Beta-1 integrin is important for the structural maintenance and homeostasis of differentiating fiber cells

2.1 Animals

All animal experiments described here conform to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Itgb1^tm1Efu/J mice, which harbor an allele of the β1 integrin gene in which exon 3 is flanked by LoxP sites (Raghavan et al., 2000), were obtained from The Jackson Laboratory (Bar Harbor, Maine) and genotyped as described (Simirskii, Wang, 2007). MLR39-cre mice on a FVB/N background expressing Cre recombinase in LFCs from E12.5 onward (Zhao et al., 2004) were obtained from Michael L. Robinson (Miami University, Oxford, Ohio) and genotyped as described (Zhao, Yang, 2004). Itgb1^tm1Efu/J mice were bred to MLR39 mice to create mice lacking β1 integrin in lens fibers (cKO) (Simirskii et al., 2013). All conditional knockout mice are mutant at the CP49 locus consistent with the genetic background of the founder strains of these mice (Alizadeh et al., 2004, Sandilands et al., 2003, Simirskii et al., 2006). The wild type mice used in these studies were of either the CP49 + C57Bl/6<Har> strain or the CP49 – FVBN strain (designated in the figure legends) (Simirskii, Lee, 2006). All mice were maintained under specific pathogen free conditions at the University of Delaware animal facility under a 14/10-hour light/dark cycle. Embryos were staged by designating the appearance of a vaginal plug in the dam as 0.5 days embryonic (E). All experiments were done using closely age matched experimental and control animals; the ages of the animals used in each experiment are noted in the figure legends.

2.2 Gross Morphology

Eyes were enucleated, lenses dissected, and placed into pre-warmed tissue culture Medium 199, 1X (with Earle's salts & L-glutamine) (Cellgro, Mediatech, Inc. Manassas, VA). Transparency was assessed by photographing lenses using a Zeiss Stemi SV dissecting microscope fitted with a darkfield base. The refractive properties of lenses were assessed by placing them on a 200 mesh electron microscopy grid then photographing them using brightfield with a dissecting microscope, optimizing focus upon the grid, as previously described (Shiels et al, 2007).

2.3 Histology

Eyes (postnatal mice) were removed immediately after sacrifice and fixed in Pen-Fix (Richard Allan Scientific, Kalamazoo, Michigan) for 2 hours. Samples were then transferred into 70% ethanol and paraffin embedded. Six-micron thick sections were prepared, stained by hematoxylin and eosin (H&E) and imaged on a Zeiss Axiophot fitted with a digital Nikon camera.

2.4 Immunofluorescence

Fluorescent immunolocalization on longitudinal eye sections was done as previously described (Reed et al., 2001). Briefly, eyes were removed, embedded directly in Optimum Cutting Temperature media (OCT, Tissue Tek, Torrance, California), 16μm thick sections created on a Leica cryostat (Leica) and mounted on slides (ColorFrost Plus, Fisher Scientific Hampton, New Hampshire). Slides were immersion fixed in 1:1 acetone-methanol at −20°C for 15 minutes or 4% paraformaldehyde for 30 minutes depending on the antibody, and blocked in 1% BSA (Bovine Serum Albumin, Sigma Aldrich) for one hour at room temperature. Sections were then incubated with primary antibody diluted with blocking buffer under conditions described in table one. Sections were washed in either 1X PBS or 1X TBS depending on antibody, and incubated with the appropriate Alexa Fluor 568 or Alexa Fluor 488 labeled secondary antibody (Invitrogen, Grand Island, NY) at a dilution of 1:200 and the DRAQ-5 nucleic acid stain (Biostatus Limited, Shepshed, Leicestershire, UK) at a dilution 1:2000 in either 1X PBS or 1X TBS as appropriate. Sections were washed again in 1X PBS or 1X TBS and then mounted in mounting media (10 ml of PBS with 100 mg of p-phenylenediamine to 90 ml of glycerol; final pH 8.0).

Equatorial sections were stained similarly, except that the sections were 20 μm thick and a 1:200 dilution of Alexa Fluor 488 labeled wheat germ agglutinin (Invitrogen, Grand Island, NY), which binds N-acetyl-D-glucosamine and sialic acids, was added to the secondary antibody solution to visualize cell membranes (Ohno et al., 1986).

