Canonical ligand-dependent and non-canonical ligand-independent EphA2 signaling in the eye lens of wild-type, knockout, and aging mice

Total EphA2 protein level decreases with age while non-canonical EphA2 activation increases with age and in ephrin-A5^-/- lenses

To determine what EphA2 signaling pathways are activated in the lens, we performed capillary electrophoresis Western blots on protein lysates from lens epithelial cells and cortical fiber cells isolated of 6-week-old, 4-month-old, and 8-month-old EphA2^+/+, EphA2^-/-, ephrin-A5^+/+, and ephrin-A5^-/- mice. We chose these ages of mice to represent young, adult, and middle-aged animals. Mice develop age-related cataracts at approximately 10 months of age [73], and thus, we did not collect older samples for this study because it would be difficult to separate the effects of very old age from the changes due to the KO. EphA2^-/- samples were used to verify antibody specificity (Supplementary Figure 3). We did not investigate the nuclear fiber cell fractions because we previously demonstrated that nuclear fiber cells do not have detectable levels of EphA2 [43]. The trends seen in EphA2^+/+ samples are comparable with ephrin-A5^+/+ samples.

In the ephrin-A5 control and KO samples, we measured the levels of pan-EphA2 (total), canonically active EphA2 phosphorylated at tyrosine 589 (pY589), and non-canonically active EphA2 phosphorylated at serine 898 (pS898) (Figure 1). Within the epithelial cell fraction (Figure 1A), Western blots showed significant decreases the level of pan-EphA2 with age, consistent with a previous report [24]. The loss of ephrin-A5 did not alter the levels of pan-EphA2 when compared to littermate controls. Canonically active EphA2-pY589 protein levels within the epithelium remained unchanged across ages and genotypes. Surprisingly, non-canonically active EphA2-pS898 protein was detected in normal control ephrin-A5^+/+ epithelial cells, and the level of EphA2-pS898 increased significantly between 6-week-old and 4-month-old and between 6-week-old and 8-month-old control samples. There was no further increase in EphA2-pS898 levels between 4 months and 8 months in control lens epithelial cells. Similarly, there was an increase in EphA2-pS898 levels in ephrin-A5^-/- lens epithelial cells with age, and unlike the control, in 8-month-old ephrin-A5^-/- lens epithelial cells, we detected much higher levels of non-canonical activation. In samples from 4-month-old and 8-month-old ephrin-A5 control and KO mice, we noticed an increase in the height of the bands for pan-EphA2, EphA2-pY589, and EphA2-pS898 on the Western results. Presumably, this is due to cleavage and post-translational modifications of EphA2 [74].

Within the cortical fiber cells (Figure 1B), the pan-EphA2 protein decreased between 6-week-old and 4-month-old and between 6-week-old and 8-month-old control and KO samples. However, there was no further decrease in pan-EphA2 levels between 4 and 8 months of age. There was no difference in pan-EphA2 levels between ephrin-A5 control and KO age-matched samples. Canonically active EphA2-pY589 protein was absent from fiber cells of all ages. Non-canonically active EphA2-pS898 protein was found to be abundantly present in cortical fiber cells, and there was an obvious increase in EphA2-pS898 levels with age. Ephrin-A5 control and KO age-matched samples had comparable EphA2-pS898 levels. Similar to the epithelial cell sample Westerns, we observe an increase in the height of the bands for pan-EphA2 and EphA2-pS898 in the cortical fiber cell samples from 4-month-old and 8-month-old control and KO samples, presumably due to cleavage or post-translational modifications of EphA2 with age. These data show that canonical activation of EphA2 was restricted to the epithelial cells and does not change with age or genotype. Surprisingly, non-canonical activation of EphA2 is detected in ephrin-A5 control and KO epithelial cells and fiber cells. The level of pan-EphA2 decreased significantly with age, but there was an obvious increase in non-canonical EphA2 activation with age and in 8-month-old ephrin-A5^-/- epithelial cells.

