Mice were maintained in accordance with approved animal protocols (State University of New York at Stony Brook and Indiana University Bloomington Institutional Animal Care and Use Committees) and the National Institutes of Health guide for the care and use of laboratory animals. Ephrin-A5^–/– and EphA2^–/– mice were generated and maintained as previously described (Okabe et al., 1997; Frisen et al., 1998; Cheng and Gong, 2011; Cheng et al., 2013). All mice were maintained in the C57BL/6J background with wild-type Bfsp2 (CP49) genes. Genotyping was performed by automated qPCR on toe or tail snips (Transnetyx, Cordova, TN, United States). Male and female littermates were used for experiments. We did not observe any obvious differences in lens phenotype between male and female mice.

Gap junction coupling conductance measurements were made as previously described (Gao et al., 2013; Cheng et al., 2015). The relationship of the “series resistance” to GJ coupling conductance is presented in Mathias et al. (1981), where the specialized electronics, necessary for wide bandwidth recordings in the lens, are also described. Control studies showing the relationship of the series resistance to GJ coupling conductance are reviewed by Mathias et al. (2010). In the current study, 3- and 12-month-old mice were used. Measurements were repeated on at least 6 lenses from 3 mice for each genotype and age. Due to partial dissection of the eyes for these experiments, images of the lenses were not obtained. Lenses without obvious severe cataracts were used for these experiments. Briefly, eyes were immediately enucleated from euthanized mice and placed in a Sylgard lined Petri dish filled with normal Tyrode solution containing 137.7 mM NaCl, 2.3 mM NaOH, 5.4 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES, and 10 mM glucose (pH 7.4). After dissection to remove the cornea, iris, and optic nerve, the sclera was cut into 4 flaps, and the lens was mounted to the bottom of a chamber with a Sylgard base. The sample was mounted on the stage of the microscope, and the lens was maintained in normal Tyrode solution. One intracellular microelectrode was inserted into a fiber cell near the lens center where a wideband stochastic current was injected. A second intracellular microelectrode was placed at various depths between the periphery and center of the lens, and the intracellular voltage and lens impedance were recorded at each depth. The distances r (cm) from the center of a lens of radius a (cm) were recorded, and the impedance (induced voltage divided by injected current) was recorded using a Fast Fourier Analyzer (Hewlett Packard, Palo Alto, CA, United States). The radial intracellular resistance of GJs (R_S) between the point of recording and the surface of the lens was estimated from the magnitude of the impedance at 1,000 Hz (Mathias et al., 1981). Four to five recordings were made in each lens from the voltage-recording intracellular microelectrode that was advanced radially into the lens. R_S values from at least 6 lenses of each genotype were pooled and curve-fitted using equations below:

Mature fiber cells generally have increased intracellular resistivity (R_MF, Ω⋅cm) compared to peripheral fibers (R_DF, Ω⋅cm) with the change occurring abruptly at b = 0.85a, the site of transition from DF to MF where organelles are degraded and Cx46 and Cx50 proteins are C-termini cleaved. GJ coupling conductance (G) per area of cell-to-cell contact (S/cm²) is calculated by:

where w = 3 μm based on the radial spacing between GJ plaques. The intracellular voltage (ψ_i) with respect to the bathing solution was recorded at each depth. In the lens, there is no single “resting voltage” as in isolated cells. The circulating current is associated with a radial gradient in the intracellular voltage, with a value of around −50 mV in central fiber cells to about −60 mV in surface cells. The transmembrane voltage is the difference in the intracellular and extracellular (between fibers) voltages, both of which vary with radial location, but we cannot measure the voltage in the tiny extracellular spaces, so we report only the intracellular voltage with respect to the bathing solution. Impedance studies at each depth were made to determine surface cell membrane capacitance (C_S) and membrane conductance (G_S), the fiber cell membrane capacitance (c_m) and conductance (g_m), and the effective extracellular resistivity (R_e) between fiber cells were calculated based on an equivalent circuit model (Mathias et al., 1981) using the equations below. Note that R_S(r) depends on radial position but not the sinusoidal frequency of the injected current, whereas Z_L(jω) is a property of membrane conductances and capacitances of all lens cells and varies with sinusoidal frequency but not location (see Figure 6 of Mathias et al., 2010, for typical data).

The surface of fiber cell membrane per unit volume of tissue is SmVT=6000⁢cm-1, and jω is the Fourier transform complex variable representing sinusoidal frequency.

