Ocular and Scleral Alterations in Gene-Targeted Lumican-Fibromodulin Double-Null Mice

Abstract

purpose. To elucidate the role of leucine-rich proteoglycans lumican and fibromodulin in the sclera.

methods. Lumican- and fibromodulin-null heterozygous mice were intercrossed to obtain wild-type (Lum ^+/+ Fmod ^+/+ ), lumican-null (Lum ^−/− Fmod ^+/+ ), fibromodulin-null (Lum ^+/+ Fmod ^−/− ), and double-null (Lum ^−/− Fmod ^−/− ) littermates. Axial length was measured on enucleated whole eyes, and ocular structural changes were examined by histology. The morphology of collagen fibrils in the sclera was examined by transmission electron microscopy (TEM).

results. Compared with the ocular axial length in wild type mice, the axial length was increased by 10% in Lum ^−/− Fmod ^−/− (P = 0.02) mice. Retinal detachment was frequent in the double-null and rare in the lumican-null animals. Compared with the wild-type sclera, the sclera in all null mutants was significantly thinner with fewer lamellae (P < 0.05). The double-null sclera contained abnormally large-diameter (120–160 nm) and small-diameter (30–60 nm) collagen fibrils, whereas the fibromodulin-null sclera was enriched for the small-diameter fibrils. The collagen fibril diameter distribution in the lumican-null sclera was similar to that of the wild-type.

conclusions. An increase in small-diameter fibrils in the fibromodulin-null sclera suggests a key role for fibromodulin in the maturation and assembly of scleral collagen fibrils. That fibril diameter distribution in the lumican-null sclera was comparable to that in the wild type, but severely disrupted in the double null, suggests a role for lumican that is crucial in the absence of fibromodulin. The eyes of Lum ^−/− Fmod ^−/− mice show certain features of high myopia: increased axial length, thin sclera, and retinal detachment. Mutations or altered expression of these proteoglycans may contribute to myopia in humans.

Lumican and fibromodulin, present in fibrillar collagen-rich connective tissues, are members of the small leucine-rich proteoglycan (SLRP) gene family. ¹ The SLRPs identified to date, form a group of 10 members, classified into three or four structurally related classes. ^{1 2 3} Lumican and fibromodulin are class II members with unique and overlapping expression patterns. Thus, lumican is present in the skin, cornea, tendon, and cartilage and, as shown in the current study, in the sclera. ⁴ Fibromodulin is a constituent of tendons, cartilage, and sclera. ^{5 6} In vitro studies have shown that these proteoglycans bind to the same region on collagen type I, with fibromodulin having a higher collagen-binding affinity than lumican. ⁷ In vivo they regulate collagen fibril structure for optimal functioning of connective tissues. Lumican-deficiency alone has a profound effect on collagen fibril structure in the skin and cornea, particularly the posterior stroma of the cornea. In addition to increased mean fibril diameter, there are numerous abnormally contoured fibrils that appear cauliflower shaped in cross section, possibly arising from lateral fusion of fibrils. ⁸ The functional consequences of these extracellular matrix defects are skin laxity and loss of corneal transparency in the lumican-null mouse. ⁴ In the fibromodulin-null mouse, the gross morphology and structure of collagen fibrils in the developing tendons are disrupted. ⁶  

We generated lumican-fibromodulin double-deficient (Lum ^−/− Fmod ^−/− ) mice to address the combined role of lumican and fibromodulin in the regulation of collagen architecture and in functions of organs and tissues where they are coexpressed. The double-null mutants have abnormal tendon collagen architecture and biomechanically weak tendons, misaligned knee patellas, and consequent gait abnormalities. ⁹ We anticipated that the sclera, normally rich in both SLRPs and very similar to the tendon by composition and ultrastructure, would be affected in the Lum ^−/− Fmod ^−/− double-null mice. 