Pan-EphA2 localizes along cell membranes throughout the lens

To further investigate the localization of these signaling pathways, immunostaining was performed on frozen lens sections in the longitudinal and cross orientations from 6-week-old, 4-month-old, and 8-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice. Sections were stained with pan-EphA2 antibody, EphA2-pY589 antibody, or EphA2-pS898 antibody and counterstained with phalloidin for F-actin and DAPI for cell nuclei. Antibody specificity was confirmed with EphA2^-/- frozen sections (Supplementary Figure 4), and the staining signal between EphA2^+/+ and ephrin-A5^+/+ sections were consistent. The staining signals were comparable between sections from 6-week-old, 4-month-old, and 8-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice. For some data sets, we will only include the 4-month-old data in the main text as representative data, and the 6-week-old and 8-month-old data can be found in the supplement.

Immunostaining of longitudinal sections revealed that pan-EphA2 localized to epithelial cells (Figure 2, arrows) and fiber cells with a band of increased fluorescence between mature and inner fiber cells that remained consistent with age (Figure 2, asterisks). There were no obvious differences in the staining pattern for pan-EphA2 between control and ephrin-A5 KO sections. To better visualize the staining pattern in epithelial cells, images were taken with high magnification of longitudinal lens sections from 4-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice, which showed localization of the staining to be primarily along the apical-apical junction of anterior epithelial and fiber cells (Figure. 3, arrowheads) in control and KO sections. At the lens equator, the localization of EphA2 expands to both the apical and basal side of epithelial cells (Figure 3, arrowheads and open arrowheads, respectively). The location of the cell membrane was judged by the F-actin staining signal. There was enriched EphA2 staining signal along fiber cell membranes. Using cross orientation sections to examine fiber cell staining patterns in detail, we confirmed the localization of pan-EphA2 protein along the short and long sides of peripheral and mature fiber cell membranes, with slightly increased fluorescence detected on the short sides of the mature fiber cells (Figure 4, asterisks). The staining signal for pan-EphA2 is low in the inner fiber cells. These localization patterns in sections from 4-month-old mice were consistent in high magnification images of sections from 6-week-old (Supplementary Figures 5, 7) and 8-month-old control and KO mice (Supplementary Figures 6, 8).

Canonically active EphA2-pY589 staining localizes to epithelial cells and is enriched at the lens equator

Next, we investigated where canonically active EphA2-pY589 signals were localized in the lens. Sections in the longitudinal orientation stained with EphA2-pY589 revealed fluorescence signals only within epithelial cells of control and KO lens sections (Figure 5, arrows). The staining pattern was very similar between control and KO sections in all age groups. Almost no signal was detected in the fiber cells, which is consistent with the Western blot results. When we examined epithelial cells in sections from 4-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice, there was much lower staining signal in the anterior epithelial cells compared to equatorial epithelial cells (Figure 6). The signal in anterior epithelial cells was mostly at the basal surface (Figure 6, open arrowheads) with weak staining at the apical-apical junction (Figure 6, arrowheads). At the lens equator, the stronger EphA2-pY589 staining signals were in equatorial epithelial cells along the apical and basal cell surfaces (Figure 6, arrowheads and open arrowheads, respectively) and between the cells at the lateral membranes. There appears to be brighter staining signal in the equatorial epithelial cells compared to that in the anterior epithelial cells. Because canonical EphA2 signaling was only detected in epithelial cells, staining of cross orientation sections, usually used to examine fiber cell membranes, is not shown. We observed similar epithelial cell staining patterns in 6-week-old (Supplementary Figure 9) and 8-month-old ephrin-A5^+/+ and ephrin-A5^-/- lens sections (Supplementary Figure 10). While previous work has suggested the canonical ligand-mediated EphA2 signaling is likely to be active in the lens, this is the first direct evidence that bidirectional EphA2 signaling is active in lens epithelial cells.

