Refractive Defects and Cataracts in Mice Lacking Lens Intrinsic Membrane Protein-2

Abstract

purpose. To characterize the optical properties of lenses from mice deficient in the gene for lens intrinsic membrane protein-2 (Lim2), which encodes the second most abundant integral protein (Lim2) of lens fiber cell plasma membranes.

methods. Lim2-deficient mice were derived from a library of gene-trap embryo stem cells. Genotyping was performed by polymerase chain reaction (PCR) amplification of tail genomic DNA and resequencing. Lim2 expression was analyzed by reverse transcription (RT)-PCR and Northern blotting of lens total RNA, immunoblotting of lens membrane extracts, and immunofluorescence confocal microscopy of lens sections. Lens morphology was assessed by light microscopy, and lens refractive properties were quantified with a laser imaging system.

results. Genomic PCR amplification and resequencing indicated that the gene-trap vector had disrupted intron 3 of Lim2, effectively resulting in a null allele (Lim2 ^Gt ), as verified by RT-PCR amplification and sequencing, RNA blotting, immunoblotting, and immunofluorescence confocal microscopy. Heterozygous Lim2 gene-trap lenses (Lim2 ^Gt/+) were morphologically indistinguishable from wild type, whereas homozygous Lim2 gene-trap lenses (Lim2 ^Gt/Gt ) consistently developed faint, central pulverulent cataracts. Laser imaging analysis indicated that rays passing through the peripheral cortex of the Lim2 ^Gt/Gt lens were more strongly refracted than normal, suggesting that the internal gradient refractive index of the lens was disturbed.

conclusions. These data show that heterozygous loss of Lim2 is insufficient to trigger cataracts in mice, and they provide the first direct evidence that Lim2 plays a critical role in establishing the correct internal refractive properties of the crystalline lens.

The crystalline lens of the vertebrate eye is a biconvex cellular structure that functions in the variable fine focusing of light onto the photosensitive retina. The lens develops from head ectoderm under the influence of eye-derived growth factors ¹ and is composed of two cell types, mitotically competent epithelial cells and terminally differentiated fiber cells. Fiber cells are formed continuously from epithelial cells located near the lens equator. Fiber cell differentiation is accompanied by rapid cellular elongation (150 μm/d in the developing chicken lens) with a concomitant increase in plasma membrane surface area (approximately 6000 μm²/d). ²  

One of the most abundant components of the newly synthesized fiber cell plasma membrane is lens intrinsic membrane protein-2 (Lim2), a 173-amino acid transmembrane protein with a molecular mass of approximately 20 kDa and variously referred to as MP17, MP18, MP19, or MP20. ^{3 4 5 6} Based on a conserved amino acid domain architecture, ⁷ Lim2 is an evolutionarily divergent member of the peripheral myelin protein-22 (Pmp22)_claudin family of mammalian transmembrane proteins, also known as pfam00822. ^{8 9} Structurally, members of pfam00822 share a common computer-predicted topology characterized by four transmembrane helices and two extracellular domains or loops. Biochemical analyses reveal that Lim2 is a glycophosphoprotein ¹⁰ that binds galectin-3 ^{11 12} and calmodulin ⁴ and forms homo-oligomers at least the size of hexamers. ¹³ Although the posttranslational glycosylation pattern supports the view that Lim2 has four transmembrane domains, alternative nontetraspan topologies have also been proposed. ^{14 15} Genetic studies have identified a missense mutation (G15V) in the mouse gene for Lim2 (Lim2) associated with autosomal semidominant congenital cataracts and microphthalmia in the total opacity 3 (To3) mouse mutant. ^{16 17} In contrast, another missense mutation (F105V) in the human gene for Lim2 (LIM2) has been linked to autosomal recessive presenile cataracts. ¹⁸ Based on its subcellular localization in the lens, Lim2 has been implicated in fiber cell junction formation or organization, ¹⁹ fiber cell adhesion, ^{11 12} and occlusion of extracellular space between fibers ²⁰ ; however, its precise role in the lens remains enigmatic. To gain further insight regarding Lim2 function, we generated a Lim2-deficient mouse strain and analyzed the resultant lens phenotype. 

Materials and Methods

Mice and Lenses

Lim2-deficient mice were generated from the OmniBank library ²¹ of gene-trap embryonic stem (ES) cells derived from a 129S5 strain background using standard procedures (Lexicon genetics.com). Briefly, chimeric mice were generated from injections of ES cells (OST98930) into C57BL/6J-albino host blastocysts, and heterozygous mice were produced by breeding high contribution male chimeras to C57BL/6J-albino females. Subsequent intercrosses among heterozygous and homozygous animals were in a hybrid 129S5/C57BL/6J-albino background. Mice were humanely killed by CO₂ asphyxiation and then underwent cervical dislocation or decapitation. Eyes were removed from F₂ littermates up to 6 months of age, and lenses were dissected in culture medium (DME/F12) at 37°C to prevent cold cataract formation, then photographed (within 10 minutes) under bright-field and dark-field conditions against a plain background with a dissecting microscope (Stemi 2000; Zeiss, Thornwood, NY) fitted with a digital camera (Spot 2; Diagnostic Instruments, Sterling Heights, MI). For certain images, lenses were placed on a 200-mesh electron microscopy grid. All procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Animal Studies Committee at Washington University. 