2.5 F-actin visualization

For whole mount analyses, lenses were dissected from the eye, fixed for 2 hours in 4% paraformaldehyde then washed in 1X PBS with 0.1% Triton X-100. Samples were stained in 1% BSA in 1X PBS with 0.25% Triton X-100, a 1:2000 dilution of DRAQ5 (Biostatus Limited, Shepshed, Leicestershire, UK), and a 1:200 dilution of Alexa Fluor 568 labeled phalloidin overnight at 4°C. Lenses were then washed in 1X PBS with 0.1% Triton X-100 and stored in the dark at 4°C until imaged.

Twenty micrometer thick equatorial sections were fixed in 4% paraformaldehyde for 30 minutes and blocked in 1% BSA for one hour at room temperature. Sections were then incubated with Alexa Fluor 568 labeled phalloidin (Invitrogen, Grand Island, NY) at a 1:200 dilution for 1 hour at room temperature then washed in 1X PBS. Sections were counterstained in a solution containing a 1:200 dilution of Alexa Fluor 488 labeled wheat germ agglutinin (Invitrogen, Grand Island, NY) and a 1:2000 dilution of the nucleic acid stain DRAQ5 for 1 hour at room temperature. Sections were washed again in 1X PBS, then mounted.

For β1 integrin and F-actin co-staining sixteen micrometer thick longitudinal sections were fixed in 4% paraformaldehyde for 30 minutes and blocked in 1% BSA for one hour at room temperature. Sections were incubated with primary β1 integrin antibody (described in table one). Sections were washed in 1X PBS then incubated with appropriate Alexa Fluor 488 antibody (1:200), Alexa Fluor 568 labeled phalloidin (1:200), and DRAQ 5 (1:2000). Sections were washed again in 1X PBS, then mounted.

2.6 Confocal image collection and analysis

Zeiss LSM780 upright, Zeiss LSM510 inverted and a Zeiss LSM5 DUO inverted confocal microscopes (Carl Zeiss Inc., Jena, Germany) were used in this study. All samples (wild type and conditional knockout) for a particular experiment were imaged using identical conditions on the same instrument. In some cases, brightness and/or contrast of images were adjusted in Adobe Photoshop for optimum viewing on diverse computer screens, however, in all cases, such adjustments were applied equally to both experimental and control images to retain the validity of comparison.

F-actin area was determined using a customized measurement protocol to find objects in Volocity 5.0.3. Volocity uses Otsu's method to perform image thresholding to exclude low intensity noise. Additional extraneous noise was removed by excluding objects at or below the size of a single pixel (0.05 μm²). The mean area of signal above the threshold (μm²) was obtained from the remaining objects found in the image and used as a measure of the amount of F-actin present. Results are presented as the mean of the sum of the area (μm²) ± the standard error of the mean and were statistically analyzed by a Student's t-test.

2.7 Super Resolution Structured Illumination Microscopy

Slides were prepared using the immunofluoresence protocols described above for equatorial sections. A Zeiss Elyra PS1 (Carl Zeiss Inc., Jena, Germany) equipped with a Plan-Apochromat 63x/1.4 NA Oil objective lens was used in this study. Image stacks were typically taken at 0.101μm z-interval with 25 images per plane (5 phases, 5 angles) and a total z distance of 5 – 15μm. Raw images were processed and reconstructed to obtain high-resolution structures and channels were aligned using ZEN software (Carl Zeiss Microimaging) based on structured illumination algorithms. Reconstructed super resolution images of single z-sections are represented in the figures.

2.8 Scanning Electron Microscopy

Eyes were immersion fixed in 0.08M sodium cacodylate buffer, pH 7.4 (Electron Microscopy Sciences, Hatfield, PA), 1.25% glutaraldehyde (Electron Microscopy Sciences), and 1% paraformaldehyde (Electron Microscopy Sciences) for five hours, the lens excised, then transferred to fresh fixative for an additional 48 hours. After fixation, the lenses were washed in 1X PBS, the lens capsule was removed from the anterior pole to the posterior pole, and around half the circumference of the lens. The superficial fiber cell layers were removed with fine forceps to visualize the cortical fiber cells at reproducible depths by measuring the distance from the lens capsule to the layer being imaged. Peeled lenses were transferred to an ethanol dehydration series (25%, 50%, 75%, 100%) followed by one overnight, and two 2.5-hour, incubations in fresh 100% ethanol. Peeled lenses were critical point dried using hexamethyldisilazane (HMDS) (Electron Microscopy Sciences) as previously described (Duncan et al., 2000a). Samples were mounted on aluminum stubs, coated with gold/palladium for 2.5 minutes and viewed with a Hitachi S-4700 Field Emission Scanning Electron Microscope (FESEM) (Tokyo, Japan).