Non-canonically active EphA2-pS898 staining is enriched in mature fiber cells

Finally, we immunostained lens sections to reveal where non-canonical EphA2-pS898 signaling is active. In low magnification images of longitudinal sections from ephrin-A5^+/+ and ephrin-A5^-/- mice, EphA2-pS898 staining showed localization to the epithelium (Figure 7, arrows) and a specific subset of mature fiber cells in the lens cortex (Figure 7, asterisks). The band of fluorescence in the mature fiber cells appeared more diffuse in 6-week-old sections, and the staining signal in those fibers compresses with age into a narrower band. Though the fluorescence signal in the fiber cells appeared to change with age, the same subset of mature cells was stained in sections from all ages. The EphA2-pS898-positive mature fiber cells are in a region where these cells have just eliminated their cellular organelles. With continued differentiation of the mature fibers as they are compacted toward the center of the lens, the EphA2-pS898 staining signal is lost. The compression of the fluorescence signal with age is due to the continuous addition of fiber cells in concentric shells in the lens cortex and the lens capsule that restricts the expansion of the tissue. This compression of signal can also be observed in the DAPI channel (blue) where the nuclei in the lens periphery were similarly compressed. There were no obvious differences between ephrin-A5^+/+ and ephrin-A5^-/- lens sections in each age group.

In sections from 6-week-old, 4-month-old, and 8-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice, high magnification images of lens epithelial cells in longitudinal sections revealed very low EphA2-pS898 signal in anterior epithelial cells, and there was an increase in fluorescence signals in equatorial epithelial cells (Figure 8 and Supplementary Figures 11, 12, arrows) with a focal increase at the lens equator. While the staining signal did appear to be near the cell membrane in some epithelial cells, there was no obvious membrane enrichment, and most of the signal appeared to be cytoplasmic. Cross orientation section staining from 6-week-old and 4-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice to examine fiber cells showed that EphA2-pS898 staining was weakly present in peripheral fibers and obvious in mature fiber cells (Figures 9 and Supplementary Figure 13, asterisks). In the peripheral fibers, the staining signal appear to be punctate and at or near the cell membrane while in the mature fibers, the staining signal is evenly distributed around the membrane of mature fiber cells without preference for either of the broad or short sides. In the inner fiber cells, the staining signal appears punctate and cytoplasmic. There were no significant differences from EphA2-pS898 staining signals between ephrin-A5^+/+ and ephrin-A5^-/- sections. Interestingly, in sections from 8-month-old ephrin-A5^+/+ and ephrin-A5^-/- mice, we observed that EphA2-pS898 signals in peripheral and mature fiber cells appeared more cytosolic (Figure 10, asterisks) without obvious enrichment at the cell membrane. These results indicate that non-canonical EphA2 signaling can be active in a physiological tissue, and the data further suggest that loss of ephrin-A5 does not significantly affect the levels or localization of non-canonical EphA2 signaling, thus eliminating the possibility that EMT in the ephrin-A5^-/- lenses is caused by abnormal EphA2 signaling.

Non-canonical EphA2 activation in ephrin-A5^-/- lenses

Curiously, in lens epithelial cells from 8-month-old ephrin-A5^-/- mice, there is a large increase in EphA2-pS898 protein compared to the age-matched control samples. We specifically selected ephrin-A5^-/- lenses without obvious anterior epithelial cell EMT and anterior cataracts for our experiments, and thus, this increase is due to changes without the involvement of cells undergoing EMT. We also performed immunostaining in ephrin-A5^-/- lens sections with anterior cataracts that displayed overgrowth of the anterior epithelial cells, and those abnormal epithelial cells do not have staining for non-canonically active EphA2-pS898 (data not shown). The anterior cataracts in ephrin-A5^-/- lenses occur by postnatal day 21 [38, 39] while the rise in EphA2-pS898 non-canonical signaling in ephrin-A5^-/- lenses is only evident in middle-aged 8-month-old KO mice. Thus, it is unlikely that the significant increase in EphA2-pS898 level in lens epithelial cells from 8-month-old ephrin-A5^-/- mice activates EMT in anterior epithelial cells to form cataracts. This evidence further suggests that EphA2 and ephrin-A5 are not an exclusive receptor-ligand pair in the lens.

