Periaxin is required for hexagonal geometry and membrane organization of mature lens fibers

Mice

All the mice used in this study were from a C57BL/6 background. Embryonic day 10.5, 11.5, 12.5, 14.5 and neonatal (d1) eyes were used to assess developmental regulation of Prx expression in the lens. Pregnant mice at specific gestational ages were sacrificed by pentobarbital injection, and fetuses removed by hysterectomy for tissue fixation of whole heads (Maddala et al., 2008). To obtain neonatal lenses, new born mice were sacrificed, whole eyes dissected out and incised at the posterior poles at the optic stalk, followed by immediate immersion in the respective fixative. Prx homozygous null mice were generated and characterized in the Brophy laboratory and genotyped as described earlier (Gillespie et al., 2000). These mouse lenses were confirmed to express BFSP2 which is known to be deleted in the mouse strain 129 (Sandilands et al., 2004). All animal procedures were conducted in accordance with the Duke University and University of Edinburgh institutional animal care guidelines and protocols.

Lens morphology by light and electron microscopy

Freshly enucleated eyes from postnatal (P24 and P90) Prx null and age-matched WT mice were fixed with 4% paraformaldehyde and 2.5% glutaraldehyde in 80 mM Na cacodylate buffer, pH 7.3, for 2 h at room temperature (RT), transferred to 2.5% glutaraldehyde, and stored in 10% buffered formalin at RT until further processing. The specimens were subsequently embedded in paraffin, and 5µm sections (equatorial plane) were cut and stained with hematoxylin and eosin. Micrographs were captured using Zeiss Axioplan2 microscopy (Zeiss Inc. Thornwood, NY. USA).

For transmission electron microscopy (TEM), buffered formalin (10%) fixed specimens were processed using a standard TEM protocol. Briefly, the tissue specimens were treated with 1% Osmium tetraoxide in 0.1 M Na cacodylate buffer, pH 7.2 at 4 °C for 1h. After rinsing with water and ethanol, the tissue specimens were treated with 2% uranyl acetate for 1 h at RT. Samples were then dehydrated through a series of ethanol grades (25% to 100%). Following this, the tissue sections were infiltrated with 1: 1 propylene oxide and spurr resin (SPI supplies), and exchanged with pure spurr and embedded at 65 °C overnight. Tissue specimens were cut (0.7 µ) in the equatorial plane using a Reichert Ultra-cut microtome (Leica), followed by ultra-thin sections (65 nm). Tissue sections were treated with 2% uranyl acetate and rinsed with 25% ethanol and water, and treated with 3.5% lead citrate. Images were captured at 8000x with a Jeol JEM-1400 equipped with Orius CCD digital camera at 60kV (JEOL, Tokyo, Japan).

Immunogold labeling

For aquaporin-0 immunogold labeling, freshly enucleated mouse eyes (P90 and P200) from both age-matched WT and Prx KO were first fixed in 10% buffered formalin at RT until further processing. The tissue specimens were labeled with immunogold by a standard protocol. Briefly, the formalin fixed lens tissue specimens were fixed with 0.125% glutaraldehyde containing 1% tanic acid in 0.1 M Na cacodylate buffer at RT. Then treated with 2% uranyl acetate for 1 h, followed by sequential dehydration using 50, 70, and 80% ethanol. Specimens were then embedded with pure LR White (55 °C, overnight). The embedded specimens were cut (65 nm) using a Leica ultra-microtome (Leica). Thin sections were blocked with PBS containing 10% goat serum and 0.1% Triton X-100 (blocking buffer). Specimens were incubated overnight at 4 °C with aquaporin-0 polyclonal antibody. Sections were rinsed with 2% goat serum in PBS, and then blocked again with blocking buffer at RT. Following this, sections were incubated with anti-rabbit IgG secondary antibody conjugated with gold (6 nm, Electron Microscopy Sciences) at RT for 2 h, then rinsed with PBS containing 2% goat serum. Specimens were then post fixed with 2% glutaraldehyde in 0.1 M Na cacodylate buffer, washed extensively with water, and treated with freshly prepared silver enhancement solution (HQ silver kit, Nanoprobes). The reaction was stopped by rinsing with water and the specimens were stained with 2% uranyl acetate, followed with 3.5% lead citrate. Photographs were captured at 8000x with a Jeol JEM-1400 TEM equipped with Orius CCD digital camera.