2.9 Impedance Measurements

Impedance measurements were taken in both WT and cKO lenses as described previously (Mathias et al., 1981). Briefly, mice were killed by intraperitoneal injection of sodium pentobarbital solution (57.1mg/kg of weight), lenses excised and mounted on a silicone base (Sylgard; Dow Corning, Midland, MI). A current microelectrode was placed in the central fiber cells and used to inject a stochastic current of selected bandwidth while a second electrode recorded the induced voltage at varying depths. Impedance was calculated in real time by a fast Fourier analyzer (model 5420A; Hewlett-Packard Co., Palo Alto, CA). Membrane conductances and the effective extracellular resistivity were determined by curve fitting the impedance data to a model for the lens. Results are presented as mean ± standard deviation and differences between groups were assessed by a Student's t-test. The series resistance at several depths was recorded in each of several lenses by advancing the voltage electrode towards the center of the lens. Data were pooled and fit with a structurally based model to determine the gap junction coupling conductance as previously described (Shi et al., 2011). The mean value of the data at each location is based on the best fit of the structurally based model, whereas the standard deviations can be estimated from the scatter of the data from the mean at each location.

2.10 Pressure Analysis

Intracellular hydrostatic pressure measurements were taken in both WT and cKO lenses using a microelectrode/manometer based system as described previously (Gao et al., 2011). Results are presented as raw pressure data recorded at several depths in each of several lenses from WT and cKO mice. The data were pooled and mean pressure at each location was estimated by the best fit of a structurally based model, whereas the standard deviations can be seen by the scatter of the data from the curve fit.

3.1 β1-integrin F/F MLR39 CRE+ mice have varying phenotypes

β1-integrin F/F MLR39 CRE+ mice (β1-integrin cKO mice) generated from our initial crosses resulted in two distinct phenotypes where lenses either exhibited an obvious cataract (Figure 1B) or were grossly normal by darkfield microscopy (Figure 1C). In contrast, the MLR39 Cre mice (unpublished observation and Robert deIongh personal communication) and the Itgb1m1Efu/J mice carrying the floxed allele (supplemental figure one) which were used to create these β1-integrin cKOs never developed obvious opacities at the ages studied, similar to what was observed for mice on the FVB/N and C57Bl/6 genetic backgrounds (supplemental figure one). Hematoxylin and eosin staining of paraffin sections from cataractous lenses demonstrated profound defects in fiber cell organization including the formation of large cortical vacuoles (Figure 1D). However, transparent lenses from β1-integrin cKO mice show no morphological defects at the light level (Figure 1E). However, the lenses of cKO mice that were grossly transparent by darkfield analysis show slightly more haze and refractive defects when placed on a 200 mesh EM grid (Figure 1G) compared to a wild type lens (Figure 1F).

Immunolocalization of β1-integrin in adult lenses revealed that β1-integrin expression was maintained in the lens epithelium of β1-integrin cKO mice (Figure 1H) at levels similar to that of wildtype lenses (Figure 1G), but β1-integrin protein was absent from cKO lens fibers (Figure 1H) consistent with the lens fiber cell specific expression of the MLR39 cre transgene (Zhao, Yang, 2004).

3.2 Time course of β1-integrin expression in the embryonic lens and its loss from β1-integrin cKO lens fiber cells

At E12.5, wild type mice express β1-integrin protein in both lens epithelial and fiber cells with levels decreasing, but still detectable, in lens fibers as development proceeds (Figure 2A, C, E, E`) consistent with prior reports (Menko and Philip, 1995, Simirskii, Wang, 2007). The cKO lenses still express β1-integrin protein in all cells at E12.5 (Figure 2B), although by E13.5, levels have decreased (Figure 2D) relative to wild type (Figure 2C) in the central lens fiber cells. At E16.5, β1-integrin is still detected in the newly forming cortical fibers of cKO lenses (Figure 2F, F`), although the protein is lost from these cells as they mature. While in some cases the protein expression of other integrins can upregulate when an integrin is lost (Walker and Menko, 2009), this is not seen here as two other β integrins are down regulated in the cKO lenses (Supplemental Figure 2). We do not believe this downregulation of β5 and β8 integrins, which function as obligate heterodimers with αV-integrin (Mamuya and Duncan, 2011), is functionally relevant to the phenotype since loss of αV-containing integrins from the lens does not affect lens morphology (Mamuya et al., 2014). However, it is possible that the phenotype observed in these cKOs arises from the simultaneous loss of β1-integrins and downregulation of αV integrins in lens fiber cells.