Localization of total and activated EphA2 in lens epithelial and fiber cells

There is a gradient of staining signal intensity for pan-EphA2, EphA2-pY589, and EphA2-pS898 at the cell membranes of lens anterior epithelial cells, equatorial epithelial cells, and fiber cells from control and ephrin-A5^-/- lens sections. There is weak EphA2 staining of the anterior epithelial cells compared to equatorial epithelial cells, and there is much stronger staining signal at fiber cell membranes. Canonical signaling localized exclusively to the epithelial cells in both control and ephrin-A5^-/- lenses, with almost no signal detected in the fiber cells. There is a gradient of EphA2-pY589 staining signal with weak staining in anterior lens epithelial cells compared to equatorial epithelial cells. Non-canonical EphA2-pS898 staining signal was present in equatorial epithelial cells and in fiber cells with increased fluorescence within the mature fiber cells. Localization of these staining patterns remained similar with age and genotype indicating that age-related or genotype-related protein level changes have no effect on the cellular localization. The exception to this was the EphA2-pS898 staining pattern in middle-aged control and ephrin-A5^-/- lenses, which we discuss below.

In the anterior epithelial cells, most of the pan-EphA2 staining is concentrated at the apical membrane where the epithelial cells would contact the apical tips of elongating lens fiber cells. This is consistent with our previous report of pan-EphA2 staining in anterior lens epithelial cell flat mount samples [39]. Interestingly, staining for canonically active EphA2-pY589 localizes to both the apical and basal sides of the anterior epithelial cells. This was not expected because staining for pan-EphA2 is not obvious on the basal side of anterior lens epithelial cells. One possible reason for this discrepancy is that the pan-EphA2 staining signal is very strong in lens fibers, and the relatively weaker signal at the basal surface of lens epithelial cells is not visible in these images nor our previous flat mounts because the intensity setting for image collection was determined based on the strong staining signal in fiber cells. In equatorial epithelial cells, we observe pan-EphA2 and canonical EphA2-pY589 staining on both the apical and basal surfaces of the cells. The staining for pan-EphA2 on the basal surface of equatorial epithelial cells is consistent with our previous flat mount data [41].

In contrast, while non-canonical EphA2-pS898 staining signal was present in equatorial epithelial cells, there was no enrichment at the cell membrane, and most of the staining signal appeared in the cytoplasm. It is possible that non-canonical activation the EphA2 leads to rapid receptor internalization. HeLa cells, human breast adenocarcinoma (MDA-MB-231) cells, human lung adenocarcinoma and squamous cell carcinoma biopsy samples, and human vascular smooth muscle cells stained for EphA2-pS897 also revealed a diffuse intracellular pattern [62, 76]. The exclusive activation of canonical EphA2 signaling at epithelial cell membranes, especially at the lens equator, suggests that ligand-mediated EphA2 activity is needed for normal patterning of equatorial epithelial cells. Canonical EphA2 activation is known to signal through Src kinase to affect the actin cytoskeleton [41, 42, 77–79]. In EphA2^-/- and Src^-/- lenses, equatorial epithelial cells fail to properly organize into neat rows of hexagon-shaped cells (Supplementary Figure 2) [41]. Further detailed studies of the localization of canonically and non-canonically active EphA2 in lens epithelial cells using flat mounts is required to better understand the subcellular localization of these phosphorylated forms for EphA2.

In lens fiber cells, pan-EphA2 is enriched along the short sides of mature fiber cells while EphA2-pS898 signal is spread along the entire mature fiber cell membrane. Interestingly, in 8-month-old control and ephrin-A5^-/- mice, the non-canonical EphA2-pS898 signal appeared to be localized more in the mature fiber cell cytoplasm than the membrane, which is an age-related change in protein localization. It is not clear whether the change in the EphA2-pS898 localization may be correlated with cleavage or post-translational modifications that can be seen in the Western blots. Lens fibers cells have very complex morphology that cannot be fully visualized in tissue sections [80, 81], and future immunostaining studies on single lens fiber cells is required to elucidate the subcellular localization of EphA2-pS898 activation in mature lens fiber cells.