The sclera—external to the choroid, retinal pigment epithelium, and anterior layers of the retina—is a layer of viscoelastic connective tissue with flattened fibroblasts embedded in it. The major extracellular matrix components of the fibrous mammalian sclera are collagens types I and III and the SLRPs decorin, biglycan, lumican, and fibromodulin. The sclera provides the necessary mechanical support to counteract intraocular pressure and maintain normal shape of the eye. Increasing evidence suggests that changes in the structure and composition of the sclera are major factors in regulating scleral integrity and axial elongation of the eye as in myopia. ^{10 11 12 13 14 15 16} A connection between myopia and abnormal collagen architecture of the sclera has been recognized as early as the late 1960s and possibly even earlier. ¹⁷ Myopia is a highly prevalent, complex phenotype involving genetic and environmental factors. The development of high myopia involves enlarging of the eye, scleral thinning, and frequent detachment of the retina resulting from stress associated with excessive axial elongation. Various lines of evidence are now beginning to suggest a key role for the SLRPs in scleral changes underlying myopia. Induced myopia in the tree shrew is associated with decreased scleral dry weight and glycosaminoglycan (GAG) levels, whereas recovery from axial myopia is accompanied by increased GAG levels and scleral dry weight. ^{12 18} Additional studies in an adolescent marmoset myopia model also indicate altered proteoglycan levels. Chondroitinase avidin-biotin complex (ABC) digestion of scleral proteoglycans and Western immunoblot analysis suggest a decrease in the SLRP core protein decorin. ^{13 14}  

The present study focused on how the morphology and ultrastructure of the sclera is affected in Lum ^−/− Fmod ^−/− double-deficient mice. The results showed that mice deficient in lumican and fibromodulin manifest certain features of high myopia of long-standing duration in humans and in animal models of myopia. These features include structural changes in collagen fibrils in the sclera, thinning of the sclera, retinal detachment, and increased ocular axial length—compared with those features in wild-type mice. 

Methods

Animals

Lumican-null (Lum ^−/− Fmod ^+/+ ) and fibromodulin-null (Lum ^+/+ Fmod ^−/−) mice were intercrossed to generate mice heterozygous at both loci. The F1 heterozygotes were intercrossed to produce wild-type (Lum ^+/+ Fmod ^+/+ ), lumican-null (Lum ^−/− Fmod ^+/+ ), fibromodulin-null (Lum ^+/+ Fmod ^−/− ), and double-null (Lum ^−/− Fmod ^−/− ) littermates. Genotypes of the progeny were confirmed by PCR analyses, as described earlier, ¹⁹ and null phenotypes were confirmed by Western blot analysis. The mice were housed in the Johns Hopkins Animal Care Facilities in microisolators and killed according to protocols approved by the Johns Hopkins Animal Care and Use Committee and guidelines set by the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Whole-Eye Measurements

Eight-week old female mice of the following number per genotype (both eyes in each animal) were examined: 10 Lum ^+/+ Fmod ^+/+ , 6 Lum ^−/− Fmod ^+/+ , 9 Lum ^+/+ Fmod ^−/− , and 5 Lum ^−/− Fmod ^−/− . Animals were heavily sedated with injections of 0.10 mL of 10 mg/mL acepromazine and were killed by cervical dislocation. Subsequently, whole eyes were isolated, placed in phosphate-buffered saline (pH 7.4), and photographed within 5 to 10 minutes of enucleation, using a dissecting microscope (Leica, Deerfield, IL) with a digital camera (Intralux Dual 5000-1; Optronics, Goleta, CA). The eyes were viewed with a 10× objective and photographed at 1.25 magnification. Images were captured with image-management software (Magnafire Application, ver. 1.0; Karl Storz Imaging, St. Louis, MO). The axial length and diameter of each globe was measured with the measuring tool available in image-analysis software (Photoshop 7.0; Adobe Systems, San Jose, CA). Axial length, defined as the distance between the anterior surface of the cornea and the base of the optic nerve, was determined after correcting for a magnification factor of 1.25. Student’s t-test was performed to determine the statistical significance of genotype-specific measurements. 

Histology

General morphology of the sclera and areas of sclera-choroid attachment were examined in paraffin-embedded sections. Eyes of 1- and 5-month-old mice were enucleated, and the whole eyes were fixed in 4% paraformaldehyde in phosphate-buffered saline overnight, serially dehydrated, and embedded in paraffin according to published protocols. ²⁰ Six-micrometer sections were stained with hematoxylin and eosin. The age of 1 month was selected as representative of the young adult stage and 5 months as the mature age. 

Western Blot Analyses

The eyes were enucleated and the sclera dissected and cut into small pieces. Tissue extraction reagent (T-PER; Pierce Chemical Co., Rockford, IL) containing a protease inhibitor cocktail (Halt; Pierce) at a final concentration of 1× was added to 20 mL/g of wet tissue, and the samples homogenized on ice. Tissue debris was removed by centrifugation for 5 minutes at 10,000 rpm. The supernatant was saved, and its protein concentration determined by the bicinchoninic acid protein assay (BCA; Pierce). The proteoglycans were visualized by SDS-polyacrylamide gel electrophoresis and immunoblot analysis with antibodies against lumican and fibromodulin. ^{4 6}  