Genotyping

Genomic DNA was prepared from tail biopsy using the DNeasy spin-column kit (Qiagen, Valencia, CA), according to the manufacturer’s instructions and was quantified by ultraviolet (UV) absorbance at 260 nm (GeneQuant pro; GE Healthcare, Piscataway, NJ). Mice were genotyped by PCR amplification (50 μL) of tail DNA (1 μg) with a three-primer (100 pmol) combination (Table 1)of two gene-specific primers (3F, 4R) and a nested vector-specific primer (Rv.3) for 30 standard cycles using a DNA engine (PTC-200 Peltier Thermal Cycler; MJ Research, Waltham, MA). PCR products (5 μL) were separated on 2% agarose gels (Mini-Sub Cell GT; Bio-Rad, Hercules, CA) containing 0.05% ethidium-bromide (Et-Br) and visualized (302 nm) with a gel-documentation system (AlphaImager 3400; Alpha Innotech, San Leandro, CA). Before colony expansion, Lim2 gene-trap mice were further genotyped (data not shown) to exclude the presence of a deletion mutation in the gene for lens-beaded filament structural protein-2 (Bfsp2) or phakinin (CP49), which is characteristic of the 129 strain ²² from which ES cells were derived for gene trapping. 

RT-PCR Analysis

Total cellular RNA was extracted from the lenses of F₂ littermates (postnatal day [P]21–P28) representing each of the three genotypes with the use of a reagent (TRIzol; Invitrogen, Carlsbad, CA) and was quantified by UV absorbance (260/280 nm). Lens RNA (1 μg) was reverse transcribed (20-μL reaction) in the presence of random hexamers, and then cDNA products (10 μL) were PCR amplified with a PCR kit (50 μL reaction; GeneAmp RNA PCR kit; Roche, Indianapolis, IN) and gene-specific (2F, 3R, 5R) or vector-specific (NeoR) primers (Table 1) , according to the manufacturer’s instructions. RT-PCR products (5 μL) were visualized on 2% agarose/0.05% Et-Br gels. 

Sequencing

Genomic DNA (100 ng) was PCR amplified (50-μL reaction) with Biolase DNA polymerase (Bioline USA Inc., Randolph, MA) and gene-specific primers (25 pmol; Table 1 ) for 35 standard cycles in a DNA engine (PTC-200; MJ Research). Genomic PCR or RT-PCR products (approximately 50 μL) were sized on 2% agarose/0.05% Et-Br gels, then purified (in 30 μL) with a gel-extraction kit (QIAquick; Qiagen). Gel-purified amplicons (7 μL) were direct sequenced (12-μL reactions) in both directions using the dye terminator cycle-sequencing (DTCS) quick start kit (Beckman Coulter, Fullerton, CA) and appropriate gene-specific primers (Table 1) , then analyzed on a capillary–electrophoresis system (CEQ8000; Beckman Coulter), according to the manufacturer’s instructions. 

Northern Blot Analysis

Total lens RNA (approximately 20 μg) was separated on 1% agarose-formaldehyde denaturing gels, transferred to positively charged nylon membrane, hybridized with a Lim2 cDNA probe, and washed at high stringency (NorthernMax blot kit; Ambion, Austin, TX). The Lim2 cDNA hybridization probe (522 bp) was reverse transcribed from total lens RNA (1 μg) in the presence of random hexamers, then amplified (GeneAmp RNA PCR kit; Roche) in the presence of gene-specific “start-to-stop” primers (Table 1) . The amplified probe was agarose gel-purified (Qiagen), then labeled with the use of a psoralen-biotin nonisotopic labeling kit (BrightStar; Ambion) and detected with a chemiluminescence kit (BrightStar Biodetect; Ambion) according to the manufacturer’s instructions. Subsequently, blots were reprobed as described with a biotin-labeled mouse β-actin cDNA probe (approximately 2.1 kb) to confirm equal loading of RNA. 