Fresh lenses were dissected and collected from 6-week-old mice and stored at −80°C until homogenization and protein extraction as previously described (Cheng et al., 2018). Briefly, two lenses from each mouse were pooled into one protein sample, and there were at least three mice used for each experiment. Using a glass Dounce homogenizer, lenses were homogenized on ice in 250 μl of lens homogenization buffer [20 mM Tris, pH 7.4 at 4°C, 100 mM NaCl, 1 mM MgCl₂, 2 mM EGTA, and 10 mM NaF with 1 mM DTT, 1:100 Protease Inhibitor Cocktail (P8430, Sigma-Aldrich) and 1 tablet of PhosStop per 10 ml buffer (04906845001, Roche) added on the day of the experiment] per 10 mg of lens wet weight. Protein samples were briefly sonicated and boiled for 5 min. Protein concentration was determined using Quick Start™ Bradford 1× Dye Reagent (Bio-Rad, Hercules, CA, United States) using manufacturer instructions. Equal amounts of protein were loaded for capillary-based (Western [WES]) immunoassay using a 12–230 kDa separation module kit according to manufacturer’s instructions (Protein Simple, San Jose, CA, United States) as previously described (Parreno et al., 2020). For each antibody, prior optimization experiments were run to determine the optimal protein and antibody concentrations. To detect Cx46 and Cx50 protein levels, we loaded 1 mg/ml of protein extracts onto WES plates and probed with rabbit polyclonal anti-Cx46 antibody [α3J, intracellular loop, 1:100, a generous gift from Dr. Xiaohua Gong (University of California, Berkeley) (Gong et al., 1998)] or rabbit polyclonal anti-Cx50 antibody [C-terminal, 1:100, a generous gift from Dr. Jose M. Wolosin (Mount Sinai School of Medicine) (Chang et al., 2002)]. To detect aquaporin 0 (Aqp0) protein levels, 0.12 mg/mL of protein extract was loaded onto WES plates and probed with rabbit polyclonal anti-Aqp0 antibody (1:50, AQP01-A, Alpha Diagnostic International, San Antonio, TX, United States). A Total Protein Separation module kit (Protein Simple) was used to detect the total protein amount for normalization. The data from WES plates are presented in electropherogram format and as virtual blots. Virtual blots convert electropherogram data into images of band densities similar to images of standard Western blots. The area under the peak of detected proteins is calculated and normalized to the total protein amount. Mean, standard deviation, and statistical significance (Student T-test, two-tailed between KO samples and respective controls) were calculated using Excel and graphed using GraphPad Prism 9. The proprietary matrix and running buffer for capillary electrophoresis differ from traditional SDS-PAGE gels and can lead to small changes in the apparent molecular weight of the detected peaks. The amount of SDS in the proprietary sample buffer (Hjemel and Chrambach, 1981; Rath et al., 2009) and the composition of the unique biotinylated molecular weight marker used for capillary electrophoresis can also influence the shift in apparent molecular weight (Sallantin et al., 1990). We only observe a single peak in each capillary, and this peak is close to the expected size of the protein that is being probed.

Freshly enucleated eyes from 6-week-old mice were fixed and sectioned as previously described (Cheng et al., 2016, 2018). Briefly, eyes were fixed for 4 h in 1% paraformaldehyde at 4°C. Samples were cryoprotected in sucrose and frozen in OCT medium (Sakura Finetek, Torrance, CA, United States) for sectioning. Twelve-micron-thick sections were then permeabilized, blocked, and stained as previously described (Nowak et al., 2009; Nowak and Fowler, 2012) using rabbit anti-Cx46 (1:200), rabbit anti-Cx50 (1:100), or rabbit anti-AQP0 (1:100) antibodies. Aqp0 detection with this antibody has been demonstrated previously in rat and mouse lens frozen sections with very low signal in the epithelium and high signal in the lens fibers (Grey et al., 2009; Petrova et al., 2015). Our immunostaining protocol may be causing slightly increased Aqp0 signal in lens epithelial cells compared to previous reports. Secondary antibodies were Alexa-488-conjugated donkey-anti-rabbit (1:200, 711-545-152, Jackson ImmunoResearch Laboratories, West Grove, PA, United States) or Alexa-647-conjugated donkey-anti-rabbit (1:200, 711-605-152, Jackson ImmunoResearch). Rhodamine phalloidin (1:100; Thermo Fisher Scientific, Waltham, MA, United States) was used to stain F-actin, and slides were mounted using Vectashield with DAPI (nuclei, Vector Laboratories, Burlingame, CA, United States). Images were collected using a Zeiss LSM800 confocal microscope. The epithelial cell thickness was used to identify cross-sectional tissue sections near the lens equator (Nowak et al., 2009; Cheng et al., 2016, 2018). Staining was repeated on at least 3 samples from different mice for each genotype, and representative data are shown. Images in the figures are from the same section going from peripheral to mature fiber cells, and these images were captured from neighboring depths without gaps or overlap using the motorized stage of the microscope for accuracy.