Transmission Electron Microscopy

Three to four mice per genotype, at 3 and 5 months of age, were killed. The eyes were fixed in situ in 4% paraformaldehyde, 2.5% glutaraldehyde, 0.1 M sodium cacodylate (pH 7.4), with 8.0 mM CaCl₂ for 15 minutes, during which time the sclera was dissected. The scleras were further fixed for 100 minutes at 4°C and processed for transmission electron microscopy (TEM), as previously described. ⁸ Briefly, the sclera was postfixed with 1% osmium tetroxide and stained en bloc with uranyl acetate-50% ethanol. After dehydration in an ethanol series and infusion with propylene oxide, the sclera were infiltrated with and embedded in a mixture of epoxy resin (Polybed 812), nadic methyl anhydride, and epoxy accelerator (DMP-30; Polysciences, Inc., Eppelheim, Germany). Thin sections were cut using an ultramicrotome (UCT; Reichert-Jung, Vienna, Austria) and a diamond knife and stained with 2% aqueous uranyl acetate and 1% phosphotungstic acid (pH 3.2). Sections from the anterior and posterior sclera of fibromodulin-, lumican-, and fibromodulin-lumican–deficient and wild-type mice were examined and photographed at 75 kV with a transmission electron microscope (model 7000; Hitachi, Tokyo, Japan). 

Fibril Diameter Analyses

For each genotype, three to four different animals were analyzed. The anterior and posterior sclera were viewed, but only the posterior sclera was analyzed in detail. For the posterior sclera, micrographs (12 per group) from nonoverlapping regions of the central portion of sclera wall were taken at ×31,680. The diameter of 85 to 300 fibrils was measured from a single region of a photographic negative. For fibrils that contained uneven contours, the minimum diameter was included in the analysis. Micrographs were randomly chosen in a masked manner from the different groups and digitized, and diameters were measured in an image analysis system (Bioquant Image Analysis System; RM Biometrics, Nashville, TN). 

Statistical analyses were preformed to compare fibril diameter distributions in the wild-type and mutant scleras, as described earlier. ¹⁹ The fibril diameter data have two levels of variations: fibrils are clustered within location (photographic plate) and locations are clustered within animal. Thus, the statistical analyses had to account for the animal-to-animal variability and location-to-location variability in fibril diameter distributions. The fibril diameter location data are generally skewed, sometimes appear to be bimodal, and do not support the assumption of normality. Consequently, percentiles of the location distributions were analyzed and compared. For analyzing genotype differences we focused on quartiles (25th, 50th, or the median and 75th percentiles). ¹⁹  

We computed the sample quartiles for each field and modeled them based on genotype, animal-to-animal, and field-to-field variability, using a linear mixed-effects model on computer (SAS, ver. 8.2; SAS, Cary, NC). Based on this model, we estimated the mean genotype quartiles and tested differences in terms of mean quartiles between wild-type and mutant genotypes. The data included 12 fields per genotype, including three to four animals per genotype and two to six samples per animal. The linear mixed-effects model assumes that the distribution of sample quartiles is normal, and based on the limited samples (12 fields per genotype), our exploratory analysis did not reveal obvious violation of this assumption. The numbers of diameter measures per location were usually higher than 200 (at least 87). Therefore, in our model we treated the sample quartiles as true quartiles of the location distributions. 

Results

Increased Axial Length in Lum ^−/− Fmod ^−/− Mice

For the eye measurements, 2-month-old littermates of the following genotypes, Lum ^+/+ Fmod ^+/+, Lum ^−/− Fmod ^+/+ , Lum ^+/+ Fmod ^−/− , and Lum ^−/− Fmod ^−/− were generated by mating double heterozygotes. The lumican-null mice had cloudy corneas as we showed before. ⁴ Based on the lumican-null phenotype alone the double-null mice are expected to have cloudy corneas. Fibromodulin is not present in detectable levels in the cornea, and its absence alone is not expected to lead to clouding of the cornea. However, to determine the effects of the single- and double-null mutations on corneal transparency, all four genotypes were examined by slit lamp biomicroscopy. The lumican-null corneas were cloudy; and, although the fibromodulin-null corneas were not as cloudy as the lumican-null corneas, they were also not nearly as clear as the wild-type mouse corneas (data not shown). The lumican and fibromodulin double-null corneas were also cloudy. Unlike the lumican-null corneas, they had a slight grainy quality to them (data not shown). The double-null mutant mice had normal iris structure and by histology the iris and lens appeared normal (data not shown). 