Immunoblot Analysis

Lenses (P21–P28) were homogenized in PBS containing a protease inhibitor cocktail (Complete-Mini tablets; Roche), and membrane proteins were extracted by centrifugation in 4 M urea (14,000g, 30 minutes, 4°C). Membrane protein pellets were resuspended in Laemmli buffer, and protein concentration was quantified (Non-Interfering Protein Assay; Geno Technology, Inc., St. Louis, MO) at 480 nm according to the manufacturer’s instructions. Proteins (6 μg) were denatured (90°C, 5 minutes) in the presence of dithiothreitol (DTT; 15 mM), then separated on 12% SDS-PAGE gels (Invitrogen) and blotted onto nitrocellulose (Novex system; Invitrogen). Blots were incubated (1 hour, 20°C) with a polyclonal antibody (IgG) to bovine Lim2 (MP20; gift from Charles Louis), raised against the eight-amino acid carboxyl-terminal peptide ¹⁶⁶CRRLSTPR¹⁷³ (see 1 Fig. 2C ), as described previously. ¹⁹ Blots were washed and blocked and then incubated with horseradish peroxidase (HRP)–conjugated donkey anti–rabbit IgG secondary antibody, and immunoreactive proteins were visualized with enhanced chemiluminescence (ECL) detection reagents (GE Healthcare), as described. ²³  

Morphologic Analysis

For conventional histology, eyes (P5.5) were fixed overnight at 4°C in formalin (Sigma, St. Louis, MO). After a thorough wash in PBS, lenses were dehydrated through graded acetones and infiltrated in methacrylate resin (H-8100; Technovit, Kulzer, Germany) according to the following schedule: resin/acetone (1:2), 1 day; resin/acetone (1:1), 1 day; 100% resin, 4 days. Blocks were polymerized for 1 hour at 4°C. Sections (1 μm) were cut and stained with methylene blue/azure blue. 

For confocal immunofluorescence microscopy, neonatal lenses (P2.5–P4.5) were fixed (1 hour, 20°C) in 4% paraformaldehyde/PBS, then embedded in agar (4%) reconstituted in PBS and cut into 100-μm sections (Vibratome; TPI, St. Louis, MO), as described previously. ²³ Lens slices were permeabilized with 0.1% Triton X-100, blocked with 10% heat-inactivated normal goat serum, and incubated overnight at 4°C with Lim2 antibody (1:500 dilution), as described. ²⁴ Slices were then washed in PBS for 1 hour and incubated with 1:500 dilution of Alexa 488–conjugated goat anti–rabbit IgG. Slices were viewed through a scanning confocal microscope (LSM510-META; Zeiss). 

Optical Analysis

The radii of curvature of the anterior and posterior lens surfaces were determined from digital images of the lens in vitro with the use of image analysis software (Metamorph; Molecular Devices, Eugene, OR). 

The refractive properties of lenses (approximately 2.5-mm diameter) from F₂ littermates (approximately 12 weeks of age) were analyzed optically with a low-power helium/neon-laser scanning device (Scantox In Vitro Assay System version 2; Harvard Apparatus, Holliston, MA), as described previously, ²⁵ with the following modifications: lenses were suspended on a circular polyester (Dacron; DuPont, Wilmington, DE) support (4-mm diameter) with a central, right-angled cross cut-out (each axis measured 3-mm long × 0. 5-mm wide) and were bathed in DME/F12 medium (37°C) supplemented with 0.01% dry milk powder (to permit visualization of refracted beams). A focused laser beam oriented parallel to the lens optical axis was stepped in 25-μm increments across the entire lens diameter. Images of the refracted beams were captured by a digital camera and used to calculate the slope and intercept of the beam with the optical axis of the lens. 

Statistical Analysis

Differences between groups were assessed using a two-tailed unpaired t-test. Differences were considered significant at P < 0.05. 

Results

Lim2 Gene-Trap Analysis

Gene trapping is a retroviral-mediated form of insertional mutagenesis that allows random targeting of genes in ES cells, irrespective of their expression status, resulting in missplicing of the trapped gene. ²¹ According to the Entrez gene database (ncbi.nlm.nih.gov/entrez/query.fcgi?db=Gene), Lim2 (Gene ID: 233187) is located on murine chromosome 7 cytogenetic band B3, approximately 43 Mb from the centromere, and is organized into five exons spanning approximately 6 kb with exon 1 noncoding. The gene-trap construct (VICTR48) was designed such that the trapped genes are oriented transcriptionally in the direction opposite that of the retroviral vector. ²¹ Therefore, sequences flanking the 3′-long terminal repeat (LTR) of VICTR48 would be the most 5′ relative to Lim2 (Fig. 1A) . OmniBank sequence tag (OST) 98930, derived from 3′-rapid amplification of cDNA ends (RACE) analysis of heterozygous mutant ES cells (http://omnibank.lexgen.com), began in exon 4 of Lim2, indicating that the gene-trap event had occurred upstream in either exon 3 or intron 3 of the gene. After the generation of chimeric mice, a genomic DNA amplification strategy with three PCR primers (Table 1)was implemented to identify the precise retroviral integration site in Lim2 and to genotype progeny with wild-type or mutant alleles (Fig. 1B) . The Lim2-specific primers (3F and 4R) amplified a 1.6-kb fragment from the wild-type allele. Sequence analysis confirmed that this genomic fragment essentially represented intron 3 of Lim2 (data not shown). The mutant allele was amplified by a combination of a Lim2-specific primer (3F) and a retroviral vector-specific primer (Rv.3), resulting in a 1.2-kb fragment (Fig. 1B) . Sequence analysis of this genomic amplicon identified a novel junction (Fig. 1D)between the 3′-LTR of VICTR48 and the third intron of Lim2 at a position located approximately 500 nt downstream from the end of exon 3, confirming that the gene-trap vector had disrupted intron 3. 