3.4 β1-integrin cKO fiber cells have abnormal morphology

Since β1-integrin is found on the lateral membranes of LFCs, we then assessed the morphology of these cells by scanning electron microscopy. The cortical fibers (150– 350 μm from the lens capsule) of wild type mice are organized in discrete parallel layers, which interdigitate with their neighbors via elaborate membrane protrusions (Figure 3A, D, G). In contrast, clear lenses lacking β1-integrin from the lens fibers (cat-; Figures 3B, E, H) have disorganized fiber cell packing, and a loss of organized repeatable membrane protrusions. Cataractous lenses lacking β1-integrin from lens fibers have profoundly abnormal fiber cell structure with no clear demarcation between fiber cells (cat+; Figure 3C, F, I).

In the lens nucleus (analyzed at a distance of 550μm –750μm from the lens capsule), wild type lens fibers (Figure 4A) normally undergo structural rearrangements including the attenuation of membrane protrusions (Figure 4D) and the formation of membrane furrows (Figure 4G) on the previously smooth broad faces. In contrast, both transparent and cataractous lenses lacking β1-integrin from the lens fibers do not have clearly distinguishable LFCs (Figure 4B, C). Instead, in both phenotypes, the tissue has become amorphous with the only obvious feature being the development of a highly organized lattice on the fiber cell membrane (Figure 4E, F, H, I) that appears to be an extreme version of the membrane furrows observed in wild type nuclear lens fibers (Figure 4G). Overall, the morphology of cataractous and transparent lenses lacking β1-integrin from the lens fibers is quite similar, although cataractous lenses have more extreme changes. Unfortunately, the mice with the cataractous phenotype did not breed well and this phenotype was lost from the colony after a couple of generations. Thus, the remainder of this investigation describes findings obtained in cKO mice with grossly transparent lenses.

3.5 β1-integrin cKO inner cortical fiber cells do not maintain an actin cytoskeleton network

Since many lens fiber cell membrane proteins are differentially localized between the long and short faces of the hexagonal fiber cell (Beebe et al., 2001, Maddala et al., 2011b, Nowak et al., 2009, Song et al., 2009, Straub et al., 2003), β1-integrin expression in adult wild type lens fibers was determined by analyzing lens fiber cell cross sections (equatorial view). β1-integrin is found on both the long and short sides of the hexagonal lens fiber cell with no apparent preference in localization (Figure 5B, E) although the staining is brighter in the outer lens cortex compared to the inner (Figure 5B). This staining is absent in all fibers from cKO lenses (data not shown) both confirming the antibody specificity and that the β1-integrin protein is lost from these cells.

The cytoplasmic tails of integrins bind to a host of different proteins, many of which are involved in cytoskeletal organization, including F-actin polymerization (Lo et al., 1997a, Maddala et al., 2011a). Thus, intact whole lenses were stained with phalloidin and imaged on a confocal microscope to visualize F-actin. In wild type lenses, F-actin was most prominent in the lens epithelium and the apical tips of fiber cells, while lower levels are seen along the lateral membranes of fiber cells (Figure 6A, B, C). In cKO lenses, F-actin levels were reduced in cortical fiber cells, although further from the lens surface, some patchy regions of F-actin are maintained (Figure 6D, E, F). The timing of F-actin loss compared to the loss of β1-integrin protein during cortical fiber cell differentiation was assessed by co-labeling sections for β1-integrin (Figure 6G, J) and F-actin (Figure 6 H, K). This revealed that the F-actin staining is reduced concomitant with the loss of β1-integrin protein from newly forming fibers (Figure 6L). The near complete loss of F-actin in ICFs (Figure 7E) is apparent by the time lens fiber cell organelle loss has occurred (Figure 6J, K, L).