Immunofluorescence analyses

For fluorescence immunostaining, eyes were fixed either for cryostat sectioning or paraffin sectioning. Cryostat sections (7 µm, sagittal plane) were obtained from embryonic heads (e10.5, e11.5, e12.5, e14.5) and neonatal eyes that were fixed in paraformaldehyde and sucrose until further processing (Maddala et al., 2008). The tissue sections were immunostained with polyclonal Prx antibody (Sigma-Aldrich) as we described earlier (Maddala et al., 2008) and images were captured with Nikon Eclipse 90i confocal laser scanning microscope (Nikon Instruments, Inc. Melville, NY. USA).

Immunofluorescence staining of paraffin embedded lens sections derived from Prx null and WT mice (P24, P90) were performed with specimens fixed in 10% formalin in PBS at RT. Five-micron tissue sections (equatorial plane) were de-paraffinized and rehydrated using xylene and absolute ethyl alcohol. Specimens were placed in the pre-heated antigen retrieval solution (0.1 M Na Citrate buffer pH 6.0) and heated with solution for 20 min at 100 °C. Then, the tissue sections were blocked for 10 min with the medical background Sniper reducing solution (Biocare Medical) in a humidified chamber. Primary antibodies were then applied (in 1% fatty acid free BSA in TBS), and incubated overnight at 4 °C in a humidified chamber. The following primary antibodies were used: NrCAM (polyclonal, Abcam), Prx (polyclonal, Sigma-Aldrich), aquaporin-0 (polyclonal, gift from Sam Zigler), β2-spectrin (monoclonal, BD Biosciences), Ankyrin B (monoclonal, gift from Vann Bennett), desmoyokin (monoclonal, Sigma-Aldrich), ezrin (monoclonal, Sigma-Aldrich), filensin (polyclonal, gift from Paul FitzGerald), N-cadherin (monoclonal, Invitrogen), β-catenin (monoclonal, Sigma-Aldrich), βIV-spectrin (polyclonal, gift from Masayuki Komada), contactin/F3 (monoclonal, Neuromab) and β-actin (monoclonal, Cell Signaling Technology). After incubation with primary antibodies, the slides were washed with TBS and incubated with either Alexa fluor 488 or 594 conjugated secondary antibodies (Invitrogen) in 1% BSA/TBS, for 2 h under dark conditions. Slides were rinsed with TBS and mounted using VectaMount and nail enamel, viewed and photographed using a Nikon Eclipse 90i confocal laser scanning microscope. For double-labeling, appropriate polyclonal and monoclonal antibodies were mixed and incubated with the tissue specimens as described above in conjunction with Alexa Fluor 488 and 594 conjugated second antibodies respectively (Invitrogen). For 3-D imaging (Fig. 2B, C (right panel), 2F (right panel)), images were acquired at RT using Zeiss LSM 510 inverted confocal microscope (Argon/2, 488; 561 nm Diode laser with motorized stage). Z Stacks (10 optical sections with step size 0.5 µm) with the 63×1.4 oil lens (Zeiss Plan Apochromat) were collected. Images were processed using Volocity 5.2.1 software and Adobe Photoshop.