Lim2 Expression Analysis

To verify that the gene-trap insertion had disrupted Lim2 at the mRNA level, thus generating a null allele (Lim2 ^Gt ), we performed RT-PCR amplification and Northern blot hybridization analyses of lens RNA. Primers specific for exon 2 (2F) and exon 3 (3R) of Lim2 (Fig. 1A)amplified a fragment of approximately 200 bp from all genotypes (Fig 1C) , consistent with normal splicing of exon 2 to exon 3. In contrast to human LIM2, ²⁶ we did not detect a major alternative splice form of mouse Lim2 containing an intron 2 sequence coding for 42 novel amino acids (Fig. 2C) . Indeed, read-through of the exon-2/intron-2 boundary in Lim2 (ATC Ggt gag cct tga) detected an in-frame translation stop codon (tga) located 6 bp downstream of the 5′-splice donor site (gt) that would likely result in premature termination of any “alternative” Lim2 polypeptide in the mouse. 

When primer 2F was paired with a primer specific for exon 5 (5R) of Lim2 (Fig. 1A) , a fragment measuring 528 bp was amplified from wild-type and Lim2 ^+/Gt lenses but not from Lim2 ^Gt/Gt lenses (Fig 1C) . In contrast, when Lim2-specific primer 2F was paired with a primer (NeoR) specific for VICTR48 (Fig. 1A) , a fragment measuring approximately 450 bp was amplified from null and heterozygous lenses but not from wild-type lenses (Fig. 1C) . Sequencing of the 2F-NeoR amplicon confirmed the presence of a novel Lim2-Neo fusion transcript in lenses from gene-trap mice (Fig. 1E) . These results are consistent with the disruption of Lim2 through missplicing of exon 2 to the Neo resistance splice acceptor site in VICTR48 (Fig. 1A) . RNA blotting (Fig. 2A)detected a Lim2 transcript of approximately 1.0 kb in wild-type lenses; however, Lim2 transcript levels in Lim2 ^+/Gt lenses were reduced compared with wild-type levels, and no intact Lim2 transcripts were detected in Lim2 ^Gt/Gt lenses, consistent with the disruption of mRNA splicing. 

To demonstrate that the gene trap generated a null allele at the protein level, we compared the apparent molecular mass (M_r) and subcellular distribution of Lim2 in wild-type compared with gene-trap lenses by immunoblot analysis and immunofluorescence confocal microscopy, respectively, using a C-terminal synthetic peptide antibody. Immunoblotting detected strongly reactive antigens of approximately 19-kDa and approximately 28-kDa M_r in urea-soluble membrane proteins from wild-type lenses (Fig. 2B) . Notably, Lim2 has been shown to form oligomers, ¹³ and the approximately 28-kDa antigen is thought to represent a Lim2 dimer that migrates aberrantly in SDS-PAGE gels. ^{4 19} In Lim2 ^+/Gt lenses, reduced levels of Lim2 monomer (approximately 19 kDa) with trace levels of putative dimer (approximately 28 kDa) were detected; however, no such antigens were detected in Lim2 ^Gt/Gt lenses. 

Immunofluorescence confocal microscopy of neonatal (P3.5) lens slices detected specific and intense staining of fiber cell plasma membranes in wild-type but not in Lim2 ^Gt/Gt lenses (Fig. 3A) . High-magnification images revealed that Lim2 was uniformly distributed throughout the plasma membranes of fiber cells located in the center (Fig. 3B)or periphery (Fig. 3C)of the lens. In contrast to previous reports, ²⁰ significant levels of intracellular Lim2 were not observed. A qualitatively similar pattern of fiber cell immunostaining was detected in Lim2 ^+/Gt lenses (data not shown). 