The different genotypes, however, showed differences in the overall shapes of the eyes: the double-null mutant eyes appeared more elliptical than those of the other genotypes (Fig. 1) . To quantify the differences in shape, we measured the axial length and diameter of the whole eyes, as viewed in Figure 1 . The diameter measurements did not show genotype-dependent variations and are not shown in the figure. Axial length of the Lum ^+/+ Fmod ^−/− eyes was significantly higher (P = 0.015) than that of eyes of the wild-type littermates. Mean axial lengths of the double-null and wild-type eyes was 5.2 ± 0.28 (n = 4) and 4.6 ± 0.2 mm (n = 10), respectively. The axial lengths of lumican-null (4.67 ± 0.11 mm, n = 6) and fibromodulin-null eyes (4.74 ± 0.26 mm, n = 9) were not significantly different from wild-type eyes (P = 0.39 and 0.146, respectively). There were genotype-associated differences in body size and total body weight as well. The double-null mice were consistently smaller and the Lum ^+/+ Fmod ^−/− often larger than their littermates of other genotypes. 

Although there is no simple linear relationship between body size and ocular length, to investigate a possible correlation between body size and axial length, we plotted axial length versus body weight (Fig. 2) . The distribution showed that the wild-type and each of the single-null mutants formed one group, and the double-null animals formed a separate group. The correlation coefficient for the comparison of body weight with axial length for each genotype was as follows: Lum ^+/+ Fmod ^+/+ , 0.2; Lum ^+/+ Fmod ^+/+ , −0.2; Lum ^+/+ Fmod ^−/− , −0.3; and Lum ^+/+ Fmod ^−/− , −0.67. The inverse correlation between body weight and axial length in the double-null mice is most likely an indication of phenotype severity—that is, the lower the body weight, the more severe the double-null phenotype and increased ocular length. Thus, the double-null animals were significantly smaller than the other genotypes (P = 0.0001) and their ocular axial lengths were significantly higher (P = 0.01). In a preliminary study, we reported increased ocular length in the lumican-null animals (Lea G, Amman S, Chakravarti S, ARVO Abstract 3934, 2000). However, because these studies were not performed in control littermates, subtle genetic background differences between the in-house–derived null mutants and the wild types, which were purchased, may have contributed to these earlier results. 

Scleral Thinning and Retinal Detachment in the Lum ^−/− Fmod ^−/− Mice

Cross-sections of the sclera from wild-type and null mutants stained with hematoxylin and eosin are shown in Figure 3 . The area of the sclera proximal to the optic nerve is defined as the posterior and the area distal to the optic nerve, but closer to the limbal area is the anterior sclera. The posterior sclera, in particular, appeared visibly thinner. Thicknesses of the scleras, measured in three animals per genotype by using the image-analysis software’s measuring tool on digital photographs (Photoshop; Adobe, San Jose, CA), were as follows: mean thickness, 30 ± 2.8 μm in wild-type; 22.1 ± 3.2 μm in Lum ^−/− Fmod ^+/+ ; 21 ± 1.3 μm in Lum ^+/+ Fmod ^−/− ; and 13.6 ± 1.8 μm in Lum ^−/− Fmod ^−/− . Compared with the wild-type, thickness of the sclera was reduced significantly in all three null genotypes (P = 0.03 for Lum ^−/− Fmod ^+/+ , P = 0.02 for Lum ^+/+ Fmod ^−/− , and P = 0.002 for Lum ^−/− Fmod ^−/− ). The Lum ^−/− Fmod ^−/− eyes had multiple areas of retinal detachment, with signs of subretinal proteinaceous debris (Fig. 3 , asterisk, Lum ^−/− Fmod ^−/− ). These structural abnormalities were evident in 80% of the double-null animals examined between the ages of 1 month (a set of images in 1-month-old animals is shown in Fig. 3 ) and 5 months (not shown). In rare instances, spaces between the retina and sclera were present in the wild-type and single null mutants. However, when images were presented in a blinded manner, Peter Campochiaro (Vitroretinal Specialist, Wilmer Eye Institute, The Johns Hopkins University) was able to differentiate wild-type with artifacts from the double-null sclera. Subretinal debris was present in more than 50% of histology preparations of the double-null eyes. 