Aqp0 puncta/aggregate mean and max area measurements were conducted on 50 μm × 50 μm regions of interest (ROI) in the peripheral fiber cells (adjacent to the epithelial cell layer) and in the mature fibers (∼350 μm from the epithelial cell layer). Grayscale images of the Aqp0 staining signal were analyzed by Volocity 6.5 (Quorum Technologies, Puslinch, ON, Canada) using the Find Object function (threshold 10, minimum object size 0.2 μm²) and Separate Touching Objects module (object size guide 0.5 μm²). Mean and maximum Aqp0 puncta/aggregate areas were obtained from three different sections of each genotype. Mean, standard deviation, and statistical significance (Student T-test, two-tailed between KO samples and respective controls) were calculated and graphed using GraphPad Prism 9.

Aqp0 staining intensity measurements were conducted on 50 μm × 50 μm ROI in differentiating fiber cells (∼250 μm from the epithelial cell layer). Grayscale images of the Aqp0 signal were used, and mean gray values at the short and long sides of six cells were analyzed and averaged on each section using FIJI software (Schindelin et al., 2012). The average intensity ratio of the Aqp0 signal for each section is reported in the graphs. Mean, standard deviation, and statistical significance (Student T-test, two-tailed between KO samples and respective controls) were calculated and graphed using GraphPad Prism 9.

While the mechanism for cataracts in humans with EphA2 mutations remain unknown (Shiels et al., 2008; Tan et al., 2011; Masoodi et al., 2012; Sundaresan et al., 2012), changes in ion/fluid homeostasis due to disruption of connexins (GJs) or aquaporins (water channels) can lead to nuclear cataracts (Gong et al., 1998; White et al., 1998; Rong et al., 2002; Wang et al., 2009; Gao et al., 2013; Kumari et al., 2017). Since mild nuclear cataracts are often present in EphA2^–/– lenses, we evaluated the localization of Cx46, Cx50, and Aqp0 by performing immunostaining in 6-week-old control, ephrin-A5^–/– and EphA2^–/– lens frozen sections. Ephrin-A5^–/– lenses were also evaluated in this study as a comparison for EphA2^–/– lenses and to learn more about whether these two molecules interact in subpopulations of cells in the lens. While Cx46 staining patterns were not significantly different between control and KO lens sections (Figure 1), immunostaining for Cx50 demonstrates that GJ plaques are reduced in peripheral EphA2^–/– fibers (Figure 2, pound signs and enlargements in dashed boxes). The Cx50 staining signal defect is restricted to peripheral EphA2^–/– lens fibers (Supplementary Figure 1). The normal F-actin staining signal in the EphA2^–/– lens section reveals that the change in the Cx50 signal in the periphery is not due to a sectioning or imaging artifact. The Cx50 staining signal was similar in other regions of the lens fibers of control and EphA2^–/– lenses. These defects were not detected in ephrin-A5^–/– lens sections (Figure 2).

Immunostaining for Aqp0 reveals that the water channel protein is abnormally aggregated in EphA2^–/– peripheral lens fibers (Figure 3, pound signs). There were also some larger Aqp0 puncta in the mature EphA2^–/– fiber cells. Quantification of the mean and maximum area of Aqp0 puncta/aggregates in peripheral fiber cells revealed that Aqp0 puncta/aggregates were increased in size (mean and max area) in EphA2^–/– lenses (Figure 4A). A similar analysis of mature fiber cells showed that Aqp0 puncta/aggregates had increased max area in EphA2^–/– lens sections (Figure 4B). Quantification did not reveal any changes in Aqp0 puncta in peripheral and mature ephrin-A5^–/– lens fibers. Interestingly, in mature EphA2^–/– and ephrin-A5^–/– lens fibers, Aqp0 is distributed evenly around the membrane unlike in control fibers where Aqp0 is enriched on the short sides of mature cells (Figure 3, asterisks and enlargements in dashed boxes). Aqp0 staining intensity (mean gray value) was measured and ratioed between the short and long sides of differentiating fibers in EphA2^+/+, ephrin-A5^+/+, and ephrin-A5^–/– lens sections (Figure 5). The Aqp0 intensity ratio could not be determined in EphA2^–/– lens fibers due to the irregular shape, leading to difficulties in reliably identifying the short vs. long sides of the cells. The ratio reveals that Aqp0 signals were more similar between the short and long sides of ephrin-A5^–/– differentiating lens fibers compared to that in control cells, suggesting a change in Aqp0 distribution around the cell membrane in KO lens fibers. These data suggest cell-cell adhesion defects and possibly an imbalance in lens fluid and ion transport in EphA2^–/– and ephrin-A5^–/– lenses.