Possible roles of non-canonical EphA2 signaling in lens fiber cell membrane interdigitations and maturation

While the role of non-canonical EphA2 signaling in the normal lens remains unclear, cancerous cells and tissues and immortalized cells with increased non-canonical EphA2 signaling are known to display several phenotypic changes ranging from increased cell motility and invasion [52, 62, 78], resistance to apoptosis [82], promotion of angiogenesis and neovascularization [83–85], and fibrotic tissue responses [9, 76]. Mature lens fiber cells undergo continuous changes in cytoskeletal structure and cell membrane during differentiation with formation of different types of complex membrane interdigitations and protrusions [43, 80, 81, 86]. Though these complex cell interdigitations have been well-described by electron microscopy studies, the mechanisms that influence membrane re-organization during fiber cell differentiation and maturation remain unknown. While different from traditional cell migration, fiber cell membranes extend protrusions and create elaborate cell-cell interlocking features that mimic protrusions from migrating cells. Our previous studies of EphA2^-/- lenses showed distinct changes in inner and perinuclear fiber cell morphology [43]. Scanning electron microscopy studies revealed abnormal cell-cell interdigitations and membrane morphology in a region of the lens that is slightly deeper than where the non-canonical EphA2 signaling is active in mature fiber cells (Supplementary Figure 2). It is possible that non-canonical EphA2 signaling plays a role in the rearrangement of fiber cell membranes in the inner regions of the lens.

EphA2 can be phosphorylated by different AGC protein kinases (Akt, RSK, PKA) resulting in different downstream non-canonical signaling changes [52, 56, 62]. Activation of EphA2 by Akt promotes changes in cytoskeletal elements, through defining the leading edge of migrating cells [52] and enhancing cell motility. Downstream effectors of Akt, like mTORC1, MEK-ERK, and Src-ERK, have been associated with cholangiocarcinoma cells [87], induction of stemness markers [88], and amoeboid motility [89]. Many malignant tumors exhibit cell motility, EMT, and drug resistance because of constitutive non-canonical activation of EphA2 through the MEK-ERK pathway upon growth factor stimulation [9]. Previous studies have suggested that Akt phosphorylation at serine residue 897 is a major mechanism of cell migration in HEK 293 cells and a variety of cancer cells [52]. Known knockouts of elements upstream of Akt display distinct lens phenotypes. PTEN knockouts cause an increase in Akt signaling and lead to lens rupture and cataracts, whereas partial PI3K knockouts lead to reduced growth and equatorial epithelial proliferation [4]. PKA activation has been shown to facilitate the formation of connexin channels within fiber cells, decreasing reactive oxygen species, and alleviating cataract phenotypes [90]. Previous studies have also shown that increased cAMP promotes PKA activity, which results in phosphorylation of EphA2 at serine residue 897, as well as other serine and threonine residues causing a phosphorylation hot spot [56]. These hotspots were enriched at leading edges of migrating cells and are believed to promote malignancy and motility [52, 91]. The PKA-dependent activation of non-canonical EphA2 signaling has further been classified in aggressive pancreatic and prostate cancers [56]. To further determine the pathogenesis of age-related cataracts, future experiments will need to be done to determine the expression levels of activated Akt, RSK and PKA to determine if there are any age-related or genotype-related expression changes and whether there is any link between the pathological and physiological functions of non-canonical EphA2 signaling.

The area of intense EphA2-pS898 staining in mature lens fibers occurs just within the organelle-free zone of the lens after fiber cells remove all cellular organelles. From our data in this study and previous work, we have not observed any obvious changes in the denucleation process in EphA2^-/- lenses. We have not examined the loss of other cellular organelles in EphA2^-/- lenses. Whether non-canonical EphA2 signaling is required for the formation of the organelle-free zone remains to be studied.

Cataract-causing mutations in human EphA2

Many human mutations in EphA2 are associated with age-related and congenital cataracts [16]. The mechanisms for these variable cataracts remain unknown, but this new data about the two modes of EphA2 activation in the lens can be useful to inform possible changes due to known mutations. Protein models can also be a useful tool in understanding how mutations affect the structural properties of the EphA2. Known cataract causing EphA2 protein sequences can be modeled alongside mutations in residues in close proximity to serine residue 897 or tyrosine residue 588 to observe if there are any significant changes to protein folding. Many human cataracts are reported in the lens cortex where there is increased non-canonical EphA2 activation with age. Future protein modeling work may reveal insights into whether mutations can affect canonical or non-canonical EphA2 phosphorylation, causing cataracts in different regions of the lens.