Abnormal Scleral Architecture in the Null Mice

The sclera is organized into layers or lamellae and the collagen fibrils within each lamella have the same orientation. Collagen fibril and lamellar organization were analyzed across the entire thickness of the sclera from 5-month-old mice by TEM (Fig. 4) . The wild-type sclera had well-defined lamellae across its entire thickness. Compared with the wild-type, there were fewer lamellae across the sclera in all three null mutants, confirming the scleral thinning detected by light microscopy (Table 1) . Furthermore, the lamellae in the Lum ^−/− Fmod ^−/− appeared thicker and disorganized compared to the lamellae of the wild-type sclera. Lamellar disorganization in the single-null mutants was less severe than in the double-null mutants. Whereas fibril-poor spaces were occasionally seen in single-null mutants, the Lum ^−/− Fmod ^−/− sclera contained large areas practically devoid of fibrils (Fig. 4) . 

Collagen Fibril Structure

Collagen fibril structure was analyzed from the midwall region of the posterior sclera of 5-month-old mice by TEM (Fig. 5) . The morphologies of collagen fibrils in cross-section were similar in the wild-type and lumican-null sclera (Fig. 5A versus 5B ). In contrast, the fibromodulin-null sclera contained increased numbers of small-diameter fibrils and fibrils of irregular contours (Fig. 5C) . The morphology of collagen fibrils was dramatically altered in the double-null sclera (Fig. 6) . It was remarkably heterogeneous, even within a single region of the sclera (Fig. 6A) . Compared with the wild-type, the double-null sclera had a larger fraction of fibrils with irregular, “cauliflower” shapes in cross-section (Fig. 6A) . Certain areas had a higher concentration of small-diameter fibrils (Fig. 6B) , whereas others had large irregular fibrils (Fig. 6C 6D) . This very aberrant fibril structure is consistent with abnormal association and fusion of fibrils. The fibril structure in the anterior sclera (data not shown) was affected in all the mutant genotypes, but not to the same extent as in the posterior sclera. 

Fibril Diameter Distribution in the Sclera

The distribution of collagen fibril diameters was analyzed from regions comparable to those just described (Fig. 7 ; Table 2 ). In the mixed model, only three points (25th percentile = 1st quartile, 50th percentile = 2nd quartile = median, and 75th percentile = 3rd quartile) were analyzed. The estimates for the quartiles with 95% confidence limits by genotype are shown in Figure 7 and Table 2 . These estimates were adjusted for animal-to-animal and field-to-field variability. There were no significant differences between the distribution quartiles of fibrils from lumican-deficient mice and wild-type mice. All quartiles for the fibromodulin-deficient data were consistently lower by 4 to 6 nm than that for the wild-type data, but were not significantly different. However, in the double-deficient mice, all three quartiles were larger than that of the wild-type mice. The mean differences in medians (17 nm) and 3rd quartiles (28 nm) were significantly different from zero, with P < 0.005 and < 0.0001, respectively. 

The cumulative proportion of observations at any specific value is shown in Figure 8 . The graph demonstrates that on average, the distribution of fibrils from lumican-deficient mice was similar to the distribution of fibrils from the wild-type sclera. The distribution of fibril diameters from fibromodulin-deficient animals consistently showed smaller diameters. In contrast, for the double-null mice, the mean values and variations were significantly greater beginning at the first quartile. 

Higher Levels of Lumican in the Lum ^+/+ Fmod ^−/− Sclera

To determine whether there is a compensatory increase in the lumican protein in Lum ^+/+ Fmod ^−/− and conversely of the fibromodulin protein in Lum ^−/− Fmod ^+/+ , scleral protein extracts were analyzed for lumican and fibromodulin levels by semiquantitative Western blot analysis (Fig. 9) . The results showed a substantial (greater than fivefold) increase in lumican in the absence of fibromodulin. In contrast, the level of fibromodulin was reduced slightly in the lumican-null background. As expected, both proteoglycans were absent in the double-null sclera. 

Discussion

Lumican and fibromodulin are SLRPs that bind collagen in vitro and play a critical role in maintaining collagen architecture in various connective tissues. ^{7 21 22 23} Decorin-null mice have abnormally thick and disorganized collagen fibrils in the skin and, consequently, skin laxity. ²⁴ Lumican-null mice contain thick collagen fibrils of irregular contour in the skin and cornea and associated functional deficiencies include skin laxity and corneal opacity. ^{4 8 25} In the fibromodulin-null mice, collagen architecture of the tendon is affected the most. Collagen fibrils have abnormal contours and increased frequency of small-diameter fibrils. ⁶ Although in vitro lumican and fibromodulin bind to the same site on collagen type I, analyses of the collagen structure in different tissues of lumican and fibromodulin double-null mouse are providing new insights into their distinctive functions in collagen fibrillogenesis in vivo. The tendon phenotype of the lumican-fibromodulin double-null mouse indicates a role for lumican in fibril assembly and limiting lateral growth of fibrils, whereas fibromodulin functions in the maturation of collagen fibril intermediates. ¹⁹  