Since Cx50 and Aqp0 localization is abnormal in EphA2^–/– lenses and Aqp0 distribution is also affected in ephrin-A5^–/– lenses, we performed quantitative Westerns of total lens homogenates from 6-week-old control, EphA2^–/– and ephrin-A5^–/– mice using the ProteinSimple WES capillary electrophoresis system. We showed the virtual lane view that is similar to traditional Western blot scans and is a representation of chemiluminescence peaks from each protein of interest and the total protein used for data normalization. The graphed electropherograms representing chemiluminescence peaks for each target protein were quantified for dot plots of protein amount normalized to total protein peaks. We did not detect any changes in Cx46, Cx50, or Aqp0 protein levels between control, ephrin-A5^–/– and EphA2^–/– lenses (Figure 6). These data suggest that Eph-ephrin signaling does not affect the overall amount of connexin proteins but may influence membrane localization and subcellular distribution of Cx50 and Aqp0.

Since GJ plaque size and localization can affect fiber cell coupling and fluid and ion homeostasis (Cheng et al., 2015), we measured GJ coupling in 3- and 12-month-old control, ephrin-A5^–/– and EphA2^–/– lenses. Surprisingly, despite the change in Cx50 localization in peripheral fiber cells, there was no change in GJ conductance between lenses from 3-month-old control and EphA2^–/– mice (Figure 7 and Table 1). We also did not detect any changes in GJ conductance between lenses from 3-month-old control and ephrin-A5^–/– mice, consistent with our immunostaining and Western blotting data. A previous report showed that EphA2^–/– lenses develop age-related cortical cataracts that progress to whole cataracts and lens rupture (Jun et al., 2009). Thus, we also tested whether there were any GJ conductance changes in lenses from 12-month-old control, ephrin-A5^–/– and EphA2^–/– mice to determine whether GJ conductance changes could account for the age-related cataracts in EphA2^–/– lenses. We found no change in the GJ conductance between lenses from 12-month-old control and KO mice.

We also determined the intracellular voltage (ψ_i) across control and ephrin-A5^–/– and EphA2^–/– lenses. In lenses from 3-month-old, ephrin-A5^–/– and EphA2^–/– mice, ψ_i is significantly more negative when compared to littermate controls (Figure 8 and Table 2), indicating that the KO lenses are hyperpolarized compared to control lenses. In lenses from 12-month-old control, ephrin-A5^–/– and EphA2^–/– mice, the values of ψ_i are very similar, showing that the membrane voltage defect is only present in lenses from young KO mice. We also measured the capacitance and conductance of surface and inner fiber cells as well as the effective extracellular and intracellular resistivity of fibers in lenses from 3- and 12-month-old control and KO mice. We did not find any statistical difference in surface or inner fiber capacitance or extracellular resistivity between control and KO lenses (data not shown), and those data were consistent with previous reports (Mathias et al., 1981; Martinez-Wittinghan et al., 2004). We did not observe any changes in conductance of surface cells (G_s) in lenses from 3-month-old control, ephrin-A5^–/– and EphA2^–/– mice or in lenses from 12-month-old control and ephrin-A5^–/– mice (Figure 9). In lenses from 12-month-old EphA2^–/– mice, G_S was increased compared to controls. Unexpectedly, loss of either EphA2 or ephrin-A5 significantly decreased the conductance of fiber cell membranes (g_m) in lenses from 3-month-old KO mice. The value of g_m was comparable between lenses from 12-month-old control and KO mice. Together these data suggest that while Eph-ephrin signaling does not affect GJ coupling between lens fibers, disruption of EphA2 or ephrin-A5 affects the conductance and intracellular voltage of fiber cells.