Concluding thoughts

The lens is a unique tissue to study the aging of cells since it contains cells that were made during embryonic development that are chronologically old and also newly made cells at the cortex that are still being made in an aged organism. Maintenance of the nuclear fiber cells for an entire lifetime to resist degradation, degeneration, and disease is remarkable, and the mechanisms of normal versus pathological lens aging remain an active area of research.

Here, we report that canonical ligand-mediated EphA2 activation is restricted to the lens epithelial cells and show the first evidence of physiological non-canonical EphA2 activity in a normal tissue. The lens, unexpectedly, utilizes both canonical and non-canonical EphA2 signaling in different cell compartments. Canonical EphA2 signaling is likely important in equatorial lens epithelial cells and may be linked to cell shape changes required for highly ordered packing of the cells to minimize light scattering. Non-canonical EphA2 signaling may also be important in equatorial lens epithelial cells, but the function is unknown due to the mostly cytosolic localization of the activated protein. In lens fibers cells, only non-canonical EphA2 activation is present, and this signaling is enriched in a subset of mature fiber cells undergoing membrane shape changes. We hypothesize that non-canonical EphA2 signaling is important for transformation of complex interdigitations in mature fiber cells.

Mice

EphA2^-/- mice were acquired from The Jackson Laboratory (strain #: 006028), and ephrin-A5^-/- mice [50] were a generous gift from Dr. David A. Feldheim (University of California, Santa Cruz, CA, USA). All mice were maintained in C57BL/6J background with wild-type Bfsp2 (CP49) genes [39]. Sporadic mutations in Bfsp2 are associated with several mouse strains and can influence lens fiber cell cytoskeleton and produce cataracts not associated with EphA2 signaling pathways or aging [80, 92–95]. Genotyping was completed using automated qPCR on toe and/or tail samples collected between postnatal days 5-7 (Transnetyx). Age-matched littermates from heterozygous mating pairs were used for experiments to prevent differences due to genetic background drift. Approximately equal numbers of male and female mice were used in our experiments.

Lens protein extraction and JESS capillary-based immunoassay

Fresh enucleated eyes were dissected in 1X phosphate buffered saline (PBS, 14190-144, Gibco) in a dissection plate as previously described [96]. Separation of the lens epithelial layer, cortical fiber cells, and nuclear fiber cell masses was also performed as previously described [97]. Samples were collected from 6-week-old, 4-month-old, and 8-month-old EphA2^+/+, EphA2^-/-, ephrin-A5^+/+, and ephrin-A5^-/- mice. At least three different mice per genotype per age were used to make biological replicates. For ephrin-A5^-/- mice, samples were only collected from pairs of eyes without obvious anterior cataracts because it would not be possible to separate the signaling that is altered by the loss of ephrin-A5 from changes that occurred after EMT is initiated in the anterior epithelial cells.

Briefly, the lens epithelium attached to the capsule was separated from the fiber cell mass, and the capsule from two eyes from the same mouse was placed into a 1.5 mL microcentrifuge tube filled with 14.75 μl of ice-cold lens homogenization buffer [20mM Tris (pH 7.4 at 4° C), 100mM sodium chloride, 1mM magnesium chloride, 2mM ethylene glycol tetraacetic acid, 10mM sodium fluoride, 1mM dithiothreitol (DTT), 1X Pierce phosphatase inhibitor (A32957, Thermo Fisher Scientific), 1:100 protease inhibitor cocktail (P8340, Sigma-Aldrich)] and 1.25 μL of 2X sample buffer [0.21M Tris-HCl (pH 6.8), 2.86 mM ethylenediamine tetra-acetic ETDA, 21% sucrose, 6.67% SDS, and 0.3M DTT in ddH2O]. DTT, Pierce phosphatase inhibitor, and the protease inhibitor cocktail were added to the homogenization buffer on the day of sample collection. Lens epithelial cell samples were then vortexed briefly with two short pulses followed by brief centrifugation at room temperature at 2,000x g for 15 seconds. To separate the cortical and nuclear fiber cells, the fiber cell mass was added to a 1.5 mL microcentrifuge tube with 250 μL of 1:1 ice-cold lens homogenization buffer and 2X sample buffer and vortexed in 30-second bursts over 2-4 minutes. Vortexing can be stopped earlier than 4 minutes if the ball of fiber cells does not become smaller with successive vortexing. The remaining ball of fiber cells are the dense and compact lens nuclear fiber cells, which were removed with clean tweezers from the cortical fiber cell lysate and discarded. Sonication (QSonica, Q55-110, with 5/64” probe, 4423, at an amplitude of 15) of lens cell fractions was performed using cycles of 3 seconds of sonication and 10 seconds of cooling on ice. Epithelial cells received one cycle of sonication while fiber cells received three cycles. After sonication, the lens lysate protein concentration was measured using the Bradford assay (5000205, BioRad) per the manufacturer’s instructions.