The present study focused on the ocular structure and scleral ultrastructure of the Lum ^−/− Fmod ^−/− double-null mouse. The results demonstrated their unique and complementary functions in regulating ocular axial length, structure of the sclera, and ultrastructure of scleral collagen fibrils. Compared with that of wild-type littermates, ocular axial length was significantly higher in the double-null mutant only. However, thinning of the sclera was evident in all the null genotypes. The low-magnification transmission electron micrographs also demonstrated a decrease in the number of lamellae associated with scleral thinning. Histology showed obvious areas of retinal detachment near the posterior sclera of the double-null mouse and, less frequently, in the lumican-null eyes as well. 

We analyzed the diameter of collagen fibrils and the frequency of different diameter fibrils in transmission electron micrographs of sections of the sclera from wild-type and null mutant mice. The range and distribution of fibril diameter frequencies was least affected in the Lum ^−/− Fmod ^+/+ mice. There was an increase in the frequency of small-diameter fibrils in the sclera of the Lum ^+/+ Fmod ^−/− mice, as we noted earlier in the tendons of 1-month-old mice. ¹⁹ In the double-null mice, the mean fibril diameters and variations were significantly greater, beginning at the 1st quartile. In the double nulls, the increase in the frequency of very small-diameter as well as large-diameter fibrils was far more than would be expected from a combination of the individual single-null phenotypes of the sclera, which suggests unique and interactive functions for these two proteoglycans in the sclera. Fibromodulin, a major constituent of the sclera, may be a key regulator of scleral collagen fibrillogenesis, whereas lumican may modulate this function. 

There is somewhat of a discord between the extent of collagen fibril changes and other ocular manifestations in the single-null mutants (such as ocular length, scleral thinning, and retinal detachment). It has been suggested that small-diameter fibrils give way to large-diameter fibrils through lateral fusion of fibrils as the connective tissue matures. ^{26 27 28} An abnormal increase in these immature collagen fibrils in the sclera has been suggested to lead to increased ocular growth and reduced scleral integrity, as seen in myopia. ^{11 17 29 30} At the ultrastructural level, the sclera of the fibromodulin-deficient mouse is clearly disrupted to a greater extent than that of the lumican-deficient mouse. Yet, of the two single-null genotypes, the lumican-null mouse only shows occasional signs of retinal detachment. Lumican is expressed early in development and has an effect during early corneal development. ^{4 24 31} These observations suggest an early developmental requirement for lumican, and although speculative, imply a defining role for lumican during ocular growth and maturation, perhaps in active remodeling of the sclera. Thus, in the lumican-deficient mouse, although scleral collagen fibril morphology is not as affected as it is in Lum ^+/+ Fmod ^−/− , gross morphology of the sclera and cell adhesion between the retina and sclera are compromised. It is possible that in a particular tissue, such as the sclera, it is critical to have a certain amount of these related keratan sulfate SLRPs, more so than any one particular SLRP. Thus, in the fibromodulin-null background, dramatic upregulation of lumican, as indicated by the semiquantitative immunoblot analysis results, may equate to enough of a class II SLRP in the sclera to prevent a severe scleral phenotype. 

The functional consequence of the collagen fibril–matrix defects in the sclera is almost certainly biomechanical weakness. ^{15 16} There is clearly a correlation between collagenous matrix defects and reduced tissue biomechanical strength in the tendon of lumican-fibromodulin double-deficient mice ⁹ and in the fragile skin of lumican-deficient and decorin-deficient mice. ^{4 24} Although biomechanical testing of the mouse sclera is difficult, a significant increase in ocular axial length in the double-null mouse implies scleral weakness. Abnormal ocular growth, axial elongation, and scleral changes are hallmark features of high myopia of long-standing duration and experimental animal models for myopia. ^{11 13 14 30 32 33 34 35 36 37 38} In the experimental models, the emmetropization process is disrupted by either deprivation of patterned vision or by introduction of refractive errors. The resultant eye enlargement in the avian sclera is achieved by increased cartilage growth and increased protein synthesis and chondrocyte density, ^{33 34} but thinning of the fibrous sclera in mammals. ^{11 12} In both avian and mammalian myopia models, increased scleral creep supports biomechanical weakening of the sclera. ^{15 16}  