Phenotype of the Lim2 Gene-Trap Lens

Growth rates of wild-type, Lim2 ^+/Gt , and Lim2 ^Gt/Gt lenses from 1 to 12 weeks of age were similar, though the mean diameter of Lim2 ^Gt/Gt lenses was slightly smaller at all ages (Fig. 4) . This difference reached statistical significance by 6 months (data not shown). For each genotype, the radius of curvature of the anterior and posterior lens surfaces was determined in 10-week-old animals (Table 2) . In all cases, the refractive surfaces were found to be spherical, and the radius of curvature of the posterior lens surface was always smaller than that of the anterior surface. Measured values for Lim2 ^+/Gt and Lim2 ^+/+ were statistically indistinguishable. The radii of curvature of the refractive surfaces in Lim2 ^Gt/Gt lenses was marginally, but significantly (P < 0.01 in each case), smaller than the other genotypes, reflecting the overall reduction in lens dimension (Fig. 4) . 

Histologic analysis of neonatal (P5.5) lenses indicated that the cytoarchitecture of Lim2 ^Gt/Gt lenses did not differ significantly from wild-type lenses (Figs. 5A 5B) . In both types of lens, fiber cells elongated normally from the equator toward the anterior and posterior poles, the nuclear bow region was well formed, and the programmed elimination of nuclei in the deep cortex occurred normally. 

We made the serendipitous finding that despite their relatively normal appearance, Lim2 ^Gt/Gt lenses differed in physical properties from wild-type (Figs. 5C 5D) . Semifixed wild-type or Lim2 ^Gt/Gt lenses were bisected and teased apart with sharp forceps. In Lim2 ^Gt/Gt lenses, this led to the complete disintegration of the tissue, as the component cells separated readily from each other. In contrast, it was difficult to dissect individual cells from the bisected wild-type lens. On dissection, the tissue broke into irregular pieces, within which individual cells were usually not discernible. These data suggest a lack of tissue cohesion in the absence of Lim2. 

Although Lim2 ^Gt/Gt lenses were largely transparent, they displayed two consistent abnormalities not observed in wild-type lenses. First, the refractive properties of Lim2 ^Gt/Gt lenses differed significantly from wild type. This was best illustrated by examining distortions in an underlying grid pattern introduced by each lens (Fig. 6) . The image of the hexagonal grid formed by a wild-type lens was relatively uniform (Fig. 6B) . In contrast, the image of the grid pattern produced by the Lim2 ^Gt/Gt lens was severely distorted, especially near the periphery (Fig. 6D) . The Lim2 ^Gt/Gt lens appeared to contain at least two refractive compartments: an inner region, which appeared qualitatively similar to the central region of the wild type lens, and an outer region, through which the underlying grid pattern was distorted. The abrupt transition between the compartments could be visualized, under oblique lighting, as a dark annulus located in the mid-cortex of the lens (Fig. 6C) . The two compartments were evident before the eyes opened at around P14 (Fig. 6E) . Second, when the mice were approximately 3 weeks of age, Lim2 ^Gt/Gt lenses began to develop faint, hazy opacities centered in the fetal nuclear region (data not shown), and, by 12 weeks of age, discrete nuclear pulverulent opacities could be discerned within the lens under strong oblique illumination (Fig. 6F) . At all ages examined, Lim2 ^+/Gt lenses were morphologically and optically indistinguishable from wild type (data not shown), suggesting that heterozygous loss of Lim2 was insufficient to trigger cataracts or refractive disturbances in the lens. Moreover, the lack of an obvious phenotype in Lim2 ^+/Gt lenses confirmed that the Lim2 gene-trap allele, which resulted in a Lim2-Neo fusion product (Fig. 1E) , did not exert a dominant-negative or other deleterious gain-of-function effect on the wild-type allele in the heterozygous state. 

Optics of the Lim2 Gene-Trap Lens

To better define the changes in optical properties in Lim2 ^Gt/Gt lenses, we quantified lens focal power and spherical aberration by laser scanning analysis (Fig. 7) . A fine laser beam was stepped incrementally through the lens, and the focal length (back vertex distance; BVD) of each beam was determined as a function of radial position. In wild-type lenses, the BVDs of eccentric beams (those located furthest from the optical axis) were greater than those of beams passing close to the lens center (Fig. 7A) . 

The ray path through the Lim2 ^Gt/Gt lens was more complex (Fig. 7B)and best appreciated in plots of average BVD (Fig. 7C) . The BVDs of rays passing through the periphery of the Lim2 ^Gt/Gt lens (more than 1 mm from the optical axis) were indistinguishable from those of wild-type lenses (Fig. 7C) . However, the BVD of beams passing though the cortex of the lens (approximately 0.7–1.1 mm from the optical axis) differed between Lim2 ^Gt/Gt and wild-type lenses. In this region, the BVD was significantly (P < 0.01) shorter in Lim2 ^Gt/Gt lenses. For beams passing through the perinuclear region of the Lim2 ^Gt/Gt lens (0.4–0.7 mm from the optic axis), the BVD was slightly, but significantly (P < 0.01), greater than that of the wild-type lens. Laser imaging analysis detected no significant difference between Lim2 ^+/Gt and wild-type lenses with respect to BVD distribution (data not shown). Thus, the laser imaging analysis indicated that, compared with wild type, the refractive power of the cortical region of Lim2 ^Gt/Gt lenses increased and that of the perinuclear region decreased. 