Lens extracts

Immunoprecipitations

To identify and characterize Prx interacting proteins in lens tissue, immunoprecipitation (IP) assays were performed using a Prx polyclonal antibody (Sigma-Aldrich), dynabeads protein G (Invitrogen), and Sigma-Aldrich IP buffer. Mouse lenses (pooled from the neonatal, P18 and adults) were homogenized (2% w/v) in 1 × IP buffer containing protease inhibitors (Roche protease inhibitor tablet, one/10 ml 1× IP buffer) and phosphatase inhibitors (Roche PhosStop tablet, one/10 ml 1× IP buffer) using a Dounce glass homogenizer on ice. Homogenates were centrifuged at 800×g for 10 min at 4 °C, and the supernatant was collected and protein content was measured using Bio-Rad reagent and used for IP reactions. IP reactions were performed by following the Invitrogen manufacturer protocol. The Prx immunocomplexes with dynabeads were separated from the sample and washed with PBST. Then 40 µl of 2X Lamelli sample buffer was added to the beads and boiled, and the protein samples were separated on the gradient SDS-PAGE (4–12%; Criterion XT precast gel from Bio-Rad) and stained with Gelcode® Blue stain reagent (Pierce). After destaining with Milli Q pure water, protein bands were excised from the gel and further characterized by mass spectrometry. Gel slices containing the protein bands were subjected to in-gel tryptic digestion using the In-Gel Tryptic digestion kit (Pierce), per manufacturer’s instructions. This digestion process included both reduction and alkylation of protein samples.

Mass spectrometry

In-gel trypsin digested protein mixes were separated by reverse phase HPLC on the C18 capillary column and samples were analyzed on the MALDI-TOF-TOF 4700 Proteomic Analyzer (Applied Biosystems) using a selection of 10 precursors per spot in a mass range of 700–4000 Da and post-source decay for MS/MS acquisition as we described earlier (Maddala et al., 2007). Fragment peak lists were generated by GPS Explorer (Applied Biosystem Inc.) and searched against the NCBI Mouse database (May 2010 release) using mascot Server v2.2 with the following parameters: trypsin specificity allowing up to one missed cleavage, carbamidomethylation at Cys residues as fixed modification and protein N-termial acetylation and methionine oxidation as variable modifications. Precursor ion mass tolerance was set to 75 ppm and fragment mass tolerance was 0.5 Da. Precursors with masses corresponding to human keratin peptides were automatically excluded from the precursors lists generated by GPS after MS analysis. Acceptance criteria for protein identification required identification of at least two peptides for each protein with confidence interval percentage (CI%) over 99.90%, which corresponded to a false discovery rate of less than 1% in searching the NCBI mouse database.

Immunoblotting

To determine the changes in protein levels or detect the presence of given protein including Prx, ezrin, aquaporin 0, β-actin, β-spectrin, ankyrin B, NrCAM, and filensin in the immunoprecipitates or in the lens tissue or sciatic nerve fractions, equal volume of 2X Laemmli sample buffer was added and boiled for 5 min and were separated either on the gradient (4–12% or 4–15%) SDS-PAGE (Criterion, Bio-Rad) or on 12 or 8% SDS-PAGE casted gels, followed by electrophoretic transfer to nitrocellulose membrane (Bio-Rad) and immuno blots were developed as described earlier using appropriate polyclonal or monoclonal antibodies (Maddala et al., 2008). Immune positive bands were scanned using Fotodyne gel doc system, densitometrically quantified by imageJ software.

Quantifications

Statistical analysis

Where required the Student’s t-test was performed to determine significance of differences noted between the Prx KO and WT specimens using Sigma plot. Values are given as mean ± s.e.m.

Periaxin exhibits lens preferred expression and developmental regulation in the eye

As evident from immunofluorescence analysis in Fig. 1A, Prx is localized primarily in lens fibers, with minimal detection in the lens epithelium, but is undetectable in the retina, cornea and other ocular tissues in mice. Given that Prx expression was previously thought to be specific to myelinating SCs (Gillespie et al., 1994), and that very little is known regarding its function in the lens (Straub et al., 2003), we sought to gain insight into the potential role of Prx in lens morphogenesis and function. To this end, we first evaluated the expression pattern of Prx in mouse eyes from embryonic day 10.5, 11.5, 12.5, 14.5 and neonatal (P1) eyes by immunofluorescence microscopy, using tissue cryosections (Fig. 1B). Prx expression is detectable as early as E10.5 in the lens pit with discrete distribution to the posterior region. During the different developmental stages evaluated, Prx expression is robustly induced in, and preferentially localized to fiber cells of the ocular lens (Fig. 1B). Moreover, Prx protein levels increased during lens maturation, as determined by immunoblot analysis of equal amounts of a lens membrane enriched protein fraction derived from P1, P7, P14, P21 and P60 mice (Fig.1C a). In the same samples, however, the levels of ezrin decrease gradually over the same time course (Fig. 1C b).