Western blot analysis was performed using the JESS capillary electrophoresis machine (ProteinSimple) to minimize the amount of sample needed for measurements. Equal amounts of control and KO proteins for lens epithelial or cortical fiber cell fractions were loaded into the JESS 12-230kDa separation module kit as per manufacturer instructions. Previously calculated protein and antibody concentration titrations were used to determine the range for antibody saturation and optimum concentrations of protein extracts. Detection of protein was performed by loading lens epithelial cells (~0.4 μg/μL) and cortical fiber cells (3 μg/μL) with 1:50 of a rabbit anti-EphA2 antibody (6997, Cell Signaling Technology), 1:50 of a rabbit anti-EphA2-pS897 antibody (6347, Cell Signaling Technology) or 1:50 of a rabbit-EphA2-pY588 antibody (12677, Cell Signaling Technology). While both antibodies against the phosphorylated forms of EphA2 were made to the human protein sequence, there is 100% homology between the human and mouse EphA2 protein sequence in the regions recognized by the antibodies. Antibody specificity was verified by lack of signal in EphA2^-/- samples (Supplementary Figures 3, 4). Target protein levels were normalized to the total protein signal from each capillary measured using the Total Protein detection kit from ProteinSimple. Calculations for average and standard deviation were completed in Excel and plotted in GraphPad 9. ANOVA (1-way) was used to determine statistical relevance, p < 0.05, between control and KO samples from different age groups.

Immunostaining

Frozen lens tissue sections were collected and prepared for fixation as previously described [81, 98]. To summarize, eyes from 6-week-, 4-month-, and 8-month-old EphA2^+/+, EphA2^-/-, ephrin-A5^+/+, and ephrin-A5^-/- mice were collected and fixed with freshly made ice-cold 1% paraformaldehyde (PFA, 15710, Electron Microscopy Sciences). For ephrin-A5^-/- eyes, sections from lenses with obvious anterior cataracts were excluded. A small incision was made into the corneal-scleral junction, allowing the PFA fixative to better penetrate the globe. The eyes were cryoprotected in a 30% sucrose solution before being frozen in OCT medium, oriented in the cross or longitudinal orientation. Twelve-micron thick slices were collected using a Leica CM1860 cryostat. Sections were stored at -20° C until staining.

Frozen tissue sections were washed with 1X PBS + 0.1% Triton X-100 (PTX) before being incubated in blocking buffer (5% donkey serum + 0.3% Triton X-100 in 1X PBS). A 1:100 dilution of primary antibody (same antibodies as described for western blots) was applied, and sections were incubated overnight at 4° C. Slides were washed in 1X PTX with light rocking before a 1:200 dilution of donkey anti-rabbit secondary antibody conjugated to Alexa-Fluor-488 (711-545-152, Jackson ImmunoResearch Laboratories) and rhodamine phalloidin (R415, Thermo Fisher Scientific) was applied to the sections and incubated. Stained sections were washed in 1X PTX and then mounted with Vectashield mounting media containing DAPI (H-1200, Vectashield). The edges of the coverslip were sealed with clear nail polish.

Sections were viewed using a Zeiss LSM800 laser confocal microscope with 20X [air, numerical aperture (NA) = 0.8] and 63X (oil, NA = 1.4) objectives, for a total magnification of 200X and 630X, respectively. Images were captured using Zen 2.3 SPI software at room temperature on the same laser and gain intensities. Comparable equatorial lens sections were selected from judging thickness of the epithelial monolayer, and presence of nuclei in the fiber cells. Staining was repeated on at least 3 biological replicates of each genotype and age. Representative data are shown.