Discussion

In this study, we have demonstrated that gene-trap mice functionally null for Lim2 (Lim2 ^Gt/Gt ) develop lenses with abnormal refractive properties and central pulverulent opacities. The Lim2 ^Gt/Gt mouse lens opacities partially resembled those associated with a homozygous missense mutation (F105V) in human LIM2 ¹⁸ ; however, the mouse opacities were confined largely to the fetal nucleus, whereas, in humans, opacities were of late onset and arranged as concentric layers spreading into the lens cortex. The recessive inheritance of LIM2-related cataracts in humans and the absence of opacities in heterozygous Lim2 gene-trap mice are both consistent with a loss-of-function mechanism. In contrast, To3 mice heterozygous for a missense mutation (G15V) in Lim2 develop congenital total cataracts, and in the homozygous state they also exhibit microphthalmia. ¹⁶ The semidominant nature of the To3 mutation is consistent with a deleterious gain-of-function mechanism, ¹⁷ and highlights the genetic complexity of hereditary cataracts in mice and humans. 

In addition to the presence of central opacities, the Lim2 ^Gt/Gt lens was characterized by refractive abnormalities in the peripheral cortex. The refractive properties of crystalline lenses depend on the shape of the lens and its internal composition. ²⁷ Because of steep (2- to 3-fold) gradients in the concentration of cytosolic proteins, known as crystallins, between the superficial and inner regions of the lens, the refractive index of the cytoplasm is not uniform. ²⁸ For example, in rats it has been shown that the refractive index varies in a near-parabolic fashion across the lens diameter, ranging from approximately 1.36 near the surface to approximately 1.50 in the center. ²⁹ This precisely defined gradient refractive index (GRIN) is thought to correct the longitudinal spherical aberration that would otherwise be present in a lens of uniform refractive index. Our measurements on isolated mouse lenses indicated that the wild-type lens was overcorrected for spherical aberration. Thus, beams passing through the lens peripheral cortex had longer focal lengths (BVDs) than those passing through the central region of the lens (Fig. 7) , resulting in a negative spherical aberration similar to that of the rat lens. ²⁷ In the rodent eye, the lens is only one element of the optical train, and it is likely that negative spherical aberration contributed by the lens is offset by positive spherical aberration from the cornea. In Lim2 ^Gt/Gt mouse lenses, the BVDs of beams directed through the peripheral cortex were shorter than in wild-type lenses, and those directed through the perinuclear region were longer than wild-type lenses, resulting in image distortion (Fig. 6) . Given that wild-type and Lim2 ^Gt/Gt lenses appeared to be of similar shape (Table 2)and size (Fig. 4) , the difference in refractive properties between the two lenses can be attributed largely to differences in their respective GRINs. 

Members of the Pmp22_Claudin family (pfam00822) share functions in cell adhesion and junction formation. Claudins are distributed widely and variably in epithelia as components of tight junctions. ⁹ Similarly, Pmp22 is a constituent of tight junctions in liver, intestine, blood-brain barrier, and, potentially, in myelinating Schwann cells of the peripheral nervous system. ^{30 31} Tight junctions form a barrier to the passage of ions and molecules through the paracellular pathway and to the movement of proteins and lipids between the apical and basolateral domains of the plasma membrane. It is unlikely, however, that Lim2 is a component of classical tight junctions in the lens because tight junctions are restricted to the lens epithelium, as shown in electron microscopic studies, ³² whereas Lim2 is known to be expressed in fiber cells. In rat lenses, it has been proposed that Lim2 forms a barrier to the extracellular diffusion of fluorescent dyes into the center of the lens. ²⁰ This proposal was based largely on the observation that Lim2 redistributes from a predominantly cytoplasmic location into the plasma membrane during the later stages of fiber cell differentiation and maturation. We did not observe this phenomenon in the developing mouse lens (Fig. 3) . However, our observation that lens fibers are readily disaggregated in the absence of Lim2 (Fig. 5)directly supports the notion that Lim2 has an adhesive function in the lens. In addition, we have shown for the first time that the absence of Lim2 impairs the internal refractive quality of the lens. The precise relationship between lens refraction and intercellular adhesion is unknown; however, generation of the Lim2-deficient mice described here provides a valuable experimental system for further investigation of the role of Lim2 in lens structure and function. 