Periaxin exhibits maturation dependent redistribution and clustering at the vertices of mature lens fibers

The cellular distribution of Prx in lens fibers was probed using immunofluorescence confocal microscopy analysis of paraffin embedded P24 mouse lens equatorial sections (Fig. 2). Prx distributes to the fiber cell plasma membrane along both the long and short arms of the hexagonal fiber cells of the outer cortex, which represents the differentiating and maturing region within the growing lens. However, as fiber cells mature, Prx distribution was found to become prominent at the short arms of hexagonal fiber cells in the inner cortical region, exhibiting a discrete, plaque-like staining pattern at the tricellular junctions (vertices), and alternating with a much reduced staining at the bicellular junctions (between vertices, both short and long arms) as determined by both single plane (Fig. 2A) and 3-D (Fig. 2B) confocal imaging. This pattern is prominent in P18 and older lenses. The en face views of Prx distribution at the broad (Fig. 2B, left panel) and short (Fig. 2B, right panel) sides of fiber cells based on 3-D images reveal clustered distribution organized in a perfect braid at the short sides (arrows). Since the staining of Prx and other proteins analyzed in the center (nucleus) of the lens was inconsistent due to the hardness of this region, our description is restricted to the cortical regions of the lens.

Additionally, Prx colocalizes with β-actin (Fig.2D) and exhibits partial colocalization with β-catenin (Fig. 2C, left: single plane, right: 3-D image) and N-cadherin (Fig. 2D) at the vertices of fiber cells as determined by confocal imaging. In contrast, at the long arms Prx was noted to colocalize with β-catenin and N-cadherin while β-actin was not detected in this region of mature fibers (Figs. 2C and D images represent the inner cortical lens region). In contrast to the codistribution of Prx and β-actin at vertices, filensin, a lens fiber specific intermediate filament protein (Ireland and Maisel, 1983; Merdes et al., 1991) required for lens architecture (Alizadeh et al., 2003), localizes to both the long and short arms of lens fibers, but displays a patch-like staining pattern at the center of the long arms of lens fibers (Fig. 2E, arrows). On the other hand, β2-spectrin, a major lens membrane skeleton protein (Lee et al., 2000), is localized to the fiber cell membrane in a pattern distinct from Prx, filensin and β-actin (Fig. 2F). β2-spectrin staining is detected throughout the short arm and approximately one-third the length of the long arm, but exhibits a much reduced staining intensity at the center of long arms (arrows) where filensin stains most strongly (Fig. 2E). At the vertices of the fiber cell, Prx colocalizes partially with β-spectrin (Fig. 2F, right panel: 3-D image). We conclude that this novel and unique spatial distribution of Prx, filensin, β-actin and β-spectrin to the tricellular and bicellular junctions of the prismoid lens fibers, as depicted in the schematic illustration in Figure 2G, is likely required for maintaining hexagonal shape and cell-cell interactions.

Evidence for Prx involvement in membrane cytoskeletal scaffolding

Since Prx is a known PDZ domain protein (Dytrych et al., 1998), we sought to explore whether Prx serves to mediate membrane scaffolding interactions in the lens by identifying Prx-interacting proteins in mouse lens tissue, using immunoprecipitation (IP) and mass spectrometry (MS) analyses (Fig. 3A). As compared to an IgG control, Prx antibody generated immunocomplexes contained several distinct proteins derived from the detergent suspended protein fraction (800×g supernatant) of whole mouse lenses (Fig. 3A). The Prx immunocomplexes were separated by gradient (4–12%) SDS-PAGE and subsequently identified by MALDI-TOF MS and MS/MS analyses. MS analysis identified several major protein constituents in the Prx immunocomplexes including, ankyrin B, spectrin (both α and β), plectin-1, Prx, desmoyokin, filensin, phakinin and vimentin (Fig. 3A). These results were confirmed in two independent experiments and the individual proteins were identified with high confidence (>99.5%) based on two or more peptides. While several minor constituents including aquaporin-0, connexin 50, synemin, lengsin, armadillo repeat gene related protein–ARVCF, catenin-α2, radixin, ezrin, actin, βb1-crystallin and microtubule-associated serine/threonine kinase (MAST) 2 and 4, were also detected in the Prx IPs, the identification of these proteins was based on either less than two peptides or with less than 99.5% confidence.