This observation was confirmed by staining equatorial lens sections with phalloidin. LFCs are normally packed into organized columns and exhibit a hexagonal cross section characterized by two long and four short sides (Figure 7A–C) and as previously reported, F-actin is localized more strongly to the short sides and vertices of the hexagonal fiber cell cross section (Figure 7B) (Nowak, Fischer, 2009, Straub, Boda, 2003). In contrast, the cortical fibers of cKO mice exhibit disorganization of fiber cell packing (Figure 7D), and the inner cortical fibers (ICFs) do not stain appreciably with phalloidin, consistent with a loss of the F-actin network (Figure 7E). Quantification of the phalloidin intensity profiles revealed peaks of F-actin along the short sides and vertices of wild type fiber cell membranes (Figure 7G), while this is absent from cKO lens fibers (Figure 7H). Further, the overall ICF F-actin network was significantly reduced in cKO lenses lacking β1-integrin in the fiber cells compared to wild type (Figure 7I). However, this is not due to loss of β-actin expression since similar levels normalized to total protein were detected in wildtype and cKO lenses (data not shown).

In other tissues, integrins regulate F-actin polymerization via interactions between their cytoplasmic tails and numerous adaptor proteins (Cammas et al., 2012, Lo, 2006). Integrin-linked kinase (ILK) is one of these integrin-binding proteins, which is expressed at relatively low levels in the lens epithelium, but upregulates upon fiber cell differentiation localizing along the fiber cell membrane with expression decreasing as these outer cortical fibers (OCFs) mature to ICFs (Figure 8A)(Cammas, Wolfe, 2012, Weaver et al., 2007). In cKO lenses, ILK levels follow a similar pattern in fiber cells, although as fibers mature, the staining becomes more diffuse suggesting a change in localization away from the membrane (Figure 8B). Using a super resolution approach via structured illumination microscopy (SIM), ILK immunoreactivity was seen to decorate WT OCFs in a punctate pattern associated with both the long and short sides of the fiber cell membranes (Figure 8D, d), while ILK is found in both the cytoplasm and at some membranes in cKO OCFs (Figure 8G, g; H, h). ILK localization in cKO lens fibers with a moderate phenotype is shown here, similar results were seen in fibers exhibiting both milder and more severe morphological defects (Supplemental figure 3)

3.6 β1-integrin cKO lens fiber cells have abnormal gap junction function and water permeability

The data above show that the lenses from cKO mice have abnormalities in fiber cell packing and loss of membrane protrusions in the cortical fibers while the nuclear fibers exhibit increased furrowed membranes. In order to assess the consequences of these morphological changes on lens physiology, the impedance and intracellular hydrostatic pressure of cKO lenses were compared to wild type. Unexpectedly, the gap junction-coupling conductance of the cKO lenses is almost two fold higher in both the differentiating (G_DF) and mature fibers (G_MF) compared to WT (Table 2). These conductances were derived from the series resistance data (Figure 9A), which are lower in cKO lenses than WT. In addition, the intracellular hydrostatic pressure is lower in cKO lenses compared to the WT (Figure 9B), as would be expected based on the higher gap junction coupling of cKO lenses. The effective extracellular resistivity (R_e), which depends on the volume fraction of the extracellular clefts, is higher in the WT than cKO lenses. These profound changes in cellular physiology may result in a defect in fluid homeostasis as cKO lenses were slightly, but significantly larger (0.0076 ± 0.0004g) than wild type lenses (0.0071 ± 0.0004g) (p-value = 0.02).

In other systems, deletion of β1-integrin is associated with increased connexin (Cx) expression (Czyz et al., 2005) however, there was no convincing change in the levels of either Cx46 or Cx50 in cKO lenses (Figure 9C–N). Structured illumination microscopy revealed that Cx46 is found mainly on the broad sides of the fiber cell membranes with lower levels at the vertices in wild type lenses (Figure 9D, d). While a similar pattern of Cx46 distribution was seen in cKO lenses, the gap junctional plaques on the broad sides may be qualitatively larger than normal (Figure 9G, g). Cx46 localization in cKO lens fibers with a moderate phenotype is shown here, similar results were seen in fibers exhibiting both milder and more severe morphological defects (Supplemental figure 4). Cx50 is also found mainly on the broad sides of the fiber cell membranes of wild type lenses although some protein was randomly scattered along the short sides as well (Figure 9J, j). Notably, in cKO lenses, Cx50 was strongly detected on both short and long sides of the fiber cell membranes, and the overall size of each gap junctional plaque may be qualitatively larger as well (Figure 9M, m). Cx50 localization in cKO lens fibers with a moderate phenotype is shown here, similar results were seen in fibers exhibiting more severe morphological defects (Supplemental figure 5)