and a piece of plastic glued to the top creating a small hole for the lens to rest on its equatorial side. A 0.8 mW, 632.8 nm helium/neon laser (Thorlabs: HNLS008L) was mounted on a vertically articulating platform and aimed through a Plano Convex Lens (Thorlabs: LA1131) which focused the beam parallel to the light axis of the mouse lens tissue. The chamber was flooded with phosphate-buffered saline (PBS) mixed with a small amount of powdered milk to increase the turbidity of the solution. Images of no fewer than 6 lenses dissected from at least 5 different animals and the surrounding media were taken following incremental raising (13–17 increments) of the platform-mounted laser such that the beam could be imaged across the equatorial plane of the lens. Measurements of focal lengths and the position of the beam within the lens were performed using Zen software (Zeiss). ANOVA analysis was used to determine whether statistical differences were observed between genotypes followed by post-hoc t-tests.

RNA was extracted from whole eyes and cerebellar brain tissue using commercially available kits (RNeasy/Qiagen) and used to prepare cDNA (Verso cDNA synthesis kit/Thermo Fisher). PCR was performed utilizing Arvcf {5′-GCT​GCT​GGC​ACC​CTG​GTC​AT (for) and 5′-GTC​TCA​GTC​CGC​CGG​GTT​GTA (rev)} and Gapdh {5′-CAT​CAC​TGC​CAC​CCA​GAA​GAC​TG (for) and 5′-ATG​CCA​GTG​AGC​TTC​CCG​TTC​AG (rev)} specific primers, Dream Taq (Thermofisher) and the following protocol: 95°C for 5 min (94°C for 30 s, 60°C for 60 s, 72°C for 30 s, 39 cycles) 72°C for 60 s to yield a 364 (Arvcf) or 110 (Gapdh) basepair band in a 1% agarose gel. Western blotting of lens or brain lysates was performed by extracting protein from two lenses or 1 cerebellar hemisphere per experimental group by first homogenizing the tissue in 1 × PBS pH 7.0 100 mM EGTA with an electric handheld homogenizer for 2 min, centrifuged (30 min 17000 × g at 4°C) and the pellet washed in homogenization buffer with 50 mM DTT and recentrifuged for (15 min 17000 × g at 4°C). RIPA buffer (200 μl/Pierce) mixed with 2 μl of protease inhibitors (HALT protease inhibitor cocktail/Thermo Fisher) was used to resuspend the pellet and incubated for 45 min at 4°C with gentle agitation. The lysate was centrifuged again (15 min 17000 × g at 4°C) and the lysate used in SDS-PAGE analysis using 12% precast TGX Stain-Free polyacrylamide gels (Bio Rad) which allows for total protein in the gel and/or blot to be visualized with UV light and is more reliable than antibody-based housekeeping gene loading controls (Rivero-Gutierrez et al., 2014). Arvcf (Bethyl Laboratories, A303-310A) and β-catenin specific antibodies (1:500, BD biosciences, BDB610153) were utilized at 1:1,000 dilution on blots and detected with HRP conjugated secondary antibodies, Clarity ECL (Bio Rad) reagent, and a ChemiDoc imager (Bio Rad).

Mouse embryo and lens tissue was fixed in 4% (embryos, whole lenses) or 2% (dissected cortical lens fiber cells) paraformaldehyde. For histological sections, tissue was embedded in OCT medium (Tissue-Tek) following 15 and 30% sucrose infiltration and a cryostat used to generate 10–20 μm sections. Dissected cortical lens fiber cells were prepared in a manner similar previously published methods (Cheng et al., 2016b). The staining protocols used were similar to those previously described (Houssin et al., 2020). The following primary antibodies were utilized followed by a combination of Alexa Fluor secondary antibodies conjugated with 488, 568, 594, or 647 fluorophores (Invitrogen). Antibodies: Arvcf (1:500 for sections; 1:250 for whole lens fiber cells, Thermofisher, PA5-64129), N-cadherin (1:500 BD Biosciences, 610,921), αN-catenin (1:250, ABclonal A15269), δ-catenin (1:500, Thermofisher, PA5-53275), p120-catenin (1:200, BD Biosciences, BDB610133), β-catenin (1:500, BD biosciences, BDB610153), β-catenin (1:200, GeneTex, GTX101435), Aquaporin-0 (1:200, Alpha Diagnostics International, AQP01-A), αB-crystallin (1:500, DSHB, CPTC-CRYAB-3). Fluorescently labeled Hoechst 33,342 (1:1,000; Sigma, B-2261) or wheat germ agglutinin (1:200, Life Technologies, W32464/W11262) were also utilized to label nuclei or fiber cell membranes (respectively). Glass-slide mounted samples were coverslipped with Fluoro-gel medium (Electron Microscopy Sciences) and imaged with a Zeiss Axio observer inverted microscope equipped with a fluorescent light source and ×40 Plan-Apochromat objective or equipped with an LSM900 Airyscan 2 super-resolution confocal system and ×63 Plan Apochromat objective.

20 4 × 4 μm regions were extracted from images taken from an Airyscan equipped confocal microscope with a plan-apochromat ×63/1.40 oil immersion objective focused within the plane of the lateral membrane of lens fiber cells co-immunolabeled for N-cadherin and β-catenin. Images of fiber cells from at least three different lenses were used. Utilizing ImageJ, each region underwent autothresholding with the “moments” algorithm followed by using the analyze particles function to measure nanocluster attributes. Two-tailed, homoscedastic t-tests were used to determine statistical significance. The area of a circle formula was used to derive the approximate diameter of individual spots using the mean area of fluorescent signal.

For all quantitative measurements, the conditions used for image acquisition were carefully selected to ensure a lack of pixel saturation and uniformity among experimental groups.

Lenses were fixed in 2.5% glutaraldehyde/0.1 M PBS pH7.4, bisected in half, incubated at 4C for 72 h, rinsed, dehydrated with a series of increasing EtOH concentrations, and dried with a critical point drier. The samples were mounted onto double-sided carbon tape on SEM stubs and underwent gold/palladium sputter coating. Samples were examined using 5 kV with a FEI Nova NanoSEM instrument.

Lens biomechanics were tested using procedures similar to those previously described (Cheng et al., 2016a). In brief, a 45^o angled mirror was used in combination with an acrylic chamber filled with PBS and stereomicroscope to view an immobilized mouse lens from the equatorial side as glass coverslips were placed on the anterior side. Images were taken 45 s following the addition of each successive coverslips (10 maximum). The axial and equatorial widths were measured with Zen 2.0 and the mean strain ((ε = measured length-original length)/original length) was determined for each mechanical load. 8-9 lenses from at least 4 different animals were utilized per genotype. Two-tailed, homoscedastic t-tests determined statistical significance. Axial and equatorial length measurements were taken of a subset of compressed lenses (n = 5–8) 2 min following the removal of compressive load to determine lens resilience. The percent difference between the pre-load and post-load measurements ((post-load-pre-load)/pre-load) were reported.