Immunoblot analysis of Prx immunocomplexes confirmed the presence of Prx, β-spectrin, Ankyrin B, NrCAM and filensin as shown in Fig. 3B. Ezrin was also confirmed to be present in the Prx immunocomplexes by Western blot (data not shown). Additionally, colocalization analysis revealed that Prx codistributes with desmoyokin and ezrin (Fig.3 C & D, respectively), and exhibits partial codistribution with Ankyrin B (Fig. 3E) at the vertices and long arms of fiber cells. In contrast to Prx (Fig. 2A), NrCAM was evenly distributed throughout the plasma membrane of fiber cells including both short and long arms (Fig. 3F).

Analysis of lens Prx immunocomplexes by gradient SDS-PAGE (4–12%) and MS consistently identified two distinct Prx proteins with molecular masses of ~160 kDa and ~180/190 kDa, with the latter being a prominent protein in the mouse lens (Fig. 3A). Both protein bands exhibited strong immunoreactivity with Prx polyclonal antibody (Fig. 3B). The predominant form of Prx present in the lens migrates at higher molecular mass relative to sciatic nerve Prx as determined by immunoblot analysis (Fig. 3G). A similar Prx protein profile has been reported for bovine lens homogenates (Straub et al., 2003).

Disrupted lens fiber cell shape and packing in the Prx KO mouse model

To understand the role of Prx in lens fiber cell shape and packing, we performed histological characterization of Prx knockout (KO) mouse lenses by light and electron microscopy analyses. The adult Prx KO mice appear to develop eyes of normal size with no obvious phenotype based on direct visual examination. Further, freshly dissected Prx KO lenses derived from 6–9 month old mice were indistinguishable from age-matched wild type (WT) with respect to lens transparency, as assessed by light microscopy. Since these mice develop severe coordination defects after 6–9 months (Gillespie et al., 2000) and were euthanized after this period, the status of lens transparency in older Prx KO lenses is not known. However, to determine the changes in integrity of lens cytoarchitecture in the Prx KO mouse, eyes obtained from P24 and P90 Prx KO and age-matched WT mice were processed for light microscopy-based histology analysis. As expected, no Prx expression was evident by immunofluorescence analysis of the Prx KO lenses (Fig. 4A). Based on hematoxylin and eosin staining, no notable differences were evident at the levels of fiber cell organization, differentiation and migration between the Prx KO and WT lenses (Fig. 4B; data for P24 not shown).

However, during the tissue sectioning of these specimens for histological analysis, the Prx lenses were found to be softer than the respective age-matched WT lenses. Moreover, immunofluorescence staining of paraffin embedded lens equatorial sections obtained from the Prx KO and WT mice (P90) for membrane and membrane-associated proteins including β2-spectrin (Fig. 4C), aquaporin-0 (Fig. 4D), filensin and β-actin revealed notable abnormalities in fiber cell shape and arrangement, especially in the inner cortical regions (data for filensin and β-actin were identical to spectrin and aquaporin-0, not shown). Although the changes in fiber cell shape and arrangement are only subtle in the outer cortical fibers (OCF), the difference was much more pronounced in the inner cortical fibers (ICF) which undergo extensive membrane remodeling during terminal differentiation and compaction (Kuszak, 2004). These changes were much more pronounced in the P90 Prx KO lenses (Fig. 4) compared to the P24 lenses (R.L., unpublished). Fiber cells in affected areas (arrows) in the P90 Prx KO lenses were, on an average, significantly wider (14.3 ± 0.21µm, n = 44) than age and region-matched WT lens fiber cells (6.4 ± 0.47 µm, n = 65; Fig. 4E). The lower panels in Fig. 4C and D show magnified areas (boxed) of inner cortical fibers from both WT and Prx KO lenses. Interestingly, this increase in fiber cell width did not impact all fiber cells uniformly and some fibers were smaller in width than WT (Fig.4 D, indicated with asterisks).