Our study on the lumican- and fibromodulin-deficient mice elucidated several phenotypic similarities with high myopia and experimental animal models of myopia. The Lum ^−/− Fmod ^−/− double-null mutant mice mimic form-deprivation myopia and pathologic human high myopia, with respect to increased ocular length, scleral thinning, and detachment of the sclera and the sclera–choroid–retina junction. At the ultrastructural level, there was an increase in the frequency of small-diameter collagen fibrils in the fibromodulin- and lumican-fibromodulin–deficient mice. These changes were most pronounced in the posterior sclera. An increase in the percentage of small-diameter fibrils has been also noted in the tree shrew form-deprivation myopia model. ¹¹ Another note of similarity between the Lum ^−/− Fmod ^−/− and the form-deprivation models is that much of the myopia-related changes in ultrastructure, biochemical composition, and biomechanical strength in these models also occur in the posterior sclera. 

Members of the SLRP family have recently been associated with ocular diseases. Thus, nyctalopin, a class IV SLRP, expressed in the retina, is mutated in congenital stationary night blindness ³⁹ and mutations in keratocan, a class II SLRP, are associated with cornea plana. ⁴⁰ Young et al. ^{41 42} identified a familial high myopia (MYP3) locus on 12q21.2-22. Lumican, decorin, and keratocan, localize to 12q21.2-q22, and are important candidate genes for MYP3. ^{42 43 44 45 46} Fibromodulin on 1q32 is not associated with a known myopia-susceptibility locus. However, it is expressed at high levels in the sclera and, as our study demonstrated, plays a critical role in scleral structural and ultrastructural integrity. ⁴⁷  

In conclusion, this study showed that, although lumican and fibromodulin bind to the same region of collagen type I, ⁷ their functional implications in vivo are unique. The double-knockout mice manifested a far more severe phenotype than the sum of the fibromodulin-deficient and lumican-deficient phenotypes. Compared with the wild-type mice, the lumican and fibromodulin double-knockout mice had increased ocular axial length and thinner sclera with altered collagen architecture. These are some of the key features of high myopia and long-term form-deprivation animal models of myopia. Although occurrence of null mutations in two SLRP genes is less likely in human populations, myopia is a multigene complex disorder. Mildly deleterious mutations in multiple connective tissue genes may lead to a broad phenotype including myopia, skin fragility, and reduced strength of tendon and bone. Changes in gene expression of one or more of these SLRPs, leading to partial or total loss of these proteoglycans, as in the double-null lumican-fibromodulin mouse, may also occur in some forms of myopia. Lumican and fibromodulin are thus potential candidate genes for susceptibility to high myopia. 

 Supported by Grants EY11654 (SC) EY05129 (DEB) from the National Eye Institute and Swedish Medical Research Council Grant MFR k99–03X-07478-14A-A.

 Submitted for publication August 2, 2002; revised November 18, 2002; accepted January 2, 2003.

 Disclosure: S. Chakravarti, None; J. Paul, None; L. Roberts, None; I. Chervoneva, None; A. Oldberg, None; D.E. Birk, None

 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.

 

Figure 1.

 

Photographs of whole eye of Lum ^+/+ Fmod ^+/+ , Lum ^−/− Fmod ^+/+ , Lum ^+/+ Fmod ^−/− , and Lum ^−/− Fmod ^−/− . The eyes were viewed with a 10× objective at ×1.25 and photographed with a dissecting microscope. Axial length (in millimeters) was determined by drawing a line along the length of the photograph (e.g., D). Bar, 2 mm.

Figure 2.

 

Axial length and body weight of wild-type and mutant animals. Image-analysis software was used to draw a line from the anterior to the posterior of the eye to determine axial length in gender-matched littermates for each of the genotypes indicated in Figure 1 and plotted versus their body weights.

Figure 3.

 

Histology of the posterior sclera (arrow) of Lum ^+/+ Fmod ^+/+ , Lum ^−/− Fmod ^+/+ , Lum ^+/+ Fmod ^−/− , and Lum ^−/− Fmod ^−/− mice. Paraffin-embedded sections were stained with hematoxylin and eosin. Note visibly thinner sclera in the sections of all three null genotypes. Scleral thickness was measured in the digital photographs on computer and compared. (✶) Area of retinal detachment and subretinal debris. Bar, 20 μm.

Figure 4.

 

Collagen architecture of the sclera in the wild-type (A) Lum ^−/− Fmod ^+/+ (B), Lum ^+/+ Fmod ^−/− (C), and Lum ^−/− Fmod ^−/− (D). Top is adjacent to the retina. These transmission electron micrographs of the scleral wall demonstrate disruption of the lamellar architecture in all three types of deficient mice (B–D) compared with the wild type (A). There was a reduction in scleral thickness and a decrease in the number of lamellae. In addition, the deficient lamellae was less defined and organized. Fibril-poor areas and altered fibril packing and organization characterized the double-deficient phenotype.