 Supported by National Institutes of Health/National Eye Institute Grants EY11411 (AS), EY09852 (SB), and EY02687 (core grant for vision research) and by an unrestricted grant from Research to Prevent Blindness (RPB). SB is a William and Mary Greve RPB scholar.

 Submitted for publication August 9, 2006; revised September 22, 2006; accepted December 4, 2006.

 Disclosure: A. Shiels, None; J.M. King, None; D.S. Mackay, None; S. Bassnett, None

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May 2004.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

PCR Primers Used for Characterization of the Lim2 Gene-Trap Mouse

Primer	Location	Strand	Sequence (5′→3′)
2F (start)	Exon 2	Sense	TGGCATGTACAGCTTCATGGGT
3F	Exon 3	Sense	TGCAGGCATCATGTTTTTCGC
3R	Exon 3	Antisense	TGAAAGCCCGGGTGGCATTCC
NeoR	VICTR48	Reverse	TCCCTTCCCGCTTCAGTGACAACGAG
Rv.3	3′LTR	Reverse	CAACAATATCACCAGCTGAAGCC
4R	Exon 4	Antisense	TCCAGTGTAAATAGCCAAGGC
5R (stop)	Exon 5	Antisense	TTATTATCAGCGTGGGGTGGCTAG

 Start and stop codons are denoted by underscoring.

Figure 1.

 

Lim2 gene-trap analysis. (A) Schematic diagram of the gene-trap vector (VICTR48) and its site of insertion into Lim2. Exon 1 is noncoding. Codon numbers are indicated below exons 2 to 5. Arrows indicate orientation and approximate location of PCR primers. (B) Genomic PCR analysis showing that primers 3F and 4R amplified an expected product (1.6 kb) specific for the Lim2 ⁺ allele, whereas primers 3F and Rv.3 amplified a novel product (1.2 kb) specific for the Lim2 ^Gt allele. (C) RT-PCR analysis showing that primers 2F and 3R amplified a correctly spliced product (202 bp) from Lim2 ^+/+, Lim2 ^+/Gt , and Lim2 ^Gt/Gt lenses, whereas primers 2F and 5R amplified a correctly spliced product (528 bp) only from Lim2 ^+/+ and Lim2 ^+/Gt lenses. Primers 2F and NeoR amplified a novel misspliced product (454 bp) only from Lim2 ^+/Gt and Lim2 ^Gt/Gt lenses. (D) Genomic DNA sequence analysis showing the gene-trap vector insertion site in intron 3 of Lim2 (arrow, right) compared with the equivalent region of wild type (left). (E) cDNA sequence analysis showing a novel Lim2-Neo fusion transcript in lenses from gene-trap mice (arrow, right) compared with the equivalent region of wild type (left). Codon reading frame is indicated by bars and by the single-letter amino acid code.

Figure 2.

 

Lim2 transcript and protein expression analysis. (A) RNA blot hybridization analysis showing the presence of intact Lim2 transcripts (approximately 1.2 kb) in wild-type and Lim2 ^+/Gt lenses but not in Lim2 ^Gt/Gt lenses. Blots were rehybridized with an actin probe as a positive control. (B) Immunoblot analysis showing the presence of Lim2 proteins (approximately 19 and approximately 29 kDa) in wild-type and Lim2 ^+/Gt lenses but not in Lim2 ^Gt/Gt lenses. Trace levels (arrow) of an unidentified antigen (approximately 75 kDa) were also detected in lens membrane extracts, irrespective of genotype. (C) Schematic diagram of mouse Lim2 topology showing four transmembrane domains, two extracellular loops, a cytoplasmic loop, and cytoplasmic amino (NH₂) and carboxy (COOH) termini. The signature motif of six amino acids (W-GLW-C-C) located in the larger first extracellular loop of pfam00822 members are filled. The four carboxyl-terminal amino acids (¹⁷⁰ATPR¹⁷³), which resemble a canonical postsynaptic density protein-95/discs large/zonula occludens-1 (PDZ)-binding motif (XS/TXA), where X is any amino acid and A is an aliphatic amino acid, ³³ are shown in bold. The location of missense mutations associated with autosomal-dominant cataracts in To3 mice (G15V) ¹⁶ and autosomal-recessive cataracts in humans (F105V) ¹⁸ are indicated. Also shown are the relative locations of an alternative splice site (Ins42) in the human LIM2 gene, resulting in the insertion of 42 amino acids, ²⁶ the retroviral vector insertion site for gene trapping (VICTR48), a consensus N-glycosylation (N-gly) site, (⁶²NXS/T⁶⁴), C-mannosylation (C-man) sites (W43, W61), and a phosphorylation (PO₄) site (T171). ¹⁰ The bar denotes the eight-amino acid C-terminal epitope (¹⁶⁶CRRLATPR¹⁷³) recognized by the bovine Lim2 antibody. ¹⁹ S170 in bovine Lim2 is replaced by A170 in mouse Lim2.