Evidence for disruption of membrane skeleton in Prx KO lenses

To explore whether the above described abnormalities of fiber cell shape and packing are associated with changes in levels of cytoskeletal and membrane proteins, we determined the levels of β2-spectrin, β-actin, filensin and aquaporin-0 in Prx KO lenses by immunoblotting analysis of soluble and membrane-rich protein fractions from P200 Prx KO and age-matched WT lenses. Since the lens defects associated with Prx KO mice show progression with increasing age, we used P200 lenses for these analyses. β2-Spectrin (73%), β-actin (70%) and aquaporin-0 (46%) protein levels were decreased significantly in the membrane-rich lens insoluble fraction of the Prx KO lenses compared to the WT lenses based on analysis of equal amounts of protein (Fig.5A & C). Filensin protein levels were elevated modestly (15%, Fig. 5C) in the membrane-rich insoluble fraction of the Prx KO mice. Additionally, we compared the total protein profiles of membrane rich insoluble fractions from the P200 Prx KO and WT lenses by SDS-PAGE analysis and Coomassie blue staining. This analysis (Fig. 5B) revealed comparable protein profiles between the Prx KO and WT lenses, with the exception of an extra protein band at ~37kDa species (Fig. 5B, arrow) in the membrane- rich fraction of the Prx KO lenses. This extra protein band in the Prx KO lenses was found to be diffuse, implying that it could be a product of proteolytic degradation, which is a common characteristic of early lens opacification (Truscott et al., 1990). The 56 kDa Coomassie blue stained protein in fiber cell insoluble fractions in Fig. 5A indicates a loading control. Additionally, we examined for changes in the distribution pattern of desmoyokin and ezrin in the Prx KO lenses by immunofluorescence analysis, since these proteins co-immunoprecipitate with Prx from normal lenses. Both desmoyokin and ezrin exhibited a distorted distribution pattern in 90 day-old Prx KO mouse lenses, with reduced staining noted at the short arms of fiber cells relative to age-matched WT lenses (Fig. 5D, E).

Disrupted plasma membrane organization in Prx KO mouse lens fibers

Additional support for the above described morphological (Fig. 4) and cytoskeletal changes (Fig.5) in the Prx KO mouse lens fibers was obtained via transmission electron microscopy (TEM). The P90 Prx KO lenses (Fig. 6) exhibited abnormal fiber cell morphology with the fiber cell membrane exhibiting extensive membrane beading and folding, especially at the short arms compared to region-matched WT lens fibers (Fig. 6, arrows) which exhibit a symmetric fiber cell arrangement characterized by a typical hexagonal morphology. Further, as noted using confocal microscopy (Fig. 4C, D), lens fiber cell shape is altered in the Prx KO lenses, exhibiting increase in width and disruption of radial column organization (Fig.6). Membrane changes were prominent and pronounced in the fiber cells of the inner cortical region compared to the outer cortical fibers of the Prx KO lenses (Fig. 6, arrows). These conclusions were based on the data derived from 4 independent analyses.

Moreover, immunogold labeling of aquaporin-0 in the Prx KO lenses starting from the P90 stage (Fig. 7A) revealed a discontinuous membrane localization and clustering pattern compared to WT lens fibers which show a continuous distribution of aquaporin-0 along the plasma membrane (see the magnified inserts; Fig. 7). In both P90 and P200 Prx KO lenses (Fig. 7A & B, respectively), and consistent with previous observations (Figs.4 & 5), aquaporin-0 staining is reduced and clustered especially where the membrane is misfolded or beaded compared (Fig.7A & B, arrows) to corresponding regions of the WT lens. Further, we counted the number of membrane folds in both OCFs and ICFs and noted a significant increase in the ICFs of both P90 and P200 Prx KO lenses (Fig. 7C & D, respectively) compared to the WT lenses. Taken together, these multiple methods of analysis confirm impairment of fiber cell shape and membrane stability in the Prx KO lenses. Additionally, the Prx KO lens fiber cell specimens reveal a smooth cytoplasmic texture relative to a course texture in the WT lens fibers indicating certain subtle changes in the fiber cell cytoplasm due to absence of Prx (Fig.7).