Table 1.

 

Mean Number of Collagen Fibril Lamellae across the Sclera in 5-Month-Old Mice

Genotype	Mean Lamellae (n)	Range
Lum +/+ Fmod +/+	16	11–19
Lum −/− Fmod +/+	13	11–15
Lum +/+ Fmod −/−	14	12–16
Lum −/− Fmod −/−	12	10–16

Figure 5.

 

TEM showed collagen fibril morphology in cross-section from the midwall of the posterior sclera of wild type (A), Lum ^−/− Fmod ^+/+ (B), and Lum ^+/+ Fmod ^−/− (C). The wild-type fibrils had a regular, cylindrical contour (A). The fibrils from the Lum ^−/− Fmod ^+/+ lumican-deficient sclera were comparable to the wild-type fibrils, but with slightly more irregular contours (B). The fibrils from the Lum ^+/+ Fmod ^−/− fibromodulin-deficient sclera were different from the wild type (C). There were more small-diameter fibrils (arrows), and the fibril contours showed irregularity (arrowheads).

Figure 6.

 

Fibril structure in the Lum ^−/− Fmod ^−/− mouse sclera. TEM showed collagen fibril morphology in a cross section from the midwall of the posterior sclera. There was some regional heterogeneity in the fibril structures within a lamella, as shown in (A). However, all regions showed fibrils with abnormal, irregular contours that were markedly different from the wild type (compare with Fig. 4A ). (B–D) Localized areas of abnormal small- to very large-diameter fibrils with irregular contour. These are examples of the highly aberrant fibril structures in the double-deficient sclera.

Figure 7.

 

Scleral fibril diameter distributions. Fibril diameter distributions from wild-type (WT), lumican-deficient Lum ^−/− Fmod ^+/+ (Lum^−/−), Lum ^+/+ Fmod ^−/− fibromodulin-deficient (Fmod^−/−), and double-deficient (Lum ^−/− Fmod ^−/− ) mice. Measurements were taken from cross sections of fibrils in the midwall of the posterior sclera. Q1 is the 25th percentile = 1st quartile, Q2 is the 50th percentile = 2nd quartile = median, and Q3 is the 75th percentile = 3rd quartile. The mean fibril diameter in nm for Q2 and Q3-Q1, with 95% confidence intervals is shown in the upper right corner.

Table 2.

 

Mean Fibril Diameter and Fibril Diameter Distribution in Each Quartile.

Genotype	Q1	Q2	Q3	Q3-Q1
Lum +/+ Fmod +/+	60 (53,68)	76 (68,84)	87 (79,96)	27 (22,32)
Lum −/− Fmod +/+	61 (55,68)	76 (69,83)	87 (80,94)	26 (22,29)
Lum +/+ Fmod −/−	57 (49,64)	70 (61,78)	82 (72,91)	25 (21,29)
Lum −/− Fmod −/−	66 (57,74)	93 (84,101)	115 (105,125)	49 (43,56)

Figure 8.

 

Genotype was related to mean fibril diameter percentiles. Percentage of the data smaller than the corresponding diameter were plotted versus fibril diameter. The wild type and lumican distributions are superimposed and appear to be identical. The fibromodulin-deficient animals showed a small but consistent downward shift in distribution of fibril diameters compared with the wild-type sclera. In comparison with the wild type, the mean distribution of fibrils from double-deficient mice had increasingly larger percentile values, with increasing variability beginning at the 1st quartile. Vertical lines: 1st, 2nd (median), and 3rd quartile data that were the subject of the statistical model.

Figure 9.

 

Immunoblot analysis of scleral protein extracts. Polyclonal antibodies against lumican, fibromodulin, and α-actin were used, as shown at left. After immunostaining for lumican or fibromodulin, the filters were stripped and immunostained for β-actin to test for comparable loading in each lane and to measure levels of these proteoglycans against a housekeeping gene product. The genotypes are shown at top. Note the more than fivefold increase in lumican in Lum ^+/+ Fmod ^−/− mice and the approximately twofold decrease in fibromodulin in the Lum ^−/− Fmod ^+/+ mice.

The authors thank Peter Campochiaro for examining the ocular histology data and for advice on retinal detachment, and Harry Deitz (Johns Hopkins University) and Henry F. Edelhauser (Emory Eye Center) for helpful discussions and comments.