Figure 3.

 

Confocal immunofluorescence microscopy of the Lim2 gene-trap lens (P3.5). (A) Composite equatorial section of wild-type (Lim2 ^+/+) lens (left) and Lim2 ^Gt/Gt lens (right) showing intense Lim2 immunofluorescence staining in the former but not in the latter. High-magnification images of Lim2 immunofluorescence in the core (B) and peripheral regions (C) of the Lim2 ^+/+ lens demonstrate that Lim2 is present throughout the plasma membrane of the fiber cells. Scale bars: (A) 250 μm; (B, C) 10 μm.

Figure 4.

 

Growth rates of Lim2 ^+/+, Lim2 ^Gt/+, and Lim2 ^Gt/Gt lenses (measured as increases in lens diameter) did not differ significantly from 1 to 12 weeks.

Table 2.

 

Radius of Curvature Measurements of Lim2 Gene-Trap Lenses

Lens Genotype	n	Anterior Radius (mm)	Posterior Radius (mm)	Diameter (mm)
Lim2 +/+	8	1.209 ± 0.032	1.173 ± 0.030	2.230 ± 0.018
Lim2 +/Gt	6	1.212 ± 0.011	1.174 ± 0.015	2.250 ± 0.022
Lim2 Gt/Gt	10	1.169 ± 0.016	1.128 ± 0.019	2.170 ± 0.026

 Measurements were taken when the mice were 10 weeks of age. Values are mean ± SD.

Figure 5.

 

Cellular structure and tissue cohesion in wild-type (A, C) and Lim2 gene-trap (B, D) lenses. The cellular organization of the Lim2 ^Gt/Gt lens is normal. In wild-type (A) and Lim2 ^Gt/Gt lenses (B), fiber cells elongated from the equator to termination points at the anterior and posterior sutures. Fiber cell denucleation (arrow) occurred normally in the Lim2 ^Gt/Gt lens. Wild-type or Lim2 ^Gt/Gt lenses were fixed briefly, bisected, and teased apart with fine forceps. (C) In wild-type lenses, dissection caused the tissue to fracture into large pieces, within which individual fiber cells were rarely observed. (D) In Lim2 ^Gt/Gt lenses, dissection led to disaggregation of the tissue into component fiber cells, suggesting a failure of intercellular adhesion. Scale bars: (B) 250 μm; (D) 100 μm.

Figure 6.

 

Light microscopy and optical properties of the Lim2 gene-trap lens. (A) Wild-type lens (6 weeks) visualized against a plain background in bright field. (B) Underlying grid (200 mesh) imaged through a wild-type lens. (C) Lens from a Lim2 ^Gt/Gt littermate (6 weeks) visualized against a plain background. Note the presence of a dark annulus (arrow) in the midcortical region. (D) The image of the underlying grid is distorted severely by the Lim2 ^Gt/Gt lens. (E) Lim2 ^Gt/Gt lens (2 weeks) showing formation of inner (i) and outer (o) refractive compartments. (F) The lens of a 12-week-old Lim2 ^Gt/Gt mouse imaged in situ in an enucleated eye shows the presence of centrally located pulverulent opacities.

Figure 7.

 

Laser imaging analysis of 12-week-old wild-type and Lim2 gene-trap lenses. A fine laser beam was projected through wild-type (A) or Lim2 ^Gt/Gt (B) lenses. The beam was moved in 0.025-mm increments across the lens diameter, parallel to the optic axis. In the wild-type lens, beams passing through the lens periphery have a greater focal length than those passing through the central region. In Lim2 ^Gt/Gt lenses, the pattern was more complex. (C) BVD (the distance between the posterior pole of the lens and the intercept between a refracted beam and the optic axis) was calculated for each radial beam location. Mean BVD of the most eccentric beams (>1 mm from the optic axis) was indistinguishable for the two genotypes. The BVD of beams passing through the mid-cortex (0.7–1.0 mm from the optic axis) of the Lim2 ^Gt/Gt lens was significantly shorter than wild type. Conversely, in the perinuclear region of the lens (0.4–0.7 mm from the optic axis), the BVD was significantly shorter for wild-type than for Lim2 ^Gt/Gt lenses. Data for each genotype represented mean ± SD for 8 wild-type and 15 Lim2 ^Gt/Gt lenses. Adjacent to the optic axis (0.0–0.2 mm), the data were less reliable because of the difficulty inherent in determining accurately the point of intersection of near-parallel beams with the optic axis. (D) Statistical analysis of the data shown in (C). Regions in which the curves differed significantly (i.e., P < 0.05) from each other are indicated in red.

The authors thank Charles Louis for sharing Lim2 antibody, Jacob Sivak for helpful advice on lens optics, and Belinda McMahan for help with histology.